This PDF file contains the front matter associated with SPIE Proceedings Volume 12256, including the Title Page, Copyright information and Table of Contents.

With the rapid expansion of power grid scale, for the mountainous areas with slow economic development, large-scale transmission lines have been exposed to the harsh environment for a long time, causing great damage to the transmission lines. In order to reduce the possibility of safety failure of transmission line, periodic inspection of transmission line is needed. Based on the power line extraction algorithm of LIDAR point cloud in the power line ecological corridor, a new multi split power line fitting method is proposed in this paper. The whole split conductor point cloud is fitted straight line in the XOY plane and catenary in the XOZ plane respectively. The curve equation obtained is used as the center line, and then the coordinates of each point relative to the center line are calculated. Experiments show that the algorithm proposed in this paper has obvious advantages in the extraction accuracy and adaptability of split sub conductors compared with the traditional algorithm. It can provide accurate positioning for the subsequent use of suspension points to simulate power lines to determine whether there are potential safety hazards, which provides technical support for the power grid construction.

OFDM system is very sensitive to synchronization error. Accurate and fast synchronization is the key to the performance of power line carrier communication system based on OFDM. In order to improve the ability of power line carrier communication system to resist harsh channel environment, a local cross-correlation operation method based on compressed binary index is proposed, which greatly reduces the amount of synchronous operation while ensuring the success rate of system communication, the complexity of operation is reduced. MATLAB simulation results show that the synchronization method has strong robustness in a variety of typical noise environments. By optimizing the synchronization sequence, when the communication performance of the system loses about 1dB, the operation efficiency can be improved by more than 50%.

Text semantic matching is an important issue in natural language processing, and it is widely used in question answering systems, dialogue systems, and information retrieval. Text matching models can be divided into representation-based text matching models and interaction-based text matching models. Aiming at the problem that the existing text matching models tend to ignore the global information, a Chinese text semantic matching model oriented to information interaction is proposed. This model uses interactive attention and self-attention to make the text's own structure for information interaction, and at the same time increases the deep semantic interaction of the two texts, and obtains abundant semantic information vectors. Experiments show that the information interaction-oriented semantic matching model proposed in this paper is superior to the traditional model on the verified Chinese text data set.

Sketch or line art coloring is a research field with great market demand. Although the effect of some previous coloring methods is effective, it is easy to have the problem of color mixing in the facial area. In this method, based on our previous work, we use the latest image conversion framework NICE-GAN instead of CycleGAN to make the network framework more compact, and apply the multi-scale discriminator structure to make the conversion effect have more information and higher training efficiency. At the same time, combined with the attention mechanism module in the previous work, the model can learn the key coloring in the facial features area through the feature map, so as to reduce the surrounding color mixing. At the same time, due to the use of unsupervised network, the whole process is fast and automatic. It can be used for quick coloring of user-defined animation avatars. It is very friendly to novice users who can't color. The final experiment also proves that this method has better effect in facial region than other previous line drawing coloring methods, significantly reduces color mixing, and has higher training efficiency.

With the development of computer vision technology, more and more robots are using advanced computer vision techniques for scene environment exploration and understanding. For intelligent robots, recognizing and understanding scenes and the objects in them is a very important task that can help robots perform more complex tasks. In this paper, we develop a scene object recognition and matching algorithm from the perspective of scene understanding, which can recognize objects in the environment quickly and stably, and can match and track objects as the camera moves. Tests on real scenes show the effectiveness and rapidity of this algorithm.

The sound signal can be transmitted over a long distance in the water environment, but there is often interference from other signal sources in the real environment, which will seriously reduce the sensitivity and recognizability of the underwater acoustic signal. At this time, it is necessary to use underwater acoustic signal separation technology to separate mixed underwater acoustic signals. Due to the time sequence of the audio signal, the feature extraction ability of the separation model for the input sound signal largely determines the performance of the model. We propose a C-RNN network model that combines convolutional and recurrent neural network to achieve the improvement of separation performance. The advantages and disadvantages of separation based on time domain and frequency domain are compared, and a hybrid coding module is proposed to achieve a new state-of-the-art for underwater acoustic signal separation.

BIM technology (Building Information Modeling,referred to as BIM) is a digital expression of the physical and functional characteristics of engineering projects,and a shared knowledge resource of relevant information of engineering projects.It is a digital tool applied to the engineering design and construction management,supporting the continuous and real-time application of various project information,which can greatly improve the quality and efficiency of the design,construction and even the whole project,and reduce the cost.With the in-depth development of new digital technology,BIM technology is more and more widely used in construction engineering.3D laser scanning technology,as a new surveying and mapping technology,combined with the application of Robot measuring and setting out technology,can obtain the three-dimensional coordinate data of the target object at a non-contact high speed and accurately,which greatly improves the production efficiency.During the construction of building curtain wall data,this technology can be used to obtain the spatial point cloud of the building and establish a real three-dimensional building model,so as to guide parametric ordering,rapid setting out,positioning and installation.We took the "magic cube" building curtain wall project as an example and introduced the in-depth application of BIM+3D laser scanning technology and Robot measuring and setting out in the construction process of building curtain wall.During the implementation of the project,3D laser scanner was used to scan the main body of the building steel structure.RealWorks software was used to process the point cloud data to obtain the accurate building model[1].On this basis,size information was extracted from the model to replace the traditional measured real quantity,and the BIM model was compared with the point cloud model[2].Conduct curtain wall whole process construction guidance and follow-up operation and maintenance.

In densely built urban cities, deep excavations are often carried out near buildings. Excavation works inevitably induce soil stress changes and soil movements in the ground. Hence, settlement and horizontal displacement may be induced in buildings. In this paper, a three-dimensional numerical analysis was carried out to investigate the effects of excavation on two nearby buildings. An excavation for a metro station in Nanning, Guangxi province of China is adopted as the background of this case study. The minimum distance between excavation and buildings is only 5.5m, which is within one time of the excavation depth. The hardening soil model with small-strain stiffness was adopted to model the behaviour of soil. The excavation procedure in the site was modelled in the numerical analysis. Based on the numerical analysis, settlement and horizontal displacement of the two buildings are presented and analysed.

For the fusion of traditional medical image fusion effect is poor, a pseudo gibbs phenomenon is complex and PCNN parameters Settings and so on, proposed a based on the next sampling shear wave transform (NSST) and particle swarm optimization algorithm (PSO), the standard differential evolution algorithm (DE) combining optimization pulse coupled neural network (PCNN) parameters of medical image fusion method. Source image in NSST domain is decomposed into the same size high frequency sub-bands of k and a low frequency subband, combines PSO and DE using spatial frequency (SF) as the fitness function of the optimization algorithm to improve the PCNN, search for the optimal parameters for fusion of the high frequency subband coefficients, low-frequency subband coefficients of energy weighted average method is adopted to improve the fusion, Finally, NSST inverse transformation is used to obtain the final fusion image. The fusion effect of medical images was evaluated and analyzed according to subjective and objective evaluation indexes. Experimental results show that this algorithm is better than other algorithms in objective evaluation index and has better fusion effect.

To solve the problems existing in the traditional networks, we have improved the existing networks. Traditional networks mainly focus on the details of images and ignore the importance of global features in the segmentation. In addition, the increasing depth of the network leads to a series of problems, such as over-fitting, decreasing convergence, drop-in speed and decreasing accuracy. Moreover, convolving with unified scale in the same layer fails in comprehensively representing the feature information. To address the above problems, a combination of deep network and corresponding shallow network is proposed to extract the image features for segmentation. In order to speed up training, a shortcut connection is employed in cascade with small kernels to facilitate gradient flow. Besides, we add a multi-scale operation to the network for extracting the both short-range and long-range contextual information. Our proposals are validated in the BRATS2013 database, obtaining the excellent performance in the Complete and Core of Dice and Sensitivity (0.89, 0.81, 0.89, 0.78) simultaneously for the Training data set.

In recent years, the collective teaching mode of digital piano has gradually become a research hotspot of piano teaching. Based on the problems encountered in digital piano teaching and actual operation, this paper proposes a kind of digital piano classroom, piano audio signal network transmission plan, realizes the network function of digital piano collective classroom, and provides a feasible method for digital piano The operating platform, in turn, helps to improve the collective sound effect of the digital piano.

The research on the form of traditional villages has always been the key content in the protection planning of ancient villages. Based on the line segment analysis method of spatial syntax, this paper analyses the external representation and internal structure form of Yumu village, which is actively developing scenic spots, and summarizes the specific expression in the village spatial form and street spatial form. The results show that: 1) the residential area and traffic trunk road of Yumu village expand outward, and the residential buildings show the type of discrete group space with large dispersion and small settlement; 2) affected by internal and external traffic, public space, and other factors, the central and eastern regions of the village are highly integrated and concentrated. 3) The internal spatial pattern of the village is scattered, there are few cultural buildings and cultural relics protection units, and the spatial passage rate and selectivity are low. 4) The reasonable date between global and local integration of the village is only 0.35, and the local space can’t better understand the whole space. According to the analysis results, the corresponding optimization strategies are put forward from the perspective of resilience theory, which provides a new idea for the protection and development of local tourism.

Power consumption forecasting is an important part of the macro planning of the industry and energy sector, and accurate forecasting of power load is very important for power grid management and power dispatching. At present, most of the power load forecasting takes the region as the object, but residents and small and medium-sized enterprise users are the basic units of electricity consumption, and their power load forecasting is as important as regional power load forecasting. compared with the regional power load, the electricity load of residents and small and medium-sized enterprises is more uncertain and more difficult to forecast. Therefore, this study combines the adaptive spectral clustering (ASC) method with the support vector quantile regression model (SVQR) to analyze the electricity consumption behavior of smart grid users and predict the residential power load. In this paper, the grid search is used to optimize the parameters of the Gaussian kernel SVQR model (GSVQR) to predict the power load, and compare it with other algorithms. From the two error evaluation index values of MAPE and pinball loss, the prediction effect of the GSVQR model is the best. In order to effectively provide uncertain information of power load, the GSVQR algorithm is used to predict the load of ultra-high energy consumption users and medium energy consumption users at any time in the future. Extensive experimental results show that: compared with other models, the prediction accuracy of the GSVQR model is higher; and the prediction results of the GSVQR model still have high reliability. Therefore, the method used in this paper can solve the problem of uncertainty of load forecasting.

LiDAR (Light Detection and Ranging) is a technology that is widely applied nowadays in distance measurement and environmental detection. By shooting out and receiving the reflected laser beam, the sensor could measure the distance between certain objects and build up a 3-D model of the surrounding environment. With the development of autonomous vehicles and other unmanned devices, the request for higher accuracy in environmental detection has emerged. The LiDAR technology is first reviewed from its working principle, development history, and application in this article. Afterwards, the current status of environmental detection using LiDAR is reviewed from its pros and cons and the up-to-date method of deep learning and neural network. Finally, the future prospect of this technology is provided.

With the development of science and technology, image processing gradually develops towards higher dimensions. High-dimensional data is usually regarded as approximated by the union of multiple low-dimensional data. That is, the high-dimensional subspace is divided into several low-dimensional subspaces, so as to provide better insights for understanding the underlying structure of the high-dimensional subspace. Most of the existing clustering methods solve the problem of Multi-view subspace clustering by constructing an affinity matrix on each view. This paper proposes a Multi-view low-rank sparse subspace clustering based on adaptive dictionary learning(ADLMLRSSC). In the multi-view low rank sparse representation model, an adaptive dictionary learning strategy using orthogonal constraints is introduced. The dictionary learns adaptively from the original data, which makes the model robust to noise. At the same time, the projection matrix and sparse low rank features are obtained by the optimization method. In theory, low rank and sparse constraints can consider both the global and local structure of the matrix.

A high-altitude parabolic identification technology for urban buildings is proposed, which aims at locating the parabolic position accurately and sending out warning signal in real-time to provide reference data for afterwards accountability. Firstly, the video image is preprocessed, and the color video image of the input is converted to the gray-scale video image. Secondly, the moving area is obtained by difference motion using three consecutive frames of images. Thirdly, the background subtraction model is used to obtain the complete moving object, and the complete moving object is obtained by performing OR operation with the former obtained moving object. Finally, the parabolic object is identified and the complete trajectory is obtained after further processing by using median filtering and morphological method. The parabolic coordinates correspond to the starting point is the corresponding floors. The algorithm not only can optimize the process of parabolic identification, but also remove the false identification of birds, leaves and other non-parabolic objects. The algorithm improve the accurate identification of multiple parabolic objects. The experimental results showed that the algorithm can correctly and effectively track the parabolic trajectory of urban buildings at high-altitude. It can locate the floors corresponding to the parabolic coordinates.

Taking a waste incineration power plant in Guangdong as the research object, this paper studies the calculation method of greenhouse gas (GHG) emission reduction of waste incineration power plant. This paper summarizes the situation of waste incineration power generation projects in China. Then the paper introduces the calculation method of emission reduction. In this paper, the green-house gas emission reductions of urban solid waste incineration facilities are calculated by Life Cycle Assessment (LCA), Life Cycle Assessment combined with monitoring method, Intergovernmental Panel on Climate Change (IPCC) recommended method and clean development mechanism (CDM) method. It is found that the calculation results of other methods except Life Cycle Assessment combined with monitoring method have different degrees of GHG emission reduction.

It is of great environmental significance to promote the deep treatment of thermal power flue gas. Therefore, this paper proposes a prediction model based on CEEMDAN-SE-RF-GWO-LSTM for the difficulty of accurately detecting the SO2 concentration at the inlet of the desulfurization system of thermal power plants. First, the process parameters related to the inlet SO2 concentration are determined as the original input variables through mechanism analysis, and a random forest algorithm (RF) is used to evaluate the importance of the variables for feature selection. Secondly, the SO2 concentration data is decomposed by complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN), and the sample entropy (SE) is used to merge and reconstruct the modal components as the final predictor variable. After determining the input variables and predictors, the long-term short-term memory network (LSTM) optimized by the gray wolf algorithm is used to establish a predictive model of the inlet SO2 concentration. The experimental results show that the grey wolf optimizer (GWO) significantly affects optimizing hyperparameters. Decomposed variables can be separately predicted and then combined to improve the prediction accuracy, while reconstruction after decomposition can improve the model training efficiency while improving the prediction accuracy.

With the rapid development of artificial intelligence, its related skills have been widely used in medical treatment. Medical image classification has become an indispensable and increasingly important part of disease diagnosis and treatment to allow accurate and rapid treatment. According to the existing neural network failed to extract local and global features significantly at the same time, this paper uses Gamma transform and the combined model of CNN and Visual Transformer to classify the images of chest x-ray patients. Our model uses convolution operation and self-attention mechanism to enhance representational learning. It adopts a parallel structure to retain local features and global features to the greatest extent. The results showed that our model has a better classification effect than Vision Transformer, which shows its availability and great potential in medical image assisted diagnosis.

The main body of a city is the urban residents. In the process of rapid development of a city, paying attention to urban residents and meeting their demands for higher quality urban life is the key to the healthy development of it. This study adopted the improved pixel method to evaluate residents' image perception of greenways. Through the analysis of the three types of greenway photos, the natural greenway photo scene is dominated by green plants, while the photo scenes of built greenways includes buildings, roads, service facilities and other scenes. In general, the improved pixel method can quantitatively analyze residents’ subjective perceptions in a relatively simple way.

In order to solve the problems of domestic satellite images such as unstable parameter accuracy of rational polynomial coefficients (RPC), difficult matching of control points and large invalid redundancy of aerial triangulation processing, the paper proposes a set of whole process solution suitable for domestic satellite image cluster optimization processing, including automatic unattended control point matching and aerial triangulation optimization, selecting the appropriate Digital Elevation Model (DEM) to assist ortho rectification, etc. The results show that the automation degree of aerial triangulation is improved after optimization, the integrated efficiency is improved by more than 1/3; the 30 meter grid interval Digital Surface Model (DSM) which fuses DSM data can meet the requirements of the ortho rectification accuracy of domestic satellite images better than 1 meter in the provincial level.

In order to solve the problem that indoor Global Positioning System (GPS) receivers cannot locate correctly, we propose a virtual satellite indoor navigation positioning system. The system uses GPS signal simulator to generate the signal of the virtual orbit satellite constellation. A variety of constellation geometries can be designed for any location at any time. The virtual orbit constellation can stably maintain a geometric configuration for 4-5 hours with low Geometric Dilution of Precision (GDOP), where the real constellation has a large GDOP and a broken line error. In indoor positioning, there are broken line errors caused by the fact that satellite, indoor antenna and receiver are not in the same line. Therefore, the design rule of the virtual satellite is to minimize this broken line error. We design a new virtual constellation with Elliptically Inclined Geosynchronous Orbit (EIGSO) satellites to reduce the broken line error. Ephemeris data of the virtual orbit satellite constellation is constructed, an indoor satellite signal simulator with four channels is developed, and a virtual orbit satellite constellation signal is broadcast. The experimental test results show that the proposed system can make the indoor satellite navigation receiver position correctly.

Urban green space is a vital building facility that provides a source of oxygen for the city to maintain the ecological environment in the process of urban construction. Its construction status plays an important role in improving the overall environmental appearance of the city. Based on Landsat 8 images, this paper extracted urban green space information in Nanjing. The images of Nanjing in 2013 and 2017 were processed and analyzed, and the green space information was extracted by using the method of supervised classification for atmospheric correction, image fusion and image cropping of the images. The final result shows that the green space area decreased gradually from 2013 to 2017 due to the rapid development of the city. It can be seen that the development of Nanjing itself has led to the continuous reduction of green space in the original city.

The impact of government regulation of nuclear prices and the reform of transmission and distribution prices requires companies to continuously improve investment efficiency and benefits in order to promote the high-quality development of power grid companies. However, the current investment budget lacks the basis for preparation and adjustment, and it is difficult to realize the coordinated management of investment plans and investment budgets. In order to achieve the management goal, this topic takes the whole process management of the power grid infrastructure project investment as the starting point, and analyzes the law of occurrence of various costs in each stage of the project construction and the characteristics of the time sequence distribution ratio of various costs on the basis of the quantitative management and control model of the investment plan. Based on the above rules, this project constructs a research on the collaborative preparation and adjustment model of investment plans and budgets, assists in supporting the collaborative arrangements of investment plans and investment budgets of companies in various regions and cities, and promotes collaborative management of various departments.

With the continuous development of society, China pays more and more attention to the protection of ecological environment. Taking an area in the Pearl River Delta as an example, this paper researches soil and water conservation control methods based on Geographic Information System (GIS) and computer application technology, designs and develops a soil and water conservation control system. The system realizes the control of the whole process and cycle of soil and water conservation prevention and control by means of information technology, which is conducive to the relevant government departments to understand the situation of soil and water erosion of the projects under construction in the area in time and make scientific decisions accordingly, effectively promoting the process of information technology of soil and water conservation prevention and control in some areas of the Pearl River Delta, reducing the amount of soil and water erosion in some areas of the Pearl River Delta to a large extent, and providing experience and reference for soil and water conservation prevention and control in other regions of China. It also provides experience for other regions in China to learn from.

Based on the analysis of the construction principle of the rolling ball model, the paper has brought forward a TIN-DEM automatic generalization algorithm about the continuous scale expression of terrain shape. By taking the radius of the rolling ball as the analysis scale, this paper analyzes the deep-seated topographic information contained in the process of topographic scale change. If the sampling point contacts the rolling ball under a certain radius, the sampling point can be regarded as a flat terrain point under the radius of the rolling ball. Otherwise, it is a non flat terrain point. The contact between the sampling point and the rolling ball can be judged by the critical rolling ball radius. Then the flat terrain points under the large-scale radius can be deleted first. All sampling point of TIN-DEM are sorted according to the order from large to small of the critical Rolling Ball radius of the sampling point determined as flat terrain points. The sample point is processed in this order to ensure the stability of the processing results. The experimental results show that the algorithm can realize TIN-DEM automatic generalization, the generalization result retains the main features of the overall landform to the greatest extent, and the processing result is stable and reliable.

In July 2021, the NDRC (National Development and Reform Committee) launched the Notice to Further Better Off the ToU (Time of Use) Mechanism [1]. The notice is aimed to create a better environment for the development of the renewables and the construction of the renewables-based new power system by deepening the reform of the power market. This paper stands in the point of view of the power plants to address how does the power supply side, especially the coal fired power plants, tackle the challenges coming out in the context of the widening tariff difference between the peak and valley loads. This paper analyzed the operational tactics of the energy storage in the power supply side. With the model analysis and the simulation of the energy storage tactics, it tried to usher the power supply side to grasp the opportunity to adopt new energy storage facilities and apply the storage operational tactics, to enhance the reliability of the system, and meanwhile increase the proportion of the renewables into the power grid, ultimately help our power industry drive a successful energy transition and form a solid new power system in near future.

In view of the data transmission errors that exist when the intelligent distribution terminal AC acquisition board and the CPU board use SPI communication between the boards, and the problems that cannot be completely solved by using only CRC check and configuration register locking methods .The paper focus on the research and analyse of uncertainty of serial communication and the problem of data transmission between boards using SPI communication during the electrical fast transient pulse group interference test, which leads to the abnormal data transmission, Three improvement schemes are proposed: 1) add a booster to the SPI bus; 2) communicate using SPI via USB-SPI conversion chip; 3) communicate using RS-485 via connecting SPI to the TIM4 microprocessor, Comparative test results show that data transmission between boards changes from SPI to RS-485,which can effectively solve the problem of data transmission error between boards.

In the face of increasingly severe air target threats, laser weapons have the advantages of fast attack speed, flexible steering, precision strikes and immunity from electromagnetic interference. They are new concept weapons that are developed by the navies of various countries. At present, laser weapons strike targets, relying on manual point selection to directional strike damage points, which takes a long time, and it is easy to miss the time of damage. In order to solve the timeconsuming problem of manual point selection, this paper proposes an automatic extraction method for damage points of laser weapons. This method divides the missile target by K_means color clustering, and then fits the center line of the missile according to the pixel position index of the missile target area, combined with the least square method, and finally extracts the damage point according to the proportion of each part in the missile. This method can accurately provide damage points in real time and improve the combat effectiveness of laser weapons

With the development of science and technology, various electronic devices of the Internet of Things have gradually entered thousands of households. However, due to the different information transmission standards adopted by various manufacturers, the access methods of the Internet of Things are also varied, such as infrared, WiFi, Bluetooth, ZigBee, VB-IoT, LoRa, LTE and so on, which makes the market fall into chaos. In view of the above situation, In this paper, a communication technology of Internet of Things using PLC encryption transmission is proposed, By setting up the narrow-band exit of the Internet of Things with the Distribution station area as the unit, and using the existing resource "power line" as the communication carrier, the network signal covers all the power terminals, so that the IoT equipment can be connected to the network when powered on, thus realizing the unified monitoring and management of various items on the IoT cloud platform and reducing the construction cost of the IoT network.

The method of image recognition and analysis has become an important means of two-phase flow research. Through the bubble evolution pictures obtained in the two-phase flow experimental device, the generation dynamic threshold image processing method is adopted to reduce the impact of image environment changes, and the appropriate threshold is automatically obtained for image segmentation. The gray level and spatial position information of point pixels are introduced to avoid large distortion of specific images, automatically identify the relevant parameters of bubble flow, deduce the bubble size and group distribution from the measured data, and analyze the bubble characteristics. Iterative dynamic threshold image processing method is complex, but it has strong anti-noise ability. It has a good recognition effect for images that are not easy to be segmented by global threshold.

The development of the Brain-Computer Interface (BCI) has provided humans with a novel channel to interact with the world directly, and the join of Augmented Reality (AR) makes it more natural and practical in an immersive experience. This paper aims to give a review of the research status of the BCI system combined with the AR environment. The fundamental introduction of the two techniques is illustrated at first. And the progress and application made in recent years are stated in three main dimensions about the interface, application field, and related testing work. Finally, it discusses the future challenge of the actual operation in a common environment covering various real-life scenarios, and summarizes its broad prospects.

In today's advanced technology, students’ hands-on ability has become a focus of every school, such as engineering, agronomy, medicine and other disciplines. The development of the system introduced in this article has greatly increased the joy of learning for college students in related majors, while also focusing on training. In order to improve the practical ability of modern college students and serve more colleges and teachers and students, the development of the Internet of Things system is essential. In the past year, according to the data issued by some colleges and universities every year, there are more excellent graduates and excellent graduation designs and papers in schools that have introduced the Internet of Things development system than when the system was not used before. The overall average has increased by %8, and high-quality employment has increased by %10. With the development and progress of the Internet of Things development system, more and more college teachers and students’ praise shows that the development and design of the Internet of Things development system is very important. This article uses STM32F429 as the core of the Internet of Things development system[1], and measured and tested in a real environment, and repeated the boot test, after comparison and analysis, under the same function, compared with STM32F401 and other series of single-chip microcomputers as the core Embedded system, indoor and outdoor detection of temperature and humidity, and toxic gas detection, vibration test, light sensitivity test, control of stepping motor, this kind of sensor[4] and equipment have higher, more accurate and reliable measurement. As a result, better operability, more stable system, and energy efficiency have been greatly improved.

ARP attacks are a major security risk in computer local area networks. This article starts with the working principle of the ARP protocol, analyzes and studies the principle of ARP spoofing attacks, uses virtual machine technology to build a virtual experimental platform on the local area network, and simulates and analyzes the process of ARP spoofing attacks. On this basis, this article proposes corresponding safety precautions.

Temperature is a vital quantity to measure the climate change. In this paper, near-surface air temperature projection of western north Pacific is analysed to show how the temperature change under different scenarios. It manifests that the average temperature is increasing and more significant under high radiative forcing level, especially in the high latitude region and over land, and temperature increment is greater in winter.

The warm phase of El Nino-Southern Oscillation(ENSO) called El Nino, originates in the Pacific Ocean. Many researchers are exploring the relationship between El Nino and precipitation, and one study shows that ENSO affects precipitation variability on a global scale. But these effects may vary in specific regions. The southeastern United States(SEUS) is far away from the Pacific Ocean where El Nino originated, and the terrain is high in the west and low in the east. Therefore, this paper takes the SEUS as an example to find whether El Nino will affect the local precipitation. All data, like the index Nino3.4 which quantifies El Nino and other parameters, are from the NOAA website. After dealing with these data through R language and Excel, the data can be visualized with Tableau's own models. Since correlation analysis only focuses on a single factor, multinomial quadratic analysis in multiple linear regression analysis is used to describe the relationship between parameters. The results show that only the two parameters of wind speed and pressure fit best with Nino3.4, and the correlation between the two parameters reaches an extremely significant level(P value < 0.0001), indicating that both wind speed and pressure influence the change of El Nino. In addition, the P value of El Nino and precipitation are greater than 0.05 and the dispersion degree is high, showing that El Nino has no effect on precipitation in a single dimension. At the same time, since wind speed and pressure have an impact on precipitation, it is reasonable to infer that El Nino is indirectly related to precipitation under multidimensional conditions, but it is not within the scope of this paper.

In order to achieve the purpose of removing indoor pollutants to ensure that children and adults can perform physical and mental activities safely.The TVOC,temperature and wind speed data were monitored throughout the year in an educational institution in Shenyang,and physical and mathematical models were established to carry out transient and steady-state simulations,and the concentration of TVOC and other data were compared and analyzed at the six measurement positions for each time period,and the TVOC flow characteristics and influencing factors were obtained according to the mathematical model simulations.The highest concentration of TVOC was found in December,with an average value of 0.85mg/m³,which is 1.7 times of the standard.Because the density of TVOC is larger than that of air.The 0.7m plane at the lowest height in the simulated field was the most polluted,and the average temperature of 293K is low in winter.Except for the wind speed at the outlet,which was more than 2m/s,but the wind speed in the rest of the area was not good.By changing the ventilation mode to open doors and windows for natural ventilation,changing the air velocity at the air supply outlet to 1.5 times of the actual measurement, and setting the number of air changes to ACH=1.5,the simulation results of these schemes were found to be effective in removing pollutants and improving indoor air quality.

In the traditional multimedia teaching system, the use of the central controller of hardware is easy to cause the inconsistency of front and back-end and intermediate transmission channels and the difficulty of system maintenance and upgrading. Based on this, the design method of interactive multimedia teaching system based on collaborative filtering is proposed. The hardware equipment and system support environment of multimedia teaching system are designed by using the concept and technology of collaborative filtering. Use the software central controller to replace the traditional hardware controller, improve the system software function, and realize the virtualization and personalization of remote control and classroom desktop operation. In operation and maintenance, it can manage all media equipment transparently, and realize fast and early warning system maintenance and barrier free upgrade. Finally, experiments show that the interactive multimedia teaching department based on collaborative filtering has high practicability and fully meets the research requirements.

In the past decade, the fractional differential equation (FDE) model has been widely used in the fields of materials science[1], electrical engineering[2], control theory[3], signal processing[4], chaos[5], etc. As a result, the research of FDE solution has become a focus of many current studies. This study devotes itself to the solution of a fractional-order partially dissipative lattice system. The paper proves firstly that the operator T(Sr) is uniformly bounded, and then that it is equally continuous before finally proves, by applying the Schauder fixed point theorem, the existence and uniqueness of the solution of the fractional partial dissipative system.

The goal of wireless multimedia sensor networks is to transmit data with the desired visual quality. To maximize network lifetime, a compromise must be found between network constraints and desired visual quality. In order to solve this problem, we propose an energy efficient multipath routing, which can predict the essential number of paths and bifurcate based on opportunistic routing according to the dependability. The proposed scheme can determine the best path and provide advisable QoS for various real-time intensive media. The embedding criterion of each objective function is adapted to determine the path from node to receiver. Simulation results show that compared with the existing routing schemes, the proposed method increases the packet reception rate, reduce the energy consumption and average end-to-end delay of sensor nodes.

In order to solve the problem of the loss of orders for group enterprises, rationally develop the functions of the customer's all-round management system to improve the level of business operations. This paper studies the operating mode and practical methods of the omni-directional management system for group enterprise customers. The customer group is subdivided through the clustering algorithm in data mining technology. And established a combination model of BP neural network and support vector machine to predict the loss of orders. Based on the analysis results, formulate customer management system operation strategies. The analysis of the practical results shows that after applying the proposed operating model, the average customer conversion rate of the enterprise is 20.44%, and the average order loss rate is 9.24%. The significant increase in data indicates that the operating model of the customer comprehensive management system has practical significance.

In order to solve the problem that the ORB-SLAM2 algorithm will cause positioning failure in a scene with a relatively single texture, the line features in the environment are extracted and matched. The point and line feature constraints are used to ensure the positioning accuracy. Firstly, the acquired image was processed by gradient density filtering. Then the feature points in the image were extracted by the ORB algorithm, the linear features in the image were extracted by the LSD algorithm and the linear features in the image were described by the LBD algorithm. In the above process, a certain strategy was used to select the keyframes. Next the point and line reprojection error term was constructed and the weighting strategy was used to weight the reprojection error. Finally, the nonlinear optimization method was used to obtain accurate pose information. The experimental results on the EuRoC dataset prove that the accuracy of the positioning algorithm fusing point and line features is higher than that of the positioning algorithm using only point features.

This paper proposed a Drone Arrangement Model. Firstly, this paper established a drone system consisting of EOCs, frontline personnel, SSA drones and Radio Repeater drones. We first determined the personnel distribution according to the heat map we made. According to the distribution of personnel, we found the location of EOCs using the k-means algorithm, ensuring that EOCs can monitor each group of front-line personnel. For the sake of building communication between personnel and EOCs, we took different radio spreadability into consideration and obtained the distribution of Radio Repeater drones. Since the drones have their maximum flight time and recharge time, we found the best arrangement of SSA drones that can minimize the cost. Finally, due to the complexity of Australia’s terrain, fire scales vary from terrain to terrain, we optimized the locations of hovering VHF/UHF radio-repeater drones and SSA drones. Therefore, we changed the flight range from a two-dimensional flat circle to a three-dimensional sphere. To assess the stability of our model, we analyzed the suitability of our model over the next decade in three ways. The sensitivity analysis shows the strong robustness of our model.

Firstly, we used multi-source data such as urban planning data, meteorological monitoring data, POI data and real estate website data to establish an evaluation index system from four dimensions of housing comfort, living convenience, external safety and community health, combined with existing evaluation indexes and methods of human settlements environment quality. Secondly, this paper took the main urban districts of Xuzhou as an example to evaluate the quality of human settlement environment and identifies its spatial distribution law by using GIS spatial method. Finally, the corresponding planning countermeasures are put forward based on the analysis results.

As a representative of a new round of technological revolution, the integration of a new generation of information technology into the education industry has promoted the continuous reform and development of the education industry. As a special discipline of art education, how dance education should adapt to the times and keep up with the new wave of "AI+education" is extremely urgent. Therefore, this article will explore how to combine artificial intelligence and other modern technologies and concepts to creatively integrate into the analysis of dance technology risks, design a set of platform models that are convenient for teachers and students to conduct interactive teaching and improve students’ learning efficiency, and propose dance technology risks.

This study focuses on the ordering and transporting raw materials in enterprise supply chain management. First of all, this paper constructs the supply capacity evaluation system of three-level and seven-category enterprises, establishes an enterprise importance evaluation model based on principal component analysis, and finally obtains the ranking of supplier importance. In addition, this paper establishes the enterprise ordering model under the minimum cost, the enterprise ordering model with the minimum transportation and warehousing cost and the enterprise ordering model with capacity maximization. Around the transportation process, this paper establishes a minimum loss transfer model based on 0-1 integer linear programming. In order to minimize the ordering cost, minimize the transportation and warehousing cost, and maximize the production capacity, this paper puts forward the order plan of the enterprise in the next 24 weeks based on the convex optimization principle and Python programming algorithm. Finally, considering the economy in the transportation process comprehensively, this paper establishes the minimum loss transfer model, ensures the minimum loss in the transportation process as far as possible, and establishes the minimum loss transfer model based on 0-1 integer linear programming. Through the Python programming algorithm, the transfer plan of the enterprise in the next 24 weeks is given.

The research frontiers, hotspots and development trends related to satellites are important data for the study of satellite intelligence. In this paper, the scientific knowledge graph is applied to the field of satellite intelligence analysis. Using the satellite research data published on the network, Citespace is used to draw the cooperative knowledge graph, keyword co-occurrence graph and keyword clustering graph of satellite research. The research results show that the research heat of satellites in various countries is increasing, and the scientific research cooperation among institutions is closer. The research frontiers focus on humanitarian applications, communication secrecy, broadband Ka-band terminal design and anti-satellite warfare and so on.

In recent years, with the introduction of "intelligent construction", a higher vision of the quality of large projects has emerged. In the "intelligent construction" system, the core elements of construction are "comprehensive sensing, intelligent decision making and automatic control", which relies on a large number of monitoring instruments embedded in the concrete as sensing tools, and through the embedded system to analyse the sensing data and make decisions. In this paper, we start with the analysis of the monitoring data inside the extra-high arch dam, extract and integrate the data through Python, and then use the visual analysis tool QH3D to make the monitoring data inside the extra-high arch dam better serve the project and promote the quality of the project.

Based on big data and the introduction of mediation model, this paper selects 29 provinces (cities) in China from 2001 to 2019 panel data for empirical analysis. On this basis, the industrial heterogeneity test and regional heterogeneity test.

Based on the collection and research of big data, this paper analyzes the characteristics of aging assets, analyzes the operating conditions of aging assets from different dimensions, determines optimized investment strategies, and formulates technological transformation and investment plans, which can further improve the accuracy of investment plans. Reasonably allocate operating costs, ensure investment capacity, and maximize the operating benefits of power grid enterprises.

In order to have a better understanding of the indoor thermal environment in the enclosed coal yard, year-round real-time onsite testing of the indoor temperature and humidity were carried out during the period of 2019.09.06 to date 2020.09.29. Based on big data monitoring, the distribution characteristics of monthly, daily, and hourly thermal parameters are analyzed and summarized to explore the actual ventilation effects of the enclosed coal yard. Suggestions to improve indoor thermal conditions are proposed.

Supply chain finance of commercial banks provides an effective way for the financing of small and medium-sized enterprises, and solves the problem of enterprise financing to a certain extent. With the rapid development of supply chain financial business, the functional design and structure of traditional business model have been difficult to meet the needs of modern commercial banks. There is an urgent need for emerging technologies to reinvigorate the supply chain financial business of commercial banks. Blockchain technology can effectively reshape the supply chain financial model and solve the development dilemma of commercial banks. In the future, it is necessary to continue to optimize the development environment of commercial banks, accelerate the improvement of the technical and legal framework, build the blockchain technology application architecture system, innovate the supply chain financial model, and promote the sustainable development of commercial banks' supply chain financial business.

Under these circumstances of big data age, environmental design specialty of vocational undergraduate can make full use of the advantages of big data to realize the sharing of teaching resources, as well as provide students with more opportunities for autonomous learning and practical application. Combined with the characteristics of big data, this paper starts with analyzing the background and current situation of environmental design specialty of vocational undergraduate, recognizes the talent training objectives for environmental design specialty of vocational undergraduate under the background of big data. It puts forward the talent training strategies of environmental design specialty of vocational education at undergraduate level, as well as the relevant suggestions to improve the teaching level and teaching quality of environmental design major of vocational undergraduate from four aspects: professional subject construction, evaluation system, digital teaching environment construction and teaching objective system construction, hoping to provide reference for environmental art talents of vocational undergraduate.

This paper develops an intelligent real-time monitory system for household waste, which comprises a house waste collection and delivery device, a dustbin load-carrying device, a garbage-dropping weighing sensor, and an information collection and supervision platform. Among them, the dustbin load-carrying device includes the dustbin load platform and the dustbin load rod. The garbage weighing device includes electronic sensor, intelligent scale device and intelligent electronic chip device. The system separates the dustbin from the ground by the load-carrying platform and the loadcarrying lever, and the quality change information of the dustbin is fed back into the smart chip through the intelligent electronic scale and sensor, which transmits the information to the online supervision platform. The study can monitor and count the generation of domestic waste in various types of communities in China in real time, realize the real-time monitoring and efficient dispatch of household waste production tasks, optimize the data information of household waste production, provide a scientific and reliable basis for the statistics of domestic waste production and the identification of information on the behaviours of residents, effectively improve the efficiency of household waste collection and transportation, and reduce the cost of household waste collection and transportation.

Hydrological forecast technology plays an important role in water resources management for agricultural irrigation, water supply and flood control in urban and rural areas. However, how to improve the applicability of runoff forecast models has always been a major difficulty in water resources research. As the largest water system in North China, the Haihe River Basin has experienced severe water shortages in recent years, which has put tremendous pressure on water supply in northern China. Based on monthly runoff observed at 15 hydrological stations from 2001 to 2018, this paper takes the Haihe River Basin as the study area and applies the simple ANN (artificial neural network) model, the GAANN coupling model, and the SWAT model to forecasting monthly runoff respectively. Four evaluation criteria (R2, NSE, RMSE, MAE) for prediction performance of models are used with a view to discussing the applicability of different models to hydrological forecast of rivers in resource-deficient areas of North China.

With the rapid development of cities, the construction of social public service facilities based on public toilets is becoming more and more important. This paper took the urban public toilets in Huangshan City as the research object, through GIS spatial data analysis, field research and other methods, carried out investigation and analysis of the current situation of public toilets in Huangshan City. Feasibility development suggestions were put forward in view of the problems in its spatial layout, service radius, external architectural design, internal facilities and management system.

This paper proposes an automatic data extraction algorithm for web pages based on noise reduction and visualization blocks' construction. In this algorithm, we first build an MD5 trigeminal tree of the web page's source files using a Message-Digest Algorithm. And then we arrange some noise reduction. Finally, we construct visualization blocks by an optimized cluster algorithm named BIRCH _A-CF which could create an area clustering feature forest and complete the clustering by dynamically changing the circle's radius according to the correction factor. We perform experiments on eight different datasets to compare our method with eight baseline methods. The experimental results show that our approach outperforms current methods by providing more accuracy and robustness, it also accelerates noise reduction and reduces the number of nodes effectively.

Explainable recommendation systems, which can produce high-accuracy recommendations and help users make quick decisions, have become a hotspot in research field. Most of existing research algorithms committed to improving the accuracy of recommendation, but ignored the interpretability of recommendation. We propose a semantic recommendation algorithm through Reinforcement Learning and weighted meta-paths which analyzes and selects the meta-path, and uses Reinforcement Learning network to train the weight of meta-path. This method can effectively solve data sparsity, improve the accuracy of recommendation, and increase interpretability. Compared with the baseline method, indicators of the recommended methods for integrating Reinforcement Learning with Heterogeneous Information Networks have been greatly improved, as verified and compared on a real MovieLens 1M dataset. Experimental results prove that the algorithm can effectively improve the accuracy and interpretability of the recommendation.

This paper introduced a method to generate handwritten digit images by using the Generative Adversarial Networks model. Alternating loss and unstable accuracy were persistent problems during the training process. Human handwritten digit image generation plays a crucial role in criminal investigation and machine learning. There are many excellent and effective ordinary deep networks in machine learning programs for recognizing different pieces of images. However, only a few models can be used to construct new patterns. The experiment used Generative Adversarial Networks as the core algorithm to generate data that look like human-written digits. A Convolutional Neural Networks model is used as Discriminator to evaluate whether an image is genuine or created. The inverse convolutional layers in the code turn input into a two-dimensional pixel values picture. The MNIST provides as much as 60,000 data samples for training the program, where thousands of handwritten single digits, between 0 and 9, are displayed by 28*28 pixels images in greyscale. The pictures created by the algorithm initially looked blurry. But after thousands of training steps, the statistics appeared much more realistic. With more iterations carried out, the output images became more and more distinguishable.

After deep learning algorithms were proposed, artificial intelligence technology applications have achieved breakthrough development. Currently, the explosive growth of data provides sufficient nutrients for artificial intelligence, and deep learning algorithms have achieved breakthroughs in speech and visual recognition, making it possible for artificial intelligence industries to land and commercialize. With the continuous development of big data, deep learning and cloud computing, we can obtain more and richer data, develop more efficient algorithms, and have more powerful computing power, laying the foundation for another artificial intelligence research boom. However, with the application of AI-related technologies in industries such as information, social governance, and transportation, the problems and challenges of algorithmic collusion and algorithmic discrimination have gradually emerged. The operating principles of algorithms differ from the risk of algorithmic collusion that may result, and they also pose different degrees of regulatory challenges for antitrust enforcement. By understanding the data-driven competitive model of the market under the influence of algorithms, the efficient information interaction mechanism and the new features embodied in the evolution of machine-driven competition can prevent the breeding of technological monopolies. The impact of algorithms on the collusion problem has two dimensions: the first dimension is to change the environment of collusion; the second dimension is to be applied directly as a tool in the collusion process. This article attempts to analyze the challenges brought to modern market competition by exploring the state of artificial intelligence technology that causes data monopoly. Thus, In order to better regulate the evaluation and regulation of artificial intelligence algorithm collusion, this article proposes related solutions based on the Chinese perspective, including: 1) broaden the extension of the competitive relationship and identify the subject of monopolistic behavior in accordance with the idea of cooperative behavior; 2) increase platform factors to identify market dominance; 3) use case analysis as the main method to define relevant markets.

Multi-hop machine reading comprehension (MRC) across multiple documents poses new challenges over single-hop MRC, because it requires reasoning several times to answer the given questions and being able to show the reasoning path to support its answer as explanation at the same time. In this paper, we propose a new model based on reinforcement learning for multi-hop MRC. Our model mainly consists of two parts: (i) a novel agent that decomposes the multi-hop question into several sub-questions implicitly, (ii) an incorporator incorporating external knowledge to enhance the answering ability of the sub-question answerer. Experimental results show that our model performs better compared to the path-based models and has higher interpretability compared to the graph-based models. In the last section, we discuss the development prospects of our model in the future.

In order to solve the problem of product defects in the mask production process. In this paper, an algorithm for target detection of mask surface defects based on YOLOv5 is proposed. The industrial-like camera and industrial-like light source are used to collect the data set, and the data set is filtered and manually labeled. The extracted features of mask defects, broken ear straps, and solder joints are trained and tested using the YOLOv5 algorithm. The experiments show that the YOLOv5 algorithm can effectively identify various defects on the mask surface, improve mask production quality and reduce mask production cost.

Traditional mining methods are easily influenced by typical factors such as nature, so the precision of data mining is not ideal. This paper presents a data mining method for energy consumption behavior of integrated energy system considering typical factors. The K-means algorithm is used to correct the damaged data and to mine and marginalize the partially disturbed data in the data. Therefore, the standardized processing method is used to quantify the data. After filtering, the characteristics of user's electricity consumption behavior can be accurately extracted. Experimental results show that the proposed method can effectively extract the accurate electricity consumption data, and the accuracy is higher than the traditional method.

In order to further improve the emergency rescue ability and emergency response speed for emergencies, and minimize the losses and impacts caused by various accidents. Starting from the emergency disposal process of emergencies, combined with the actual situation of emergency command information construction in Dadong District of Shenyang, this paper analyzes and designs the overall architecture and software platform of digital emergency command and dispatching system for emergencies, and realizes the dynamic visualization of command. At the same time, this paper proposes to use analytic hierarchy process to build a scientific plan task evaluation system to provide scientific decision support for emergency command and rescue.

On the basis of traditional hidden danger management, the system adopts WebGIS and Baidu API organic integration technology to achieve accurate management of hidden danger. Based on traffic information analysis of huge amounts of data in the cloud resource platform sorting, find out the relevance, and to analyze hidden danger and accident related factors, explore the development trend, and hazard warning there will be a valuable data pushed to Scott, Baidu navigation service, give full play to the function of hazard warning role, play a dual role of the prevention of accidents and service people. The scientific leapfrog development of "Internet + hidden trouble investigation and rectification" has been realized. The development and application of the system, on the one hand, effectively solve the hidden perils and regulation between technical barriers, by changing the traditional hidden perils in road traffic safety management mode, greatly convenient screening governance concerns, the grassroots police broke the limitation of the public security network "confidential", implements the linkage effect of work, should outside threats. On the other hand, the investigation and management of hidden dangers, social cooperation and joint treatment, early warning and emergency rescue at accident hotspots should be organically combined to highlight the social benefits. At present, the system has realized the regular work about hidden danger information management, data sharing, joint review and co-governance.

Civil-military integration is the in-depth interaction between national defense and military construction and the national economic and social development. Strengthening the construction of civil-military integrated management information system under the background of big data is an important starting point for promoting the construction of national defense and military informatization. This paper introduces related concepts of military big data and management information system, analyzes the challenges that big data faces in the military field, and puts forward countermeasures and suggestions to improve the construction quality of civil-military integrated management information system from three aspects: standardizing top-level design, strengthening framework structure, and expanding application fields.

Human neuroimaging studies mostly combine data from many subjects to infer general patterns of brain activity shared between people. However, certain activity characteristics of the human brain have a high degree of inter-individual variability and intra-individual (cross-state) stability, which can be used as an individual's identification index (i.e. neural fingerprint). Extracting brain activity data through fMRI technology can analyse the neural fingerprint of each person, and realize the transition from group-level research to single-body research, so as to accurately identify a single subject from a large group. In this study, functional magnetic resonance imaging (fMRI) technology was used to record the brain activity data of two narrators and multiple subjects when they were telling/listening to stories. Researching and extracting brain activity data can analyse each person’s "neural fingerprints", which will ultimately be used in the study of brain pathology, and even improve diseases such as Alzheimer's disease and autism. Using independent component analysis (ICA), the 50 independent components in the extracted brain fMRI data are used as the nodes of the network, and the correlation coefficients between the time series of the independent components are used as the connecting edges to construct the functional connection network. In this paper, the functional connection matrix is used as a kind of "neural fingerprint" to identify and match individuals. The functional connection matrix obtained by using the data of the unknown subject's state is used as the target matrix, and the functional connection matrix obtained by using the data of the known subject's state as the database matrix. Compare the functional connection matrix of the target set with each functional connection matrix in the database to find the most similar matrix. Similarity is defined as the Pearson correlation coefficient between the target matrix and the vector obtained by vectorising the upper triangular elements of the data set matrix. The correlation coefficient corresponding to the identified identity is the maximum value among all correlation coefficients, and the ID that the data in the target set should match can be obtained from the maximum value.

For the purpose of exploring the situation of ship collision accidents, study and analyze the factors affecting the situation of ship collision accidents, establish the ship accident situation evaluation model based on principal component analysis and the ship accident situation prediction model based on SVM support vector machine BP neural network, comprehensively judge the severity and development trend of ship collision accidents, and based on the situation model, Combined with the collected accident data, the corresponding emergency rescue decision-making schemes are put forward for different ship collision accident situations.

Logistics distribution is an important part of chain operation, reducing the cost of this part can improve the competitiveness of enterprises. Considering the time window limit, the VRP model of terminal distribution is established to minimize the logistics distribution cost of Y chain stores. The genetic algorithm is used to solve the problem in MATLAB. Compared with the initial delivery, the optimized delivery mileage can be reduced by 1.3 km, the vehicle loading rate of the first line after optimization can reach 97.5%, and the optimized distribution cost can save 6.5 yuan per delivery.

During the dredging process, suspended sediment will increase rapidly in the sea area and affect the surrounding marine ecosystem. Therefore, studying the transport and diffusion of suspended sediment in dredging project has important reference significance for marine ecological environment protection and ecological impact assessment. Based on MIKE hydrodynamic and sediment transport module, the suspended sediment diffusion during the construction of Shuangyu Island channel dredging project in Zhangzhou is simulated. The results show that the diffusion range and morphology of suspended sediment are mainly controlled by tidal current, and the total envelope area of suspended sediment concentration over 10 mg/L is about 4.38 km² during the whole construction period. When antifouling curtain is added to the dredger, the total envelope area of suspended sediment concentration over 10 mg/L is about 3.56 km². The diffusion of suspended sediment (>10mg/L) will affect the nearby recreational areas and aquaculture, and have a certain impact on the Marine environment. The results can provide scientific basis for quantitative calculation of biomass loss and ecological compensation caused by suspended sediment.

In order to effectively improve the IT service quality of IT service management center, on the basis of considering load balancing, an IT service quality perception algorithm of IT service management center based on principal component analysis is designed. Based on the analysis of IT service process of IT service management center, the system software module is designed, including information management module, external perception evaluation module, internal perception evaluation module, comprehensive perception evaluation module and perception evaluation result analysis module; In the comprehensive perception evaluation module, the IT service quality perception evaluation system is constructed, the IT service quality perception evaluation indicators are obtained, the IT service quality perception value and the IT user expectation value of the IT system are determined, the principal component analysis method is used to calculate the difference between the IT user expectation value and the actual perception value, and the IT service quality perception evaluation model is constructed, Complete the design of reliability awareness algorithm of IT service management center considering load balancing. The simulation results show that the difference between the evaluation results obtained by the system and the actual survey results is small, and the perceived evaluation effect of IT service quality is good.

Cardano is an open-source and decentralized public blockchain platform, with consensus achieved using proof of stake. It can facilitate peer-to-peer transactions with its internal cryptocurrency, ADA, with no third-party involvement. In recent years, machine learning has been proliferating and has made many theoretical breakthroughs that find its application in many fields. The study of the machine learning approach in price prediction in Bitcoin and Ethereum has gained much attention, while relatively little research focuses on ADA forecasting. The experiment objective is to investigate the prediction of ADA's short period future prices dealing with real-world data. A comparative study of the results produced by different machine learning models, data visualizations, and statistical approaches. The experiment indicates that Gradient Boosting is the best-suited algorithm that can be selected to predict future ADA prices for short-term trading strategies.

The appropriate acquisition and processing of water quality data were crucial for water resource management. In this study, multivariate statistical analyses were performed to the assessment of the spatial-temporal variation on water quality in Jingyan section of Mangxi River basin (China). The 3 main assessment indicators of water pollutants, including COD_Mn, NH₃-N and TP were analyzed. Water quality data was collected at 18 sampling sites from different monitoring sections monthly over a two-year period. Based on the similarity of water quality characteristics, 18 sampling sites were divided into 4 groups by cluster analysis (CA). The results showed that the highest levels of pollution were identified in cluster 2, which was consistent with the intensive areas for non-point source emission. In addition, the main reason for the significant difference in the water quality characteristics was the environment of different spatial location. The results illustrate the usefulness of multivariate statistical techniques for analyzing and interpreting complex data sets, identifying pollution sources and understanding variations in water quality.

Making full use of bus dedicated lanes can not only ensure the priority of public transportation and enhance the attractiveness of ground public transportation, but also make full use of road resources to achieve the purpose of intensive use of resources. This paper establishes a comprehensive evaluation method system of bus dedicated lanes based on multi-source big data. Using the method, from the angles of physical network of bus dedicated lanes, the degree of matching between bus dedicated lanes and bus passenger flow, and the utilization of bus dedicated lanes, the paper evaluates the use of bus dedicated lanes of Beijing’s main bus passenger flow channels, analyzes the problems, and gives corresponding suggestions, which can provide references for the planning and management of bus dedicated lanes in big cities.

When the high-efficiency clarifier was used in the water plant to treat drinking water, the sediment flow from the flocculation zone to the sedimentation zone was quite different. Based on CFD, the high-efficiency clarifier was modelled as a whole to analyse the reasons for the sedimentation to one side. The results showed that the distribution of water volume and flow velocity was not uniform. The sediment was biased to one side and the blades rotated so that the fluid had an uneliminated circumferential velocity loop at the outlet of the lifting cylinder. The sediment concentration and flow velocity distribution in the transition section of the flocculation tank and the sedimentation tank were biased to one side, resulting in an uneven distribution of sediment concentration in this section.

A face feature extraction algorithm combining Gabor wavelet transform and principal component analysis (PCA) was proposed. Firstly, face features of different directions and scales were extracted by Gabor wavelet transform to form feature vectors, and then dimensionality of extracted features was reduced by PCA. Experimental results show that the algorithm has good robustness, especially in large sample database.

This paper is based on big data technology, the distribution network cable line construction project, the use of fuzzy Petri network for cable line project construction cost control information modeling technology research, and derives the cost control indexes of distribution network cable line construction, the deviation confidence of control indexes, the correlation confidence of control indexes and the deviation path of control indexes. It is conduciv e to promoting the collaborative management of distribution network investment projects, improving the level of lean cost control in the whole process and enhancing the efficiency of operation.

It is a common method of using machine learning methods to analyze data in water quality assessment. When faced with different data and different research, different machine learning methods perform differently. In this study, in the water quality assessment problem based on water color, the effects of SVM(support vector machine) and DT(decision tree) method were compared. Through modeling, training and testing, the experimental data of the two methods were obtained. By drawing the confusion matrix and calculating the evaluation indicators, it’s found that the accuracy of DT method was 0.927, which was higher than the SVM method of 0.78. Especially in the F1(harmonic mean) value, in which the DT model was 0.721, while the SVM was only 0.485, so the decision tree method performed better. Although there are still some shortcomings, this research provided a reference for the selection of machine learning methods in water quality assessment.

With the gradual improvement of my country's informatization construction, people need more intelligent and accurate information retrieval and automatic question answering and other services in the field of artificial intelligence. In order to continuously improve the performance of the algorithm to provide more efficient and comfortable services, a large number of researchers have invested in the research of natural language processing.Text matching is the core and basic problem in the field of natural language processing. It has experienced from the early traditional text matching methods based on statistics to the deep text matching methods in recent years. This paper studies several popular deep learning text matching methods, including single-semantic text matching, multi-semantic text matching and attention mechanism text matching. On the basis of the currently widely used algorithms, a multi-channel matching pyramid model, a text matching model of cyclic attention mechanism and a model stacking integration algorithm of dynamic parameters are proposed, and the integration is realized by using natural language processing technology.

Helmet wearing is one of the most effective means of ensuring the personal safety of workers at job sites such as construction sites. Improving the accuracy of helmet wearing detection is one of the key technologies for intelligent helmet wearing supervision. To address the problem that the YOLO v5 target detection algorithm fails to focus on important features in the process of extracting features, a YOLO v5 algorithm based on the attention mechanism is proposed to pay attention to important features to improve the detection accuracy. Then, the model is optimized based on the idea of stochastic weight averaging to further improve the model detection performance. The specific method is as follows: After the training iteration until the model accuracy is stable, the learning rate is adjusted to train multiple model parameters, and the final weight model is obtained by stochastic weight averaging. The improved YOLO v5 target detection method has higher detection accuracy than Faster R-CNN, SSD, YOLO v3, YOLO v4, and other detection algorithms, with about 2.3% improvement over YOLO v5, which prove that attention mechanism and stochastic weight averaging are effective methods to improve the performance of helmet detection.

Thermal and visible videos can provide complementary information in the moving object detection and recognition. However, most previous approaches focus on the detection and recognition of moving objects from visible videos. In this paper, we present a two-stage approach to moving object recognition by jointly utilizing the thermal and visible videos. In the first stage, we extract the static appearance and the optical flow of moving objects from both sources of videos based on deep networks and generate the bounding box proposals of moving objects. In this stage, two sources of video frames need to be first registered to cover the same scenes with the same resolution. In the second stage, we design a deep network to recognize the categories of the object proposals generated in the first stage and thus obtain the recognition results. Combining the thermal and visible information for recognizing moving objects can improve the performance especially in the low light conditions. To evaluate the proposed approach, we build a thermal-visible video dataset consisting of 200 video pairs. Experimental results demonstrate the effectiveness of the proposed approach.

In the north area of a new town in Guangdong, the basement adopts special-shaped shear wall. In this project, a variety of design software are used to conduct multi-level analysis on the structure, and the load, internal force, reinforcement and crack checking calculation of the special-shaped basement shear wall are calculated and analyzed. According to the actual situation of the basement shear wall, all the calculated values are within the requirements of the national code and meet the functional requirements, It provides a reliable basis for the safe and correct construction of the basement shear wall, and also provides a reference for the construction of the same type of basement shear wall.

In the 1990s, the computerized Chinese input and output of the "light and power" technology revolution, and the integration of computer graphics technology have completely changed Chinese font art and brought new changes and challenges to Chinese font design. The article describes hand-painted art characters to computer input and computer-generated fonts, explores electronic screen flowing fonts to multimedia presentation of three-dimensional dynamic Chinese font image forms. The evolution of Chinese font design is progressive and stable, and has formed a large number of practical works and some theoretical research results, which are sorted out, analyzed, refined and explained to provide material for future Chinese font design research.

Due to the popularity of social media, rumors spread rapidly on social network, which has been damaging the public trust system and social stability. In recent years, the research on rumor monitoring and early warning has attracted much researchers’ attention. In order to make full use of social network interaction information, researchers use this information to enhance tweets text to improve the performance of rumor monitor. However, most of these methods have to obtain static and complete network structures before their algorithms work, while nodes and edges are constantly evolving in practice. These methods are not only ineffective in evolutionary network structures, but also time and memory consuming, which will directly affect the feasibility of dynamic monitoring of Internet public opinion. In view of the above challenge, this paper proposes a dynamic graph learning method based on interactive information of social networks, which integrates network structure, context semantic and sequential information, especially simplifies the computational complexity of social network evolution. We have carried out a large number of experiments on a largescale real dataset. The experimental results show that our proposed model is better than the existing rumor detection methods.

The low-voltage power line broadband carrier (HPLC) communication field test device is mainly used for the collection of power line channel characteristic information such as field power line noise and impedance, as well as the synchronous playback attack of laboratory power line communication. Low-voltage power line broadband carrier communication synchronous acquisition and replay device needs to be synchronized with the industrial frequency signal sampling, can generate distortion signals in the power line carrier test system synchronized with the industrial frequency, to achieve accurate replay attack test of the test signal collected in the field, in the laboratory environment as realistic as possible to reflect the operating conditions of the communication system to effectively evaluate the performance of different vendors.

This paper predicts the potentially suitable area, grade division and adaptability evaluation of Alternanthera philoxeroides. It comprehensively analyzes the relationship between Alternanthera philoxeroides and climate, soil and altitude, which provides a scientific basis for monitoring and controlling the spread of Alternanthera philoxeroides. Using a combination of the MaxEnt niche model and ArcGis technology, this study is predicted potentially suitable distribution regions for Alternanthera philoxeroides. The results show that the total areas of highly suitable regions, suitable intermediate regions and low suitable regions for Alternanthera philoxeroides were 52.29×104km², 87.04×104km² and 103.35×104km², respectively, representing 5.45%, 9.07%, and 10.77% of China's land area, respectively. The highly suitable regions were mainly located along with the belt of Anhui Province, Jiangsu Province, Hubei Province, Hunan Province, Zhejiang Province and Guangdong Province. The dominant environmental factors affecting the distribution of Alternanthera philoxeroides were, sequentially according to the magnitude of influence, the precipitation during the driest month, altitude, the lowest temperature in the coldest month, the wettest quarterly average temperature and sediment concentration. The distribution of Alternanthera philoxeroides is affected by many environmental variables.

No matter in financial theory research or financial practice application, the long memory time series problem is of great significance. The ARFIMA model is one of the commonly used long-memory financial models. It has problems such as the lag of the long-memory relationship caused by the untimely update of fractional parameters, and the large deviation of parameter values caused by a small number of data sub-intervals. Based on this, this paper proposes a new method with a time-varying Hurst exponent. The new definition of fractional calculus is used to combine with the long memory model ARFIMA to explore the long memory relationship of Chinese financial products and predict the changing trend of financial products. The improved method can effectively solve the problem of interval selection of the R/S analysis method, and the Hurst exponent is also more in line with the changing characteristics of time series. The experimental results show that the new method improves the model performance, better fits the forecast trend curve and effectively reduces the forecast error of financial products. It provides a new research perspective for the non-linearity and complexity of the current financial market, and the empirical results show that the method has practical significance.

Link prediction is a technology to find existing but unobserved links (static) or predict new links (dynamic) by studying the network topology. In real life, most networks change with time. Some recent works focus on the dynamic prediction of networks, but they do not solve the weight problem in link prediction well. At the same time, the methods based on matrix decomposition and node embedding used in most link prediction tasks have some problems, such as huge amount of calculation or can’t better represent the time evolution. In this paper, we extend the method of predicting the weight between nodes in dynamic networks and propose a generative adversarial model based on multi-layer graph convolution and recurrent neural network. The model consists of a generator and discriminator trained during an adversarial process. Multi-layer graph convolution used to approximate the high order similarity of network graph, so as to compensate the time evolution characteristic that RNN network can only deal with low order similarity. Using the process of confrontation training, the model can learn the representation of robust time evolution and predict with high accuracy. Experiments on several real data sets show that our model has good adaptability better than several baseline models.

Using physical cables, branch boxes, switch boxes and other low-voltage power equipment to build a physical environment simulation device, low-voltage power line broadband carrier communication verification environment for testing station identification, phase identification and other functions at the same time, to solve the problem of certain differences between laboratory testing and field testing. The platform can simulate different cable types, different power supply topologies, flexible access to low-voltage power line broadband carrier communication equipment at branch and end points, flexible access to low-voltage power equipment by setting signal injection points, and noise signals that can be played back synchronously by the actual environment.

In the field of video super-resolution, multiple frames provide more scene information, not only in intra-frame spatial dependence, but also inter-frame temporal dependence (for example motion, brightness, and color changes). Therefore, existing work mainly focuses on making better use of spatio-temporal dependence, including explicit motion compensation (for example based on optical flow, based on learning) and cyclic methods. After years of research by predecessors, video super-resolution has achieved good results, but there are still some difficulties and shortcomings that need to be resolved. Generally speaking, SR algorithm series using deep learning technology are different from each other in the following main aspects: different types of network architectures, different types of loss functions, and different types of learning strategies. In this paper, the spatio-temporal fusion module is improved on the basis of a two-stage 3D convolutional network. The adjacent frame and the current frame are merged in space and time, and the generated residual block is added to the feature map. The two-stage 3D convolution is applied to the residual block in the backbone network DenseNet to better process the pre-deblurring module.

In recent years, the Markov model has been widely used in the field of economic management, and one of its characteristics is its ineffectiveness.The stock prediction framework is based on the coupled hidden Markov model, and also makes a 3D and interconnected improvement on the correlation features of predicted events. The model combines stock quantification information and stock news event information, which can effectively alleviate the problem of sparse data, and propose a two-maintenance positive algorithm based on time and space to further modify the results of the coupled hidden Markov model and the results show that our method can effectively improve the model prediction accuracy.

Link prediction aims to find the missing link in current networks or estimate the likelihood the link will appear in the future. In many real-world scenarios, networks can be massive and drastically evolving. Many recent works concentrated on how to solve the procession of massive data, which cares less about the effects of the temporal information. In some dynamic networks, the edge that appears a long time ago will have fewer effects on the appearance of edges in the future. The accuracy of link prediction in dynamic networks can be further improved by adding temporal information in prediction measures. In this paper, we mainly explore the effects of the link temporal information in the graph stream link prediction. We design a graph stream-based framework to solve link prediction problems in dynamic networks which can provide a convenient platform to implement different link prediction methods for dynamic networks, and evaluate these methods on a fair basis. Then, we proposed the Time Decay-based Link Prediction (TDLP) method to improve the efficient of link prediction problem in the dynamic networks. TDLP extends the neighborhood-based measure with time decay functions in graph stream scenarios which can handle temporal information and massive data. Experiment results demonstrate that the accuracy of link prediction for dynamic networks can be efficiently improved in our TDLP method.

With the development of openness, sharing and interconnection of computer network, the architecture of enterprise network becomes more and more complex, and various network security problems appear. Threat Intelligence(TI) Analysis and situation awareness(SA) are the prediction and analysis technology of enterprise security risk, while intrusion detection technology belongs to active defense technology. In order to ensure the safe operation of computer network system, we must establish a multi-level and comprehensive security system. This paper analyzes many security risks faced by enterprise computer network, and integrates threat intelligence analysis, security situation assessment, intrusion detection and other technologies to build a comprehensive enterprise security system to ensure the security of large enterprise network.

Accurate prediction and judgment of the future labor force participation rate will be conducive to the scientific formulation of future employment policy, actual population policy and social security policy by the authorities. In order to propose a suitable labor participation rate prediction model, this paper selects 12 indicators to build the model from four aspects of population structure, economic development, social environment and living conditions based on the analysis of China's social environment. The combination prediction model of neural network model (ANN) and grey model GM (1,1) is constructed to empirically analyze the labor participation rate of China's population in the past 20 years. The average absolute error (MAE), root mean square error (RMSE) and mean absolute percentage error (MAPE) are used as application evaluation index to compare the accuracy of the prediction model. The results show that : (1) The RMSE and MAE error parameters of the combined model based on ANN-GM(1,1) are smaller than those of single prediction. (2) The ANN-GM(1,1) combination forecasting model with weights of ( 0.982,0.018 ) has the best prediction effect.

Based on the combination of BIM technology, cloud computing, the internet, database, and other technologies, this paper studies the design of the function, the logical flow, and the module design of the construction monitoring information management system. The cantilever construction method is mainly used for long-span continuous beam bridges. The larger the segment size, the more complex the construction monitoring will be, especially the alignment control of the completed bridge. Considering the characteristics of the cantilever construction method, combined with BIM + GIS technology, a construction monitoring information management system based on the web is developed. This system can be applied to the construction monitoring of long-span continuous beam bridges with extra-long continuous segments, help the construction monitoring decision-makers to realize the intuitive and efficient management of massive data and documents in the construction monitoring, and realize the management of intelligent information in construction monitoring.

In recent years, with the further development of big data analysis and artificial intelligence technology, many intelligent methods of machine learning have been used in actual production links. The alloy absorption rate refers to the ratio of the weight of the alloying element absorbed by the molten steel during the deoxidation alloying to the total weight of the added element. The main work of this paper is to predict the absorption rate of C and Mn in the iron and steel industry, and realize the automatic optimization and cost control of alloy ingredients in smart factories. However, in the process of molten steel deoxidation and alloying, the alloy absorption rate is affected by many factors, and it is difficult to determine it by an explicit expression. First of all, this paper analyzes the factors that may affect the absorption rate of C and Mn from the perspectives of mechanism and quantification. Secondly, we utilize the canonical correlation analysis method (CCA) to screen out all factors with a correlation coefficient greater than 0.8 as the main influencing factors of element absorption. Finally, comparing the results of the BP and Elman network models, this paper employs the generalized regression neural network (GRNN) to accurately predict the absorption rate.

In this paper, we propose a method of Chinese speech synthesis. In order to achieve the purpose of synthesizing fluent and natural speech, two processing methods, rule-based and statistics-based, are used in the developed system. Firstly, a module design method with the function of text language recognition is introduced in this paper. This module can classify and recognize the text of Chinese, English and special symbols, and deal with the recognition problem that the input text contains Chinese, English and special symbols. Secondly, the Chinese speech synthesis methods used in the system are explained. In the prosody control module, we use a prosody structure prediction model combining neural network and decision tree; when concatenating speech, we propose two smoothing method. One is smoothing algorithm based on spectrum analysis and the other is smoothing algorithm based on dual-threshold. Experiments prove that the synthesis effect of our system is great, and the synthesized speech is clear and natural. Due to the moderate amount of system data and high code efficiency, it is suitable for application systems in mobile Android.

Dental caries is one of the most common diseases suffered by modern people. If early diagnosis and treatment are not carried out in time, dental caries may greatly affect the quality of life. However, the doctor’s diagnosis often misses tooth decay that is not obvious. In our research, we propose a caries detection model that combines attention mechanism and transformer. A new attention mechanism is introduced to represent features in different channels better. The AP50 of the model reached 63.81%, and even unlabeled caries in the X-ray image could be found.

With the widespread use of the mobile Internet, people began to share information on social media more and more. Predicting the popularity of posts can help users quickly discover highly readable content, which is more important in times of crisis. In this paper, we propose a novel model that combines Bidirectional Encoder Representations from Transformers (BERT) and Conditional Random Field regularized Topic Model (CRFTM), referred to as BERT-CRFTM. We first leverage CRFTM to extract the hidden topic information of posts, which can help the model better understand the background knowledge. Secondly, we proposed a method to integrate topic distributions into textual representations obtained by BERT to predict the popularity during the crisis. Experimental results show that, compared with baseline methods, CRFTM-BERT achieves the best results in all three evaluation indicators.

As the shortage of energy resources and the deterioration of the ecological environment continue to intensify, distributed energy is an indispensable part of the future energy system. Distributed energy sources such as solar energy, wind energy, photovoltaics, etc., which are in line with the "dual carbon" goal, have low pollution, high efficiency, and high operational reliability, and will initially become an important source of power supply. However, there are many problems with distributed energy transactions in the traditional electricity market. With the continuous expansion of distributed energy transactions and the continuous increase of information and data, the clearing process has become very cumbersome. For the electricity supplier and consumer, there are problems in the electricity market transaction that the transaction volume caused by power loss does not match the actual received volume, and users do not obtain information in a timely manner. At present, most distributed transaction processes lack a central supervision mechanism. How to confirm and verify various information of both parties to the transaction to ensure transaction security is an urgent problem that needs to be resolved. This article proposes a centralized transaction method based on blockchain technology. Both parties of electricity purchase and sale write data to the blockchain through the blockchain client. After the block link receives the quotation submitted by the user, it will match and settle the two parties through continuous bilateral transactions to realize the centralized transaction clearing of distributed energy. The method proposed in this paper improves the efficiency of distributed energy transactions, and at the same time ensures that the information is authentic and cannot be tampered with.

Aiming at the development status of the Internet of Things, the system architecture of the mainstream Internet of Things is analyzed, and the technical characteristics of the interconnection for different levels are given based on the communication protocols. The access layer and application layer are analyzed for the feasibility of interconnection technology. Especially for the detailed decomposition of application layer protocol, the application of interconnection is analyzed in the industry. The implementation method and future development direction of protocol interconnection are proposed.

Numerical simulation is an effective tool for solving geotechnical engineering problems, and it is more and more widely used in the prediction of mine roadway support effect. However, comparing previous numerical simulation results with practical results, it is found that the additional stress field stress generated by bolt prestress in roadway support generally differs by a large order of magnitude from the original rock stress field value of the roadway. The additional stress field generated by the anchor bolt prestress is much smaller than the original rock stress. Therefore, it is impossible to simulate the stress field generated by the prestress of the anchor rod and cable under the condition of the original rock stress. Therefore, in order to better visualize the compressive stress concentration area around the roadway under the prestressing action of the bolt and anchor cable combined support, firstly, the distribution state of the bolt prestress in the surrounding rock of the roadway is carried out under the non-original rock stress. After completing the simulation, the original rock stress, vertical stress and horizontal stress were added to the model. Where бx = бy = 37MPa, бz = 31MPa. After comparison, a simulation conclusion is drawn. On-site monitoring of roadway separation found that this simulation method has less error and more accurately simulates the effect of roadway support.

BIM can be increasingly utilized in wharf engineering as technology advances, however there are still certain issues owing to the cohabitation of specialies. It is investigated and summarized the application of BIM in wharf engineering Investigation，design, construction and management based on the wharf in Jiangsu, so it solves the issues of collaborative design, simulation construction, real-time sharing of the collaborative platform, and so on. It provides an effective reference basis for the application of BIM technology in wharf engineering and improves the quality and efficiency of wharf engineering Investigation, design, construction and management.

Function, especially the complex system function, is a key point to produce tones of troubles. The complex logic and interface make the natural language and the traditional design methodology based document can not meet the requirement of new aircraft. The MBSE is introduced. the model is a standardized, visualized and simulative. In this paper, the authors present a methodology of the function description and analysis based SysML model and validation it with simulation.

In order to ensure the stability of the quality of the tobacco in the loosening and conditioning process and the accuracy of the measuring instruments, verification can be carried out by constructing a mass transfer prediction model and an online verification process. Statistically analyze the input and output of materials in the production process of loosening and conditioning process, and obtain the prediction model of dry matter quality, water matter quality, moisture steam conversion coefficient in tobacco leaves and outlet moisture gas. After production experiment verification, the relative error range between the predicted value of the outlet moisture of the tobacco leaves and the actual measured value is maintained at ±0.48%, and the average relative error is 0.21%, which is less than 0.5%, which meets the requirements of process standards, and the model is accurate and reliable. Based on the material balance forecasting model, the theoretical value is compared with the actual value to realize the online comparison and verification of the moisture meter, which has a certain guiding role in improving the accuracy and verification efficiency of the measuring instrument.

The time-varying law of design flood is the theoretical basis of sponge measure construction safety guarantee and measure effect evaluation. Based on the 2016-2018 rainfall and flood data in Wufeng River Basin in Pingxiang City, this paper uses the time-varying moment method to evaluate the time-varying process of flood designed by computer simulation. The analysis shows that the measured flood volume decreases linearly with time, the relative time difference between rain peak and flood peak increases with the logarithm of the time history, and the peak discharge firstly increases and then decreases with time. The optimal ratios of skewness coefficient and variation coefficient of total flood volume, peak discharge and relative peak current time difference are 2.5, 2 and 2 respectively, and the Nash coefficients between the element theory and empirical frequencies are 0.98, 0.79, and 0.93, respectively. After the transformation of the sponge city, the total amount of design floods decreased, and the total amount of once floods in the main flood season and the post-flood season from 4 to 100 years decreased by 16% and 8.5%. The relative time difference between rain peak and flood peak increased by 101.5% and 79.5% respectively in the main flood season and the post-flood season. The peak discharge during the main flood season increased by 4.8%, and then decreased by 9.5% during the post-flood season. This article can provides a scientific basis for the scientific and construction of sponge cities.

Software simulation of building energy consumption has become one of the normal means of measuring building energy consumption. This paper uses Openstudio energy simulation software to construct passive residential building models from the perspective of envelope, internal loads and air conditioning systems, and future climate change. Selecting the SRESA2 emissions scenario downscales to generate weather files for the 2050s and 2080s, forecasts future passive residential building energy consumption, and analyzes the influencing factors of passive residential energy consumption. The software simulation results show that the lighting density power and equipment power of the internal load, as well as the COP of the air conditioning system and the interior design temperature, all affect the total energy consumption of the building to varying degrees. The heat transfer coefficient of the exterior walls and the temperature of the COP air conditioning system and the interior design temperature have the greatest influence on passive residential buildings. To provide reference for strategies related to passive residential building design under future climate change.

Safe production is related to the company's economic benefits or even survivals, and an important goal pursued by the company. It is basis to predict hazard precisely for controlling hidden dangers, eliminating hidden dangers, and improving the company's safety production capacity effectively. In this paper, the example of safety production hazard investigation is studied based on cloud service. First, the company's existing hazard data is used to classify statistics, normalize data and convert to grade. Second, the matrix decomposition method is introduced to predict the hazard degree of the company with hazard data of company and safety supervision department. By comparing the prediction data with the actual hidden data, it is show that the matrix decomposition method can be well used in hazard prediction in the enterprise to better eliminate hidden dangers.

With the development of science and technology, the application of artificial neural network and computer vision has become more extensive. Traditional river patrol methods have some inefficiencies and disadvantages in the quality of river patrol. To solve this problem, in this paper, we proposed an approach combined the high-performance YOLO v3 model with the SURF algorithm to detect illegal constructions and behaviors around the river bank. We first take image as the input of YOLO v3 model, then feed the output of YOLO v3 into the SURF algorithm to make a further detection. With this method, we can improve the quality of the river patrol, and enhance the ability to deal with illegal incidents around the river.

Since the 21st century, with the rapid development of science and technology, China's economy has ushered in a leap through reform and opening up. With the increasing maturity of information technology, the construction industry, as one of China's key pillar industries, has also begun to apply information technology. As a new information technology, BIM Technology has attracted the attention of the architectural design industry once it came out. Its emergence not only brings a new design method to the construction industry, but also promotes the development of the construction industry and solves many difficult problems in the past. This paper will analyze the significance of applying BIM Technology to the construction industry.

The Yangtze River Delta region is characterized by hot summer and cold winter, mild climate in spring and autumn, significant precipitation in plum rain season, high relative humidity throughout the year and relatively sufficient light and heat. In this climate environment, people use a large number of air conditioners to improve the indoor comfort of buildings, which also greatly increases the building energy consumption. Combined with the natural environment and human economic characteristics of the Yangtze River Delta, this paper simulates the energy consumption of passive buildings, and the significant improvement of energy efficiency fully explains the applicability of developing passive buildings in the Yangtze River Delta.

DEA method was used to calculate the Malmquist-Luenberger index of agricultural energy, and the resilience effect of agricultural energy was further calculated according to the ML index to study the agricultural energy efficiency and energy conservation. The results show that :(1) the national average agricultural energy efficiency is greater than 1 in most years, indicating that energy efficiency is on the rise. (2) The rebound effect of agricultural energy is generally between 0 and 1, without obvious adverse effect, and the energy saving effect is fair. (3) The agricultural energy efficiency and resilience effect of major grain-producing provinces are lower than that of non-major grain-producing provinces, and there is a large space for energy conservation and efficiency improvement.

Few studies have explored the relationship between green transition, carbon emissions and non-fossil energy consumption. To bridge this gap, a green transition index based on four indicators is first constructed: resource endowment, socioeconomic, industrial pollution and environmental governance. Secondly, data from 2000 to 2019, vector error correction model is used to analyze the relationship between green transition level, carbon emissions and non-fossil energy consumption. We find that green transition is imminent in China. Green transition is conducive to non-fossil energy consumption, and the reduction of carbon dioxide emissions will promote the use of non-fossil energy. The variance decomposition results show that the contribution of green transition to non-fossil energy consumption reaches 36.66% in the tenth period, and carbon intensity gradually increases to about 22.08% in the whole period. Finally, based on the results, this study puts forward suggestions to focus on controlling fossil energy consumption and implement total carbon emission control as a supplement, and provides new ideas for promoting the country's green transformation.

Using the method of acoustic experiment, this paper analyzes the acoustic characteristics of diphthongs in Shigatse dialect, such as formant, fundamental frequency and duration. It is concluded that there are two types of diphthongs in Shigatse Dialect: True diphthongs and false diphthongs come from the reduction of syllables and the falling off of the stop ending, respectively; True diphthongs are long vowels and false diphthongs are short vowels; True diphthongs are paired with long keys, with keys 44 and 14, and false diphthongs are paired with short keys, with keys 51.The discussion of diphthongs in Shigatse dialect can provide a certain reference for the further study of diphthongs in Weizang dialect in the future.

With the development of artificial intelligence technology, knowledge graph has been applied in more and more fields. In this paper, agricultural data is constructed as graph structure data set, and the graph structure is mapped to low dimensional vector space through the first and second order similarity of the graph. The attention mechanism is used to constrain the information propagation in the process of graph convolution, and finally the classification of each data node is obtained by SVM. In the process of information transmission, the accuracy of the model is improved to some extent by effectively utilizing the higher-order information of the graph. Compared with GCN and GAT models, the accuracy of the experiment in this paper is increased by eight percent and four percent.

In the process of continuous casting slab production, serious defects will have an adverse impact on the subsequent rolling process. The target detection of cracks using machine vision algorithm has been increasingly applied in industry. The detection of defects in hot billets is of great significance. Adjusting the flow and flow rate of mould in advance can prevent more defective billets from being produced. In this paper, the detection system of hot billet is constructed by combining YOLO (You Only Look Once) and public data set, which can realize the defect detection in industrial production. Combined with the two algorithms of YOLO V3 and YOLO V4, the system detection results are compared, and a comparative conclusion is drawn. YOLO V4 algorithm uses multi-scale detail boosting at the input for image enhancement, and the part of neck adopts SPP module and FPN + PAN mode, and on this premise, the definition of partial loss function is changed. These changes eventually make YOLO V4 faster, more accurate and lighter. According to the experimental results of this paper, the following conclusions are drawn: This system can realize the defect detection of hot continuous steel casting in industry. The maximum value has not been greater than 0.1, while GIoU is lower than 0.02 when the epcho is greater than 200. The accuracy of YOLO V4 training prediction framework is much higher than that of YOLO V3, and the target detection is also more accurate. In terms of recall rate and average AP value of various categories, YOLO V4 is better, with a maximum increase of 0.1. At the same time, among the samples divided into positive examples into all crack categories, the average proportion of actual positive examples is also higher.

Based on the research and exploration of the construction process of the steel reinforced concrete composite structure, this paper summarizes a set of systematic improvement methods for the design and construction of the section steel concrete composite structure. The BIM technology is employed to build a three-dimensional model to optimize the construction layout of the core area of the steel composite structure. In addition, the precise positioning of beams, columns, splicing plates, stiffeners, bolts, and welds can be obtained in advance. Therefore, the whole process of visualized construction can be realized to reduce construction cost and improve construction efficiency. The research results in this paper can provide reference for similar projects.

Using four-year exam scores of Class 2018 students majoring in business administration in an independent college as a case study, the paper performed a comprehensive evaluation of the students based on principal component analysis (PCA) and application of MATLAB. In addition, the paper also compares the ranking based on PCA with the ranking gained through the average score method. It is found that the PCA method is more reasonable and more in line with the academic goals of independent colleges and universities.