This PDF file contains the front matter associated with SPIE Proceedings Volume 12085 including the Title Page, Copyright information, and Table of Contents.

In recent years, perovskite solar cells have received extensive attention due to the advantages such as high light capture coefficient and long carrier diffusion distance. The photoelectric conversion efficiency of perovskite solar cells has increased from 3.8% to 26.7%, which has enormous development potential. The preparation process of the absorption layer is studied, and the influence of the preparation process on the photoelectric properties is investigated. The development of the preparation methods are put forward, and the stability of perovskite materials is prospected to improve the photoelectric properties.

Azimuth super-resolution is generally realized by real beam scanning deconvolution and array antenna super-resolution algorithms, especially by using amplitude envelope or phase information. Aiming at the situation that echo information is not fully utilized, this paper proposes a forward looking imaging method of array radar scanning based on compressed sensing, combining the real beam scanning by transmitter and the echo receiving by array antenna. Under the condition of sparse scene, the range dimension super-resolution is achieved by pulse compression, and the azimuth direction complex observation matrix is constructed based on the array antenna echo model, and the super-resolution is achieved by improved L1 norm smoothing approximation algorithm. The 2d super-resolution imaging simulation results show that the proposed method can effectively improve the performance.

With the development of science and technology, the military micro-UAV, as a new combat force, is used more and more frequently in missions. In order to improve the informatization and intelligence level of the military micro-UAV operator training and assessment, and meet the requirements of equal height, uniform speed and small deviation in the actual flight process, we use RTK Positioning Technology, After satellite positioning and real-time dynamic difference calculation of the military micro-UAV, important flight control data such as real-time flight trajectory and altitude change and result reference score can be displayed and stored on the user's screen, which can effectively solve the problems of strong subjectivity of evaluation results and great impact on the environment in the existing the military micro-UAV training assessment. Ensure that the evaluation of the military micro-UAV training assessment is objective and reasonable, and effectively ensure and improve the training level.

As the core key equipment of the power system, power transformers transmit AC power to the power grid. They are an indispensable part of long-distance power transmission and play a vital role in the safe, stable and reliable operation of the power grid. In order to perform on-line detection of partial discharge of transformers and classify faults, this paper proposes an intelligent spectrum monitoring method based on power transformers. This article first designs a power transformer electromagnetic spectrum signal acquisition system, and then performs data preprocessing on the collected power transformer spectrum data using wavelet transform. Wavelet transform can check the frequency domain characteristics of the local time domain process and the time domain characteristics of the local frequency domain process. After that, this paper use a convolutional neural network model, which is a residual network. The model makes a reference to the input of each layer and learns to form a residual function instead of learning some functions without reference. This residual function is easier to optimize and can greatly deepen the number of network layers, Input the preprocessed I/Q data into the network model for fault classification of power transformers. Based on the above, the frequency spectrum of the power transformer can be effectively monitored and fault detection can be performed. Experimental results show that our method can classify power transformer faults with an accuracy rate of 98.4%, and has a higher accuracy rate than traditional transformer fault detection techniques.

The Transformer has achieved tremendous success in computer vision, natural language processing, and graph representation learning. However, the transformer cannot effectively encode the topology information of the graph into the model, while it is the advantage of the graph convolution network (GCN). Therefore, we propose a model GTGC combining transformer and GCN for graph classification tasks. To this end, we take the result of graph data passing through multi-head self-attention and feed-forward blocks as the input of the graph convolution module. By increasing the number of neighbors for each node’s feature matrix, the nodes with more neighbors are more important in the attention mechanism. We validate the validity of the model on multiple data sets, including social network datasets and bioinformatics datasets. Experimental results demonstrate that our model achieves advanced accuracy

Energy is an important component of a country's economic development, and power resources are also an indispensable energy in people's production and life. There is also a large amount of energy waste caused by low lighting management efficiency in universities. This paper has designed a new intelligent light control system based on IPV6 network, using the network camera installed in the classroom to transfer the classroom image to the server through the IPV6 network. Through image processing, the data in the classroom is obtained and sent to the light control module to control the switch of the classroom light, supplemented by the manual control module, clock control module and photosensitive module, etc., and finally realize the intelligent management of the classroom lighting. Through the test of the intelligent light control system based on IPv6 network, the system has high sensitivity to moving images. For classrooms with different lighting conditions, large changes in pedestrian density and large randomness of use time, the control of the system can achieve accurate regulation. The whole system has low cost and high reliability. It is very suitable for classroom lighting control in teaching units and has good application value.

Using remote sensing technology to extract urban impervious surface information has a wide range and high efficiency and plays an essential role in monitoring urban expansion and urban ecological environment change. Generally speaking, if a place is an impervious surface, its surface temperature would be higher. The traditional impervious surface information extraction is based on a single data source and a single-phase remote sensing image, and the error of the extraction results is large. Based on the multi-temporal Sentinel-2 remote sensing image and LJ1-01 nighttime light data in 2018, the impervious surface information was extracted in the main urban area of Qingdao. Combined with the surface temperature retrieved from Landsat OLI/TIRS data, the spatial relationship between the urban heat island effect and impervious surface coverage was analyzed. The results show that multi-temporal and multi-source data can accurately extract urban impervious surface information, and some regions between impervious surface coverage and urban heat island effect cannot show spatial correlation.

Plotting system refers to a software system that can draw a series of vector symbols, typesetting, saving, reading and so on. The traditional plotting system uses key and mouse operation to draw. With the development of electronic equipment touch screen in recent years, the traditional menu selection is not convenient enough, but the way of using hand-painted sketch and recognized by the system is more natural and fast. Traditional image recognition methods need huge amount of data for training, but the vector symbol set in the plotting system is not a common symbol set, so it is difficult to obtain a large amount of data. Therefore, this paper proposes a β-TCVAE-Attention hand-drawn plot recognition method which can obtain better recognition accuracy in the case of small samples. The model first uses the Image dataset to pre-train βTCVAE, and then uses the small sample data of the plot sketch to input the β-TCVAE encoder to obtain the sketch features. After the sketch features are enhanced with the attention module, they are input into the Softmax classifier based on cosine similarity. The neural network is fine-tuned according to the classification results of the classifier. Finally, when using network prediction, input the support set and query graph into the network for classification. In this paper, a 20-way-5- shot plotting data set is collected for experiment, and the experiment proves that the model has certain accuracy.

Speech enhancement technology has been applied to many research fields of speech signal processing, such as speech coding, speech recognition, noise suppression and so on. However, in practical applications, there is always noise in the environment, voice in the transmission, to avoid noise interference, and the quality of the environment noise has different effects on the quality of the speech signal. Therefore, speech enhancement algorithm can realize the original clean speech from noisy speech signal. The traditional single microphone in the process of dealing with environmental noise, the received target signal and noise signals are often overlapping in the time and frequency spectrum. In view of the traditional method, using microphone array can improve the quality of voice communication, and can make full use of the time and space information of speech signal. Therefore, the use of a microphone array can overcome the shortcomings of a single microphone.

China Energy Investment Corporation (China Energy) conducts research on energy management and control in photovoltaic buildings and building energy efficiency control, establishes a comprehensive energy control system, and has established demonstration pilots throughout the country. However, the research on the security of energy management and control systems is relatively weak. For this reason, this article analyzes the application background of trusted computing and establishes a comprehensive energy control system security architecture based on trusted computing, focusing on the realization of trusted guidance for embedded devices and trusted operation of system applications. And equipment status monitoring and other functions, provide a reference for the construction of safety protection measures for the integrated energy management and control system in the future.

Under the background of world energy crisis and increasingly serious environmental pollution, how to collect and use kinds of mechanical energy in nature has attracted more and more attention. TENG can be used to collect wind energy, ocean energy, vibration energy and other kinds of energy. In this study, we have fabricated a TENG for collecting impact energy, it consists of two independent generating units. In addition, this paper also designs a power management circuit, the circuit is mainly composed of rectifying part and voltage stabilizing part, the rectifying part is stacked by rectifying bridges, the AC signal generated by the TENG is rectified into a pulsating DC signal, after capacitance filtering, the voltage stabilizing part uses a voltage stabilizing diode to convert the output voltage into a stable DC voltage. It has been tested by experiment, when generating unit is working, the voltage can be stabilized at 3.7V by single tube circuit.

Objective According to the characteristics of Fuzhou Shadow Play, digital reconstruction and innovative display are carried out through somatosensory interaction to enhance the audience's senses of participation and immersion, provide a new technical path for Intangible Cultural Heritage Inheritance and digital protection, and make the audience feel the charm of Chinese excellent traditional cultures. Methods On Microsoft Kinect 2.0 equipped with a CMOS depth sensor, depth data are acquired through the time-of-flight (TOF) ranging algorithm, which is based on the two-way ranging information technology of asynchronous transceivers. By collecting the depth image within the field-of-view of an infrared camera, the sensor uses a high-speed generated image stream with depth information to reproduce the scene in three dimensions. The optimized preprocessing shortens the evaluation of images with depth information, which is used to identify various parts of the human body. In the depth image, a specific mask area is created for each tracked viewer. Each pixel in the mask is transmitted to a machine learning system to identify various parts of the human body. Each recognized part is marked with the possibility of a specific pixel for the part, and a complete skeleton system is generated according to the coordinates of the 20 joint points tracked. Then the three-dimensional space data of the human body limbs are sent to the computer. Results Subsequent interactive reconstruction shows the human body movement and the joint point shadow puppet model movement match well, without model interspersion or bone damage. Conclusions The threedimensional digital preservation of shadow puppet entity realized by the construction of a joint point shadow puppet model eliminates time and space limitations against preservation of shadow puppet entity, highlights the immersion experience and reconstruction of drama time and space, creates a new interactive relationship of traditional drama performance, and changes the way of performing, viewing and creating Fuzhou shadow puppet art.

For fast telescope, when the observation object is directly above the reference sphere, this paper first analyzes that the adjustment of the reflection panel is affected by the radial expansion range of the actuator, the variation amplitude of the main cable node, the ideal parabolic equation, the working aperture and the mean square error rate, so as to list the constraint conditions and objective function. Then the problem is transformed into a single objective optimization problem by using ergodic algorithm and genetic algorithm. Finally, the equation of the ideal paraboloid and the distance amplitude between the paraboloid and the reference sphere are determined.

In this proposed paper, a microstrip line patch antenna with Rogers RT5880 substrate is designed to meet the demands of a 5G millimeter wave mobile network's demands. The presented antenna has a 10mm × 10mm × 0.57mm dimension with four L-shaped slots on the patch. The antenna resonates at 25.82 GHz, which is in the n258 band and has a realized gain of 7.85 dBi. In this paper, all geometry dimensions of the antenna and other parameters are presented.

With the advent of the 5G era, people's requirements for communication quality are gradually increasing. Compared with the previous centimetre wave antenna, the millimeter wave antenna has a wider bandwidth and ultra-high-speed with low latency. In present, the use rate of the millimetre wave frequency band is less, so there are few interference sources. In this respect, the reliability of millimeter wave communication is improved. Patch antennas have been widely used in recent years because of their small size and good integration with other circuit components. In the design of this article, we will use CST software to simulate a 5G millimeter wave microstrip line patch antenna.

The high reliability and low latency of 5G communication provide effective technical support for data transmission in the substation grid, but the delay and jitter of 5G technology will affect the effect of grid differential protection. In order to reduce the delay and jitter caused by the transmission of the data collected by the substation through the 5G network, this article first studies the current 5G network usage scenarios in the substation, analyzes the delay minimization method to optimize the demand for substation network communication protection, and proposes to address the delay The research plan for optimized access of 5G communication slices in the scene meets the application scenarios of the delay constraint of the communication protection of the substation.

Due to the characteristics of large volume, multiple types and fast speed of power grid data, traditional methods have been unable to process power big data. Therefore, this study proposes the information security terminal architecture of mobile Internet of Things in power industry based on big data analysis, and presents the architecture of the whole system. Based on the Yarn cluster computing system that supports Spark, Hadoop technology, wireless communication technology, Internet technology, sensing technology and so on are used to provide the solution of big data storage and terminal collection of the system, and the existing difficulties and problems are analyzed. The main difficulty of big data construction lies in that HadOOP platform itself is only one of the technical means to achieve big data platform. Therefore, the main difficulty of this paper is how to construct a more reasonable data organization mode and determine the source and organization form of associated data from the actual application requirements of the power industry, so as to truly mine as much value-added information as possible from the large database.

The research of heterogeneous data collection devices and methods for mobile networks is an important research topic in the field of mobile communication nowadays. After the development of 3G, 4G and 5G, mobile communication networks are becoming more and more complex, the number of network elements is increasing, and the types of services are becoming more and more diversified. Traditional mobile networks encounter problems such as poor filtering flexibility and lack of good load balancing mechanism in the process of data collection. This paper optimizes the design of heterogeneous data collection devices and methods for mobile networks based on DPDK technology, which can greatly simplify the signaling collection scheme, increase the utilization, make the network more flexible, and improve the versatility of data collection equipment.

Bayesian network can effectively deal with uncertain reasoning and realize data statistical analysis, and it has been widely used in current power system research. Because of its strong logic and fast data processing, Bayesian networks have great advantages in solving the faults caused by the uncertainty and relevance of complex systems. The construction and operation and maintenance of the power Internet of Things are faced with various uncertainties. In order to improve the ability of the power Internet of things to deal with various uncertain and unexpected factors, a Bayesian network-based reliability evaluation model of the power Internet of things is established. This model integrates the level indicator reliability requirements of the perception layer, network layer, platform layer and application layer of the power Internet of Things, and establishes a Bayesian network node and network structure containing its threat factors. In addition, the Bayesian probability solution method is used to describe the reliability level, and quantitatively analyze the failure and accident consequences of the power Internet of Things. From this, it is possible to assess the emergency response level of the power Internet of Things to formulate corresponding risk response methods, to curb any power Internet of Things security incidents at the source in the future, and to increase the security construction and operation and maintenance of the power Internet of Things.

YOLOv3 is one of the target detection methods based on DNN, which has good performance in terms of speed and accuracy. However, with the gradual increase in application scenarios and the increasing demand for detection, the missed detection problem of YOLOv3 is deeply criticized. To further improve detection performance, we propose an improved model YOLOv3-A based on YOLOv3. Firstly, YOLOv3-A introduces channel and space two-dimensional attention modules in the feature extraction backbone network to achieve feature focusing on different dimensions, thereby reducing the missed detection rate. Then, in order to avoid network degradation, YOLOv3-A uses a residual module to replace the multi-layer convolutional layer at the rear of the feature extraction backbone network. Finally, YOLOv3-A introduces two different loss functions for positioning and confidence, respectively, using GIoU loss for bounding box positioning and Focal loss for confidence to improve model accuracy. On the PASCAL VOC data set, the mAP of YOLOv3-A is 59.13%, and the mAP of YOLOv3 is 50.92%.

With the advancement of technology and the updating of information, more people choose to watch movies on the Internet. In the world of brust data, users often encounter the difficulty of finding their favorite movies. Implementing the movie recommendation system of the media platform is one of the most effective ways to solve this problem. In order to avoid the imbalance of user data in the actual operation process, the content-based recommendation model is adopted in this research, aiming to find their similar variables in each movie, to calculate similarity index between each movie through the matrix and Pearson formula. The advantage of this method is that the potential interest of users for each movie can be discovered, and the probability of the movie being recommended will not be caused by the problem of the popularity of the movie.

With the outbreak of the Covid-19, people’s reliance on electronic devices has become more and more serious. Especially in the special period in present, the mode of online learning through electronic devices has been widely adopted by young students at different stages. There are various supports for electronic devices, but most of them can only be placed in a fixed position. Some display supports can be manually adjusted to a suitable position, which cannot provide more intelligent functions. Based on the current circumstances, by using computer software to build a 3D bracket model, and make it by 3D printer, with raspberry pi as the main control chip and stm32 as the system auxiliary chip, a set of intelligent display screen bracket based on IOT is developed. The smart bracket can independently adjust the position of the user and the screen, and can set auxiliary functions in real time according to needs, such as reminder of learning hours, sitting posture adjustment, voice interaction, etc. and can control core functions through the mobile terminal of the mobile phone. This paper aims to alleviate the problems of eye fatigue and high myopia rate caused by the improper posture of young people when using electronic display devices.

This paper first introduces the concept of DevOps and microservice architecture, then combines a year of domestic and international research on DevOps and microservice project management, and proposes a DevOps-based microservice project management technology selection, including the selection of microservice environment, automated deployment environment and automated operation and maintenance environment. DevOps is a software development management method that has received wide attention because of its DevOps enables continuous delivery of software products from development to O and M services by integrating the development and O and M phases, thus allowing enterprises to improve their effectiveness. Microservices are a relatively small independent functional unit or a group of them, and this paper combines the two and proposes a technology selection idea.

The task of Chinese named entity recognition (CNER) is closely related to Chinese word segmentation, because most Chinese entities are composed of words. The CNER model with words as the minimum input unit is called word based CNER model. However, wrong word segmentation will lead to wrong entity recognition results, so the CNER model with single word as input (char based) has gradually become the focus of research. Recently, many efforts have begun to integrate external word information to improve the performance of word based CNER model. And softlexicon are excellent word information fusion methods. When using softlexicon method for word fusion, too much matching word set information will be introduced, resulting in additional "noise" interference, resulting in entity recognition errors. Aiming at the problems of softlexicon method, this paper proposes a word information Selected-Softword-Concat method (SSC), which effectively solves the problem of "noise" in softlexicon method by introducing word segmentation information. The experimental results on four general CNER data sets show that compared with softlexicon method, the F1 value of the model constructed by SSC method is increased by 0.94, 0.92, 0.78 and 0.25 respectively, which effectively solves the "noise" problem of softlexicon.

With the gradual advancement of globalization and the rapid development of computer network technology, many aspects of people's life are being strongly impacted by the network era. Traditional foreign language teaching is facing unlimited impact and challenge in the new environment and opportunity of computer network. How to guide the efficient integration of resources, namely, give full play to the advantage of the information age, the computer network to adapt to the needs of the development of modern education and really become an organic part of course, to promote our country university art teaching from the traditional to the modern transformation of important issues, but also constitutes the research focus of the current domestic foreign language teaching reform. This study conform to the time background of university art education reform in our country and the development trend, the research scope limited to institutions of higher learning, to observe the evaluation of the school in college art teaching current situation of the development of computer network environment, summarizes the traditional art teaching imbalance in the development of the modern education technology integration of performance, analyzing from the Angle of education ecology imbalance causes, countermeasures of innovation for alleviate and eliminate disturbance. The research results are expected to play a role in teaching practice and play a reference and complementary role in China's current foreign language teaching reform.

As an important part of China's socialist market economy, the non-public economy is an indispensable force in our country's national economy. This paper introduces a set of designs for constructing a knowledge graph of the non-public economy. After cleaning, converting, fusing and extracting multi-source heterogeneous data related to the non-public economy, a special domain knowledge graph is constructed for storage, which can improve search, storage and update efficiency, improve the accuracy and completeness of data, better mine the information carried in the data, etc. So as to help the All-China Federation of Industry and Commerce to better supervise and guide the non-public economic market to promote the healthy, orderly and prosperous development of the non-public ownership economy.

The popularity of the 5G network makes People's Daily life more convenient. When facing different situations of transmission and receiving, through the design work in 28 GHz 5G millimeter-wave frequency patch antenna, by changing the different parameters of the antenna. By comparing the working conditions of antennas under different parameters, we can find the most suitable parameters that applies millimeter-wave to 5G signal transmission. Because 5G signal transmission is faster and more convenient for users' daily life.

An edge computing offloading system based on container-layering is proposed, and the task is divided into several discrete offload able subtasks aiming at the computation-intensive and time-sensitive applications in the IIOT and the resource heterogeneity of edge computing devices. For low offloading latency, a SPMOC-ABC (Seven-Phase and Mutation-andOriginal-Crossover Artificial Bee Colony) algorithm based offloading approach is proposed which include Seven-Phase optimization mechanism and Mutation-and-Original-Crossover updating mechanism. Numerical results demonstrate that the proposed container-layering offloading system can significantly improve the utilization of computing resources, and the SPMOC-ABC based offloading approach can effectively reduce latency.

With the development of modernization, the grain logistics system also requires more complete and supporting infrastructure, more scientific and standardized operation methods, and more efficient and accurate information services, so as to improve the management level of the grain industry and reduce the cost of grain logistics. This paper, through the research of policy grain information high concurrent collection technology, the construction of grain quantity and quality information data collection platform, and the collection of grain information data, provides common data unified access service for the grain information service cloud platform, provides more powerful data service ability for the application of the grain industry, and provides technical basis for the informatization and intelligent development of the grain industry.

The container terminals and carriers are regarded as a challenging wireless communication enviroment, caused by the metalic walls of container and the corners with the resulting multipath and shadow effect. Accurate propagation modeling of it is helpful for node deployment and channel estimation. However, existing researches mostly focus on obstacles on the macro-scale, and rarely analyze the small-scale propagation characteristics, such as the propagation characteristics along the gaps between containers. Thus there is also a lack of engineering practical path loss model. In this paper, we first study the propagation environment of the stacked containers. Then we propose a path loss model for gaps between containers based on geometric analysis. The proposed path loss model matches well with ground truth value.

This paper mainly studies the optimal cost planning of production enterprises in supply chain management. The research on this problem can effectively reduce the risk of enterprises in the procurement process, enable enterprises to obtain the maximum benefits with the lowest cost, and enhance the market competitiveness of enterprises. First of all, the data in Annex 1 are analyzed and preprocessed to extract three kinds of indicators: supply volume, stability and credit value, and then comprehensively evaluate the quality of suppliers from these three angles. Then the double-objective programming model is used to solve the problem, and the genetic algorithm, subjective weighting method, MATLAB programming calculation and SPSS statistical analysis are used to get the ordering plan and transfer plan in the next 24 weeks, which can meet the production demand and the lowest production cost at the same time. Combined with the premise that the enterprise has sufficient room for capacity improvement, the total supply of all suppliers calculated is the weekly capacity appreciation of the enterprise. At the same time, the order quantity is the supply quantity, so when making the ordering and transshipment plan, we only need to consider the economy of the transshipment plan, through the 0-1 programming model with the lowest loss rate as the objective function, get the raw material ordering plan and transfer plan of the enterprise in the next 24 weeks. Finally, the advantages and disadvantages of the model are analyzed and evaluated. on the basis of improving the shortcomings, it is concluded that the model can also be extended to the sales planning problems faced by enterprises and the scheme selection problems in life.

Antenna design is significantly vital in IoT (Internet of things) communication, arousing many researchers' interest in studying this area. In this paper, a microstrip patch antenna for a 5G millimeter wave is proposed. We base on some IoT applications and 5G standards to calculate the ideal parameters. After that, CST is used to simulate the results and modify our calculated parameters to fit the designed demand well. The proposed antenna configuration with a maximum gain of 7.98 dBi at 25.87GHz will cover some 5G wireless systems.

In today's rapid development of internet of things technology, the application of internet of things technology to port logistics construction management will help improve port resource utilization and productivity, and promote port intelligent development. Analyzes the advantages and mode of application of networking technology in port logistics, and on this basis the aid of radio frequency identification technology, global positioning system and cloud computing technologies elaborated the application process of things container yard at port.

With the development of fifth-generation wireless communication, 5G has attracted more and more attention. 5G is an essential link for the power system in developing the future smart grid, whether in terms of various advantages of 5G or low delay internet of things (IoT) based on 5G support. 5G will show its benefits more comprehensively in the smart grid. The current power system incorporated with 5G will be more reliable, secure, flexible and durable. This paper briefly discusses the contribution of the IoT to the development of power systems from generation, transmission line and distribution side, and users perspectives. This paper provides an analysis of the role of 5G in the growth of IoT technology and smart grid. Huge amounts of data will be generated due to the introduction of renewable energy, deployment of smart grids and improvements in electricity markets. Therefore, applying the IoT to the power system is crucial to connect the IoT to consumers, things to things, and people to people in the power system. However, 5G also faces some challenges during deployment. Finally, this paper outlines the role, impact and challenges of 5G in the power system.

The torpedo car (TPC) plays a vital role in the safe transportation of molten iron in steel manufacturing. Due to its longterm operation in high-temperature environments and frequent use, it is prone to a variety of faults. However, traditional manual judgment methods are not effective in fault diagnosis and have low safety factors. Through engineering practice and data visualization analysis, it is found that the temperature data of the torpedo car have a strong time dependence and are closely related to the occurrence of its faults. Therefore, this paper selects the temperature data from the working status data of a torpedo car in a large steel plant. On the basis of effective data preprocessing and sample construction, the multilayer ATT-GRU network model based on the attention mechanism using timing features is proposed for its fault diagnosis. The model uses a multilayer GRU network to extract the temperature sequence characteristics of different time segments, and the attention mechanism is used to focus on the part of the input that contributes a greater amount to the fault diagnosis results, so as to improve the diagnostic effect of the model on the type of the torpedo car faults. Compared with traditional methods, experiments show that this method significantly improves the convergence speed of model training and diagnosis accuracy. It provides a friendly solution for real-time and accurate fault diagnosis of the torpedo car and other industrial equipment.

Fuzzing has become one of the most commonly used methods for exploiting software program vulnerabilities because of its ease of use and effectiveness. With the increasing complexity and importance of network services, fuzzing in parallel mode can greatly improve the efficiency of finding vulnerabilities. Typical representatives are Google's Clusterfuzz and the parallel mode of fuzzers such as AFL, Libfuzz. However, due to the lack of effective control in the process of mutating seeds in different fuzzing instances, the existing parallel fuzzing methods have problems such as high redundancy of generated test cases and low comprehensive coverage. In response to this problem, we proposes a parallel fuzzing method based on two-stage mutation. First, we use random mutations to perform fuzzing test within a period of time, divide the detected state space of the target program into independent state subspaces, and then distribute the corresponding tasks to each fuzzing instance, and use the same ordered mutation on each instance to ensure the generated test cases are lowredundant. We use this method to conduct experiments on the LAVA-M test set and open source software OpenJPEG. The experimental results show that it can improve the efficiency of fuzzing and find more crashes.

The application of 5G millimeter wave communication is becoming more and more widespread. This paper analyzes the rectangular patch antenna based on the transmission line model, and designs a mobile terminal antenna for 5G millimeter wave communications through CST software, which effectively reduces its size by using centre-side feed and embedded feed technology. The simulation results suggest that the central frequency of the antenna is 43GHz, the effective working frequency band covers 40-47GHz, the width of the band is 7GHz, the port reflection coefficient when it works at the central frequency is less than -20dB, the maximum gain is 5.451dB, and the antenna also has a good sense of direction and low loss. Furthermore, the design is quite flexible and simple in structure. Therefore, it is much likely that the antenna can be used for the future 5G millimeter wave communication terminal and the MIMO antenna element of the 5G communication base station.

C4 olefins play an important role in the production of chemical products and medicine, and ethanol is used as the raw material for the preparation of C4 olefins in this paper. the catalyst combination and temperature will have an effect on the selectivity of C4 olefins and the yield of C4 olefins. This paper mainly studies the optimal scheme of C4 olefin selectivity, which is solved by correlation analysis. In this paper, a correlation analysis model based on Pearson correlation coefficient was established to determine the selective linear regression of temperature to ethanol conversion and C4 olefins, which conformed to the first-order linear regression model. Then using SPSS software under different catalyst conditions, it is concluded that the conversion of ethanol at temperature and the selectivity of C4 olefins accord with the first linear relationship of positive convergence. Then, by making a scatter diagram, the changing trend of the selectivity of each product is determined, and a reasonable regression is carried out to determine the changing trend of the selectivity of each product with time. Then a control model based on the method of control variables was established. First, the data were preprocessed to separate different types of catalysts in the catalyst combination. The effects of Co loading, Co/SiO2 content and ethanol concentration on ethanol conversion and C4 olefin selectivity were removed by control variable method, and a partial correlation model was established to supplement and improve the control model based on control variable method. The effects of HAP content, the quartz sand content, Co/SiO2 and HAP loading ratio on ethanol conversion and C4 olefin selectivity were determined.

In this paper, the relationship between ethanol conversion, C4 olefin selectivity and temperature is studied. Firstly, the initial three-dimensional surface is drawn with catalyst combination and temperature as independent variables and ethanol conversion and C4 olefin selectivity as dependent variables. Then, in order to more accurately obtain the change rate of dependent variables, take the catalyst combination as independent variables, perform linear, logarithmic and exponential fitting on the other variables respectively, obtain the fitting optimal curve and total parameter table, and draw the final three-dimensional surface diagram to further optimize the fitting results, The differential equation is established to consider the influence of the change of product concentration and reactant concentration on the dependent variable. Then, this paper analyzes the changes of substances at different time points in an experiment, in which the independent variable is time, and carries out linear, logarithmic and exponential fitting with ethanol conversion and product selectivity respectively. The fitting parameters are determined when the degree of fitting is the highest. Thus, the relationship between reactants and products with time in the reaction process can be analyzed.

The existing vehicle network system of Beijing Metro Line 5 adopts the form of RS485 train bus, which is relatively backward compared with the current mainstream bus form of metro trains, and it is difficult to expand and upgrade the network function on the current basis. Aiming at this problem, the domestication of vehicle network system of Beijing Metro Line 5 is studied, and the domestic train network system (TCMS) based on Multifunction Vehicle Bus (MVB) technology is used to replace the original RS485 bus network system. Based on the functional characteristics of the original vehicle network, the existing vehicle space structure and the driver's usage habits, the network system topology, hardware equipment, and the software functions are optimized and improved, and the main advantages of the domestication and replacement of the network system are analyzed. The research results prove that all the performance parameters of the domestic network system meet the requirements of existing vehicles and have better function expansion and upgrading capabilities.

With the continuous evolution of domestic power grid construction technology, my country's power grid has built a diversified composite power transmission and communication network that integrates information networks, computer networks and power networks. The existing electronic information detection technology is mainly for power generation, transmission and transformation. , Lack of technical means to detect power communication equipment. Therefore, this paper proposes an intelligent troubleshooting method for power communication equipment based on PRA technology, analyzes key technologies such as equipment abnormal locations, failure causes, and power outage restarts, and establishes an effective guarantee for protecting power communication equipment based on PRA, and realizes key technologies Autonomous and controllable, improve work efficiency and quality, make digital technology In-depth integration with the business to help the digital transformation of enterprises.