This PDF file contains the front matter associated with SPIE Proceedings Volume 12346, including the Title Page, Copyright information, Table of Contents, and Conference Committee listings.

Surveying and mapping tools are indispensable in human production and life. This research integrates the defects and functional limitations of traditional surveying and mapping tools, and designs an intelligent surveying and mapping tool. The Electronic Workbench simulation software is used for modeling and simulation. The ranging is realized by emitting laser. The Bluetooth transceiver module connects the mapping and calculation operation. The LCD has the function of inputting parameters and displaying the measurement results. This intelligent surveying and mapping tool integrate the functions of measurement and mapping, which can speed up the surveying and mapping, improve the accuracy of Surveying and mapping, bring convenience and comfort to users, and has the value of use and promotion.

With the development and progress of unmanned aerial vehicle (UAV) technology, UAVs have become the main force in performing various complex tasks, but at present, there is little related research on the application of UAVs in oil pipeline inspection, especially where there are continuous snowy peaks in alpine mountainous areas. Ravines, harsh natural conditions, and complex electromagnetic environments have adverse effects on the launch, recovery, use efficiency, command and control, and maintenance support of UAVs. By analyzing the characteristics of the alpine environment and its impact on UAVs, this paper puts forward solutions from three aspects: strengthening the support guarantee, enhancing the adaptability and innovating the methodology of designing alpine UAVs. It aims to provide a reference for UAVs to carry out the oil pipeline inspection tasks in alpine mountain areas.

The traditional granary ventilation system generally adopts upper-cage ventilation and floor trough ventilation, which leads to serious vertical stratification of grain moisture. In order to overcome the defects of the existing technology and fully consider the special storage environment of the dry bulk terminal, this paper studies and designs an intelligent ventilation system suitable for the grain storage bin of the dry bulk terminal. The combined method, equipped with an external mobile fan on the basis of the internal axial flow fan, solves the problem of poor heat dissipation effect of the ventilation system in the special environment of the dry bulk terminal through the structural cooperation of the annular main air duct and the horizontal air duct.

With the construction and continuous growth of the smart grid, the maintenance of traditional manual transmission lines has been gradually replaced by software robots or intelligent equipment for maintenance, and the trend of intelligence and digitization has become increasingly clear. need. This paper analyzes the application of RPA technology in the reconciliation process of material outbound on the power grid, and studies how RPA technology is applied in material reconciliation, helping the power grid to better apply RPA technology.

Facing the problem that the isolation execution time of multi-user file resource isolation method is too long in the application process, a multi-user file resource isolation method based on Docker container is designed. It identifies the characteristics of multi-user file resources, expands the storage space, divides the locking module using Docker container, clarifies the subject-object relationship in the host, introduces the concept of time sequence and sets the resource isolation mode. Experimental results: The mean isolation execution time of the multi-user file resource isolation method in the paper, compared with three other multi-user file resource isolation methods, is 62.183s, 82.983s, 82.660s and 83.393s respectively, indicating that the designed multi-user file resource isolation method is more effective when combined with the Docker container.

In order to promote the development of display technologies such as desktop computers and notebooks, a graphics module based on the E9171 display chip was designed. The hardware design and software adaptation method of the graphics module are introduced, and the performance of the module is tested. The 3DMark11 performance test results show that the performance of the graphics card is about 50% higher than that of the previous generation. The performance-to-power ratio of the graphics card is excellent, which can meet the application of most embedded occasions, and is a very good realization of graphics module.

During initial target tracking by radar, the sighting telescope is usually used to guide the antenna to achieve the initial target intercept. When the radar changes from the guidance state of the sighting telescope to the self-tracking state of radar, it’s easy to lose the target due to the inconsistency between the movement direction of radar antenna and that of target. To solve this problem, a regenerated feedback compensation control strategy is proposed based on speed memory extrapolation and compensation timing selection in this paper. By loading the algorithm into the real equipment and tracking the Unmanned Aerial Vehicle, it’s verified that the method can achieve the stable tracking of the target at a faster speed.

Quadrotor aircraft system is strongly coupled and underactuated, which is a nonlinear flight control model. It is easy to b e affected by system uncertainty and external disturbance during flight. Therefore, an improved fuzzy PID control with e xpert control method is proposed. After establishing the mathematical model of the quadrotor aircraft, the fuzzy PID cont rol method is optimized and improved, and the fuzzy control is optimized by using expert rules and experience. Through Matlab/Simulink simulation comparison, the results show that the proposed control algorithm can optimize the control ef fect of the quadrotor aircraft.

Cameras and LiDARs are widely used in the important problem of visual coverage due to their advantages. However, most existing results consider the coverage metrics for cameras and LiDARs separately, and few studies focus on the modeling of camera-LiDAR sensor. In this paper, a geometric model of camera-LiDAR sensor is established firstly, and six effective criteria are then proposed to address the coverage performance of the camera-LiDAR sensor. A generic coverage task of camera-LiDAR sensor network for a target object is presented, and a gradient ascent algorithm is proposed to optimize the deployment of the camera-LiDAR sensors for good coverage performance. A simulation is provided to illustrate the effectiveness of the constructed model and the proposed method.

In the ballistic parameter return system, due to the high dynamic changes of the projectile speed and trajectory, a large Doppler frequency shift will occur, which makes the acquisition of the spread spectrum signal more difficult and the acquisition time increases. The effectiveness of shells on the battlefield will be affected. The use of partially matched filtering combined with fast Fourier transform (PMF-FFT) can effectively improve the capture efficiency and greatly reduce the capture time. The selection of the number of partially matched filters affects the characteristics of the algorithm. The theoretical performance and capture performance are analyzed in detail in this paper, which has guiding significance for theoretical analysis and engineering practice.

Multiple object tracking is complex and this essay will analyse a method to build flexible tracking system, contributing to people detection. The system can perform automatic detection using Kalman filter’s prediction and update function, and bboxOverlapRatio calculates the cost of matching detection and track. Then, referring to the cost calculated, Munkres’ version of the Hungarian algorithm matches detection and track. Also, the method uses customized non-maximum suppression and approach to discover and eliminate false alarm tracks. The simulation tracks people from a moving car, and finally, under common circumstances, the method can help detect people successfully and rapidly. Although, there are still some problems existing, like that it can’t recognize the static people and the distance between pedestrians and car will influence the result. The method still provides new ideas of detection and is useful to solve the problems of shelter among objects from one another, the effect of motion, including vagueness and capture.

Corona effect and audible noise caused by transmission line in UHV transmission project have always been the key problems worthy of attention. At present, the measurement of audible noise of transmission line is usually measured by direct detection method, but it is easy to be affected by background noise in complex environment. Based on the correlation model of audible noise and corona current, aiming at the need of audible noise detection of Ultra-High Voltage Direct Current (UHVDC) transmission line, a realization method of software, which is based on C # Net and MATLAB, is proposed to obtain the sound level of audible noise by corona current measurement data. The software includes corona current time domain waveform measurement model, corona current processing module, audible noise calculation module and so on. The test results show that the software has complete functions and can meet the audible noise detection method.

In this paper, the partition control technology of ammonia injection is adopted and optimized to study the cooperative control of dust removal systems and denitration systems. Based on the existing intelligent feedforward control strategy, its prediction models are modified, and an advanced control strategy for ammonia injection is formulated to realize the rapid response and accurate controlling of the ammonia injection under variable load conditions. The main conclusions can be summarized as follows: The intelligent partition control system of ammonia injection can realize the dynamic adjustment of ammonia injection quantity and improve the automation level of operation and the refinement level of management of the denitration systems. In terms of the future work, the cooperation control of ammonia removal in coal-fired power plants should be carried out further to enhance the denitration efficiency of the technology.

Based on the traditional turbine vibration fault diagnosis expert system, the rough set theory is introduced to solve the bottleneck problem that the expert system is difficult to obtain complete knowledge. The system starts from the decision table formed by historical fault data, uses rough set theory to reduce, and constructs an expert system knowledge base model. The confidence level of the diagnostic rule is expressed by calculating the membership roughness of the rule. Using inference engine and fault case base, the dynamic maintenance of knowledge base is realized.

The traditional maintenance mode is gradually unable to meet the needs of today's subway development, with the key structure of the UPS equipment in the subway power supply system as the research object, under the data provided by the Shanghai subway intelligent monitoring system platform, the system design of status analysis is conducted, help for the whole-process optimization, monitoring and management of the operating status of the UPS equipment, and provides a reliable theoretical reference for the troubleshooting of UPS equipment in the subway power supply system in the future.

The test is done to test the original incorporated microfluidic system. It can distinguish motile sperm from tiny sampling hard to manage by traditional sperm sorting approaches. This is in the light of the ability of motile sperm to through streamlines disposable, which is an refined design accomplish sample inlets, outlets, various channel, and a novel is difficultly compelled pumping design that suppliess a stable strem of liquid. The instrument meets the requirement in tiny installations where potty numble of sperm require to be classified. A micro-scale integrated sperm sorter (MISS) is an apparatus that harness the mechanism of osprey sorting onto a straightforward disposable polymetric micro channel device. The MISS separate motile sperm from normal semen cases that are hard or impossible to handle via conventional sperm sorting techniques. The MISS is economical and can function and can be used simply. By applying the method MISS, the motile sperms are separated from the non-motile sperms successfully. By using the MISS system, the percentage of Motile sperm increase to 90%. The MISS maybe able to decrease risks24 related to ICSI (Intracytoplasmic sperm injection) by means of separating viable sperm with a mild, biomimetic sorting system in line with sperm motility. Further, it also eschews centrifugation and case compaction, leading to sublethal ruin sperm.

By combining new energy technology and environmental protection technology, the high-efficiency and energy-efficient food waste recycling device is manufactured. To guarantee intelligent system performance without manual operation or external power supply, we design remote concentrated control, PLC control, and energy control systems by using solar power for energy supply. These systems enable us to complete bio-fermentation and transform kitchen waste into botanical organic fertilizers. To achieve dynamic energy supply and expenditure balance, the system uses photovoltaic technology to replace conventional power sources and smart control to reduce energy use. The system can be employed in densely-populated areas like factories, educational facilities, tourist attractions, etc., which is of great significance for energy conservation, emission reduction, and environmental pollution reduction.

Traffic flow is one of the most important urban operation and land development indicators. Effective traffic flow prediction is of great significance for optimizing urban transit systems and the improvement of land use planning. In this paper, a method based on area aggregation combined with long and short-term memory neural network + graph convolutional neural network is proposed to overcome the situation that the traditional travel matrix with zero elements makes it impossible to use deep learning methods for prediction. Experimental results show that the method proposed in this paper effectively improves the situation that the direct use of deep learning methods makes the prediction fall into local optimum. In addition, this paper has essential references and implications for traffic volume prediction.

The traditional fire monitoring system mainly collects data through smoke, temperature and humidity sensor, carbon dioxide sensor and so on, and simply uses a fixed threshold to monitor or predict disasters. However, this method uses sensors to detect the fixed threshold, which will lead to different judgment accuracy in different regions, different installation environments and different seasons, and is often limited by bandwidth and power. It is easy to lead to problems such as misjudgment or failure to trigger in more and more complex building environments, excessive reliance on cloud servers, and excessive data delay. Therefore, a building environment security early warning method based on tinyml technology in cloud computing environment is proposed. The system analyzes and collects smoke data in real time through real-time tinyml technology, so as to judge whether there is danger in the environment. Combining cloudml technology to analyze and intermittently collect temperature and humidity environmental data to predict future environmental safety conditions can improve system efficiency, optimize data delay and achieve more accurate prediction accuracy.

Global warming has had a huge impact on our lives, and carbon dioxide, as one of the greenhouse gases, has drawn enormous attention. This paper briefly introduces the urban transportation carbon emission monitoring and analysis platform. First, this article briefly introduces the platform architecture and functions. After that, the mobile carbon emission monitoring system is introduced, including the carbon emission and driving characteristics monitoring module, the traffic and road condition information collection module, and the meteorological monitoring module. After that, the transportation carbon emissions are analyzed from the micro, meso, and macro levels. At the macro level, bottom-up and top-down methods are selected to calculate carbon emissions quantification. Finally, through an example analysis, it is concluded that small passenger vehicles and heavy trucks are the two most important emission sources in urban traffic.

Type-reduction is the key step of interval type-2 fuzzy c-means clustering(IT2FCM) algorithm, which determines the accuracy and efficiency of IT2FCM. In order to solve the problem of inefficiency of traditional IT2FCM algorithm which adopts iterative type-reduction method, an adaptive type-reduction(A-TR) method based on the result feedback was proposed, which was used to optimize the interval type-2 fuzzy clustering algorithm. The proposed A-TR method can find equivalent type-1 fuzzy sets which can instead of corresponding interval type-2 fuzzy sets according to the clustering results, so as to achieve the purpose of rapid type-reduction and improve the efficiency of traditional IT2FCM algorithm. We conducted several experiments with remote sensing images. The experimental results show that the proposed method not only improves the efficiency of IT2FCM, but also ensures the clustering effect.

The uncertainty of SAR platform position or non-ideal propagation mediums in synthetic aperture radar (SAR) imaging will cause phase error, which can result in image defocusing. To solve this problem, autofocusing technology in SAR imaging is used to correct the unknown phase error based on the collected data, and hence obtain the well focused image. In this paper, we propose an L₁-norm regularization based sparse SAR autofocusing imaging method. The method uses an iterative algorithm to implement both phase error estimation and L₁ regularization reconstruction. Firstly, the proposed method introduces the phase error term into SAR imaging observation model based on the autofocusing principle. Then it will estimate and correct phase error during the sparse imaging by solving an L₁-norm regularization problem. And thus obtain well-focused sparse SAR images. Experimental results based on real data verify the effectiveness of the proposed method.

We propose a novel model for 3D reconstruction. At present, most learning-based reconstruction methods process highresolution input images from coarse-to-fine manner, and have achieved good reconstruction results. However, these methods usually use the plane sweep volume with fixed depth hypothesis to construct the cost volume, which may miss the small boundary and the depth mutation region. We use the uncertainty estimation module to adapt to the uncertainty of depth prediction per-pixel. In addition, we introduce a double self-attention module into the feature extraction network, so that the network has the ability to capture the long-range dependent features related to depth inference tasks. Experiments were carried out on DTU datasets. The results show that our method achieves excellent performance.

Synthetic aperture radar interferometry (InSAR) exploits two complex-valued images of the same target area to extract interferogram. However, due to the periodicity of trigonometric functions, the interferometric phase is only measured modulo 2π and wrapped between [-π,π). Therefore, phase unwrapping is necessary in InSAR processing for absolute terrain phase. The phase unwrapping method has a great influence on the accuracy of phase recovery. The phase difference between wrapped phase and unwrapped phase, usually called phase ambiguity cycle, can be estimated as an integer multiple of . The phase unwrapping problem is essentially the problem of estimating the phase ambiguity cycle. Since the phase ambiguity cycle is equal to zero in most cases, its estimation can be regarded as a sparse signal recovery problem. In this paper, an L₁-norm regularization-based SAR phase unwrapping method is introduced to estimate the phase ambiguity cycle and obtain the high-quality absolute phase. In the proposed method, we firstly construct the phase unwrapping model based on the relationship between wrapped and absolute phase. Then the phase ambiguity cycle will be recovered by solving an L₁-norm regularization problem. Experimental results based on simulated and real data validate the proposed method.

To address the lack of effective models and data in field combat path planning and position planning of armored vehicles, and to solve the slow real-time calculation speed due to the high precision and large data volume, an overall correction method is proposed targeting the strong coupling relationship between field environment characteristics and vehicle maneuverability. Based on this idea, a vehicle passing probability model under different layer data is constructed with the known field elevation data, vector data, and the maneuvering parameters of the reference vehicle. This model produces a grid of passing probability for each layer, and then weights and sums the passing probabilities of all layers to obtain the comprehensive probability of the whole area. In addition, this paper performs a three-level buffer calculation on the road layer, which generates different passing probabilities in the outer layer, the middle layer, and the inner layer. The calculation method effectively avoids the typical problem of A* path planning, which is prone to producing path results closing to the edge of the obstacles; for the situation where multiple geometric elements of a layer occupy one grid at the same time, the passing probability of the grid is comprehensively calculated through the area proportion, making it more suitable for the actual situation. Finally, the effectiveness of the method and model is verified based on the QGIS platform.

Based on the development trend of electronic parking brake technology, combined with the actual application requirements of the current electronic parking brake system, this paper completes the in-depth research of this topic for the purpose of promoting the application of the cutting-edge technologies and high-tech products related to automobile safety. After the introduction and analysis of the basic structure and working principle of the electronic parking brake system, this paper will focus on the development and research of the electronic control unit of the electronic parking brake system after determining the function and demand relationship of the electronic parking brake system. Through the analysis and design of hardware and software, combined with bench experiments, the simulation results of the electronic control unit will be obtained, and reasonable suggestions will be put forward for the development and research of the electronic parking brake system, so as to effectively improve the effectiveness of the development and research of the electronic parking brake system. At the same time, it also makes a positive attempt for the further innovation and development of electronic parking brake technology.

With the popularity of SMS services, telecommunication fraud incidents occur frequently, which seriously threaten people's life and property safety. The existing fraudulent SMS recognition techniques have the problems of feature selection relying on manual and low recognition. To solve this problem, a fraudulent SMS identification method (referred to as BERT-SIG) that combines BERT (bidirectional encoder representation from transformers) and Sigmoid is designed. The method performs automatic feature extraction by multi-headed attention mechanism, improves the quality of recognition by Sigmoid activation function, and performs the task of recognizing fraudulent SMS by fine-tuning to improve the efficiency of fraudulent SMS recognition. In the experiments, two open-source datasets, "Spam Email" and "E-Mail classification NLP", are used to check the persistence and robustness, and the BERT-SIG is compared with various machine learning models. The experimental results show that BERT-SIG has the best performance accuracy and F1 score, reaching 99.72% and 99.53% respectively, and can effectively identify fraudulent SMS.

The design scheme of high-speed network communication equipment based on space network is proposed, and completed a ten gigabit network communication equipment with high-speed space communication rate. This equipment is proposed for the high-speed transmission speed of space communication network required by large spacecraft. The hardware design includes power supply module, monitoring management configuration module, gigabit network module 1, gigabit network module 2, ethernet switching and three-layer routing module, network connection module. The actual measurement shows that the ten gigabit network communication equipment can realize gigabit network communication and ten Gigabit network communication, and the communication rate can reach 10Gbps. It has high reliability, can adapt to the complex space environment, and can speed up the spatial data transmission.

Facing the development trend of the surge in the production capacity of a certain key aircraft type in China during the 14th Five-Year Plan period, a large-component intelligent attitude adjustment and alignment system based on the Inovance motion control system was designed. Carry out research on key technologies such as coordinate self-transformation technology based on piecewise straight line fitting and dynamic tracking measurement technology based on actual value prediction. Solve a series of problems such as the difficulty of installation and debugging of large parts of the aircraft and the low success rate of automatic measurement. System equipment is fully localized. Inovance AC812 is used as the core controller, which effectively reduces the risk of imported equipment brands getting stuck. The application shows that the system has high assembly precision and assembly efficiency, and can meet the requirements of use.

After Daqing Changyuan oilfield entered the ultra-high water cut period, the reservoir heterogeneity was further enhanced, the remaining oil was highly dispersed, and it was more and more difficult to tap the potential. In order to realize the continuous potential tapping of the old oilfield, measures such as well pattern infilling and subdivision water injection were adopted, the water injection quality was continuously improved, and the increase of water cut and decline of production were effectively controlled. However, the number of oilfield development wells is increasing, the layered water injection is more and more fine, and the layered test workload of water injection wells is also increasing. According to the traditional layered measurement and adjustment mode, it is necessary to continuously increase the investment of testing personnel and equipment, so that the cost of oilfield development continues to rise and the benefit gradually decreases. In order to explore a new way for the high-quality development of old oilfields, aiming at further improving the qualified rate of water injection and improving the development efficiency, the layered test management system of water injection wells based on PC end and mobile end is developed by using information technology, and the information-based and intelligent test and adjustment operation management mode is established. The system includes five modules of "test operation, safety management, cost reduction and efficiency increase, staff training and team construction", and realizes "informatization of production operation, remote quality supervision, networking of training and learning, and intelligent allocation of water wells". It improves the layered test efficiency of water injection wells and reduces the test operation and management cost, which is of great significance for further prolonging the oilfield development cycle and realizing high-quality sustainable development.

In view of the harsh environment of the construction site, and the difficulty of directly measuring the height and volume information of the mound unloaded in the construction site, a volume measurement method based on binocular vision is proposed. The experimental results show that the error rate of volume measurement is less than 6% and the error rate of soil height measurement is less than 3.5%, meeting the construction requirements. It can achieve the volume estimation, soil height information extraction and soil positioning of the bulldozer in the process of travel, and provide information support for the planning of the bucket movement trajectory of the bulldozer.

In view of the inaccurate statistics of quality information, the accumulation of measurement tasks can not be measured in time, and the traceability of abnormal parts can not be realized in the process of parts inspection in the metering room of a manufacturing enterprise, a measuring room parts measuring system was designed based on B/S architecture. In view of some problems that exist in the process of measuring parts in the measuring room, this system provides solutions through modules. The system uses Java programming language and selects MySQL database for data storage. The front-end measurement system page uses Thymeleaf template engine and the back-end uses SpringBoot framework to realize the business logic. Through the practical application in a manufacturing enterprise, the system can meet the business needs of the enterprise to solve the problems of low efficiency and inaccurate data filling in the measurement process.

This paper is aimed at the problem of large campus networks with too many applications and different security levels for authentication. Based on discussing and analysis the characteristics and shortcomings of traditional authentication systems, this paper introduces the development trend of current authentication platforms. Moreover, it designed and implemented a unified campus identity authentication platform based on CAS. Judging from setting up an experimental platform for the trial operation and analysing and evaluating test data, this paper proves that the system is safe and feasible, simple, and efficient in access management and maintenance, so the system can be deployed in large-scale campus networks. The process of designing a unified authentication platform will analyze the campus network, design an authentication platform suitable for large campus networks based on the analysis results. The use of unified authentication in the case of multiple network applications is widespread, and it is because of these needs that numerous authentication technologies have been born. These authentication techniques are compared one by one to select suitable solutions, and finally the solutions are selected and then integrated for use, with the aim of complementing each other. Both Shiro and CAS used in this design can be used for authentication but they have their own advantages, and their integration can complement each other. A series of tests are required for this project to see if it meets the design requirements and to plan for follow-up maintenance.

With the rapid development of computer network technology and digital technology, network education has entered a new era of digitalization, intelligence and individualization. New communication technologies and network teaching methods provide more network education resources and bring unprecedented development to network teaching construction. In recent years, people's demands have been rising. They are no longer satisfied with browsing and downloading texts, but need media resources that can be used and interacted in real time. Streaming media is born from this and has developed rapidly. Streaming media is limited by bandwidth in the process of transmission, and cannot maintain a large number of clients to connect at the same time. For this problem, this paper adopts the P2P mode to improve the parallel speed. On the basis of the experiment, the P2P network topology of the abstract layer of the network is proposed, and a P2P-based network teaching system is established. The system has been developed and tested, and has been applied in the relevant school local area network. Experiments show that the system solves the limitation of the resources of a single server in the traditional system. It makes reasonable use of idle resources and greatly improves the capability of the whole system.

This paper uses data mining technology and artificial intelligence technology to establish a crime risk prediction model for urban scenes. We provide a low-cost and large-scale crime risk prediction method. We combine scenario analysis of crime with scenario risk prediction. The method presented in this paper can effectively identify the scenario variation characteristics of crimes and has a good effect on identifying the elements and layout characteristics of urban scenes with high crime risk. In addition, we consider the scenario influence factors of crime and dig deeper into the causes of crime, providing a reference for multi-scale crime geography research. This study helps promote the interaction between other subjects and criminal geography.

Aiming at the pain points of low efficiency of garbage classification, low intelligence, large personnel investment, and lack of supervision in most communities, an intelligent garbage classification control system is designed based on vision and STM32 controller, including voice broadcast module, multi-sensor Collaboration module, wireless network access module, visual recognition module, remote control and diagnosis module, based on the authorized patents, the odor filter module was developed, and the automatic bucket changing function module was redesigned, which improved the efficiency of garbage classification and reduced manpower and material resources. Input costs. The test results show that the control system can effectively improve the work efficiency of sanitation personnel, the operation process is simplified and humanized, and it is suitable for promotion and application in various occasions, and has good market prospects and economic benefits.

With the improvement to China's living standards, the production of household garbage also increases, soda cans, plastic water bottles and other beverage packaging accounted for 30 percent of the household garbage. At present, the recycling of beverage bottles mainly relies on manual sorting, which is low in efficiency and easy to cause secondary pollution, resulting in the low utilization rate of beverage bottles. In this paper, a new type of beverage bottle recycling device is proposed, which integrates three functions of compressed beverage bottle, classification of beverage bottle according to the material, and centralized treatment of the waste liquid in the bottle, making the recycling more efficient. The mechanical design and partial structure of the device are analyzed in this paper. The new installation can reduce the transportation cost of beverage bottles, improve residents ' environmental awareness, improve the urban landscape, and reduce the burden of waste transfer stations.

Cloud computing uses cloud resources to provide users with computing and storage services It has changed the processing mode of information resources for individuals and enterprises But The existing security reinforcement methods of cloud computing environment can not deal with external attacks caused by the singularity of cloud computing environment services and its own vulnerabilities This article starts from the cloud service itself the security threats of cloud computing platform are studied from three aspects: cloud service interface and network interface In order to deal with these security threats, the cloud service itself is analyzed and compared Randomization method of cloud service interface and network interface next. A randomization deployment model with multi-level, inter layer randomization awareness and distributed collaboration is proposed. The problem of randomization parameter awareness in model deployment is discussed the cooperation of the same layer services on different virtual machines The corresponding solutions are given. Finally, the security measures and limitations of the randomization method are discussed. Point out the future research direction.

The adaptive deep convolutional neural networks (ADCNN) with wide first-layer kernel algorithm model for processing one-dimensional bearing vibration signals has low diagnostic accuracy for data noise and variable load. Therefore, the adaptive deep convolutional neural networks with wide first-layer kernel rolling bearing data diagnosis algorithm based on adaptive batch normalization (ADABN) adaptation is proposed. The algorithm first builds 64 × 1 the first wide convolution kernel layer, and then construct all 3 × 1, and finally build the full connection layer and SoftMax layer. At this time, the mean and variance of the target domain samples are used in each BN layer of the network to replace the mean and variance of the source domain samples used in the original BN layer, so as to achieve the purpose of data domain adaptation. The research on the adaptive deep convolutional neural networks with wide first-layer kernel fault diagnosis algorithm based on adaptive batch normalization is carried out on the ship propeller bearing data set provided by CSIC. The results show that the fault diagnosis accuracy reaches more than 99% in the case of variable data domain, which is better than other algorithms, and improves the adaptive ability of adaptive deep convolutional neural networks with wide first-layer kernel algorithm in the data domain.

In this paper, a novel deep reinforcement learning algorithm Proximal Policy Optimization (PPO) based on F-divergence is proposed to realize the motion control of unmanned surface vehicle. Aiming at the nonlinear and underactuated characteristics of unmanned surface vehicle system, the new reinforcement learning algorithm can overcome the problem that PPO algorithm falls into local optimization in the training process, and improve the diversity of algorithm exploration. Based on the Open AI simulation environment, this paper analyzes the motion law of the unmanned surface vehicle on the water surface, and establishes a three-degree of freedom kinematics and dynamics mathematical model. An improved reinforcement learning algorithm is used to design the motion controller of unmanned surface vehicle, and a compound reward is designed, which effectively improves the learning efficiency of the network. Simulation results shows the effectiveness of the improved Proximal Policy Optimization algorithm in the motion control of unmanned surface vehicle, and verify the superiority of the improved algorithm.

Multi-Label Text Classification (MLTC) is a task to assign documents to one or more class labels. Most of the latest approaches are based on deep learning. However, among the various types of methods, there are still defects in capturing fine-grained information in the text and in modeling the dependencies between labels. In recent years, neural network model-based approaches have made great progress, among which the BERT language model can capture bidirectional contextual information well but cannot capture label-specific semantic information, so we propose a new multi-label text classification framework LAGAT by combining attention mechanism for feature learning. This framework is devoted to mining inter-label dependencies by constructing graphs of label-specific text semantic information and using the graph attention network (GAT) to learn the relative importance between labels by updating the weights on the edges, thus mining the inter-label dependencies. In further experiments, we confirm that the GAT is weaker than the self-attention mechanism in mining text labels by aggregating neighborhood information at moderate label size, and propose an improved model framework LA-Trans, which also utilizes label association relations with text semantic information, focuses on the target region and ignores irrelevant information, and achieves the performance of state of the arts.

The quadrotor UAV is an aircraft with excellent performance such as vertical take-off and landing, hovering in the air, low-speed flight, etc. It is flexible and easy to operate. Therefore, the application of quadrotor UAVs in the military and civilian fields such as aerial photography, express transportation, plant protection, pipeline inspection, etc. has shown unique advantages. The application scenarios are being rapidly developed and have broad prospects. The control system of the quadrotor UAV has the characteristics of multi-variable, high coupling, nonlinear, and underactuated, which brings challenges to the design of the controller. For this reason, this paper established the dynamics model of the "X" quadrotor UAV, designed an adaptive controller based on backstepping, and completed simulation experiments. The simulation results show that in the presence of disturbance, the designed adaptive controller still has good control accuracy, response speed and robustness, which proves the feasibility of adaptive control in the field of quadrotor UAV.

In the actual control process of unmanned aerial vehicle (UAV), the current intelligent control method of the UAV cluster has insufficient binding force on the common control of multiple UAV clusters, which easily leads to the problem that the UAV is out of control and can not continue the task. Therefore, an intelligent control method for UAV clusters based on neighborhood following is proposed. The adaptive following control model of the UAV group is designed, and the mathematical model of the UAV group motion is built. The task scheduling between individuals is optimized to enhance the frequency of information exchange between UAVs. Based on the domain following, the intelligent control law of following target between individuals in the UAV cluster is established. The comparative experimental results show that the method can ensure the close combination of UAV groups, and can reach a stable state in about 60s. The control efficiency is greatly improved compared with the traditional method, which has practical significance for strengthening the intelligent control technology of the UAV groups in the future.

With the coming of intelligent warfare, drone swarm warfare will become an important combat style. At the same time, the drone swarm system will also face some security problems, such as the destruction of centralized management nodes, the data destruction of enemy invaded node, information leakage, data theft and so on. Due to the decentralized, tamperproof, traceable and other characteristics of blockchain technology, it highly meets the requirements of drone swarm warfare application, so it can effectively solve the business pain point of drone swarm. We first introduce the operational process of drone swarm and discussed the pain points in drone swarm warfare application. We then present the solution of drone swarm warfare application based on blockchain technology. The research objective is to provide design ideas for promoting the integration application of blockchain technology and drone swarm warfare.

In this research, A wool microscopic image recognition and real-time tracking system based on LABVIEW is developed. The program block diagram of wool image acquisition, preprocessing, feature value extraction, real-time tracking, quantity and field density was designed. The results show that the image recognition and real-time tracking of wool detection process were successfully realized in this design. It could be screened for specific needs, and the wool count and field density were calculated. In this paper, the real-time tracking of microscopic images and image processing were of great reference significance.

The technology of deep fake face detection in videos is particularly important in the field of internet and finance. In order to solve the significant problem of the spread of misinformation, it is essential that deep fake face detection methods robust manipulation. This paper proposes a Bidirectional-LSTM method based on temporal features for deep fake face detection in videos. Bidirectional-LSTM method is one of deep learning methods, which has positive effect on exploiting the temporal information in videos. Facial expressions and muscle movements play an important role in an individual’s speaking style. The continuity features of temporality suffer from the working process of the fake face in videos so that deep fake detection can be used for the fake face videos. Compared with previous methods such as SVM, BP neural network, the proposed method takes less time to train the model and have a good performance on accuracy. The experimental results show that the accuracy reaches 82.65% on DFDC data set. The proposed method has an excellent effect on the robustness in detecting compressed videos with noise interference.

With the rapid development of intelligent vehicle technology, the intelligent requirements of parking space recognition technology are getting higher and higher. This paper intends to develop an automatic parking system based on the STM32F767 embedded system, which is used for automatic parking in the parking lot. The automatic parking system uses third-party smart forklifts to autonomously patrol lines and identify vehicles, and transport them to designated parking spaces. Through the flexible movement of omnidirectional wheels, a series of auto-matic parking functions such as autonomous line inspection, environmental perception, networking processing, parking space planning, lifting vehicles, and omnidirectional motion are realized. Aiming at the difficulties in automatic parking, such as the kinematics of vertical parking spaces, horizontal parking spaces, and inclined side parking spaces, research is carried out to ensure that the vehicle can accurately move to the designated position without obstacles. The automatic parking system has the characteristics of high recognition rate, good safety, and fast parking speed. Combined with the human-computer interaction interface of the Internet of Things, the parking situation of the parking lot and the occupancy of the parking space are clear at a glance, which greatly improves the operation efficiency of the whole system.

Different path-finding algorithms are chosen in different environments, according to their advantages and disadvantages. This study investigates the characteristics of three of the popular heuristic path-finding algorithms — A*, IDA* and JPS. The investigation focuses on their performance of time efficiency, memory efficiency and performance on whether could give out optimal solution, and was carried out in 2D orthogonal grids. These algorithms are shown to be affected by density of barriers in grids, complexity of routes, map size and whether the map has boundary. This study analyzed the data of performance of three algorithms based on these aspects and concluded the environments that are suitable for each algorithm. In general, A* was found to be more universal when navigating, JPS was less universal and IDA* has the most limited universality.

Cluster UAV autonomous cooperative decision control is used for UAVs with autonomous control ability to achieve a high degree of autonomous cooperation through organizational rules and information interaction, so as to achieve rapid and effective acquisition, processing, transmission and multi platform control of all kinds of information. This paper mainly focuses on the cooperative control of intelligent multi UAV systems. It analyzes dynamic cooperative task allocation, online cooperative path planning with time and space constraints, cooperative control strategy in cluster cooperative formation control and formation reconfiguration. The results are verified by simulation analysis.

In this paper, the development and application have been studied for the simulation technology. Combined with the characteristics of the fly-by-wire flight control system, the different configurations and functions of Hardware-in-the-loop simulation systems have been researched at different development phases of the civil aircraft. The accuracy, completeness, and practicability of the Hardware-in-the-loop simulation system have been verified through the pilot-in-the-loop simulation test. In addition, compared with the iron bird test data, engineering simulator test data and test flight data, the accuracy of Hardware-in-the-loop simulation system and its practical value of engineering have been further improved. The hardware-in-the-loop simulation system has been proved as a useful tool for the design and development of civil aircraft flight control system.

Greenhouse control systems are extensively used in the production of crops. However, temperature and humidity control pose the problems of multivariate coupling, nonlinearity, and high inertia. In this study, a multivariate decoupling fuzzy proportional-integral-derivative (PID) control model based on genetic algorithm (GA) optimization was developed. First, by using the dynamic mathematical model of the greenhouse control process, the coupling between temperature and humidity was reduced by decoupling the greenhouse model by using a multivariate pre-feedback decoupler. Next, GA was used to optimize the quantitative and proportional factors in fuzzy PID controllers to determine the global optimal value. Finally, the designed control model was simulated in MATLAB/Simulink. The output of the GA-optimized temperature controller was 98.87% lower than that of the unoptimized temperature controller. Upon optimization, the humidity controller output, temperature transition time, and humidity transition time were 89.67% 16.67%, and 38.46% lower than when not optimized, indicating increased system stability. Therefore, the proposed model effectively improves the control accuracy of greenhouse temperature and humidity.

A facial expression recognition method with improved residual network is proposed to address the problems of low recognition accuracy and loss of detailed features in the discriminative task due to inappropriate pooling in current facial expression recognition models. First, multi-scale importance pooling with adaptively learned input feature weights is proposed to improve the model's ability to retain and recognize global features. Second, asymmetric convolution is used to augment the convolution kernels to improve the characterization of local features, and the set of asymmetric convolution kernels trained to convergence is fused and equivalently converted to the original network structure to avoid introducing additional parameter optimization in the inference stage. Finally, the expression recognition accuracies of 85.53% and 72.72% were achieved in two publicly available datasets, RAF-DB and FER2013, respectively, and the experimental results demonstrate the effectiveness of the proposed method with certain application prospects.

To improve the response time of the power big data management and control service, the paper builds a more flexible and diversified management service mode. It creates a stable and reliable processing environment, combines with cloud computing technology, and designs a power big data management analysis and service method. The management and control environment is preprocessed, and the hierarchical processing mechanism is constructed in combination with the cloud computing model. Based on this, the cloud computing integrated power big data management and control model is designed, and the cloud service processing is used to realize the management and control. The final test results show that compared with the traditional multi-dimensional power big data management test group and the traditional positioning power big data management test group, the response time of the management and control service are obtained by the cloud computing power. The Big Data Management test group designed in this paper is relatively short, which indicates that it has a better management effect for power big data and has important practical significance.

In this study, by examining the effects of the optimal carbon-nitrogen ratio and other factors on the composting effect, the optimal composting parameters of municipal sludge and kitchen waste were determined, and the carbon emissions were calculated at the same time, It had reference value for carbon neutralization and carbon peak. On the basis of reduction, harmlessness and resource utilization, a coordinated aerobic composting process for municipal sludge, kitchen waste and agricultural and forestry waste was established.