This PDF file contains the front matter associated with SPIE Proceedings Volume 12704, including the Title Page, Copyright information, Table of Contents, and Conference Committee listings.

Short-term power prediction is one of the important tasks in the daily operation of the energy system. With the popularity of smart grid, it brings more difficulties to the accuracy of electricity prediction. The purpose of this study is to explore the characteristics of short-term power data to achieve efficient and accurate power prediction and provide a reference for power grid operation management. Firstly, according to various characteristics of power prediction, this paper builds shortterm power prediction models by using SVM and LS-SVM, respectively. Then, this paper compares the prediction results of the two single models with the test values. It is found that the prediction curve of the LS-SVM model is better than that of the SVM model, and the LS-SVM prediction model has better potential for prediction short-term power consumption. This indicates that different machine learning models are suitable for different types of electrical power predictions.

Keyboard and LED driver is a very widely used integrated circuit. Taking the ZLG7290CS chip as an example, the structure and working principle of the keyboard and LED driver are analyzed, the timing and electrical characteristics of the keyboard and LED driver are studied, and the test method of the keyboard and LED driver based on J750Ex integrated circuit automatic test equipment is proposed.

Microgrid construction and the industrial Internet of Things will be an important part of future economic development and industrial restructuring. Based on the current development status of the field, this paper sorts out the actual needs of the system in order to solve the traditional power equipment management mode that is due to man-made errors and omissions. The system can complete the connection of power equipment, and compare the parameters collected by each sensor with the actual value. The acquisition error meets the actual needs. The power equipment monitoring and management platform’s functional modules, which include a communication service module, an operation module, a database module, a monitor module and a data transmission module, can be used normally. The working flow is complicated. The system not only realizes the management of power equipment but also monitors its state parameters. The simulation experiment tests the basic measurement function of intelligent power monitoring. The accuracy test meets the expected requirements, which can realize the intelligent monitoring of the power grid equipment parameters, and can be promoted in the power system with high practical effect.

An adaptive filtering strategy is proposed to address the impact of high-frequency chattering, filtering delay and harmonic interference on the estimation accuracy of the conventional sliding mode observer based rotor angular frequency estimation strategy for permanent magnet synchronous motors. At the same time, to solve the problem of narrow passband of the adaptive filter, a frequency-locked loop is introduced to reduce the frequency tracking error, and a quadrature phase-locked loop is introduced to superimpose the fundamental frequency speed on the angular frequency estimation to improve its tracking performance. Finally, the strategy is verified through experiments that the strategy has better dynamic and steady-state performance.

In this text, the proposed liquid momentum ring is aimed at the space mission of attitude control of satellite. The characteristic distribution of magnetic field about researched magnetohydrodynamic liquid ring is numerically simulated. The numerical calculation indicates that the magnetic field distribution design about liquid ring is appropriate, and electromagnetic force generated makes the liquid metal to rotate. The calculation results offer a data support basis to the following liquid ring optimization design and a numerical reference for following torque test experiment.

This paper focuses on the non-linearity and the self-oscillation of H-bridge converters. Firstly, the nonlinear model and characteristics of the H-bridge converters are proposed based on the describing function method while considering the delay caused by digital control. Then, the stability and the self-oscillation of current loop control systems are analyzed according to the extended Nyquist stability criterion method. For verification, a prototype of the current loop control system is developed and the experiments regarding to current response affected by the open-loop gain, load characteristics and digital delay are carried out. The results show that the proposed model is accurate and effective.

A finite-time adaptive pseudo inverse output feedback dynamic surface control scheme (DSC) with output constraints is proposed to address the consensus attitude tracking problem of the photovoltaic power generation servo system. Combining the DSC technique with a command filter to avoid repeatedly differentiating virtual control signals and filter errors. Approximating the unknown parameters and uncertainties in the system by using radial basis functions. Solve the hysteresis problem of the servo motor by adaptive implicit inverse method to improve control accuracy and stability. It is proved that all the signals are semi-globally uniformly ultimately bounded, and the tracking errors converge can be made arbitrarily small in finite time.

With the continuous development of radar technology, radar systems in civil and military fields raise urgent requirements for miniaturized, low-cost, and lightweight radar front-end T/R components. All the T/R components designed herein adopt FR4 as the circuit substrate, which greatly reduces the cost compared with T/R components of the same type. Meanwhile, the requirements of miniaturization, low power consumption, and high isolation are met. The method and idea for C-band T/R component design through MMIC are proposed, in which, the transmission power is 15dbm, the dimension is only 70mm×60mm×18mm, the overall power consumption is only 2W, and the weight is only 120g.

In mimic architecture devices, programs with equivalent functions are usually implemented on heterogeneous executors. However, due to the differences in the scheduling order of threads among executors, the outputs of programs are also different. Usually, the scheduler in the mimic device needs to compare the outputs of different executors to decide the final output. The difference between the outputs will have a significant impact on the arbitration of the scheduler, causing executor's output cannot pass the arbitration, and the program cannot function normally. In order to solve this problem, some researchers proposed a state synchronization method based on relative time. This method can effectively solve the problem of program state synchronization in multiple heterogeneous executors, but it can be improved in completeness and running efficiency. Based on this method, this paper proposes an improved program synchronization method based on targeted reply, which considers the synchronization problem in more scenarios, saves the communication overhead caused by synchronization, and improves the operation efficiency of the system. The experimental results show that the method is effective to some extent.

An integrated analysis method of design cost based on modular 3D view of distribution network is proposed in order to reduce the complexity of distribution network design cost. The design cost integration system of distribution network project is designed through which the project data processing, design result display is realized. Design modularization, material modularization, and budget modularization as the core unit with professional modularization as the guiding ideology, a centralized management operation platform is established to strengthen the professional control of key links, so as to effectively organize and promote the implementation of the project. The 3D view is lightweight in order to make the 3D view production process easier. Finally, the cost integration management method is designed to realize the intensive and professional management of distribution network construction. The analysis and practice results show that this method effectively reduces the complexity of distribution network design cost and improves the effect of design cost integration.

In order to verify the accuracy of the results of the virtual simulation experiment system on the gravity dam stress experiment line in the experimental teaching process, this study specially carried out the online experiment of gravity dam stress experiment and got the results, and further carried out the online manual experiment in the laboratory. The results of the two experiments were almost the same, which verified the feasibility of the virtual simulation experiment system on the gravity dam stress experiment line and led to the advantages of the three-dimensional virtual simulation experiment. Changing the teaching mode to online preview and offline experience of authenticity experiment aims to make up for the shortcomings in teaching and achieve good teaching results.

Push forward converter is constrained by its switching loss under high frequency conditions, so it has to use heat sink to increase the volume of the converter, which is contrary to the design concept of small volume and high power density. Based on the traditional push-pull forward converter, this paper adopts phase-shifting control to realize ZVS of the primary side switch through LC resonance generated by the parallel capacitor of the primary side switch, the primary side leakage inductance and the secondary side filter inductance. At the same time, the dead time is set by calculating the duration of each mode in half a cycle. Finally, saber simulation is used to verify the feasibility of realizing ZVS of the push-pull forward converter, so as to provide a basis for the high power It provides a reference for the application under high frequency conditions.

In terms of the problem of solving the three-phase voltage errors in the transformer, a method for solving the three-phase voltage errors in the transformer is proposed in this paper based on the genetic algorithm. By encoding the numerical groups of the three-phase voltage errors in the transformer and setting constraints such as the three-phase imbalance rate and voltage difference value, errors are solved by multiple rounds of iteration. The results show that the method proposed in this paper can effectively solve the three-phase voltage errors in the transformer, providing support for the metering equipment management to transform from “on-site operation and maintenance + periodic rotation” to “online operation and maintenance + precise rotation.”

In order to build the broadband model of electrical equipment, it is necessary to measure the broadband impedance characteristics of the equipment. However, the large size of high-voltage electrical equipment requires long measurement leads to measure the broadband impedance characteristics. At present, there is a lack of effective and universal compensation measures, which leads to unacceptable measurement errors in high frequency band. In order to solve this problem, a method for measuring broadband impedance characteristics of large-scale electrical equipment is proposed. In this method, impedance analyzer is used as measuring equipment, and the coupling calculation model of the long leads over the earth is established by using the partial element equivalent circuit (PEEC). The two-port network parameters of the long leads measurement loop are obtained, and the network parameter matrix is used to compensate the influence of the measurement results by the long leads. A measurement experiment is carried out in the laboratory, and the accuracy of the measurement method is proved by comparing the results measured directly by short leads with those measured by compensated long leads. The measurement method can accurately measure the broadband impedance characteristics of large-scale electrical equipment in the range of megahertz.

In December 2020, affected by the cold wave and extreme low temperature, the output of regional large-scale wind power clusters fluctuated significantly. The maximum fluctuation of wind power in the whole grid accounted for 37% of the total installed capacity, which seriously threatened the safe and stable operation of the grid. In order to explore the generation mechanism of this large-scale wind power cluster output fluctuation process, this paper uses conventional terrain data and 1°× 1°reanalysis data ( provided by NCEP/NCAR) to analyze the weather situation, meteorological element characteristics and relevant physical quantity fields of this process. The results show that in the early stage of the process, low heat pressure and inverted trough develop on the ground, and the ground temperature rises significantly. Later, the mid latitude ladder trough guides the northern cold air to the south, and the cold high on the ground develops to the intensity of cold wave. The narrow pipe effect of terrain accelerates the movement of cold air, and the range of surface cooling increases significantly. In addition, there is a significant correlation between the change process of surface temperature and relative humidity before the event and this large-scale wind power abnormal operation event. Before the cold wave, the near surface temperature rises, the relative humidity increases, and the atmosphere is in a state of easy saturation. This provides favorable background conditions for the rise and condensation of water vapor in the cold wave weather, leading to a large number of wind turbine operating condition alarms in hilly areas, and significant changes in the output of the whole grid wind power cluster in a short time.

The digital transfer of power grid engineering is faced with problems of multiple links, multiple objects and multiple data. During the transfer process, various resources are difficult to locate, correlate and trace, and the data of each link is repeated, which consumes a lot of manpower. In this study, we propose a two-dimensional code for power grid engineering based on QR technology, establish a unified basic identity library for basic elements of power grid engineering such as personnel, units, projects, equipment, and atlases, and realize the organic correlation of various objects of power grid engineering and the information penetration of various power grid majors. Research and build a code management and control technology platform for power grid engineering, take power grid engineering management as the main line, combine State Grid chain(Blockchain) to realize the whole process of code generation, authorization, code use, and cancellation of power grid engineering code on the chain, digitally empower grid management of power grid engineering, and accelerate the digital transformation of power grids.

Nowadays, all cloud platforms support the live migration technology of Docker containers, which plays an important role in improving the stability and quality of services. In this article, we will use the JNLP algorithm to determine which migration method to use in pre-copy and post-copy based on the actual network bandwidth and real-time monitoring of memory changes. Moreover, this article analyzes the storage mechanism of docker containers, using the hierarchical structure of docker to reduce the data that needs to be transferred during migration, thereby reducing the total migration time. Experimental results show that the live migration technique proposed in this paper is more general and reduces container data migration time by 40% and total migration time by 30% compared with state-of-the-art techniques.

The traditional road condition information collection methods mainly include direct access to the user's GPS positioning and sensor indirect access to two ways, the GPS positioning method has the problem of privacy disclosure cannot be ignored, in case of illegal use may threaten the user's personal and property safety. How to collect accurate traffic information in real time on the premise of protecting user privacy is an urgent problem to be solved. In order to solve this problem, this paper proposes an auxiliary road condition detection scheme to protect the privacy of service users under the framework of crowdsourcing. Paillier homomorphic encryption algorithm is adopted to encrypt the user's driving direction, which can effectively protect the user's driving track. The crowdsourcing server delivers road condition collection tasks to users with the help of roadside unit (RSU), improving the availability of road condition information collection and realizing rapid auxiliary monitoring of real-time road conditions while protecting privacy.

With the continuous development of online education platform, knowledge tracking (KT) has become a key technology to help online education platform provide personalized education. However, the existing knowledge tracking model based on recurrent neural network is difficult to be used for the input of long sequence, and has the problem of long-term dependence. Secondly, although the knowledge tracking model based on Transformer does not have the problem of long-term dependence, it is difficult to capture the input sequence information. Therefore, this paper proposes a knowledge tracking model based on recurrent neural network and transformer. A new position coding method is designed, and LSTM is used to replace the position coding method of Transformer to encode sequence features, so that the model in this paper can not only capture the input sequence information, but also get rid of the long-term dependency problem based on the recurrent neural network, and use GRU network to capture the context information. In addition, an adaptive fusion gate is designed to fuse the global features and context features obtained by Transformer, and use the fused features to predict the students' answers to the next question. In addition, an adaptive fusion gate is designed to fuse the global features and context features obtained by Transformer, and use the fused features to predict the students' answers to the next question.

Including process modelling, deployment of BPMN process definition, task flow, process monitoring and more, workflow engines are dedicated to improving the efficiency of process development. The traditional workflow platform is a monolithic architecture, which is embedded in the application as a library to build a unified unit that couples all business concerns together, with low code reusability, unable to cope with continuous business changes, and a single architecture resulting in poor system scalability and unable to cope with high concurrent user access requests. To meet these requirements of reusability, scalability and high concurrency, this paper provides an all-encompassing, scalable distributed workflow platform based on the microservices architecture for automating enterprise business processes to improve efficiency and productivity. It is a service-oriented architecture, built as a collection of different small services rather than a complete application. By being distributed and loosely coupled, the microservices architecture allows for rapid, linear scaling.

Spectroradiometer is mainly used to measure the spectrum properties of radiation sources and radiation intensity. The accuracy of the instrument's measurement data results is a prerequisite for the determination of radiation sources, and the accuracy of the data processing method in the measurement process will directly affect the measurement results of the target to be measured. The circular variable filter type spectroradiometer uses a circular gradient filter with linear relationship between transmission wavelength and angle for spectroscopy, with a spectral range of 1.3-14.3um and a resolution of 2% of the wavelength. In this paper, the data processing method of this type of spectroradiometer is studied, the basic principle of the instrument is introduced, and the processing of the data of circular variable filter type spectroradiometer is carried out. The data processing is mainly divided into three parts: data pre-processing, drift correction and wavelength re-alignment. The raw sampling data is obtained after the data pre-processing of the original acquired data package. The drift correction is calculated using the inter-correlation method, and the calculated offset value is used to correct the acquired raw data by cyclic offset. The corrected raw samples are converted into spectral profiles by wavelength realignment. A standard blackbody is used as both the target and reference source to verify the measurement results. The accuracy of the data measurement method was demonstrated by comparing the CO2 absorption peaks in the obtained spectral data with the CO2 transmittance calculated by the atmospheric radiative transfer model MODTORAN, based on experimental data of blackbody measurements by spectroradiometer. It effectively ensures the accuracy of the measurement results and lays the foundation for the later radiometric calibration of the instrument.

The measurement and control of harmonics in extra high voltage direct current (EHV DC) transmission lines is an important guarantee for the safe operation of the EHV DC transmission system. According to the operation it is concluded that when the grid frequency fluctuates, the frequency of AC harmonics will also fluctuate. In the face of frequency fluctuations, the measurement results obtained from the raw data have a large error and have exceeded the design requirements. In order to explore the proper measurement method, further research on voltage harmonics in AC-DC networks under wide band is needed. The new wide-band high-precision voltage divider is the main equipment for measurement and control protection in extra-high voltage DC transmission lines. In this paper, the new wide-band high-precision voltage divider is described from several aspects, such as development method, stray capacitance additional error solution method, experimental test, and experimental result analysis. The experimental results show that the new wide-band high-precision voltage divider can accurately measure the harmonics in the AC-DC transmission network and is verified by Fourier analysis, and the measurement accuracy can meet the accuracy requirements of less than 0.2% at 50Hz and less than 1% from 50-3000Hz.

To study the characteristic frequency of the vibration and noise source of the linear induction motor (LIM) in the mediumlow speed maglev train powered by inverter, this paper measured the three-phase current powered by inverter and input the current into the LIM finite element model. The LIM air-gap magnetic field was calculated by the finite element method, and the main harmonic frequency of the air-gap magnetic field in inverter power supply was analyzed from its spectrum. The LIM electromagnetic force and its characteristic frequency were analyzed and calculated by finite element method and analytical formula. The natural mode shape and frequency of the motor stator were calculated by the finite element method, finding out that the natural mode frequencies of the motor have resonance frequency bands with the harmonic components of the electromagnetic force. The tests of the LIM vibration and noise under inverter power supply showed that the frequencies of vibration and noise are consistent with the frequencies of current and electromagnetic force harmonics, indicating that the current harmonics under inverter power supply is the main reason for the LIM vibration and noise.

In order to address the growing demand for vector operations in embedded systems and the urgent need to enhance vector operations in embedded systems, this paper designs a 128-bit vector operation unit based on the open source RISC-V instruction set. In order to verify the correctness of the vector operation unit, we experimentally verify the vector operation unit in the ModelSim simulation environment, and all modules meet the correctness requirements. In addition, we integrated the vector operation unit with the open-source processor Hummingbird E203 and tested it on a 16MHz FPGA development board. The results show a 1.2 times performance improvement of the demo application compared to the scalar processor.

The GasP circuit is a high-speed controller that has been used more often in the field of asynchronous circuits. However, in the conventional GasP design method, the matching process of its internal delay lacks the consideration of the circuit application environment. In view of this, this paper proposes three delay adjustment methods by combining the usage scenario of the circuit and the logical effort. The method first adjusts the pulse width of the digital circuit, the ability to drive the DFF and the maximum equivalent frequency of the circuit, then customizes the delay and drive of the GasP, and finally verifies the performance with a NoC built with the GasP. The experimental results show that the maximum equivalent frequency of GasP can reach 2.28GHz in 110nm CMOS process, which is about 10 times of the classical synchronous design.

An optical fiber sensor was designed and fabricated for the determination of 2-Chlorophenol (2-CP) based on 2-CP oxidation by oxygen with the catalysis of iron(III) tetrasulfophthalocyanine (Fe(III)PcTs). A lock-in amplifier was used for detecting the lifetime of the oxygen sensing film by measuring the phase delay of the sensor head. The linear detection range of the biosensor was 1.0×10^-6 -9.0×10^-5 mol/L and the response time was 226s. The sensor displays high selectivity, good repeatability and stability, which shows good application prospect in the analysis of 2-CP concentration in actual water samples.

Under the ground that the next generation of power systems is constructing, our research work focuses on the core capabilities of power communication networks for transmission lines, including coverage, service-bearing capacity, flexibility and convenience, and technical security. We propose a convergence solution of heterogeneous networks to form a stable and reliable communication link based on 5G, ad hoc networks, and other technologies to solve the problem of poor network coverage in remote areas and realize data collection and remote control in areas without signal coverage. This system supports the establishment of intelligent monitoring and inspection of overhead transmission lines, offers an early warning service network system, and promotes the improvement of power digital space holographic perception and interconnection capabilities. Finally, we will realize an all-time communication named “anytime, anyplace, always online.”

Aiming at the current situation that the extended Kalman filter (EKF) method, which is mainly used in power system dynamic state estimation, has poor adaptability and limited application scope, a new dynamic state estimation method is adopted. Firstly, the variable parameter exponential smoothing method is used to construct the state transition function; Using micro synchronous phasor measurement technology（μ PMU) and distribution network data acquisition and measurement system (smart meter) mixed measurement data to build three-phase mixed measurement equation, so as to build a state space model. Secondly, according to Bayesian probability principle, the maximum likelihood posterior probability density likelihood function is constructed for the state space model, and the optimal solution of the state variable is obtained by maximizing it. Finally, the conditional posterior Cramerol lower bound (CPCRLB) of the mean square error of the estimation error is derived to determine whether the result of the state estimation is optimal. Through simulation analysis in three-phase unbalanced distribution network, the results show that the algorithm proposed in this paper meets the accuracy constraints and has higher estimation accuracy than the traditional EKF algorithm, which verifies the effectiveness of the proposed algorithm.

Different types of distributed generation have different grid-connected modes, control links and basic models, and each type has different effects on 10kV power grid, so corresponding models should be established to analyse it. This paper focuses on the research of inverter-based distributed generation, and analyses the impact of different capacity, installation position and fault location on the protection device.

In this paper, the customized algorithm to calculate interval beam pointing accuracy has been proposed. Based on the algorithm, the effect of interval excitation deviations of array antennas on the beam pointing accuracy is studied. The 10×10 array operating within X band with half wavelength is taken as an example to investigate the corresponding influence. The results show that compared with the traditional definition of interval beam pointing accuracy, the improved interval beam point accuracy can better reflect the influence of interval excitation deviations on the array performance. The conclusion can provide valuable suggestions for the mechanical design of phased array antennas.

In order to obtain the best construction scheme for the arrangement of high-voltage cables in-phase and multiple parallel cables and ensure the power supply quality to the greatest extent, the BIM construction model of the arrangement sequence of high-voltage cables in-phase and multiple parallel cables is designed. Prepare all relevant data of multiple parallel cables in the same phase of high-voltage cables, build the model of multiple parallel cables in the same phase of high-voltage cables, combined with 3D simulation results, describe multiple parallel cables in the same phase of highvoltage cables and their occupied space, repeatedly introduce the time dimension, identify the construction conflict in this space, and generate the layout sequence of multiple parallel cables in the same phase of high-voltage cables. The test results show that the model can complete the simulation of multiple parallel cables in the same phase of high-voltage cables, and input the cable parameters efficiently and reliably.

The diagnosis accuracy of smart meters is very low due to the uneven distribution of fault data. In order to improve the fault diagnosis accuracy of smart meters, a fault diagnosis method for smart meters based on the fusion of improved gray wolf algorithm and deep forest classifier (SCGWO-DF) is proposed. Firstly, the daily operation data of intelligent ammeter is obtained, and the data is classified according to the fault type, and the training set and test set are divided. Secondly, the training set is put into the deep forest classifier to train the diagnostic model. Thirdly, the improved grey wolf optimization algorithm is used to optimize the three key parameters: the number of features, the number of random forests and the number of completely random forests. Finally, the trained model is verified by using the data of a power plant smart meter. The experimental results show that the SCGWO-DF diagnostic method proposed in this paper has a higher accuracy rate than the traditional SVM, DBN and random forest methods, and the accuracy rate reaches 98%.

As a type of sensor intended for the real-time feedback and dynamic monitoring of angular velocity and angular displacement, encoder is widely used in electric vehicles, servo closed-loop control systems, household appliances, cameras and many other fields. In order to reduce the temperature drift caused by ambient temperature to the magnetic encoder, a self-tuning temperature compensation algorithm is proposed in this paper based on the calibration principle. The rationale of this algorithm is that error function is designed to calculate the error at different temperature and room temperature, and the partial derivative is calculated to update the weight coefficient. Finally, the least square model is applied to establish the relationship between Hall output amplitude and temperature coefficient, thus correcting the temperature drift of the encoder. According to the experimental results, the self-tuning temperature compensation algorithm is applicable to compensate for and correct the temperature drift caused by the amplitude of Hall signal at different ambient temperatures. The ultimate accuracy of temperature drift is improved by 0.2° for the encoder.

The electroencephalography (EEG) signal acquisition process is inevitably affected by a variety of physiological noise signals, including electrooculogram (EOG), electromyography (EMG). The traditional methods of removing EOG and EMG rely heavily on the subjective experience and prior knowledge of the user. However, the ambiguity of artificial judgments can lead to erroneous and misleading interpretations that are insufficient for qualitative analysis. This inaccurate denoising may affect the true information of the signals in the time domain and spectral domain, leading to a decline in the accuracy of the BCI system. In recent years, a variety of EEG denoising methods based on deep learning have been proposed, but their denoising performance needs to be further improved. In this paper, we design a novel autoencoder (AE) neural network to remove artifacts in EEG. The network includes an encoder and a decoder module. The encoder contains five convolutional layers with increasing feature dimension as depth increases, which are responsible for detecting and suppressing artifacts. The decoder contains five deconvolution layers, whose feature dimension decreases gradually, and is used for EEG reconstruction after denoising. The experimental results on semi-synthetic EEG datasets demonstrate that the proposed algorithm outperforms the four benchmark models.

This paper proposes a short-term load prediction model based on the improved whale optimization algorithm (IW0A) optimized gated recurrent neural network(GRU) to address the issue of strong unpredictability of electric load and low forecast accuracy. First, the whale population is initialized by S chaotic mapping to enhance the population diversity and improve the quality of the initial solution; second, a nonlinear convergence factor is proposed to balance the global and local search ability of the algorithm and improve the convergence speed in order to avoid the defects that the standard whale optimization algorithm is easy to fall into local optimum and slow convergence speed when solving the GRU parameter optimization problem. Finally, WOA is used to automatically determine the best parameters and create the IWOA-GRU load prediction model by optimizing the number of layer neurons, learning rate, and other factors. The results show that when compared to the prediction methods used by LSTM, GRU, PSO-GRU, RSO-GRU, and WOA-GRU, the proposed model may successfully increase convergence speed and prediction accuracy.

High-quality customers play an important role in the sustained development of power enterprises. Power customer value analysis is the basis for power enterprises to implement differentiated services for customers of different values and maximize benefits. Based on the analysis of power customer value, this paper constructs a customer value evaluation index system including current value and potential value, and determines the combined weight of evaluation index by using Analytic Hierarchy Process (AHP) and entropy weight method, and then combines with Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method to construct an power customer value evaluation model based on combined weight-TOPSIS. Lastly, 15 customers of a prefecture-level city power supply enterprise are selected for specific analysis to verify the effectiveness and practicability of the customer value evaluation model.

In order to realize the rapid detection and imaging of dangerous goods with high dielectric constant at small and mediumsized express points, a three-dimensional imaging method for packaging inspection based on planar ECT technology was proposed; Firstly, the electrode matrix of traditional planar ECT is optimized, and a planar ECT sensor with linear detection electrode array is designed; Then, the simulation model of the package to be tested under different dielectric constants is established. During the detection, the electrode plate is placed on the top of the package and moved slowly to obtain the capacitance signal of the detection electrode plate in different areas of the package; Finally, the iso-surface method is used for partial and full 3D imaging of the object in the package. The experimental results show that the designed detection method can realize three-dimensional imaging of high water content in packaging and metal dangerous goods.

Power knowledge graph of the great application potentials for the power utilities, has become one of the interesting research topics in academia and industry. The power big data onto the deployment of information management system, poses a challenge to the current power grid company's administrative duty system, at the same time, the opportunity for/in applying intelligent management notion into the power grid company has come into being. Based on these above descriptions , the application and exploration of knowledge graph in power grid company administrative duty has been put forward in this paper. Firstly, the on-duty text data is pre-processed; secondly, entity extraction and relationship extraction are carried out on the processed data; finally, the data is stored in the graph database to build the knowledge graph.

In the southwestern United States, Lakes Powell and Lake Mead are taking the responsibility of supplying water and generating electricity to the surrounding five states of Arizona (AZ), California (CA), Wyoming (WY), New Mexico (NM), and Colorado (CO). Accordingly, we have established three models, and have established them and conducted sensitivity analysis combined with changes of various conditions in reality. In combination with multiple water supply needs, Model 1 optimizes the economic benefits from water and power supply of the five states as objective functions. In setting the constraints, the model takes into account such factors as water balance, minimum demand for water and power supplies, lake level requirements, and issues of sovereignty in downstream Mexico. Considering the influence of seasons on various factors, the decision variables are the water supply flow and power supply flow of each state on a quarterly basis. When solving the model, IA-PSO is used for optimization, reducing the possibility of local optimal solution.The optimization result is the optimal distribution of the power and water supply flows provided by the two lakes to the states over the four quarters of a year. In the case of CO,when the supply exceeds the demand, the time to provide the optimized water supply flow to meet its one-quarter water consumption demand is calculated to be 57.5 days; while when the supply fails to meet the demand, the additional water supplement required is 0.8×109m3.

With the continuous development of wind power generation, solar power generation and other new energy generation, how to improve the utilization rate of new energy and increase the consumption of new energy is the current problem to be solved. This paper proposes a virtual power plant aggregation method based on cloud model, analyzes the response characteristics of its internal resources, establishes the peak regulation and frequency regulation indexes, and firstly clusters and then aggregates them to improve the utilization efficiency of virtual power plant. The simulation results show that the response time and daily load fluctuation after aggregation are reduced, and the peak regulating virtual power plant is preferred in the peak regulating scenario.

With the expansion of the coverage of the electricity market, the impact of the electricity market operation on the security of electricity supply has gradually increased, and the importance of electricity market operation monitoring has also risen. To this end, based on the summary of previous research results, this paper proposes a set of monitoring index system suitable for the existence of multiple trading time order electricity markets in four aspects: power deviation situation, tariff deviation situation, quotation strategy effectiveness situation and arbitrage situation, and constructs a comprehensive index for electricity market operation monitoring by combining hierarchical analysis method to provide reference for market managers.

Based on ontology knowledge base technology，this paper mainly proposes two aspects for building the ontologies knowledge base of robot in the intelligent space. On the one hand, this paper proposes the ontologies knowledge base model of intelligent robot. On the other hand, it proposes automatically construct the intelligent robot ontologies knowledge base by information system.This model integrate information about elements in the intelligent space. This model could help intelligent robots quickly understand and adapt to unknown environments, resulting in smarter interconnections. A unified information model helps information sharing and coordination of work between robots and between robots and other devices, thus providing better services to users. Therefore, forming an ontologies knowledge base model of intelligent robot.

According to the requirements of reliability analysis of different AC/DC hybrid microgrid grids, this paper proposes an analysis method based on flexible reliability index. Firstly, two kinds of universal low voltage AC/DC hybrid microgrid grid structure schemes are proposed; Secondly, the matrix model of microgrid topology is established; Then the flexible reliability index of microgrid is proposed; Finally, by comparing and analyzing the impact of different component failures on the system reliability of a real microgrid under different grid structures, it is verified that the proposed method can comprehensively and accurately reflect the reliability of the microgrid, and has high flexibility and scalability.

In order to obtain the influence of distributed photovoltaic on the distribution network. Based on the Monte-Carlo method and IEEE33 model, the influence of photovoltaic capacity, access location and transmission line length on the maximum voltage overlimit probability, network loss and total operation cost of the distribution network is investigated. The results reveal that the photovoltaic access location has little influence on the maximum voltage overlimit probability, network loss and total operating cost. The maximum voltage overlimit probability decreases first and then increases with the increase of photovoltaic capacity. Once the transmission line length is longer than the critical value, the maximum overlimit probability of voltage will increase with the increase of transmission line length.

In order to reflect the impact of source-load uncertainty on the overall operation risk of the power system in the future time interval, this paper establishes a multi-dimensional risk assessment system for renewable energy distribution networks based on value-at-risk. Firstly, the output probability model and load probability model of wind power photovoltaic are established, and then the sampling combination and power flow calculation are carried out based on the Monte Carlo method, and the corresponding node voltage and branch power flow results are obtained. Calculate the two risk indicators of branch power limit and node voltage limit at each prediction time of the distribution network, and use them as the first dimension risk assessment index; secondly, based on the value-at-risk theory to quantitatively evaluate the system risk, calculate the second dimension risk assessment Index; Finally, calculate the third dimension risk assessment index, calculate the comprehensive risk index through the entropy weight method, and comprehensively and reasonably analyze the operation risk of the distribution network under the access of different amounts of renewable energy. Finally, the rationality of the proposed model and method is verified by IEEE33 example system.

At present, the conventional line loss location prediction method mainly adopts the way of stratified location analysis to realize the line loss prediction by calculating the line loss value in layers, but there is a problem of low prediction accuracy. In this regard, a feature vector-based prediction model is proposed. The similarity of line loss sections of distribution networks is calculated, and the actual situation of theoretical line loss and communication line loss is discussed by situation to construct the line loss prediction model. The experimental results show that the output of the proposed prediction model has a low root-mean-square error and has a more desirable prediction effect.

Building a new power system is an important measure taken by China to cope with climate change problems. Establishing a flexible and perfect electricity market and price mechanism is an important means to ensure the safe and stable operation of the power system. As one of the important electricity price mechanisms, the change in real-time electricity price (RTP) is a key factor in the operation of the electricity market, and each market participant can formulate a response strategy according to the RTP. However, the uncertain, stochastic, and fluctuant characteristics are definitely difficult problems for the RTP prediction. With the aim of solving this issue, this paper proposed a RTP prediction method based on a window-based XGBoost model. Through the input conversion of the proposed model, it can help to reduce the complexity and capture the autocorrelation effect of the RTP. The case study is conducted through the actual load data of a province in China and the superiority is proved by comparing with several state-of-art models. The result shows that the window-based XGBoost model applied in this paper can decrease the prediction error by 69.48%-95.67% and greatly enhance RTP's prediction performance.

Traditional power load or electricity forecasting methods have used only limited meteorological information. Specifically, these methods either only select one of maximal, minimal and average temperature as the influencing indicator, or only capture the very basic meteorological information, instead of exploring the nonlinear relationship between temperature and loads in depth. To overcome the shortcomings of the traditional methods, this paper proposes a new medium- and long-term load forecasting method based on sensitivity analysis of multi meteorological indicators. This method constructs the encoder and decoder between weather features and loads. Moreover, this paper combines this method with XGBoost and SVR, which can improve the accuracy of medium- and long-term forecasting algorithm in the power grid field effectively. The simulation results of this synthesized model show that average daily prediction accuracy rate of the year is 95.37% and the average monthly prediction accuracy rate of the year is 98.02%.

In order to guide the demand-side electricity consumption behaviour of users, after investigating the whole social electricity consumption load of a county in southern China, according to the demand response, time-of-use electricity price and the reality of China's electricity market, this paper adopts the fast non-dominated sorting genetic algorithm with elite strategy to conduct a more in-depth optimization study on the demand-side management measures of three different users, and the conclusions are as follows: Combined with the practical problems in the electricity market transaction, a mathematical model with the minimum peak-valley difference of the load curve and the maximum power consumption comfort as the objective function is built, and the bundle of variables is established. The optimization results show that the multi-objective optimization on the demand side is slightly insufficient in smoothing the load curve, but considering the comfort of users' electricity consumption, the multi-objective optimization will improve its comprehensive performance and have stronger practicability, which can play a guiding role in the demand response behaviour of users and power trading in the power market, and provide a theoretical basis for actual trading.

The article proposes a PV grid-connected inverter system based on a two-stage structure. The two-stage structure is composed of the front-stage boost circuit combined with the disturbance observation method for maximum power point tracking and the rear-stage full- bridge inverter circuit. The control strategy adopts voltage and current double closed-loop control, and at the same time adds grid voltage feedforward control, which can effectively suppress the impact caused by grid voltage fluctuations. Finally, using Matlab/Simulink to build a system model and conduct simulation experiments to verify the feasibility of the system.

With the development of China's power industry entering a critical period of mode transformation, structure adjustment and power change, the state pays more and more attention to the promotion of new energy microgrid.The development and utilization of abundant wind, hydropower, solar and biomass energy resources in rural areas is an effective way to solve the problem of rapid increase in power consumption in rural areas and power supply in remote mountainous areas.This project analyzes the complementarity of the four renewable energy combined power generation in many aspects, and proposes a new rural "scenery aquatic" multi energy complementary power generation microgrid power supply system.The rural micro grid structure model of the small hydropower and solar photovoltaic power generation system with water storage reservoir is designed, and the scheduling strategies of the two operation modes of the micro grid island and grid connection are analyzed.

Aiming at the weak voltage support of the long-chain power grid, an optimal configuration method of condensers considering system stability constraints is proposed. To add condensers to key buses to improve the voltage support strength of the power grid. The basic data of Southern Xinjiang Power Grid in winter operation mode was collected, and statistical prediction and analysis of load and power surplus were made. After preliminarily selecting the location for the 750kV backbone grid, the location of the condenser was determined. Further, the weak area of 220kV bus voltage stability to achieve the accurate location of the condenser was determined. With the stability of the power grid as the constraint and economic optimization as the goal, the condenser capacity configuration is realized. The power grid model is built on the PSASP software platform, and the power flow calculation and voltage stability analysis are carried out. The results show that the proposed method can effectively improve the

In the power system, various energy sources such as cooling, heating, and electricity are coupled and complementary. As a multi-energy supply and demand balance method, power demand response can further improve the flexibility and economy of the system. In order to clarify the influence of demand response on the optimal operation mode of multi-energy cooperative system, this paper proposes an optimal operation model of multi-energy cooperative system based on market-elastic price demand response. Firstly, the main body of the multi-energy cooperative system is analyzed, the multi-energy coupling characteristics are studied based on the energy hub model, and a price-based demand response model based on market elasticity is constructed. Secondly, the optimal operation model aiming at the lowest daily operating cost of the multi-energy cooperative system is established. Finally, combined with the calculation and analysis of actual cases, it is verified that the adoption of price-based demand response measures can significantly improve the economy of the power multi-energy collaborative system.

In order to fully mobilize user-side resources in an increasingly open energy trading market, this paper proposes an optimal allocation strategy for electricity/heat/gas shared energy storage based on the probability prediction method. The proposed optimized configuration establishes an energy hub structure with electricity/heat/gas shared energy storage, and a twobody optimized model with two-layer from the view of users and providers participating in the shared energy storage business model is established. The bottom layer describes the uncertainty of new energy output based on the probability prediction method based on long-term and short-term memory and Bayesian neural network, a user-side shared energy storage charging and discharging model, which is optimized aiming to minimize the user's total cost, is established, and the decision information will be informed to the shared energy storage provider. At the top level, aiming to minimize the investment and construction cost of shared energy storage providers, concentrates on optimizing the allocation of energy storage power and capacity of decision-making entities. The big M method is adopted to relax and linearize the nonlinear part of the objective and constraints, and then it is transformed into a mixed-integer linear optimization problem. Finally, three typical application scenarios are established. As to the verification of the superiority of the strategy, the CPLEX optimization solver is called through the YALMIP toolbox in Matlab to solve the models in different scenarios, and the overall costs and benefits are jointly compared. From the case analysis, we can draw the conclusion that compared with the traditional buying and selling model, the shared energy storage business model in this paper effectively reduces the investment and construction scale of user-side energy storage, correspondingly reducing the investment and construction cost of user-built energy storage and the time cost of operating and maintaining physical energy storage.

At present, the EHV transmission system in China has entered a new stage of large-scale construction, which is different from the general EHV transmission lines. The distributed current in the EHV transmission lines has increased substantially. To realize the overall balance of long-distance transmission in the power grid, a differential protection method for EHV transmission lines based on current traveling wave prediction is proposed. The fault current traveling wave of the EHV transmission line is predicted. The fault current traveling wave section is distinguished, and different schemes are selected to realize the traveling wave differential protection for internal and external faults. The simulation results show that the proposed method is reliable and sensitive for the protection of UHV transmission lines, and has certain practical value for ensuring the stable operation of the smart grid.

Based on the actual engineering perspective, this paper presents the site of pumped storage power plants, which is determined by collecting and processing elevation data, river system data, road data, meteorological data, building a river system model, optimizing the dam axis model, calculating indexes and matching upper and lower reservoirs to propose a site for pumped storage power plants. We have constructed a comprehensive evaluation system to determine the location of pumped storage plants, wind farms, and photovoltaic plants to provide reference suggestions and research ideas for the initial selection of a new energy base that integrates wind and storage.

In recent years, object tracking attracts much attention as a upstream task in many specific area. Multi-object tracking is one of them and its challenge are ID switch and ID re-matching due to object occlusion and image quality. In this paper, two possible methods are introduced to improve the baseline. In ReID model, a data augmentation method is built. Its main idea is to discuss and reduce the influence of background by retraining the ReID model. As ReID focuses on the features of objects in most cases, backgrounds can affect the feature metric significantly. In this situation, different objects in different backgrounds are captured and extracted as train data. Specially, a pure background, or background without objects are used in triplet loss to maximum the similarity distance. Besides, an updated SORT is implemented as CFSORT. In CFSORT, features are extracted by retrained ReID and two new processes or units are created. Comparing with the baseline, CFSORT requires lower confidence objects and emphasis a “coarse to fine” process. Those two new units are corresponding to similarity and IOU measurement. Currently, all new detection will be treated equally at first matching action. Similarly, more detection and tracked objects will match in new IOU unit which is stricter than the normal one. As a consequence, the result indicates that CFSORT has an improved performance on several evaluation scores in different scales. Meanwhile, this work can also prove the importance of accepting low confidence.

Deep neural networks (DNNs), which have high accuracy prediction and stable network performance, have been widely deployed in various fields. However, the adversarial example, a sample of input data which has been modified very slightly in a way, may easily cause a DNN to maximize loss. Instead of white box attack being able to obtain gradient information, most DNN based systems in actual use can only be attacked by multiple queries. In this paper, we regard face recognition (FR) system as target, and propose a new method named SA-Attack to generate adversarial samples which cannot be distinguished by human within very limited queries. Experiments show that SA-Attack can successfully attack advanced face recognition models, including public and commercial solutions, which proves the practicability of our method.

In the wind resources assessment, it is very important to get the representative wind speed. The routine method used the liner relationship between weather station and wind farm to calculate it. The roughness and shape of the terrain sometimes takes into poor correlation. This text takes a wind farm in Shandong as example to show the limitation of this method.

Power quality data mining is of great potential in both supply-side and demand-side energy management system. In recent decades, with the wide application of flexible AC/DC power grid and grid-connected renewable energy generation, power quality data has been unified as a generalized model for improving power quality. Meanwhile, power quality monitoring system has also been deployed on a large scale. In order to further highlight the availability and usability of power quality data, the paper integrates various types of information to support power quality analysis. A multimodal data system is constructed to process information collected in different forms into a multi-dimensional data model, which can be pretrained to provide integrated features for various power quality analysis tasks. Firstly, the three data types of voltage waveforms, texts and images are embedded through feature extraction, low-dimensional spatial representation and CNNbased representation, respectively. Then all information is fused with the interaction model based on Attention mechanism. The output of the data model can be sent to networks specific to certain downstream tasks.

Different unmanned aerial vehicles teams often work independently, leaving a “stove-piped” isolation problem, and the cost of collaboration among them is relatively high. Interoperability has been proposed to study the ability of unmanned aerial systems in working together, of which smooth communication is the fundamental requirement. To construct communication ability of different systems, the protocols they support should be essentially concerned. A novel, lightweight and flexible protocol, named as LAB protocol, is developed in this paper, where we study it and STANAG 4586 and MAVLink for comparision. We propose a bridging method to break isolation and verify interoperability among UASs using different protocols through experiments.

In order to solve the power frequency oscillation problem in the virtual synchronous generator (VSG) multi-machine parallel system, a power-frequency "admittance" model is proposed, which can comprehensively describe the powerfrequency response characteristics, but does not consider the primary frequency modulation function. With the massive access of new energy and the continuous penetration of power electronic equipment, there will inevitably be a system with new energy as the dominant power source, which mainly uses VSG technology to connect the power system. At this time, VSG must play a primary frequency modulation function, and the above model is no longer applicable. Based on the multimachine parallel system with VSG as the main power supply, the power-frequency "admittance" model of VSG participating in the primary frequency control of the system is established. The factors and mechanisms that affect the power frequency oscillation of the VSG in the system are analyzed through the transfer function, and the effectiveness of the model and frequency oscillation characteristics is verified by PSCAD/EMTDC simulation.

Laser is one of the greatest inventions in modern times. With the development of laser and the updating of technology, the research on laser technology has made professional exploration in many aspects such as military application, remote sensing, industrial production, etc. However, in various applications of laser technology, the detection of laser spot image and the extraction of center position directly determine the reliability of the application. However, in the process of facula image acquisition, it will be affected by various factors such as background illumination, atmospheric disturbance, camera thermal noise, etc., which makes the facula unclear, image degradation, facula scattering, and facula local loss, which has a serious impact on facula detection and center extraction. Pointed out the difficulties in the field of spot center detection technology, combined with the actual situation in the application, sorted out the main solutions to the current problem of spot center detection technology, analyzed the advantages and disadvantages of the solutions, and provided some ideas and inspiration for relevant researchers.

Automatic people tracking and identification have a lot of application prospects in access control, intelligent monitoring, personalized service, etc. Although the primary sensors used now are cameras, they are challenging to cope with low light conditions, adverse weather conditions, and clothing changes. The privacy risks brought by cameras cannot be ignored with people's increasing awareness of privacy. In this paper, we use a commercial millimeter-wave radar to track and identify multiple people indoors. The mmWave radar can "see" objects even in the dark and protects people's private information. We propose PPMM mechanism to solve the problem of tracking multiple people walking at close distances. What's more, we design transformer for mmWave radar pointclouds (TMP) based on transformer architecture. Finally, we evaluate our model and demonstrate the results on our dataset, which involved 6 people. Our method can track up to 3 people simultaneously. We achieve the best identification accuracy of 86.34% overall different numbers of people scenarios, and the accuracy of single, two, and three people scenarios are 87.93%, 87.00%, and 64.98%.

A non-delayed full link high precision internal calibration scheme of radar system based on built-in calibration module is introduced in this paper. By adding internal calibration module and necessary calibration signal coupling and separation circuit to the existing radar system, this scheme can realize the transmitting, receiving and reference calibration of the radar system in working condition with low cost and low system complexity. It has the characteristics of simple system composition, comprehensive calibration mode, large dynamic range of calibration signal, high precision calibration in full temperature range. It can be used as a feasible scheme for non-delayed calibration of radar system.

The paper firstly introduces the new energy resources and the existing energy storage systems in Southern Xinjiang, and then discusses the hybrid energy storage system that are of the potential to solve the low-inertia issue of the Southern Xinjiang power systems, namely the combination of battery, supercapacitor and flywheel, the method to define their equivalent virtual inertia are proposed. Finally, the paper summarizes the existing problems and future plans to develop the hybrid energy storage in Southern Xinjiang.

In order to cope with the risks of energy supply equipment in integrated energy systems (IES) and reduce costs at the same time, this paper conducts a study on the optimization of multi-market-multi-objective two-tier transactions in integrated energy systems. The IES multi-objective optimization model is developed by considering the operating cost, risk value and bilateral transaction cost as the optimization objectives, and considering various constraints such as electricity and natural gas. The results of the analysis show that the model can better balance the safety and economy of system operation.

The key nodes in complex networks are of great significance for understanding and controlling complex networks. This paper proposes a local centrality node recognition algorithm. This method comprehensively evaluates the characteristics of the node itself, the characteristics of the second order neighbor node and the Ks value of the node. Experimental results show that this method does not rely on other parameters, and can effectively improve the resolution of node recognition.

In order to solve the problem that the single core DSP digital communication system can not meet the requirements of the complex APF(Active power filter) algorithm, a dual core digital communication system based on DSP and FPGA is proposed. The digital communication principle, chip selection and communication between each sub module of the system are described. On this basis, the software flow chart is designed, and an APF experimental platform based on DSP and FPGA is built to verify the feasibility of the scheme.

Against the backdrop of large-scale wind power generation solar energy networking, the problem of peak regulation of power systems is becoming more prominent day by day. Under the premise of considering sufficient absorption of wind power generation and photovoltaic power, the text uses stored energy to assist the depth control peak of the thermal power generation unit, constructs a two-layer optimization model of the fire storage association regulation peak, and optimizes the storage energy charging and discharging plan with the goal of minimizing the valley difference between the load peaks. The lower layer aims to minimize the peak adjustment cost and carbon emission of thermal power generation unit, and optimizes the combined adjustment of thermal power generation unit by considering the difference in peak adjustment depth and cost of different units. The simulation revealed that the addition of an energy storage system reduces the peak adjustment cost of thermal power generation unit.

With the development of load modeling technology, the accuracy of load model is increasingly required. In addition, as the end user of power, the accuracy of load model has a great impact on the simulation of power system. In the load modeling of large power grids, 220KV or 110KV substations are generally used as load nodes, and enterprise load modeling is rarely carried out. In view of the shortcomings of the current comprehensive load model in the enterprise load grid structure, due to the large proportion of the enterprise motor load, this paper uses the comprehensive statistical method based on statistical data to divide the motors with different load characteristics into multiple motor groups, and then aggregate the motor clusters into equivalent motors. Then, according to the characteristics of motors and the influence of distribution network impedance, the motors in the enterprise are aggregated into two or more motor groups to realize the integrated load modeling of the enterprise load grid.

The National Development and Reform Commission and the Central Network Information Office pointed out for the first time in 2020 that digital twin(DT) is one of the seven new generation digital technologies, which means that the digital twin technology has been incorporated into the national development strategy system. Similar to the conception of digital twin, Human Digital Twin(HDT) is referred to as the replica of a physical-world human in the digital world. Human Digital twin has shown considerable potential and attracted extensive attention from industry and academia since it was put forward. This paper discusses the knowledge of HDT and proposes an HDT system based on non-contact sensor Kinect, and briefly introduced digital twin and HDT. This paper discusses the latest HDT technology by combing the references and proposes a system for human digital twin interaction using non-contact sensor Kinect. The future work plan is to develop and implement digital twins in the clothing manufacturing industry, and finally evaluate the application of these digital twins in the real world.

In view of the influence of N-1 disconnection on the maximum load margin and system stability of power-gas interconnection system, a continuous power flow model of power system and natural gas system considering the dynamic piping characteristics of natural gas was established. Based on the increase of system load, the system N and N-1 load margin of power system and natural gas system is calculated by SRSM continuous hybrid power flow. The power flow transfer factor after N-1 branch of the electric-gas interconnection system is defined, and the power flow transfer is predicted for the system so as to obtain the off-limit branch of the system. The weak nodes and EH coupling degree of the power system, natural gas system and electric-gas coupling link under the electric-gas coupling are further evaluated.

We consider the problem of synthesizing optimized circuits for reversible functions using NOT, CNOT and Toffoli gates (NCT), which has important applications in circuit design and quantum cryptography. By defining the state space and state transitions from NCT gates, we transform the synthesis problem into a reachability problem of finding the shortest path connecting the identity function and the targeted function that is to be synthesized. Using Bounded Model Checking (BMC), we propose an algorithm for synthesizing the optimized circuit for reversible functions. For the synthesis of linear reversible functions, we design another method by directly parametrizing the linear invertible matrix. Our methods can produce optimal synthesis of reversible circuits when the number of bits is small, and also can be used as a subroutine of heuristic methods for large bits. Experimental results show the effectiveness of our methods.

The power source of electric vehicle is lithium-ion battery. Due to the influence of aging degree, the driver's estimation error of lithium-ion battery power is caused, so it is very practical to accurately estimate its state of charge. In order to solve the problems of Gaussian white noise and poor robustness of adaptive extended Kalman filter algorithm, this paper adopts double adaptive robust extended Kalman filter algorithm for online joint estimation of model parameters and SOC. The simulation results show that, compared with AEKF estimation, the maximum absolute error, mean absolute error and root mean square error of battery state estimation can be reduced by 1.14%, 0.49% and 0.62% respectively.

Coherence is crucial in quantum information processing, and quantum walk provide an excellent model for studying fundamental quantum theory and quantum algorithm design. The coherence of the initial state, system boundaries, coin operators, and evolution time are all relevant factors in the evolution of coherence systems. Using l1 norm as the coherence measure, the evolution characteristics of coherence in the three-state quantum walk with boundaries in a 2D lattice were investigated, and the effects of broken links, phase damping, amplitude damping, and unitary noise on the coherence of the three-state quantum walk were analyzed. The results show that the boundaries suppress the growth of coherence of a single-point initial state of the entire system, but does not affect the coherence of equal-weight superposition states. Small noise can slightly enhance the coherence of the entire system and position subsystems for broken links and unitary noise, while larger noise suppresses the growth of coherence.

optimal estimation of the attitude, an attitude solution system based on low-cost IMU sensors is proposed. The system uses quadratic coordinate conversion to fuse the gyroscope, accelerometer, and magnetic compass data, and the attitude deviation is corrected by the PI controller, and the attitude information of the vehicle is solved by a complementary filtering algorithm. The experimental results show that the proposed algorithm not only effectively improves the accuracy of the attitude solution to meet the flight requirements of small UAV, but also suppresses the influence of random deviation of the attitude sensor and the noise of the three-axis acceleration measurement on the attitude solution, and achieves the requirement of accurate and reliable attitude data output for a long time.

DUT board is an interface between the Device Under Test system for the IC test. This paper describes how to design a DUT board based on the test system’s configuration. To enhance the quality of performance on the DUT board, it also focuses on some methods of electrical design about layers distribution, trace routing, signal filtering for mixed-signal, high-speed signal, and special components for the V93000 test system.

With the rapid development of deep convolutional neural networks, the use of deep convolutional neural networks to extract features instead of manual features has become one of the current research hotspots. However, deep convolutional neural network can not understand image features well, and there is a “semantic gap”. Contrastive Language-Image PreTraining (CLIP) model is a pre-training neural network model based on matching image and text. Use the pre-trained CLIP model to extract the high-dimensional feature vector of the image data set to be retrieved, and the Local Sensitive Hash (LSH) algorithm was used to extract the retrieval speed to complete the retrieval task based on the image content and text. Experimental results show that compared with other content-based image retrieval algorithms, the proposed algorithm can also understand the text information in the image to complete the retrieval task, and has a wider retrieval range.

Accurate prediction of the surrounding traffic environment is crucial for the safety of autonomous vehicles. However, the limitation of onboard system resources and the complexity and diversity of driving scenes hinder the deployment of scene understanding in the auto-drive system. This paper optimizes the backbone network, uses deep separable convolution to reduce the complexity of network operations and uses multiple attention mechanisms in the decoding stage. On this basis, this paper adopts the shared strategy for the feature extraction module and jointly trains the semantic segmentation and Object detection, which can reduce the network parameters, improve the reasoning speed, and improve the accuracy. We have evaluated the proposed method on the public data set. The results show that our method achieves the most advanced performance and can balance speed and accuracy.

In order to solve the problems of difficult control of cable wellhead, loss of manhole cover, illegal invasion, cable theft and cutting, etc., an intelligent electronic manhole cover is developed, which integrates multiple unlocking methods, edgeconnected agent, regional linkage and power supply module, and an intelligent electronic manhole cover management system is designed. The intelligent manhole cover adopts Internet of Things technologies such as edge calculation, AI image recognition, linkage control, etc., and realizes the functions of alarming abnormal conditions such as illegal opening, tilting and vibration of the manhole cover, monitoring toxic and harmful gases and temperature and humidity, and uploading the perceived data to the management system. This not only solves the problems of weak management of cable channel entrance and safe entry into the well, but also improves the intrinsic safety level of cable lines and channel equipment, and promotes the grass-roots transportation, inspection, quality and efficiency.

The clamp, as a fixed connection element of hydraulic pipeline system, its condition affects sustainability, reliability and safety of the whole system. Therefore, it is essential to identify level of loosed clamp caused by wear and tear. This paper proposes a clamp looseness extent recognition method based on multi-scale feature fusion convolutional neural network (MFF-CNN). The 2-D input constructed by small sample data of 7 distributed FBG sensors is useful for integration of temporal-spatial feature of clamp looseness. Furthermore, multi-scale fused features in MFF-CNN are more effective to overcome problem of small sample data caused by transient nature of mechanical faults and improves efficiency of CNN model training. The proposed model provides better performance on clamp looseness extent recognition compared to 1DCNN, 2DCNN and MSCNN. The average recognition accuracy reaches up to 96.3%.

Real-time monitoring of key data of underwater target in navigation status is an important basis for mastering and intervening the navigation status of underwater target in the whole process and ensuring the navigation safety of underwater target. With the help of CAN bus communication technology, this paper designs a key data monitoring software for underwater target based on CAN bus, which can monitor and display the key data of underwater target such as their attitude, depth, motor voltage, motor speed, driving plate temperature, etc. in real time. On the premise of ensuring real-time performance, the software makes full use of CAN bus communication features, and has the advantages of good human-computer interaction, simple operation and high visibility.

The exploration of frame animation technology based on WinCE platform frame animation is a common animation form that decomposes animation actions in consecutive key frames, that is, draws different content on each frame of the timeline and makes it play continuously into animation. This paper introduces the basic concept of WinCE, shows in detail how to build the development environment of WinCE, and studies the implementation technology of frame animation based on WinCE embedded operating system, and gives a detailed schematic diagram. Finally, the main process of realizing frame animation is introduced in detail with a complete example. Finally, a complete example is used to introduce the main process of frame animation in detail.

Under the background of "carbon peak and carbon neutrality", the scale of the power grid is increasingly expanding, the intelligence level of the power grid is continually improving, and the dependence of the power system on the communication network is increasingly increasing. The existing power communication network has numerous problems, such as network architecture, upper business support, network operation and maintenance management. This paper first introduces the development status and conceptual principles of SDN (Software Defined Network), based on the advantages and disadvantages of SDN architecture, the common attack types and defense methods are introduced. By combining the advantages of SDN architecture and traditional power communication network, a new security defense framework of power communication network is designed. Finally, a new power communication scheme with open, numerical control separation and controller centralized management is constructed, and the new power communication scheme is summarized and prospected.

Images are a way for humans to get information, and they are ubiquitous in our daily lives. Compared to other media, we tend to get information through images. Most current image enhancement methods are based on Retinex theory. However, this method can easily lead to the loss of image information. More importantly, the theory is based on the assumption that light changes smoothly. And this assumption is very unfriendly to bright images. To solve this problem, an unpaired enhancement network for strong light image is proposed. The network consists of illumination curve, illumination loss function and illumination attenuation network. Experiments show that our network can obviously improve the image contrast caused by strong light environment, and significantly improve the quality of strong light image.

The university laboratory is the place for teaching and experiment, where a variety of instruments and electrical equipment and some flammable and explosive chemicals are placed. Due to improper use or failure to operate according to strict procedures, it is easy to cause safety accidents. These accidents will cause the loss of school property, and even threaten the life safety of teachers and students. Therefore, the safety and protection of university laboratories has attracted the attention of university leading departments. Based on this, a laboratory security alarm system is designed. The system selects the GEC6818 platform, carries the Samsung Cortex-A53 series high-performance eight-core processor S5P6818, uses ubuntu, selects the flame sensor, DS18B20 temperature sensor and MQ-2 smoke sensor to monitor the environmental parameters in the laboratory, and uses the buzzer and indicator light as the alarm device. In case of a safety accident, The buzzer and indicator light work simultaneously. After debugging, the system has reached the expected goal, and the system can play a good reference role for security in other places.

Based on the linear BLG tree storage structure, a new multi-scale representation method for map curves is proposed. Firstly, the feature point sets under different thresholds are calculated by DP algorithm, after which data points are stored in a certain order by using the linear BLG storage structure, the map scales corresponding to different feature point sets are calculated according to the minimum visible distance of human eyes and combined with the average distance between adjacent feature points, finally, according to the preset map scale, the corresponding feature point sets are selected, the B spline curve is used to generate the final map curve by fitting. The experimental results show that, unlike the frequency filtering method which shifts the local features of the simplified curves, this method can take into account the overall shape of the simplified curves and the accuracy of the local features, and the multi-scale expression results with obvious differences can be formed by choosing different scales, which verifies that this method has better effects on the multi-scale expression of map elements.

With the continuous development of social economy and the continuous increase of transportation, bridges play an increasingly important role in transportation. Bridges are the basis of accelerating urbanization, and also the key to ensure safe and smooth transportation. With the increase of traffic load, the safety problems of bridge structures also appear. Due to the limitation of construction level, the understanding of structural complexity and the influence of external unpredictable environmental factors, people can not effectively understand the damage of the bridge structure and accurately evaluate the operation and maintenance of the bridge, resulting in a series of traffic accidents. In view of the above problems, this paper carried out the research of bridge health monitoring system based on "cloud edge". It takes acoustic emission (AE), capacitance, impedance, optical sensor, etc. as the basic sensing unit, and combines edge based big data processing with edge computing model as the core and centralized big data processing with cloud computing model as the center. A bridge health monitoring platform based on the cloud-edge-end architecture is designed, which can effectively process data in real time and realize cloud backup, so as to achieve real-time assessment and diagnosis of bridge operation safety without interrupting bridge traffic functions.

Based on the analysis of the classic routing protocol AODV, a link-state-information-based routing protocol MNLSI for wireless MESH networks is proposed to meet the requirements of flight test. The protocol calculates the path with the minimum cumulative weight from the source to the destination by link stability factor and node state information, and introduces packet definition, route discovery and route maintenance. Finally, the protocol is simulated and analyzed by Network Simulator 2. Simulation results show that MNLSI outperforms AODV in terms of packet forwarding rate and end-to-end delay.

Since the implementation of the "Tenth Five-Year Plan" digital seismic observation network project, the daily duty of the seismic network of Jiangxi Province Seismological Bureau has changed greatly from the past, and the workload has increased obviously. In this paper, the design and implementation of the operation and management platform of the seismic network of Jiangxi Province Seismological Bureau are realized by means of information technology, which effectively improves the operation and service quality of the network.

According to the national policy of proposing water pollution prevention and control regulations for the Chaohu Lake basin, to solve the current problem of chemical pollution in the Chaohu Lake basin, the relevant environmental data of the Chaohu Lake basin provided by Chengxin Environmental Testing Company were used, and a random forest algorithm based on high correlation filtering was proposed to construct a prediction model for the concentration of chemical pollution such as ammonia nitrogen in the Chaohu Lake basin. The method uses high correlation filtering to eliminate the strong correlation between water pollution data and adjusts the parameters of the random forest model to achieve the best prediction effect for the training data. The model was tested by real data sets, and the Bayesian ridge regression model, ordinary linear regression model, elastic network regression model, support vector machine model, and correlation vector machine model were used for comparison experiments, and the average relative error was used as the evaluation index. Finally, the random forest algorithm model with high correlation filtering achieved the best prediction results in the study of quantitative analysis of eutrophic chemical pollution such as ammonia nitrogen in the Chaohu Lake basin.

The safety of student dormitory has always been the most important issue of university management，Traditional dormitory security has always been done by administrator, This paper designs a dormitory security management alarm system based on wireless transmission chip CC2530，This design detects the concentration of combustible gas in the dormitory through MQ2 smoke sensor，The RE200B sensor can detect whether anyone enter through the window, Call for help through the SOS button, he above data is transmitted to the management center through the wireless communication module for display and alarm.

As a bridge of the transmission system, flexible DC transmission line plays an important role. If a fault occurs, it will bring serious consequences to the entire transmission system. It is very important to study the fault location of the flexible DC transmission line. According to the system structure of flexible DC transmission line, the mathematical model, operation mode and control strategy of MMC are analyzed. According to the characteristics of transmission line with bipolar short circuit fault, a fault location method based on BP neural network is proposed. Combined with PSCAD/EMTDC, a twoterminal flexible DC transmission system is built, and the fault data of the flexible DC transmission line is obtained by simulation. After preprocessing the data, a sample set is constructed. The neural network analysis uses the two-terminal fault location method, and finally a high-precision fault location model is obtained, which verifies the accuracy and effectiveness of the location method proposed in this paper.

Detection of apple fungi infection is significant to provide the customized prevention and control strategies and ensure food safety. In this study, an identification method of infection of Botrytis cinerea and Rhizopus stolonifera was developed using the RGB images and monochromatic images (MIs) of effective wavelengths (EWs) of hyperspectral imaging. RGB images converted by CIE 1931 colour matching functions, and MIs of EWs were screened by random frog from hyperspectral images. U-Net combining data splicing strategy was adopted to segment the region of rot (ROR). Network features of RGB images and MIs of EWs of ROR were extracted by VGG16 and adopted to develop the classification models of fungi infection by using SVM, RF and KNN. The fused features of two-type images obtained the better classification, outperforming the other one-type image, and the optimal accuracy in prediction set of 99.25% was gotten from the SVM model. The proposed method provides the accurate detection of apple fungi infection and is beneficial to improve the quality of apple fruit.

Realizing energy transition involves developing new clean and low-carbon power systems. The article combines the meaning and fundamental qualities of the new power system under the double carbon goal, analyzes the difficulties and opportunities encountered during the construction of the new power system, and on the basis of this, establishs a new type of power system prediction model and proposes the development direction of the new power system from three perspectives, such as the power side, grid side, and energy storage side.

A load ground station system that supports intelligent charge management is designed. The load ground station is a ground measurement and control equipment for UAV monitoring system. It integrates computer, display, power management unit and photoelectric pod control rocker and other components, which can realize the real-time display of UAV airborne mission load video and real-time control of the pod. The system can support intelligent charge management, and has the functions of complementing mains power, automatically switching power supply path, over-charge and over-discharge protection. It integrates the ground station computer and pod control rocker into one, which can be widely used in rescue and disaster relief, homeland security, urban management and other fields. It has been verified that the ground station can adapt to the requirements of long time operation in the field, and has the advantages of convenient carrying and easy going out operation.

The thesis proposes a new methodology of cost dynamic management to realize the whole process of power transmission and transformation project cost dynamic management. Based on historical cost data to modify expected cost control, building cost goal deviation classified warning system, using BP artificial neural network method for the cost of the deviation of the dynamic monitoring, power transmission and transformation project cost target deviation classified warning model is established. In the case of a substation project, to make a validation method. The results show that the proposed method is feasible and can achieve dynamic monitoring and early warning of power transmission and transformation project cost target deviation.

Multi-MCU system architecture is often needed in applications that require parallel completion of multiple tasks or matrix control applications. When multiple Flash-less MCUs are used in the system, in order to make the system start normally, the general practice is to configure the special program Flash for each MCU to start, which not only increases the hardware cost, but also brings challenges to the mass production of products and subsequent program upgrades. In this paper, we introduce how to use MCU's I2C ISP to realize multiple MCU to share the same Flash to boot, and describe the implementation process and difficulties in detail. The results show that the method of multiple flash-less MCUS sharing a single Flash boot is feasible in practical application.

Data deduplication can effectively reduce data redundancy. However, its write performance is insufficient for existing storage systems, due to the additional calculation and I/O operations. In order to improve the deduplication speed in a distributed storage system, we propose FastDedup, a fast and effective deduplication scheme that focuses on the stored data. FastDedup improves deduplication speed through deduplication task distribution model and multi-container pool technology. Specifically, the deduplication task distribution model maintains the correctness for multiple deduplication nodes working simultaneously. The multi-container pool technology saves the operation time on the data merging stage. Evaluation results on three real backup datasets demonstrate that, compared to the unimproved technique, FastDedup increases deduplication throughput by 3.2% - 69.1%.

With the continuous increase of the power grid scale, the amount of information handled by regulators is increasing. In order to solve the problem that all kinds of information and data cannot be exchanged, and to effectively control all kinds of events, the State Grid Ruian Power Supply Company has developed a set of intelligent control auxiliary system. Based on this, this paper proposes an image recognition method for power regulation and control system based on BHS-CTPN and AA-CRNN to detect and identify the content of the power regulation and control interface. The experiment shows that the method proposed in this paper can automatically identify and process the alarm information and dynamic operation data in combination with the intelligent regulation and control auxiliary system, and achieve full response to all kinds of information. Under the premise of ensuring security, the optimization of power grid operation management mode is realized, and the dispatching ability and response efficiency of power grid alarm are improved.

Aiming at the problems of ID switching and tracking performance degradation caused by frequent occlusion and similar appearance of the tracked objects in dense scenes, a multi-object tracking method named TPFairMOT based on trajectory prediction and FairMOT is proposed in this paper. In the trajectory prediction branch, the object position of the future frame is predicted by using the object bounding box of the past frame and the velocity information learning network parameters, which overcomes the prediction failure caused by the uncertain motion state after the object is occluded in the tracking process. Secondly, the joint learning framework is used to combine the trajectory prediction branch with the detection and re-identification branch, and the tracking error caused by the high similarity between multiple objects in the tracking process is solved by integrating the appearance features and motion features of the tracked objects. Finally, MOTChallenge benchmarks (IDF1, IDs, MOTA, MT, and ML) are introduced to evaluate TPFairMOT, and different trajectory prediction strategies (FairMOT_KF and TPFairMOT_RNN) are used on FairMOT and TPFairMOT for comparative analysis. It is proved that the accuracy and ID switching times of trajectory prediction in this paper are better than other strategies. In addition, TPFairMOT, TPFairMOT_RNN, and FairMOT were compared on the public data sets MOT16, MOT17, and MOT20. The results show that TPFairMOT reduces the number of ID switching when the object is occluded, maintains the long-term validity of the identity information, and demonstrate good anti-occlusion performance.

This paper presents a multi-agent structure and optimization algorithm to enhance energy utilization in a flexible jobshop environment under Time-of-Use (TOU) electricity pricing, in the presence of system changes. The proposed control structure based on multi-agent systems (MAS) takes into account re-scheduling in the workshop under TOU pricing. A multi-objective rescheduling model was developed to minimize due dates, makespan, total power consumption, and energy consumption costs. An improved genetic algorithm (GA) was designed to solve this model by converting the multi-objective into a single objective by optimizing each objective individually and combining the results. Simulation results demonstrate the effectiveness of the proposed algorithm and the application of the MAS structure.

Crash report analysis is a necessary step for software maintainers before fixing software bugs. Fuzzing and symbolic execution tools are often used to automate the generation of crash reports, but there are a large number of duplicate crash reports. Although it uses heuristics such as stack hashing, crash points, etc. to reduce duplicate crash reports, the false alarm rate is high and still requires manual analysis to identify them, which requires considerable effort. In this paper, we propose a crash clustering method based on program crash state extraction, which mainly uses fuzzing techniques to explore the program state space, selectively performs path reduction to generate the core behavior of the crashed program runtime, collects the path information for core behavior and generates feature vectors. Then discriminates whether different crash samples are caused by the same root cause through similarity comparison. We evaluated our tool Diccriminate in real software with real vulnerabilities, and the experimental results show that it can perform crash clustering with 94.31% accuracy, which is more than 20% improvement over existing advanced crash clustering techniques, validating the feasibility and practicality of the method in this paper.

In this paper, the ratio of non-fossil energy consumption is determined, and the case of energy transformation is analysed according to "double-carbon" goal restriction, and the system combination is also discussed. On the basis of the dynamic coupling of power supply structure and energy structure, this paper derives the process of decarbonization of power system by using the ratio of non-fossil energy. Based on this, the paper further divides the power supply structure into three phases, and puts forward the strategy of "Three Steps" to construct a new electric power system.

Anomaly detection in unlabelled and highly imbalanced high-dimensional monitoring data is one of the most urgent and challenging industry problems in the energy industry. Based on the powerful high-dimensional data analysis capabilities of autoencoders, the use of autoencoders for anomaly detection is becoming more and more popular. This paper proposes an anomaly detection method based on deep learning and support vector data description. First, feature engineering is built based on an optimized serial deep autoencoder; second, different feature combinations are studied and compared; finally, anomaly detection based on support vector data description. In this paper, experiments are carried out on the actual operating data of a real steam turbine to verify the effectiveness and accuracy of the proposed method.

With the continuous advancement of smart grid and the rapid development of computer and information technology, massive amounts of redundant and noisy information are connected to the power grid. Therefore, this paper proposes a key stable feature selection method based on improved ant colony algorithm. First, use the artificial intelligence model to solve the optimal number of key features. Then use Filter based on the minimum redundancy-maximal relevance (mRMR) algorithm and Wrapper based on the improved ant colony algorithm to select the optimal key feature subset in two stages. Finally, example analysis with IEEE39 nodes system is used to verify the availability and effectiveness of the selection method.

Short-term power load forecasting is a very significant content in the operation and dispatch of power system, and it is a significant side to make certain the secure and economic operation of the power system and realize the scientific administration and dispatch of the power grid. By way of eliminating the matters of difficult parameter selection and insufficient forecasting accuracy in traditional forecasting methods, this paper use an improved sparrow search algorithm to optimize gated recurrent unit neural network. Firstly, Preprocess the raw load data.Secondly, use the processed data to train the model, and optimize model parameters with firefly sparrow search algorithm. Finally, carry out the power load forecast on the day to be forecasted, and comparative analysis with other two models, SSA-GRU and GRU , the results of the example indicate that the model established in this paper can advance the prognosis preciseness degree effectively and is effective in the application of short-term power load forecasting.

With the increasing popularity of substation inspection robots, the localization and detection of outdoor inspection pictures gradually becomes the focus and difficulty of research. For outdoor substation meters, in different weather conditions, for the current stage of detection algorithm model complexity is high, easy to the detection of the target caused by low recognition rate, is not conducive to the rapid deployment of practical applications and other problems, proposed an improved YOLOV5-based substation detection algorithm, the algorithm uses YOLOv5 + CBAM attention mechanism module to detect meter dials, for different weather conditions The algorithm uses YOLOv5+CBAM attention mechanism module to detect meter dials, which is of great significance to identify meter dials accurately and quickly under different weather conditions and further realize intelligent inspection of substations.

During fault diagnosis of rolling bearings of rotating machinery in nuclear power plant, there are different data processing methods for original vibration signals, and different data processing methods have different diagnostic accuracy under the same diagnosis model. [Purpose] When the performance of the model is limited, in order to improve the diagnostic accuracy of the model, it is necessary to study the data processing method. [Methods] Therefore, the original vibration signal is processed into five ways: time-domain diagram(TD), time grayscale diagram(TGS), fast Fourier transform frequency-domain diagram(FFT), short-time Fourier transform time-frequency diagram(STFT) and continuous wavelet transform time-frequency diagram(CWT) for the initial feature extraction of the original vibration signal, and the five methods are compared and analyzed in the original data, data with added noise and data with less sample size. [Result] The experimental results show that the FFT data processing methods has a more obvious test accuracy under the noisy data and small sample size, and its accuracy is 1.9% and 6.7%(data under 50 sample size) higher than the suboptimal methods in both cases. [Conclusions] Therefore, in the case of noisy data and small sample size, the diagnostic performance of the model can be further improved by adopting the FFT data processing.

As an important means of mining network connection information, link prediction can not only master more hidden information from it, but also discover missing information in the research. It is a common way to complete the incomplete network. In the process of being included in the prediction technology research, researchers have put forward a variety of application algorithms, among which the most representative is the link prediction method based on similarity. Therefore, on the basis of determining the link prediction and technology research status of complex network, this paper puts forward a new method to calculate node influence and resource path, and combines them together for experimental analysis. The final results show that the algorithm is more accurate and effective in practice.

Due to increasingly serious environmental issues, there are growing concerns about the environmental impact of automated container terminals (ACTs). Therefore, the ACT needs to strike a balance between effectiveness and environmental sustainability. AGV scheduling is a vital issue that plays an important role in the efficiency improvement and energy conservation of ACTs. This study examines the AGV scheduling issue in ACTs while taking energy usage and AGV efficiency into account. First, a multi-objective mixed integer programming model with the goals of energy consumption and makespan is built. Then, an improved whale optimization algorithm (IWOA) is proposed to solve this problem, in which task combination and adaptive parameter adjustment strategies are presented to enhance the algorithm. Finally, case study examples are conducted to verify the effectiveness of the proposed method.

The background information of garbage images is complex, and the difference within the class is large. It is difficult for traditional methods to extract significant classification features from garbage images, resulting in poor classification results. Aiming at the above problems, a garbage image classification method based on attention mechanism and SPP model is proposed. First, EfficientNetV2 is used as the basic network, and an improved SE attention module is introduced to make the network pay more attention to the important features in the feature map, suppress the background information, and reduce the impact of intra class differences; Secondly, in order to deepen the network depth and integrate multi-scale information, a feature pyramid structure is introduced; Finally, the experimental results show that on the extended Huawei dataset, the classification accuracy is improved by about 4.8%, reaching 93.5%, which proves that this method has certain advantages in garbage classification tasks.

Semi-supervised learning (SSL) has attracted much interest for its ability to increase model performance using unlabeled data in recent years. The mainstreaming SSL frameworks are built on the teacher-student model. However, the structural reparameterization (SR) model has not been thoroughly studied in the teacher-student model. The SR technology enhances model capacity and improves performance during the training phase. Therefore, we introduce the SR model in the SSL framework, where the SR model is adopted to construct the teacher-student model for image classification. The consistency regularization is applied to the teacher and student models, and the teacher model’s weights are updated based on the exponential moving average (EMA) strategy. To verify the effectiveness of our approach, we conduct experiments using the CIFAR-10 and Food-101 datasets. Compared to the supervised learning of the SR model, our SSL framework has achieved better performance on these datasets.

The maglev power supply system is realized by the loop formed by the contact rail. Hard points on the contact rail are the key factors that affect the continuity of power supply, and even affect the safe operation of the track in serious cases. The hard points height characteristic are related to the separation time and distance between the collector shoe and the contact rail. In order to explore the characteristics of the height data, a hard-points platform for simulating the contact rail is built. The characteristics of the acceleration signal passing through the simulated hard points platform are extracted by time-frequency analysis. Using the neural network model to explore the correlation, the regression prediction of the contact rail height value can be realized. Based on this method, the prediction error of simulating hard points inversion at a specific height is within 2%, and the effect is good. At present, it has been used in engineering practice, and it has played an important role in the detection and maintenance of the hard points of the medium and low speed maglev contact rail.

Current applications of the small target detection algorithm include enterprise logo detection, small face detection, pedestrian detection, traffic sign detection, automatic driving, remote sensing image detection, and criminal investigation. Practical applications are primarily still in the research phase. Consequently, this paper proposes an optimized small target detection algorithm on the basis of Faster R-CNN. Utilizing ResNet-18 with fewer convolutional layers in place of ResNet-50 as the backbone network of the original Faster R-CNN algorithm, the feature extraction network is strengthened, reducing the number of parameters of the algorithm. Moreover, by analyzing the characteristics of small targets, we can eliminate missed detection and false detection of small targets, Faster R-CNN algorithm uses feature fusion to amplify the contrast between small target features and background features in the feature pyramid network; Enhance prediction box deviations caused by pooling regions of interest by utilizing ROI Align. Lastly, the improved algorithm is experimentally verified on the MSCOCO data set. The faster R-CNN small target detection algorithm proposed in this thesis improves the optimized algorithm’s average accuracy from 89.14% to 93.39%, which is 4.25% higher than the original Faster R-CNN algorithm. Intelligent monitoring significantly enhances the detection and recognition of smaller and medium-sized targets.

Key of sparse subspace clustering is to solve an optimization problem based on sparse penalty term to obtain sparse representation coefficients. Ideal sparsity penalty term is ℓ₀-norm, but the optimization problem based on the ℓ₀-norm is NP-hard. At present, most methods for solving sparse coefficients use the convex relaxation of the ℓ₀-norm, ℓ₀-norm as a penalty term, but it can not well describe the sparsity of the representation coefficients. Therefore, In this paper, a nonconvex φ_α energy functional is used to replace the ℓ₀-norm in the objective function and a sparse subspace clustering algorithm based on non-convex φ_α energy functional is proposed, compared with the traditional ℓ₁-norm, non-convex φ_α energy functional increases the sparsity of the representation coefficients and obtains a better similarity matrix, where α ⪆ 0 is a parameter that regulates the degree of non-convex constraints. In addition, the alternating direction method of multipliers is used to solve the optimization problem with non-convex constraints. Experiments on synthetic datasets and face datasets show that the proposed algorithm reduces the error rate of clustering.

Algorithms that model and estimate noise based on statistical properties, such as spectral subtraction, can estimate the distribution of stationary noise, but their performance degrades when suppressing non-stationary noise. Dictionary learning and sparse representation algorithms have made great achievements in solving non-stationary noise suppression. However, the multi-channel speech enhancement algorithm based on dictionary learning needs to manually estimate the parameters of spectrum reduction threshold in practice. In order to obtain optimized noise reduction results, the adaptive estimation of spectrum reduction threshold is of great significance. According to the power spectrum of the signal, the algorithm of spectral subtraction threshold is defined and the spectral subtraction threshold is used to optimize and enhance the quality of speech. The experimental comparison shows that the spectral reduction threshold calculated based on the power spectrum is closer to the optimal result compared with the fixed threshold. In the -10dB noise environment, the multichannel dictionary learning algorithm based on improved power spectrum improves the segmental signal-to-noise ratio by 1-2dB compared with spectral subtraction and non-negative matrix decomposition, and improves the perceived speech quality assessment and short-term intelligibility by an average of 2.3 and 0.11 points respectively. The experimental results show that the multi-channel dictionary learning algorithm based on the improved power spectrum can effectively remove additive noise under both unsteady and steady state noise conditions.

Due to the complex lighting environment in space, the surface coating material of the spacecraft is uneven and has strong reflection characteristics, resulting in the complex problem of extremely bright or extremely dark in the image under sunlight, and it is difficult to extract stable features. In addition, for the actual measurement system, the amount of data for image feature processing and image matching is large, which makes the speed of the key algorithm unable to meet the real-time nature of the application. Therefore, based on the above problems, this paper proposes a set of threedimensional measurement algorithms based on binocular vision, mainly has four parts, camera calibration, Image correction, stereo matching and pose solving. We use traditional image processing algorithm for satellite image, using M software for camera calibration, and the calibration results are further optimized and corrected, due to the requirements of real-time, based on ORB algorithm, the use of Fast algorithm for feature point extraction and descriptor calculation by Brief, and the polar constraint is fused for feature matching. Improve the robustness and real-time performance of the algorithm, and perform pose solving by rotating angle. Experiments show that the algorithm can quickly and accurately achieve three-dimensional pose measurement.

The modular inversion operations in the prime fields play an important role in the whole elliptic curve cryptosystem, and its operation speed directly determines the execution efficiency of the whole cryptosystem. The binary extended Euclidean algorithm in the existing general modular inversion algorithm is improved to obtain an optimized new algorithm. The new algorithm improves the shift efficiency of the original algorithm and reduces the cost of hardware resources. The 256-bit adder used in the new algorithm is split and reassembled, and the carry lookahead logic is used between groups. The addition and subtraction components are pipelined to improve data throughput. The algorithm is implemented using Verilog hardware description language and verified successfully on Virtex - 7 FPGA development board. The verification results show that the design can correctly handle 256-bit modular inversion operations, and the single calculation only consumes 1.12 μs.

In order to replace the complicated operation of the traditional vector network analyzer, the portable magnetic resonance probe performance detection system has become a research hotspot. However, its measurement accuracy has not been high. This research aims to improve the traditional calibration algorithm. While eliminating the directional error, source matching error and reverse tracking error in the traditional single-port network model. The deviation of measurement results caused by the lack of standard calibration parts is eliminated. Compared with the existing algorithms, the accuracy of experimental results are significantly improved.

As a place to receive and distribute electrical energy, substations are an important part of the power system. Therefore, the safe and stable operation of power equipment in substations is very important for the power system, and the safety of power equipment is related to the safety and stability of people's lives and society and the normal progress of industrial production. If the scene of the substation can be displayed using image stitching technology, it will greatly facilitate the maintenance and operation-level engineering construction personnel to understand the substation environment and equipment operation status. We collect images of substations through cheap equipment, use algorithms to construct substation panorama of the collected images, and combine real-time monitoring dynamics in the substation to understand the basic situation of the power equipment in the substation to prevent the occurrence of emergencies in the substation. Therefore, it is necessary to carry out a panoramic construction of the substation.

Robot vision technology has been widely used in many fields since its emergence, and real-time control of robotic arm control systems has long been a challenging problem. When using optical flow algorithms for tracking in conventional vision control, its accuracy and large movements are affected by the size of the integration window to track fast-moving objects, making it difficult to achieve real-time control in robotic arm vision control. By using a smaller integration window, the Pyramid Lucas-Kanade (LK) method can solve the problem that large motions cannot be tracked, but the basic Pyramid LK method is not very accurate. Therefore, an improved LK optical flow method is proposed for the practical application of the traditional LK method for robotic arm control with poor real-time performance and accuracy, applying a combination of the improved FAST (Features from Accelerated Segment Test) corner point detection and the pyramid LK optical flow algorithm. With the improved FAST algorithm, the corner points with the strongest grey-scale variations can be extracted quickly. This approach allows better estimation of optical flows with strong corner points to track moving objects. The corner points calculated by applying the improved FAST feature point detection are first used as candidate feature points; then the candidate feature points are re-extracted by setting filtering conditions with the information obtained from the robotic arm; and finally, the target feature points are tracked using the LK optical flow method.

With the proposal and implementation of the "double carbon" goal in China, it is necessary to further improve the dynamic energy efficiency of the operation process of combined cooling, heating and power (CCHP) units to maximize the economic efficiency of the system. In this paper, the optimization algorithm based on the combination of chaos search of Tent map and nonlinear adaptive particle swarm optimization combines the schedulable resources of energy production, energy storage and energy consumption into a CCHP "source storage" system, which can simultaneously meet the power, heat and cooling needs of the user side. Taking the operation cost and pollutant emission of CCHP system as the objective, a multi-objective optimization model is established. Under the constraint conditions of equipment output, power balance and so on, the equipment operation hourly output with the best economic and environmental benefits is obtained. The calculation results show that the CCHP "source storage and load" system reduces the operation and maintenance costs by 22.31%, and carries out the economic and environmental advantages.

At present, there have been many studies to design various CNN hardware accelerators to accelerate the inference of deep neural network models. The FPGA-based CNN reasoning accelerator can provide sufficient computing power support with flexible data accuracy, lower energy consumption and lower application cost, and has received a lot of attention in the application field of IoT terminal devices with limited computing power and energy consumption. Widespread concern. However, although the current FPGA-based CNN accelerator has greatly improved the speed of model reasoning through various methods, most of the methods cannot be effectively applied to actual terminal scenarios due to limitations in memory and energy consumption. In response to this situation, we designed an acceleration framework that takes into account both inference acceleration and energy consumption. Aiming at the limitation of computing power in the terminal environment, optimize a large number of multiplication operations in the convolution operation that consumes the most computing power in the CNN inference stage, by using local cache and matrix transformation formulas, and skipping pairings by zero values in the calculation process the model inference operation is further accelerated while reducing energy consumption. The experimental results show that compared with the current advanced neural network accelerator, not only the computing power has been significantly improved, but also the energy efficiency ratio has achieved better results. Moreover, this method can not only be implemented in FPGA, but also be migrated to other embedded terminals.

In order to further improve the efficiency and reliability of the active distribution net-work in the process of failure recovery, based on the IBPSO-SA algorithm, the failure recovery analysis of the active distribution network with a high proportion of distributed power sources is carried out. On the basis of preliminary isolated island partitioning, the three objective functions of the largest total amount of important load recovery, the smallest network loss, and the smallest number of switching operations are considered comprehensively. The IBPSO-SA algorithm is used to solve the optimal fault recovery strategy. The IEEE33-node distribution network model is used for simulation verification. The results show that the time-varying and load characteristic models of photovoltaic energy storage systems are superior, and the active distribution network fault recovery strategy based on the IBPSO-SA algorithm can reliably restore power supply.

Graphs, as a type of data structure that exists in numerous scenarios, are applied in various fields of the internet, such as recommendation algorithms, community detection, path search, and so on. In recent years, with the rapid development of the internet, the amount of graph data has greatly increased, and analyzing and utilizing these graph data to establish graph network models and apply them to various real-life production and living scenarios is of significant importance, such as using user-item data to establish a recommendation algorithm model for recommendations. Currently, many graph neural network models have achieved good results, but there is still room for improvement. Better algorithms can more accurately understand the user's needs and bring a better recommendation experience to the user. The paper designs a new way of constructing graphs and improves the graph network message-passing algorithm. A new graph neural network algorithm, GAFM (Graph Attention Factorization Machines), is proposed. Compared with traditional algorithms, this algorithm will construct graph structures of users and items based on original data, and can also construct item graph data by linking to a knowledge base to introduce new items. When this algorithm performs message passing between nodes, it will consider the first-order neighbor set of the central node and the second-order cross-item of the first-order neighbor set, and use an attention mechanism to regulate the weight in the message passing process to capture high-dimensional neighbor information and improve accuracy. The experimental results on real-world datasets show that the algorithm has better performance compared to existing graph recommendation algorithms.

For the sensor drift problem in electronic nose systems, we propose a novel orthogonal constraint-based subspace learning technique in this study. First, the method minimizes the central distance between two domains and preserves the geometric structure of data in the subspace after projection. Second, a linear regression function based on the l_2,1 norm is used to represent the mapping relationship between the subspace and the label space, allowing the correlation relationship between data before and after projection to be maintained, feature extraction ability to be improved, and noise robustness to be achieved. Third, the orthogonal constraint is utilized to encourage geometric interpretation and data reconstruction. Finally, we conduct experiments on typical sensor drift datasets with long-term drift, and the results demonstrate the effectiveness of the method.

Accurate prediction of short-term power load can ensure the safety of power system and reduce the cost of power generation. Aiming at the problems of large fluctuation, strong randomness and difficulty in peak value forecasting, a short-term load forecasting method of gated recurrent unit (GRU) based on improved sparrow search algorithm optimization is proposed. Firstly, In the process of load forecasting, the influence of temperature, humidity, electricity price and other factors on load forecasting is considered, and they are taken as input variables of the forecasting model. Secondly, the improved Levy Sparrow Search Algorithm (LSSA) is used to optimize the parameters of GRU network and fully mine the characteristic information of load data, thus improving the forecasting performance of the model. Finally, the load power of a place in Australia is analyzed as an example, and compared with similar forecasting algorithms, the results show that the forecasting effect of this method is better than other methods, and the forecasting accuracy of shortterm load is effectively improved.

For the path planning problem of overhead transmission net sealing robot in the process of sealing network, a visual perception and decision method based on deep reinforcement learning is proposed. By combining the perceptual capability of convolutional neural networks with the decision-making capability of reinforcement learning, the method achieves direct output control from the visual perception input of the environment to the action through end-to-end learning, forming a closed loop between the system environment perception and decision control directly, and obtaining the optimal decision strategy by maximizing the cumulative reward return of the robot's interaction with the dynamical environment. Simulation experimental results prove that the method can meet the requirements of multi-task intelligent perception and decision making, and better solve the problems of traditional algorithms such as easily falling into local optimum, oscillating in narrow passages and unreachable targets near obstacles, which greatly improve the real-time and adaptability of trajectory tracking and dynamic obstacle avoidance of the net sealing robot and ensure the safe operation of the net sealing robot in transmission line sealing operations.

The IEEE SystemC standard is widely used in Electronic System-Level design, and this high-level abstraction enables more efficient architectural analysis, design, and redesign. Although it often runs on multi-core processors, SystemC itself is written to execute sequentially and cannot use multi-core resources. Based on Out-of-Order PDES for SystemC, this paper provides a more fine-grained data race detection algorithm, which provides more accurate conflict detection results without losing extra time in compilation. The experimental results show that the accuracy of data race detection based on array element granularity is better than that of field granularity detection, meanwhile it can effectively improve the execution efficiency of the program.

The detection of click fraud in online advertisements on the Internet for the purpose of extracting advertising fees is one of the important aspects of machine learning applications. In this paper, using the data information of 400000 ad click cheating cases, we use recursive feature elimination method to determine the predictors and use five algorithms of gradient boosted decision tree (GBDT), random forest (RF), Adaboost, KNN and LGbmclassifier to train a single classifier, compare the prediction performance of each type of classifier, and the first three with better prediction performance The top three with better prediction performance were fused with multiple algorithms for prediction. The experimental results show that the random forest, Lgbmclassifier and Adaboost algorithms have the highest prediction accuracy, 87%, 83% and 79%, respectively, with AUC values of 0.90, 0.87 and 0.81. The prediction accuracy of the multi-algorithm fusion model taken in this paper can improve by 3% compared to the single algorithm with the best prediction performance, reaching 90%.