This PDF file contains the front matter associated with SPIE Proceedings Volume 12301, including the Title Page, Copyright information, Table of Contents, and Conference Committee Page.

Vehicle re-identification aims to retrieve and match vehicles under non-overlapping cameras. Although this technology has made some progress in recent years, the problem of vehicle appearance ambiguity caused by perspective shift still can not be well resolved. To alleviate the above problem, this paper proposes a vehicle re-identification method based on keypoint-guided semantic feature alignment and graph matching strategy to enhance complementary information under the transformer framework. The method first uses the pre-trained vehicle attitude detection model to extract keypoints, and proposes an information mapping strategy. Through the coordinate information of keypoints, the corresponding local feature tokens under the transformer framework are extracted and given semantic content to achieve the same semantics. Feature alignment for attribute tokens. Then, a graph matching network is constructed to realize the transfer of semantic information between similar samples, and the expressive ability of features is further improved through the multivariate interaction of information. The proposed model achieves state-of-the-art performance and superiority compared to similar methods on two large vehicle re-ID datasets.

Aiming at the problem of low efficiency of pumping wells in the later stage of high water cut development of the oil field, the reasonable allocation method of motors is studied. Through theoretical calculation and field test, the power calculation model of oil well motors is established, the reasonable load rate limits of different motors are analyzed, the motors are matched step by step on site, and the calculation method model is verified, the optimal configuration of pumping well motor is realized. The reasonable matching technology of pumping well motor further improves the energy-saving theory of pumping unit lifting technology.

Taking the front cover of a certain automotive model as the research object, a new structure scheme of the front cover inner panel was proposed by carrying out the lightweight design and analysis of the front cover inner panel through material optimization and structural topology optimization based on the static stiffness and elastic deformation capacity object. The lightweight designed aluminum alloy component met the requirement of static stiffness, and the static and dynamic deformation resistance was improved while the weight reduction effect reached 36.5%.

In 2020, China has repeatedly expressed its views on "carbon peaking, carbon neutrality" and "building a new power system" on the international stage, and the construction of a new power system is accelerating in China. The important prerequisite for leading the low-carbon development and transformation of the power system is to build a new power system, accelerate the construction of a new power system, by accelerate the conversion of the power structure. It is also an urgent need to accomplish the goal of carbon peaking and carbon neutrality. This paper expounds the background, characteristics, problems and emerging technologies of building a new power system, and provides a reference for the planning and decision-making research of the new power system under the low-carbon environment goal.

In order to improve the problem of slow torque and speed response of permanent magnet synchronous motor, a strategy combining model predictive control and particle swarm optimization is proposed in this paper. By analyzing the error of the input parameters in the model predictive control system, the Particle Swarm Optimization (PSO) was introduced to improve the speed loop. When the system model is changed, the particle swarm algorithm is repeatedly iterated to obtain better PI parameters, thereby improving the given current accuracy of the model predictive control system and improving the stability of the system operation. The simulation results show that this method shortens the adjustment time, improves the running stability of the permanent magnet synchronous motor, and reduces the torque ripple problem and the steady-state error of the system.

By improving the realization method of inverter, a realization scheme of SPWM (Sinusoidal Pulse Width Modulation) controller based on Field Programmable Gate Array (FPGA) is proposed. The design of the controller is mainly composed of a waveform generator and a SPWM generator. It has the function of frequency modulation and amplitude modulation. The dead time can be set according to the hardware environment, and it has the characteristics of closed-loop adjustment. The modular design method makes the whole control system simpler and more reliable. The function and timing simulation of the designed controller is carried out, and the simulation results verify the correctness and feasibility of the design.

The evaluation and evaluation of the company's employees by the personnel department is an important part of the personnel management work. Although there are many general office software systems on the market, these general software cannot meet the business needs of the personnel department due to the different assessment contents and evaluation rules of each company. In view of the business needs of a company's employee assessment and evaluation, this paper realizes the development, testing and deployment of the assessment and evaluation system in a short development cycle based on a professional rapid development platform under the premise of low investment and rapidity. After practical use, the newly developed system has met the actual requirements of the personnel department.

Ball and plate system is a typical multiple-input and multiple-output system, which has the problems of slow response speed, large oscillation amplitude, long oscillation time and other problems of poor stability. In this paper, a fixed time backstepping controller is proposed to improve the existing problems. Firstly, the mathematical model of the ball and plate system is established under the condition of ignoring the interference, and a fixed time convergent backstepping controller is designed according to the mathematical model. Secondly, the backward step method is used to analyze the system, and the command filter is used to process the virtual control parameters for derivation. Then, Lyapunov theory proves the stability of the whole system, which indicates that the system converges in the fixed time. Finally, the simulation results show that the proposed controller not only improves the accuracy and stability of the trajectory tracking, but also ensures that the ball can track the desired trajectory in a fixed time.

Uneven temperature distribution within the heated material is an important factor limiting the widespread use of microwave heating. This paper proposed a dynamic power regulation method to improve the uniformity of microwave heated materials. First, the microwave heating model was constructed by the finite element method (FEM), and the electromagnetic field model and temperature field model were solved to obtain the temperature distribution inside the heated material. Second, the effects of constant power heating methods and dynamic power heating were compared under the same conditions of total energy fed in. Finally, the simulation results showed that the dynamic power heating method was able to improve the uniformity in the horizontal and vertical sections by 69.1%-71.9% and 64.1%-71.8%, respectively, compared to the traditional constant power heating method. The energy conversion efficiency of the two heating methods was almost identical. These results showed that the dynamic power heating method can improve the uniformity of microwave heating and save energy.

To solve the complicated parameter tunning problem of bridge crane sliding mode controller, a sliding mode controller based on IGWO (Improved Grey Wolf Algorithm) was designed. Different from the existing methods, the proposed method can obtain good control effect without going through the complicated manual adjustment process. Specifically, a new switching function is designed to reduce chattering. In addition, the population diversity of traditional gray Wolf algorithms is poor. To enhance its search ability, an IGWO algorithm based on neighborhood learning is proposed and the controller parameters are set with it. The simulation consequences show that the proposed controller can position and anti-swing well.

Wavelets are referred to as "mathematical microscopes" which can effectively extract the effective state information from the signals of mechanical systems. Multiwavelet is made up of two or more scale functions. It is easier to extract from mechanical system signals than single wavelets. However, in the decomposition of the multiwavelet mechanical system signal, it does not have obvious high and low pass characteristics, and its equivalent filter does not have the ideal frequency domain characteristics. As a result, a frequency band mixing problem will inevitably arise during the decomposition of the mechanical system signal. In order to solve the problem and get a purer decomposed signal, this paper proposes a harmonic wavelet-based band aliasing suppression method. The harmonic wavelet is tightly branched in the frequency domain. It has well-defined functional expression and possesses a fully boxed spectrum that allows the mechanical system signal to be processed in the frequency domain to obtain the desired frequency band information. Therefore, the harmonic wavelet transform can be performed on the mechanical system signals of different frequency bands after multiwavelet decomposition to obtain a purer mechanical system signal in that frequency band. The feasibility and validity of the method are demonstrated through the simulated and measured mechanical system signals, and the comparison with the decomposed mechanical system signals without harmonic wavelet transform shows the superiority of the method and its potential value for application in intelligent maintenance and health diagnosis of energy equipment.

Counter the anti-swing positioning problem of two-dimensional underdrive bridge crane, an adaptive super-spiral bridge crane controller based on coupled signals was designed. Firstly, a composite signal is introduced to strengthen the coupling between the trolley displacement and the swing Angle, which improves the control performance of the bridge crane system partly. In addition, the adaptive superhelical sliding mode control can not only reduce chattering of crane control system effectively, but also achieve fast convergence of state quantity, and suppress system matching disturbance well. The stability of the controller is analyzed by lyapunov theory, and the simulation results show the effectiveness and robustness of the proposed method.

Wuhan University of Technology FSC racing car adopts CBR600 four-cylinder engine, which is the decisive factor body of racing power. Considering the Rules of China Formula Student Car Competition (2021), the air intake system shall be composed of air filter, throttle, restrictor valve, pressure stabilizer and injection base. The intake system is particularly important to the analysis of engine power performance and stability, especially the pressure stable box through the simulation optimization of the intake pressure stable box, can increase the intake flow and reduce the intake uneven degree, so that the comprehensive performance of the intake system improved by 34%。

During the driving process of the racing car, it is necessary to complete the prescribed complex actions such as winding the pile and avoiding overtaking. During this process, the brake converts kinetic energy into heat energy, and the temperature of the brake caliper, friction pad, and brake disc rises rapidly, which will cause thermal recession. The braking effect of the car is getting worse and worse. In this paper, the ventilated brake disc is designed, its mechanical structure is analyzed, and the structure of the brake disc is optimized through finite element analysis to improve the heat dissipation effect of the brake disc, thereby improving the braking performance of the formula car as a whole. The research in this paper can make the braking system of the racing car optimized.

To solve the problem that plantar point cloud cannot be directly obtained in the digital twin foot model building system which simultaneously collect the 3D point cloud of foot and plantar pressure distribution data, a plantar point cloud repair method based on plantar pressure distribution data is proposed in this paper. First of all, the contour points of the foot point cloud and the plantar pressure distribution data are extracted, and then the two contours are matched by searching the optimal rotation and translation transformation, secondly, the plantar pressure distribution data is smoothed by using the mean and gaussian filters, then on the basis of the contour of the foot point cloud clipped the plantar pressure distribution data, finally, the plantar point cloud is optimized by constructing a mapping function to convert plantar pressure data into the height value of the plantar point cloud. Experimental results show that the proposed method is significantly better than the planar repair method.

In order to realize the visualization of various index data parameters of college student formula racing, this paper designs a hardware design scheme of formula racing sensor based on AT89C52, in which a variety of sensors are used to distribute the temperature in the engine coolant of the racing car and the torque of each driving wheel, so as to accurately measure the speed of the racing car in various states and ensure the stability of the racing car in the running process, It is convenient for the driver to adjust, correct and judge. In this paper, the signal processing of the sensor, including filtering, noise reduction and other steps, and then the analog signals are ad converted and stored through serial communication. In addition, this paper also designs the upper computer window of LabVIEW for driver information exchange, and displays and processes the parameters in combination with temperature sensor and speed sensor. The sensor data is processed through AT89C52 to correct the parameters of LabVIEW upper computer.

According to the statistics released by China Maritime Search and rescue center, at present, the success rate of maritime search and rescue in China can only be maintained at about 95%, and some ships and personnel in distress will still not be able to get effective rescue 1. In the process of maritime rescue, how to quickly and accurately find the target in distress in the water surface and underwater environment is a crucial step. It is the key to solve the "invisible" problem in the process of rescue and salvage, which is also the primary factor restricting China's improvement of maritime search and rescue efficiency. In this project, the ground effect wing ship is applied to high-sea rescue. The ground effect wing ship is equipped with unmanned equipment to assist search and form a cooperative search and rescue intelligent formation to integrate the search information of UAV and underwater vehicle into the ground effect wing ship in real time for comprehensive information summary, and then the ground effect wing ship assigns tasks according to the actual situation. In order to establish the search and rescue model of air sea level underwater three-dimensional regional information sharing, combined with the 3D visual sea air integrated intelligent multi-level search platform system, the system provides high-precision, high reliability and intelligent comprehensive information support for wading operations, and improves the rescue efficiency and success rate.

In recent years, many wireless network security risks are caused by information explosion and the interconnection of all things. Data mining technology has been applied in the field of wireless communication to solve the problems of wireless network security risks. This paper introduces the construction of wireless communication network security risk prediction model and the implementation of prediction scheme based on recurrent neural network(RNN), which provides an effective method for mobile wireless network security data prediction.

When the high-speed train runs in the track near the electric shield and the protective wall, the edge of the electric shield is less than 3.5m away from the high-speed train, so when the high-speed train passes the electric shield and the protective wall, it will produce a huge air pressure difference between the front and back of the electric shield. The flow field finite element analysis software Fluent was used to analyze the air pressure on the electric shield when CRH3 EMUS passed the electric shield at different speeds, and the distribution law was found, which provided a reference for analyzing the structural mechanical characteristics of the electric shield and the protective wall.

In order to ensure the safety of construction personnel and equipment, it is necessary to comprehensively analyze the mechanical properties of the protective structure in the process of high-speed train passing, so as to provide a scientific basis for safe construction. This paper established the high-speed railroad construction protection structure finite element analysis model, and then set the boundary conditions and calculating load conditions, the static strength analysis and the simulation results show that the local wind load points to the protective wall, under various load cases, protective wall and prevent panels maximum lateral strain on the control panel, and the wind load increases; The vertical strain of the protective wall and the electric plate is basically the same.

In the Chinese Formula Student Electric Racing (FSEC), the monocoque is favored by most teams because of its higher stiffness and lighter weight than the steel tube frame. Due to its stable structure, the carbon fiber load-bearing body meets the lightweight design requirements, but there are also many unstable connections between the chassis and electrical parts of the lifting lugs and the body, and the bolt connection of the truss steel tube frame cannot guarantee the connection strength. Embedded parts are connectors pre-embedded in carbon fiber products, which can be positioned and protected by a single shell. In order to enhance the stability and safety of the racing car, and to maximize the lightweight goal, this paper designs a The monocoque of the composite material was embedded, and its structure and strength were analyzed, checked and analyzed. The results proved that the composite material improved the performance of the racing car and met the requirements of the racing car.

With the expansion of the scale of high-rise buildings, the internal structure is becoming more and more complex. Fire, earthquake and other emergencies easily lead to people in the building unable to choose the best escape path because of panic. The slow escape speed and untimely treatment greatly increase the casualty rate, so it is very important to plan a reasonable escape path. Aiming at the problems of low accuracy and slow speed of traditional emergency escape path planning methods, the building environment model is built, and the distributed firefly algorithm is designed for the new intelligent building platform, and then the path planning method is studied. The algorithm sets the shortest escape path as the objective function. The simulation results show that the algorithm can converge faster and find the optimal path compared with the traditional firefly algorithm, and the accuracy rate is improved by 8% compared with the standard firefly algorithm, which verifies that the algorithm has a guiding role in emergency evacuation guidance.

At present, many FSC racing cars still use truss steel pipe frame. The welding positioning and accuracy of steel pipe frame has always been a difficult problem to be solved by each team. In view of this phenomenon, this paper designs an aluminum profile frame, and designs some general fixtures to improve the local welding accuracy and accuracy. This can save a lot of production time and cost, and greatly improve the production efficiency of weld products. At the same time, a strain sensor is designed at the welding part, and the safety and strain of the welding point of the frame are monitored by using signal processing and signal analysis technology, so as to evaluate and display the overall vehicle condition.

Currently, there are two hot issues on the research of Internet application technology, which are peer to peer (P2P) and Voice over IP (VoIP) based on IP technologies respectively. Recently, there is a trend that the technology of P2PVoIP is integrated by the above 2 technologies. The essay based on the defect of traditional VoIP system discusses on the future development trend of VoIP system and purposes a pure P2P VoIP structure based on Kademlia system. Finally a P2P VoIP system-ppPhone is designed and completed.

In recent years, with the continuous exploration of marine resources, the research of underwater robot has been paid more and more attention. Starting from the development history of underwater robots, this paper introduces the research status and problems to be solved of existing underwater robots, and independently designs an underwater automatic depth determination intelligent robot according to various needs. Through the sealing structure design based on sealing chamber and O-ring, the good water tightness and safety of the device are ensured; The system is equipped with Jetson nano embedded platform, on which relevant programs based on PID control algorithm and yolov5 recognition algorithm are written to ensure that the robot can complete the specified tasks underwater; The underwater robot system uses gyroscope, electronic compass mhr3500, camera and other relevant sensors to collect the relevant motion data of the device, and can feed back the data to the main control board. The main control board matches the corresponding output to the horizontal and vertical propellers according to the relevant algorithm, so that the robot can move at any height; The system uses the combination of software and hardware to ensure the stability of the device during operation.

In order to study rotating speed influence on internal flow characteristics of ESP (electric submersible pump), the electric submersible pump under different rotating speed schemes are simulated. The results of the simulation show that the gas is unevenly distributed in the blade passage and mainly concentrated on the inlet side of the passage near the front cover plate as the impeller rotates. In the blade passage, the gas phase shows periodic aggregation and diffusion. When the impeller rotating speed increases, the cycle of periodic accumulation is shortened. In the meanwhile, diffusion of gas phase in the blade passage is shortened. With the gas phase concentration in the impeller decreases, the overall flow velocity in the blade passage, the inlet pressure difference, outlet pressure difference increase. With the impeller rotating speed increases, the pressure difference between two sides of blade and the fluctuation frequency of blade surface load increase gradually like it.

With the continuous development and progress of the times, college English courses should adapt to the requirements of the development of the information age in terms of teaching concepts, teaching contents, and teaching methods, so as to adapt to the changes of learners' personal needs and social needs. Because the traditional teaching mode ignores students' experience and perception, the development of students' language literacy is limited, which restricts the training efficiency of foreign language talents. The introduction of smart classroom can change this situation. The system can develop one-to-one learning methods based on students' mastery of foreign language knowledge and students' own strengths, so as to help students improve their English learning efficiency. This paper mainly analyzes the overall design of ICAI system on Massively Open Online Courses platform: the system design takes SSH structure as the framework and Back Propagation, and focuses on the basic principle, specific steps of training and heuristic rules of evaluation and application of BP algorithm. Smart classroom can truly realize personalized learning and teach students according to their aptitude, so that learners can get wisdom development. The Particle Swarm algorithm-Back Propagation prediction model is designed for student behavior data mining. The representative data are selected as the input of the neural network, and the course grade is selected as the output of the neural network. The grade prediction error is 12%, which provides a new idea for improving the teaching quality.

In the future, information-based warfare will gradually become swarm intelligence and unmanned, which puts forward higher requirements for battlefield environment situational awareness based on artificial intelligence. In this paper, based on the requirement of the intelligent system of unmanned combat application traction, in view of the limited information interaction and observation ability of Multi-agent Cooperative reconnaissance and perception mission for battlefield environment situation, analysis the current mainstream of Multi-agent Reinforcement Learning method, put forward a kind of Distributed Reinforcement Learning method of online knowledge migration, provide solutions for the problems of the topology changes dynamically of Coordination Graphs and corresponding extensibility, computational complexity problem caused by agent moving in the battlefield environment, effectively improve the collaborative perception reconnaissance search ability of intelligent unmanned cluster.

D2D(Device-to-device) improves communication quality by reusing cellular users' spectrum resources, becoming one of the key technologies for super-large data processing and massive devices accessing in the 5G. Due to load imbalance and co-frequency interference caused by distance change or increase in the number of users, this paper introduces interference management strategy in the 5G scenario where D2D communication and cellular communication coexist. By establishing up and down link resource model of D2D user reuse cellular communication to measure user throughput, and then control transmitting power and the distance between user and base station to reduce the influence of interference on the system. Simulation results show that compared with other traditional algorithms, the network performance of D2D communication based on interference management is significantly improved.

This paper design and developing a quality management system based on ISO 16949 according to the actual situation of companies. Competition in all industries is fierce. The industry has entered the stage of quality management. Enterprises can enhance their competitiveness through improving of product quality. The unification among Autonomous Maintenance (AM), Statistical Process Control (SPC) and Overall Equipment Efficiency (OEE), is suggest to successfully bring about continuous improvement in the production capacity. The aim of this paper is to suggest OEE as the guide to measure the equipment efficiency. In this study, a methodology is presented to analyze OEE of automatic cable machine and efficacy amelioration are perform to practice Define, Measure, Analyze, Improve and Control (DMAIC). An approach is developed to identify and address the losses and failures which are responsible for lowering the OEE. Result of the study presents that the implementation of AM has successfully reduced from 1.75% to 0.23% of the defect rates of automatic cable machine. OEE has been improved from 43.1% to 80.8%.

Artificial intelligence is a hot topic in current information technology research. Undoubtedly, the progress of artificial intelligence technology needs to strengthen the cultivation of talents, and the development of technology is essentially the development of talents. For college teaching, its own teaching mode and teaching philosophy have also been adjusted accordingly. As far as the teaching of string chamber music is concerned, it is a problem that contemporary music teachers should pay attention to the practice reform and innovation of the whole teaching. Influenced by the traditional teaching concept, there is a big gap in the mode and equipment of contemporary music teaching. In today's era of rapid development of social science and technology, the society is gradually moving towards an intelligent development trend. At the same time, the appearance of artificial intelligence system has brought great convenience to the society and met people's daily needs in all aspects. The application of artificial intelligence system in music education not only breaks the traditional music education mode, but also promotes the development of music education in China. Therefore, the department of artificial intelligence has become an important auxiliary means in current teaching work. The teaching equipment of artificial intelligence system effectively combines images, words, sounds and other elements, which can be vividly displayed to students. For the current string music education, the application of artificial intelligence system can make string students absorb and transform music knowledge more quickly, and achieve twice the result with half the effort.

In gene data analysis, feature selection can retain original information, reduce redundancy, remove irrelevant features, and select the feature genes that are beneficial for classification. In order to select better minority genes, a category-based feature selection algorithm is proposed, which mixes univariate and multivariate methods. First, the Relief-F algorithm is used to calculate the score of each feature, and the genes with high scores are retained, and in order to remove redundancy, the gene categories with low classification accuracy obtained under the support vector machine classifier are removed by combining K-means algorithm, and then by Spearman rank correlation coefficient set conditions and thresholds to select genes with obvious differences between categories. Finally, support vector machine, k-nearest neighbor and decision tree classifier are used to test feature subsets. The experiment show that the algorithm can select better genes, achieve gene data dimensionality reduction and improve classification accuracy.

Harris hawk optimization algorithm (HHO) is one of the population-based algorithms proposed by in 2019. It has received great attention from researchers. However, HHO algorithm still has some problems, such as the exploitation ability is too large compared with the exploration ability, which leads to low optimization accuracy, slow convergence speed and so on. Therefore, the collaborative strategy and quantization strategy are introduced. With the support of the two strategies, the algorithm can avoid falling into local extremum in the early phase of iteration and improve its optimization accuracy at the late phase of iteration. Through the test of four representative functions, compared with the other three basic algorithms, the proposed algorithm greatly improves the optimization accuracy and convergence speed of the optimal solution, and the ability to find global extremum is greatly improved.

For garment automatic production line, using the localization of the configuration king software Kingview to design and develop a computer monitoring software, which can complete the real-time monitoring of clothing automatic production line. The monitoring platform combine communication, data acquisition, related equipment monitoring and alarm functions combination. The experimental results show that the computer monitoring software has a good human-computer interaction interface. In addition, it also has reliable performance.

Aiming at the problems of sparrow algorithm (SSA), such as easy to fall into local extremum, uneven initial population distribution and slow convergence in late iteration, a sparrow search algorithm (CMSSA) was proposed, which combined cosine similarity and random multi-chaotic disturbance. The algorithm firstly integrates the reverse learning strategy and initializes the population by using the population uniform adjustment strategy of cosine similarity to ensure the uniformity and richness of the population, so that the algorithm can better search for the global optimal solution. Secondly, a random selection mechanism of multi-chaos local search strategy is used to take advantages from different disturbance states of multiple chaos models. Randomly selected chaos maps are used in each iteration to perturb individuals and help SSA get rid of local extremums. Simulation results show that compared with other intelligent algorithms, CMSSA achieves better results in robustness and optimization accuracy.

An adaptive random reverse learning mayfly algorithm is proposed to overcome the shortcomings of mayfly optimization. Specifically, hyperbolic tangent function is used as adaptive curve to flexibly optimize the personal experience and social experience coefficients of the algorithm, so that the algorithm can better balance the global and local search capabilities through the evolutionary iterations. In addition, on the basis of elite mayfly in each generation, random reverse learning mechanism is used to expand population diversity and speed up algorithm convergency. The calculation results of several test functions show that the algorithm presented has good convergence accuracy and improved algorithm performance.

Battery production is a long and complex production chain. The cell production section has been developed for more than ten years. Basically, the maturity of process and equipment standardization is relatively high, which is suitable for discrete process production. The latter power battery assembly section is an industry that has only developed in recent years. It is mostly in the exploratory stage, which is more suitable for assembly line process production. This paper explores the intellectualization of the battery pack test line. Due to the differences in the structure and assembly of the battery pack, it is the future development trend to configure intelligent equipment such as robot, conveyor line, visual positioning, special equipment, and special detection instruments to complete the battery pack test. This paper will explore the application of intelligent technology in the battery pack test and verification, Explore the intellectualization and automation of assembly and testing production line. Solve the big problem of large number of personnel and low work efficiency for enterprises and provide high-quality test data.

Aiming at the problems of low experimental efficiency and little traceable data in big data in the mechanical and electrical training rooms of higher vocational colleges, the use of Internet technology, hardware-in-the-loop controller (HIL), CAD model data technology, etc. It is an open and shared digital design and simulation intelligent training platform that combines virtual and real. The construction of the smart training platform adopts a combination of simulation experiments and physical training, which can manage the training and virtual simulation debugging of hardware equipment at the same time. The training methods are rich and flexible, which can not only drive students to train independently, but also allow teachers to achieve Intelligent management and control of education provides platform support for improving students' comprehensive skills and teachers' teaching quality, so as to realize the sharing of practical training teaching resources inside and outside the school, and meet the needs of multi-regional, multi-school and multi-disciplinary experimental teaching.

Art is the spice of life and another interpretation of life. Rich emotions make art more vital and appealing. In the art of performance, emotional expression is very important, which can not only enhance the expressive power of performance, but also sublimate the emotion contained in the work. In the eighteenth century, the harp was first played in opera orchestras. From the mid-nineteenth century, the harp was widely used in symphony orchestras. The harp became an important large-scale plucked instrument in modern orchestras. Composers are even more fond of her. No matter in solo or in band, the artistic charm and performance value of harp can not be replaced. Because of its unique charm of timbre, it is very infectious and unforgettable to the audience. This paper tries to analyze the artistic characteristics of this work more objectively, and discusses the creative ideas of combining the work with performance, so as to provide good experience for the performance practice of this work to a certain extent. At present, the recognition and classification system in the field of music information retrieval mainly extracts music features manually, then trains the model with the classifier, and finally uses the built model to recognize and classify the test music samples. At present, however, there is a bottleneck in extracting music features manually. As a new feature extraction technology, deep learning has achieved great results in image processing, natural language understanding and other fields. Therefore, this paper aims to use the powerful feature extraction ability of deep learning to find more suitable music features for music genres. And the problem of identification and classification of traditional Chinese musical instruments has been studied. In this paper, a harp recognition and classification algorithm based on deep learning is proposed. The music samples of each instrument are preprocessed and Mel frequency cepstrum coefficients are extracted, which are then input into the depth confidence network, and then adjusted for training. Finally, the trained model is used to predict the types of test instruments.

The design of this timer, its function is the timing control circuit required to work for 1 second, and then stop for 3 seconds. This is a working cycle. When the timer works, the light-emitting diode lights up, when the timer stops, the light-emitting diode puts out, The number of working cycles of timer is displayed by digital tube. Stay after a period of seven shows on eight. NE555N is used as the clock control circuit, SN74LS161AN is used to count, CD4511BCNis used as the digital tube driver, and the common cathode digital tube is selected as the display. Two ways are used to design, that is, digital electronic technology and PLC control, and the actual circuit board, and have been successfully implemented according to the requirements.

With the rapid development of the Internet of things (IOT), the cloud side-end 4-tier architecture of the IOT is becoming more familiar. With the realization of massive downstream terminal adaptive access and network communication pipeline, the application value of the IOT system is reflected in the cloud platform and application construction. In the business application of generation, transmission, transformation and distribution of power system, there is also the problem of core capacity construction of platform layer. It is necessary to solve the problems of massive heterogeneous data collection, storage and integrated analysis application in the south, and the requirements of various application system access and required business data sharing and call in the north. Through in-depth study of the overall architecture and functional design of the power Internet of things (PIOT) platform layer, this paper provides value for the cloud construction of PIOT.

Internet of things identification analysis is an important foundation of Internet of things and an infrastructure to promote the integration of new generation information technology and real economy. Deepening the "Internet plus advanced manufacturing" development of the Internet of things guidance, put forward to promote identification analysis system construction, build identification resolution service system, use logo to achieve precise docking between global supply chain system and enterprise production system, and promote information resources integration and sharing. This paper studies the application and practice of Internet of things identification analysis industry from three aspects: the importance of Internet of things identification analysis, the application of Internet of things identification analysis, and the problems and measures of pushing the Internet of things identification analysis system.

In the general assembly and development of satellites in China, the mechanical overturning trolley with six degree of freedom is generally used for the installation of satellite decks. In this mode, a lot of equipment is needed, the cost is high, the number of operators is large, the assembly efficiency is low, and the assembly quality is difficult to guarantee. In order to solve this problem, the satellite deck installation method based on industrial robot is studied, and the operation mode of mobile robot combined with simple support vehicle is adopted to reduce the cost of tooling; The robot compliance control technology is used to realize the automatic pin guided assembly of the satellite deck, which eliminates the manual precise alignment adjustment operation in the previous manual adjustment mode, and greatly improves the assembly efficiency. Furthermore, due to the application of compliance force control technology, the assembly force can be controlled within the safe range in assembly, so as to ensure the assembly quality. The research results have been applied to the deck installation of Some satellite. The assembly time of a single deck has been shortened from more than 1 hour to less than 30 minutes, and the number of operators has been reduced from 5 to 3. More than 400 deck assemblies have been completed in practical application, which has won precious time for the intensive development of the satellite system.

At present, there are two methods for the detection of long diversion tunnel, one is manual detection, the other is underwater robot detection. Manual detection is dangerous and costly. Underwater robots can be divided into two types: ROV with cable and AUV without cable. At present, the development of ROV is relatively mature, but the cable also greatly limits its scope of activity. ROV cable is easy to knot and break, and it is also easy to wind on obstacles such as underwater aquatic plants, stones and dead branches, affecting the work of ROV and even making it unable to return. The research of AUV started late and there is no cable constraint, which greatly improves its activity range. However, AUV is limited by energy and information transmission when working underwater. Therefore, it is of great significance to study an underwater equipment which has the advantages of ROV and AUV and can make up for their defects. Based on the detection of medium and long tunnels in water diversion project, an underwater group intelligent equipment is proposed in this paper. The main body of the equipment is composed of ROV and AUV, which cooperate with each other. ROV provides energy for AUV and carries out information interaction with AUV, while AUV is responsible for the main detection work. The technologies used mainly include electromagnetic induction charging technology, SLAM algorithm navigation technology, image analysis technology, laser scanning technology, sonar technology, etc.

In order to make the collector voltage distribution more balanced when multiple IGBT devices are connected in series, so as to realize the better application of IGBT in high voltage, large capacity and high power occasions. Based on the research of various IGBT series voltage equalization technologies, a new circuit combining inductance clamp circuit and active clamp circuit is proposed to realize static and dynamic voltage equalization, and the working principle of each part of the circuit is discussed. Detailed analysis and explanation. And using LTspice simulation software, a new voltage equalization simulation circuit is established. The simulation results show that the new voltage equalization circuit not only achieves a good voltage equalization effect when the IGBT drive signals are synchronized, but also maintains the voltage equalization well when the signals are not synchronized, and can also significantly suppress the overvoltage in the switching state. Feasibility of new voltage equalization circuits.

With the development of Automotive Intelligence and networking technology, vehicle-borne intelligent devices and systems have been widely used, and it can collect operation data in real time. Data has the characteristics of large amount of data, many types and wide coverage. Through the analysis and identification of these operation data, it is of great significance for monitoring vehicle operation status and evaluating vehicle technical performance. Especially the road transport vehicles, which have high running intensity and high safety risk, need to carry out targeted research. This paper collects and combs the operation data parameters of road transport trucks, refines and extracts various signal parameters by using principal component analysis theory, so as to form the data set of on-line monitoring of vehicle operation safety and provide basic data support for subsequent vehicle safety management.

An infrared absorption CO₂ sensor with high stability is proposed. According to the practical application characteristics of long-time operation and unable to be calibrated regularly, the research is focused on improving the long-term stability of the sensor. The sensor uses high-precision machining and high-precision temperature compensation technology to improve the stability of infrared CO₂ sensor. After 1000h experiment, the test results show that the output voltage of the sensor is stable at ±1.53% FS without startup self calibration. Thus, the sensor has the characteristics of high precision, high reliability and good long-term stability.

The temperature of the incoming flow in the intake duct of the engine not only requires accurate measurement, but also has a harsh operating environment and strict requirements on the performance of the temperature sensor. When the measured engine airflow is fast and the temperature is relatively high, the reliability of the thermocouple wire will be reduced. If the temperature sensor is not well sealed, the hot air entering the sensor will cause physical damage to the sensor itself. In this paper, a Z-shaped shell and a sealing structure of glass slurry sintering method are designed to solve the reliability and sealing problems of temperature sensors in high temperature environment. After testing, the basic error of the sensor is better than 1% FS within the measured temperature (-60 ~ +700) ℃. This design has the characteristics of good sealing, strong insulating ability, and is more suitable for high temperature gas measurement, which can ensure that the sensor can perform temperature measurement with high reliability for a long time in harsh environments.

With the development of modern communication technology, people pay more and more attention to family safety detection. At the same time, due to the improvement of people's living standards, there are also various potential gas safety hazards in the family, including flammable, explosive, toxic and harmful gases such as natural gas and methane, which are endangering the health of users. Based on this, this paper designs a home version harmful gas detection system with embedded GSM communication, which mainly includes hardware detection part and software GSM communication alarm part. Through the high-frequency sampling of indoor gas, the timeliness and accuracy of the communication system are realized. For the software part, the user terminal is designed to process data and output at commands to control the operation of GSM communication system. At the same time, the integrated sensor module is used to accurately monitor a variety of harmful gases, so as to ensure the air quality of users' families.

In recent decades, China's economy has developed rapidly, people's living standards have been greatly improved, and people pay more and more attention to medical care, which also promotes the rapid increase in the demand for medical equipment and intelligent medical treatment in hospitals and families. Combined with the above background, this project comprehensively analyzes the intelligent medical products existing in the current market, and designs an intelligent medical monitoring and early warning system based on intelligent perception. It uses a variety of sensors to collect information, uses the Internet of Things to ensure the detection of patients' vital signs by medical staff, and uses intelligent infusion module and infinite vital sign monitor to detect patients' vital signs and daily medical activities. Then collect, process and store the collected data through the Internet of Things, and update the data of the detection system in real time, so as to realize the exchange and sharing of hospital information and resources, so as to greatly improve the efficiency of hospital management and service quality.

Common information model (CIM) is an important part of IEC61970/IEC61968 series standard and grid data standard. In this paper, the method of analyzing and detecting the power grid model file compiled by CIM/XML model standard is studied. The data mining of CIM file is carried out by the structure of Resource Description Framework (RDF) data. Based on the XML files, the CIM files are completely expressed and the CIM/XML model file is parsed. According to the IEC61970/IEC61968 standard, the input CIM model data streams are effectively detected. It depends on whether they conform to the grid standard and have smooth interoperability.

In order to reduce the maintenance cost of China's EMUs, improve the maintenance efficiency, and optimize the maintenance process, a general terminal electrical simulator that can be configured with a debugging interface on the host computer is designed and adapted to different types of EMUs. STM32F105 is used as the control core; STM32F103 is used as the expansion MCU , and the pin expansion is realized through the CAN1 bus, and the GPIO interface and the configurable circuit are combined to realize the collection and control of the digital signal of the electrical debugging interface; the CAN2 bus analog debugging Data; use the wireless Wi - Fi module for data transmission, so as to realize the remote control and feedback signal of the electrical equipment under test.

A certain type of equipment information recording system has many shortcomings such as long information transcription time and poor recording quality, which affects the efficiency of equipment maintenance and security. In response to the above problems, through the study of the system circuit, a convenient information recording system improvement scheme was proposed using modern digital acquisition technology. The overall scheme of the system was elaborated and designed, and hardware development was carried out. The recording device was redesigned using a digital signal processor as the processor, and the system was simulated and debugged. The digital recording device can achieve long time and multi-sortie data acquisition quickly and greatly reduce the data transcription time, effectively ensuring the reliable reproduction of the flight process. The research of this topic strengthens the digital processing capability of this type of equipment information, gives full play to the benefits of this type of equipment control system, and is of great significance to improve the effectiveness of this type of equipment.

Based on the research on the dynamic test results of the piggyback transportation system, this paper analyzes the impact of the vibration acceleration generated during the operation of the semitrailer suspension system under different loading conditions on the safety of the piggyback transportation vehicle. The results show that the maximum lateral vibration acceleration of the semitrailer is about 0.3g, the maximum longitudinal acceleration is about 0.6g. The vibration acceleration of the semitrailer under full load is larger than that of the semitrailer under no-load conditions, and the value is approximately doubled. It shows that when the total load of semitrailer goods increases, the vibration of the suspension system will increase accordingly. If the overload occurs, it may affect the operational safety of the piggyback transportation system.

The trajectory planning optimization can improve the performance of robot, and therefore the artificial life system method is applied in arc trajectory planning of robot. Firstly, the mathematical model of arc trajectory planning of robot is analyzed. Secondly, the basic theory of artificial life system is studied, and the corresponding algorithm procedure is designed. Finally, the simulation analysis is carried out, and the optimal trajectory of robot is obtained, and results shown that the artificial life system optimization method can be applied in the trajectory planning of robot effectively.

With the rapid development of 5g large-scale commercial and edge computing, cloud edge collaboration technology is widely used in various industries. This paper studies and analyzes the key technologies of cloud edge collaboration in realizing resources, data, applications and services at different levels of IAAs, PAAS and SaaS, and introduces the application scenarios of cloud edge collaboration in many fields such as industrial Internet, smart transportation, smart city and medical imaging.

In order to analyze the modern pharmacological and clinical effect of Schisandra chinensis effectively, the big data is used to analyze it. Firstly, modern pharmacology of Schisandra chinensis is analyzed. Secondly, the big data analysis algorithm is designed by combining possibility fuzzy C-means algorithm and particle swarm optimization. Finally, a case study is carried out, results show that the proposed method can obtain the pharmacological and clinical effect of Schisandra chinensis correctly.

In order to improve the efficiency of production and commissioning of high-speed EMU and the ability of statistical analysis of production data.Based on Web mobile communication technology, taking the commissioning production line as the test carrier, the SSM framework is used to realize the design of digital commissioning and management platform, and create a set of commissioning scenarios applied to high-speed EMUs, which can realize the functions of mobile operation, digital commissioning and management, big data analysis and so on.Using the characteristics of automatic inspection, automatic judgment and data management of the platform, manual judgment can be reduced and operation efficiency can be improved.At the same time, real-time monitoring, quality control and intelligent adjustment of the whole commissioning process can be realized through the platform storage and management function.Through field application verification, the platform can effectively solve the waste of human and material resources in the commissioning process, At the same time, it improves the production efficiency and product quality, and provides a portable intelligent debugging and management scheme for high-speed EMU manufacturing and maintenance enterprises.

Swarm intelligence is inspired by Ant Colony, Bee Colony and Bird Flock from nature and it builds distributed and scalable algorithms which can employ a large amount of homogeneous and straightforward agents to accomplish the vast and complicated tasks that cannot be fulfilled by a single individual agent. Swarm robotics is a typical application of Swarm intelligence in which researchers all over the world are investing heavily to build the initial theoretical framework and effective models. This article mainly summarizes the latest development of Swarm robotics and attempts to understand and analyze its applications especially in gathering in a team, spatial dispersion, combination forming, combination forming and group movement.

With the rapid development of information technology, the application of data analysis and processing technology is becoming more and more extensive. In the data processing stage, necessary tools and technical means should be used to identify and process erroneous or missing data. The problem of missing positioning information data and picture additional information data will occur in the data recorded by the driving track of the car. This paper discusses the complementation methods for the above data information missing, and uses fragment interpolation, Lagrangian interpolation and other methods to achieve the complementation of missing data.

Image denoising is a classical problem in the current field of computer vision. The goal of the task of image denoising is to use techniques to preserve as much clear detail of the original image as possible when the image has external noise. The essence of the image denoising process is to reduce the noise in the digital image and to recover and reconstruct the original clear image. The reason for image noise is that during image transmission and acquisition, the integrity of the image cannot be guaranteed due to environmental, acquisition equipment, human and other factors, so the image will inevitably be damaged by different degrees of noise. In medical, military, and optoelectronics fields, there is an extremely high demand for image realism, so the task of image denoising becomes very important. In this paper, we will discuss a history of image denoising techniques and analyze the denoising methods into traditional denoising and deep learning based denoising.

With the arrival of the information age, we can see images in every corner of our life, and images convey a lot of information to us. But not all images are clearly visible. And the image obtained after image transmission is often more blurred than the original image or the lack of information in the image. Therefore, the image we see may have been more or less disturbed, resulting in some damage to the image. This interference is mostly composed of noise, Such as AWGN and Poisson noise and so on. Aiming at the influence of noise on the image and preserving the integrity of image information to the greatest extent after eliminating noise, wavelet threshold denoising is one of the most important methods of image denoising. It is a relatively simple and less computational wavelet denoising algorithm. This paper summarizes the related research of wavelet transform in the field of image denoising. Firstly, the general principle of wavelet threshold denoising is explained. Furthermore, three different wavelet threshold image denoising methods are described. The characteristics and problems of three kinds of wavelet threshold denoising are summarized. Finally, the prospect of wavelet transform is also given.

Aiming at the shortcomings of manta ray foraging optimization, such as facile to sink into local optimal and sluggish convergence velocity, an improved manta ray foraging optimization for high-dimensional conundrums was proposed, which is named Chaotic Manta Ray Foraging Optimization based on Elite Hybridization and Reverse Learning (ERMRFO). Firstly, the initial population is initialized by Circle mapping, which makes the initial population traverse the space more randomly. Secondly, a multi-elite hybridization strategy was used to improve the poor particles. Finally, an adaptive reverse learning strategy combining Lévy flight and Gaussian variation was used to increase population diversity. Five high-dimensional test functions are selected for experiments, and the results show that the improved algorithm has better ability of escaping local optimum and faster convergence speed.

With the rapid development of the times, big data has started to affect all aspects of our life, and more and more products in industries are becoming carriers of big data. For example, CNNs with more implicit layers are proposed to meet our needs, although they have a more complex network structure. Compared with traditional machine learning, it has more powerful feature learning and feature representation ability, which can better meet the needs in the era of big data. Since its introduction, deep learning has achieved remarkable results in several large-scale recognition tasks in the field of computer vision. This paper presents a review of target detection methods based on a survey of domestic and international literature on the subject. Firstly, the current state of research in the field of target detection is introduced. Next, the development and rise of deep learning and convolutional neural networks are briefly introduced, and the basic structure of convolutional neural networks, convolutional feature extraction, and pooling operations are outlined. The analysis and discussion of deep learning-based target detection algorithms are focused, and the current shortcomings are pointed out. Finally, the applications of deep learning-based target detection and the future development directions are summarized.

This article through the establishment of quantitative model and the corresponding prediction model orientation of network information mining, through the database to adjust some of this information and integration, the database of the directivity classification work, can according to the requirements in terms of system analysis of data at the same time, through the data to forecast its trend, get the best prediction model and the prediction results. In this paper, a data set is established based on the stage new information indicators of database information and the corresponding rise and fall trend broken line graph, and a quantitative Model and trend prediction system based on two-brain Strategy Model are established to predict and evaluate the situation of directional information. And then to predict the trend of fluctuations is relatively accurate prediction results quantitative optimization model is established, the grid strategy and hedge strategy is adopted to improve the decision-making and judgment, decision variable coding as operands, can converge to the optimal solution with large probability, data analysis and forecast the dynamic change of process, with different weights allocation scheme to give corresponding risk prediction.

With the progress and development of science and technology, mobile phone signaling data and new statistical methods bring new research methods and directions for population statistics. Regional grid algorithm is an algorithm that further refines the sector, divides the sector into a single and equivalent grid, counts the population in each grid, and further improves the statistical accuracy of population density and distribution. K-means clustering algorithm adopts Euclidean distance as the evaluation index of similarity, which is a typical distance-based clustering algorithm. Based on the instant call record data of mobile communication operators, using the location, time, personal attributes and other information in the mobile signaling data, combined with the base station engineering parameters, and based on the demographic behavior characteristics, this paper judges and measures the mobility of the population from the user behavior represented by mobile communication big data. Regional grid algorithm and K-means clustering algorithm are used to process mobile communication data, and the results are classified, so as to analyze the permanent population size of each region in Wuhan. Through case analysis and verification, the error of classification results is far less than 0.05, and the classification is basically in line with the population distribution of each region in Wuhan, which can verify the validity of the results.

Shopping review information is a comprehensive evaluation of the quality, price, after-sales service and other dimensions of the product after the consumer purchases the product. This information implies the consumer's satisfaction with a certain product and emotion for a type of product. This plays a vital role in the later product push of merchants and private customization services. Therefore, it is particularly important to intelligently analyze consumer evaluation data and perform fine-grained sentiment analysis on product characteristics to help users understand a large number of consumer evaluation data. Therefore, in this paper, combined with the data background, in the implementation process of SVM, the threshold between the support vector and the point vector is fully utilized for denoising, which improves the accuracy of the system. In terms of operation, the calculation time is saved by decomposing the original point vector, and the time efficiency is improved.

Heavy-duty wreckers generally use special operating devices such as lifting mechanisms and crane mechanisms to achieve the functions of supporting, lifting, and centering. The telescopic box beam is the most important working part of these mechanisms. Starting from the deflection differential equation, this paper derives the accurate recurrence formula of the out-of-plane instability critical force of the crane telescopic boom. Through the simulation analysis of ANSYS software, it is checked whether the strength of the telescopic box beam meets the design requirements.

With the development of machine vision and various image generation algorithms, various physical signals can be transformed into corresponding visual images, and valuable information can be obtained by analyzing and processing the visual images. In view of the instability of bearing vibration signal and the development trend of on-line detection of fault diagnosis, this paper proposes a bearing fault diagnosis method based on symmetrical polar coordinates and gray level co-occurrence image processing algorithm. When applied to the diagnosis of bearing wear fault, it can maintain a better diagnosis effect under the instability of bearing vibration signal, which is more intuitive and faster, which is conducive to on-line fault diagnosis.

In order to solve the problem of insufficient ability of campus achievement analysis, this paper puts forward a complete set of intelligent campus achievement analysis methods. Firstly, the method collects and preprocesses the students' achievement data of each subject, uses the Apriori association rule algorithm to select the main pre courses that affect the students' comprehensive assessment results, and filters the irrelevant pre courses; Then, the neural network algorithm is used to construct the score prediction model, convert the scores corresponding to the main pre courses into feature vectors, take the students' comprehensive examination scores as labels, and use the feature vectors and labels to train the score prediction model; Third, use the trained model to predict the students' comprehensive examination results; Finally, if the predicted comprehensive assessment result fails, send an early warning notice of the failure of the comprehensive assessment result to the student in advance, and remind the student to strengthen the study of the main pre courses. Experiments show that the method proposed in this paper performs well in achievement analysis and prediction, and is conducive to promoting the improvement of students' achievement.

All kinds of distributed devices and intelligent devices are connected to the power system, which makes the power system more and more sensitive to the fluctuation of power, which leads to the identification and processing of power quality disturbance (PDQ) becoming more and more important. Aiming at the problem of composite disturbance classification and identification with multiple single power quality disturbances, a composite power quality disturbance identification method based on improved S-transformation is proposed in this paper ¹. First, for higher time-frequency resolution, an improved S-transform with new window width adjustment coefficients is introduced. Then use S transform and wavelet transform to extract the features of disturbance signal, and compare the effect of three methods on feature extraction. Finally, the simulation results show that the method can effectively classify the interfering signals, and the energy concentration and resolution are greatly improved.

In order to solve the lack of edge precision of traditional Canny algorithm and the defects of Gaussian filtering and gradient calculation, we propose an algorithm based on wavelet transform to realize image denoising, which improves the detection accuracy by replacing the original gradient template algorithm with the Scharr operator in four directions, and then uses the reference threshold, combined with the Zernike moment algorithm to perform sub-pixel pinpointing of all possible edge points thus The edge detection is achieved. It is confirmed by many experimental results that the algorithm improves the image edge positioning accuracy and processing efficiency, and has better anti-noise ability than the traditional Canny algorithm and Zernike moment algorithm.

The current common 3D human pose estimation algorithms achieve good results in representation learning, but there are still problems such as poor estimation accuracy at the human skeleton joint points, so how to use redundant 2D pose sequence spatio-temporal information from monocular RGB images to estimate the human pose in an effective way is a research challenge, this paper proposes a 3D human pose estimation based on spatio-temporal multi-feature fusion network. The algorithm specifically combines a spatio-temporal multi-feature fusion hierarchical method of image appearance information and motion timing information, which uses a compact convolutional neural network to learn spatio-temporal information to model 2D joint point position information as 3D joint point position. Experimental results show that the proposed method currently achieves more advanced end-to-end pose estimation accuracy and does not require any post-processing stage of the pose optimization method, and the experimental results show that the pose estimation obtained in this paper is effectively improved in terms of average accuracy, proving that the method in this paper can effectively improve the accuracy of human pose estimation.

This paper mainly proposes a message push method suitable for iOS mobile terminals. In this method, APNS is used for offline iOS terminal message push and MQTT is used for online iOS terminal message push. This method can reduce the number of messages sent through APNS. It can also solve the disability of storage-and-forward and the dissatisfaction of delivery rate under large amount of concurrent messages or over extent of message length. At the same time, different push strategies can be selected according to the online state of the target terminals. The experimental results about bandwidth and message length show that when the concurrency of pushing messages is large or network resources are insufficient, this strategy can dynamically balance the load or expand the bandwidth of network devices and servers to increase throughput. This strengthens the data processing capability for message push and improve resource utilization.

According to the primary and secondary loop resonance theory of Tesla transformer, the output characteristics of nonideal Tesla transformer under different coupling coefficients are studied. The MATLAB software was used to calculate the mutual inductance of the primary and secondary coils of different physical structures. The optimal primary and secondary coil structure and capacitance and inductance parameters corresponding to the high frequency AC voltage output with frequency of 300 kHz and amplitude of 255.9 kV were obtained. Then the model was simulated and verified by the Simulink module. Based on this, the physical prototype of the Tesla transformer was built. The output voltage of the frequency of 298.7 kHz and the amplitude of 241.1 kV is obtained by measurement, and the energy transfer efficiency was about 27.1%. The output voltage amplitude of the physical prototype deviates from the simulation value by 6.3%, which basically meets the design requirements. The Tesla transformer designed in this paper has certain application value for high voltage test, insulator string and other insulating materials in the performance detection under high frequency and high pressure environment.

To solve the problems of missing information and masking features in traditional pooling methods, a new Gaussian stochastic pooling (GS-pooling) method is proposed in this paper. The elements in the pooling domain are assigned corresponding weight values, and then the multinomial distribution sampling is sampled according to the weight values. Finally, the pooling result is obtained according to the corresponding Gaussian kernel. We conduct comparative experiments on the MNIST, CIFAR-10, and Market-1501 datasets. The results demonstrate that our pooling method achieves the best performance.

The rapid development of Machine Learning (ML), Machine Vision and imaging technology has greatly promoted medical imaging and Intelligent Medical Engineering. Radiomics combines medical imaging with Big Data, Machine Learning and other technologies to realize the diagnosis and treatment of Corona Virus Disease 2019 (COVID-19) by obtaining and analyzing lung image characteristics. This paper systematically reviews the realization process of radiomics in COVID-19, the latest research on radiomics in COVID-19's diagnosis, classification and prognosis, as well as the problems and challenges faced in this research field. By and large, radiomics provides great potential and application value in the diagnosis, classification and prognosis of COVID-19. It makes up for the deficiency of doctor's diagnosis and Reverse Transcription-Polymerase Chain Reaction (RT-PCR) test, and provides an effective and feasible method for improving the diagnosis and treatment level of COVID-19 with low cost, high efficiency and accuracy.

County Energy Internet can solve the problems of low comprehensive energy efficiency and insufficient utilization of clean energy in rural energy system. Reasonable and scientific evaluation is an important condition to promote its steady development. This paper constructs a comprehensive benefit evaluation system of county energy Internet covering reliability, economy, ecology, elasticity and sociality. Then, considering the relationship between the indexes, the G1 method is modified to obtain the subjective weight, the coefficient of variation (CV) method is used to obtain the objective weight, and the minimum information identification method is used to obtain the combined weight. Finally, the idea of approaching the absolute ideal solution is introduced, the absolute ideal solution is set, and the generalized grey absolute correlation evaluation method is formed. An example is used to verify the effectiveness of the county energy comprehensive benefit evaluation method.

Based on Chinese human body characteristics and seat comfort evaluation model, this paper investigates the intelligent regulation model and validation method of car seat comfort applicable to Chinese human body. Consumers input the basic human characteristics parameters (height, weight, gender, age), and the seat position information (slide position, high adjustment position, angle position) can be calculated by using the regulation algorithm, so that the system can control the adjustment state of the seat. In the model building stage, the BP neural network algorithm, Gaussian regression algorithm and SVM prediction algorithm are studied in this paper, and the comparative analysis reveals that the prediction error and the stability of the prediction results can be optimized by using SVM algorithm for the prediction of the three position parameters, which can achieve the optimal prediction effect. In order to verify and optimize the model, a universal body platform is developed in this paper, which combines software and hardware together to verify the effectiveness of the regulation algorithm, and also feeds back and records the micro-secondary regulation data of consumers for subsequent optimization of the model.

Target of underwater environments classification and distinguish using active sonar is very complex signal processing work, therefore, many research institutions in the world study the classification and recognition methods of underwater small targets using active sonar.The recognition methods of static underwater small targets can be roughly divided into imaging method and echo method. The imaging method is based on high-frequency narrow beam to obtain the acoustic image of the target and its shadow while suppressing the reverberation interference, so as to realize the classification and recognition of the target. Imaging sonar has high working frequency, limited sonar range, and can not effectively identify false targets with similar geometry. Bright spot model and resonance scattering theory are important theoretical basis of target echo method. The two theories agree that the echo is the result of the interaction between the incident signal and the underwater target, which must contain the characteristics of the target, so it is feasible to detect the target by analyzing the characteristics of the echo signal. However, in the actual underwater target detection process, due to the weak target echo signal, complex seabed environment and target geometry, uncertain sound propagation conditions and so on, the target recognition based on echo method still has great challenges.

Due to the short duration and low intensity, the efficient feature learning is a big challenge for robust facial microexpression (ME) recognition. To achieve the diverse and spatial relation representation, this paper proposes a simple yet effective micro-expression recognition method based on multiscale convolutional fusion and capsule network (MCFCN). Firstly, the apex frame in a ME clip is located by computing the pixel difference of frames, and then the apex frame is processed by an optical flow operator. Secondly, a multi-scale fusion module is introduced to capture diverse ME related details. Then, the micro-expression features are fed into the capsule network for a good description about spatial relation. Finally, the entire ME recognition model is trained and verified on three popular benchmarks (SAMM, SMIC and CASMEII) using the associated standard evaluation protocols. Experimental results show that our method based on MCFCN is superior to the works based on pervious capsule network or other state-of-the-art CNN models.

With the deepening of reform and opening up, the aquaculture industry has developed rapidly, and the demand for aquaculture risks has also changed. Due to the large individual differences of aquaculture groups and the high difficulty of research, there are few articles exploring this field. With the further development of science and technology, machine learning has become the mainstream of current analysis technology and has extended to the field of sentiment analysis. To this end, this paper proposes an insurance demand analysis method based on CNN. Combined with the sentiment dictionary method and the construction of a co-word network, an insurance demand sentiment analysis model is established. Through the method of cluster analysis, the survey data was sorted and analyzed, and the results proved that farmers with different identities have differences in the demand for insurance for crayfish breeding.

In this paper, we try to find two methods based on deep learning and machine learning to detect phishing websites from those legal ones. In terms of machine learning models, first, both phishing and benign URLs of websites are gathered from online resources, and required URL features and website content-based features are extracted and selected to form the data set. 5 different training methods are used and analyzed. In terms of deep learning model, a character - embedded Bidirectional LSTM is used for training. And the model is testified then.

[Objective] According to the shortcomings of existing detection techniques, this paper proposes a new method based on SVM (Support Vector Machine) and weighted Euclidean distance detection to improve the detection accuracy of sprouted potatoes. [Method] Firstly, the original image of potato was obtained based on industrial camera, gray image and median filtering were used to ensure the image quality of the image, then B and H component training SVM classifiers were extracted in RGB color space and HSV color space respectively. After this, the well-trained SVM classifier was used to segment the potato image and background. Finally, the weighted Euclidean distance and morphology method were adopted to detect and mark the potato germination site. [Results] Under the help of the MatlabR2014a software platform, the weighted Euclidean distance and the traditional Euclidean distance method were employed to test the I and II potato samples. The experimental results reveal that the average recognition rate of the weighted Euclidean distance method is 90.6%, compared with 88.4% of the traditional Euclidean distance which indicates that the recognition rate of the method in this paper is higher, and the detection effect on the germinated potato is better.

This paper reconsidered the Two-Dimensional Bin Packing Problem from the perspective of practical application based on actual needs of small furniture companies in China. Although there have been many researches on this problem for many years, most of the existed researches put their foremost attention on developing an algorithm with low complexity or approximately best performance which means that it can nearly work out the best solution most of the time. This is the common situation when we do research on such problems, but unfortunately it will leave out abundant practical needs. For example, the computational resources of those small companies are seldom, which makes it impossible to use algorithms that almost gets the best solution but takes much time. The significant goal is to find out a solution satisfactory enough within a short time, so some complex algorithms which aim to get a nearly best result aren’t suitable here. Meanwhile, the cutting times are also a rather critical matter for them. Cutting too many times will cause high wastage to machines and is therefore unaffordable for small companies, but this was often ignored in previous researches especially in classic meta-heuristic algorithms since they frequently lead to complicated packing patterns with massive cutting times. This paper reviewed previous related researches and summarized their shortages and some necessary characteristics of a practically efficient algorithm. Based on classic algorithms, it gained a two-stage heuristic algorithm which performed better than current used ones in those small companies, as well as meets their special needs stated in this paper which is a rather important issue. And this algorithm itself may be used to meet needs of those factories or be used as a starter of the further optimization search.

For large delay problems that exist in industrial process control we put forward the improved model-free adaptive control algorithm. The advantage is that the model does not require the specific object ,the system can be controlled just by the output and input. In the improved algorithm by introducing a constraint entry of the rate of change of the output with large time-delay to reduce the impact of large delay on the system. Verify that the improved algorithm can control the delay system effectively by MATLAB and SMPT-1000 simulation in the system response speed and system stability.

Aiming at the situation that CFNet, the back propagation filter tracking algorithm based on the siamese network, is likely to cause the model drift tracking effect to decrease when it encounters the interference of similar objects or the background information is similar to the foreground target, a siamese network target tracking algorithm fused with semantic feature network is proposed. In image processing, through the deep network of deep convolutional neural network, rich semantic information can be extracted. These semantic information can cause similar interference, motion blur, severe target deformation, etc. In situations, it is very useful to identify the target. In the proposed algorithm, a semantic feature network is added to the original network structure of CFNet, which is complementary to the appearance feature network of CFNet. The training of the two feature networks is independent to maintain the heterogeneity of the two features and obtain their respective response maps. Later, the fusion is performed by calculating the confidence of the two response graphs, which improves the discriminative ability of algorithm. Tests show that, compared with the CFNet tracking algorithm, the proposed algorithm performs better in situations such as analog interference or the background information is similar to the foreground target.

Aiming at the imbalance of the midpoint voltage of the DC bus of the T-type inverter, this paper proposed an optimized fuzzy VSVPWM control strategy based on ant colony algorithm. Firstly, based on the traditional VSVPWM modulation strategy, this new control strategy refined the sector distribution, redistributed the time sequence of output voltage vector, and increased fuzzy control factor D to control the action time of vector. Then, the three memberships of triangular membership function, fuzzy control rules, as well as quantization and scale factors are simultaneously optimized by ant colony algorithm. Moreover, in the path construction, the algorithm optimized the probability transition function and effectively prevented the ant colony from falling into the local optimal state. Finally, the simulation results proved that the parameters of the fuzzy controller optimized by the ant colony algorithm were applied to the T-type inverter system, so that the point potential fluctuation difference in the system is controlled within the allowable range, and the optimized system prove to be great in dynamic response and static properties.

An algorithm is presented for the special solution, it can be obtained within finite iterative steps for a special matrix. Its best approximation solution of a given matrix can be obtained in this solution set, and the minimum norm solution can be given when a special initial matrix is chosen. The examples show that the algorithm of the paper is effective.

There are many health status parameters for aeroengine, leading to partly information overlap. The accuracy of commonly used evaluation methods is seriously restricted, due to the reliance on human subjective experience. The PCA-Kmeans combination algorithm for aeroengine health status evaluation is constructed, the specific steps are proposed, and the result is evaluated and verified by RBF neural network. Taking NASA public dataset as an example, the experimental results suggest that the PCA-Kmeans combined algorithm is well suited to health status clustering based on PCA dimension reduction, and is basically consistent with the evaluation results of RBF neural network. It provides a reference for large scale objective evaluation of aeroengine health status, comprehensively mastering the overall performance degradation of engine and scientifically making maintenance decisions.

The temperature control of heat exchanger is characterized by complex, nonlinear and hysteresis. In order to improve the temperature control effect of outlet working medium, a temperature control method based on improved sparrow search algorithm was proposed to analyze the internal dynamic heat transfer process by establishing the distributed parameter model of heat exchanger. The parameters of PID controller are adjusted in real time by using the good searching performance of improved Sparrow algorithm to achieve the purpose of adaptive control. The temperature control system is simulated under different working conditions and compared with traditional PID control. The results show that the algorithm can reduce the outlet temperature regulation time, small overshoot, high control precision and strong anti-interference ability.

With the development of Automotive Technology in the direction of intelligence, networking and electrification, the number and complexity of on-board sensors and electronic devices are gradually increasing, so that the traditional automotive maintenance business is no longer limited to mechanical fault maintenance, and then turns to the diagnosis and maintenance of automotive electronic systems and components. According to the development trend of modern automobile technology, this paper puts forward a calculation method of automobile reliable life to realize the real-time online evaluation and prediction of vehicle reliable life. It is helpful to scientifically judge the technical status of vehicles in use, and plays an important supporting role in reasonably carrying out vehicle maintenance, improving vehicle owners' habits, and prolonging the reliable service life of vehicles.

In order to solve the problems of permanent magnet synchronous motor under disturbance conditions, the motor overshoot is too large and the anti-interference ability is not strong. Based on Matlab\Simulink, a vector control system model of PMSM based on active disturbance rejection control and double closed-loop PI control is established. Under the same motor parameters and simulation conditions, the speed, electromagnetic torque and three-phase current response of permanent magnet synchronous motor under ADRC control were studied and the simulation results were compared with the traditional double closed-loop PI control system. The experimental results indicate that the ADRC has obvious advantages in restraining speed overshoot and improving speed response of PMSM. It has better robustness and control performance.

[Purpose/Significance]The contents, description and storage format of the "12316" data resources under the Ministry of Agriculture are not uniform, and the sharing of resources is difficult, which is not conducive to the "12316" business services. It is necessary to develop a set of effective sharing construction methods to meet the needs of industry data resource publishing and sharing, and to realize data exchange with different fields. [Method/Process] The scheme puts forward the "shared resource semantic description model" exchange mode, combined with the "12316" sharing platform construction, gives the "12316" exchange mode design ideas and implementation, and defines the data sharing metadata set among them. [Result/Conclusion] This scheme provides a standard, flexible and concise description method for the construction of "12316" sharing platform, and proposes an effective information integration interaction mechanism for the distributed, multivariate and heterogeneous "12316" data resource sharing exchange.

The characteristic of far-field radiated noise of underwater vehicle is an important index to evaluate its stealth performance. Therefore, the level of radiated noise measurement technology has become a key factor affecting the ability of target acoustic recognition and stealth. This paper summarizes and analyzes the related progress of measurement technology of underwater vehicle radiated noise at home and abroad, including foreign PVLAMS, STAFAC measurement system, vector hydrophone measurement technology, civilian ship radiated noise test and standard, far-field estimation method based on actual measurement and model, etc. It provides a comprehensive reference conclusion for the research of radiated noise measurement technology and engineering practice.

The safe and stable operation of the power system is related to the quality and efficiency of national production. With the mature development of the detection technology of large power equipment, online monitoring has gradually become an important means of detecting the insulation status of equipment. Casings, as the core component of power transformer, bears the electromagnetic environment of high voltage and strong electric field in operation, and has multiple interfaces (oil-air interface). The complex working environment of casing leads to the occurrence of faults, so it has high engineering value to study the on-line monitoring system of Casing. Firstly, this article designed casing shrinkage ratio model, and make tail damp and metal particle discharge fault artificially. Secondly, we extract five kinds of characteristic value in the laboratory including the capacitance variation, the dielectric loss tangent, the maximum discharge amplitude, the maximum amplitude and the average discharge pulse number. At last, we used the analytic hierarchy process (AHP) to build the fault tree model for fault diagnosis and classification of two kinds of discharge.

The matching and fusion of insulator multispectral images is an important prerequisite for the operation condition monitoring and fault diagnosis of transmission line insulators using multispectral cameras carried by drones. There are a lot of image matching algorithms, and each algorithm has its advantages and limitations. At present, no scholar has proposed a decision model for insulator multispectral image matching. The drone and the multispectral camera were used in this paper to shoot insulators at different shooting distances and angles to obtain multispectral images of insulators. Comprehensively considering the characteristics of insulators, the image matching algorithms based on feature point detection were analyzed and compared, and the region of interest mutual information method was used as the evaluation system of matching degree. Finally, the decision model of insulator multispectral image matching was constructed, which provides conditions for the subsequent insulator multispectral image research.

Thickness accuracy is one of the important quality indexes of strip products. With the development of social economy, the quality requirements of strip products are higher and higher, and the requirements for precision are also higher and higher. Many original production lines have been unable to meet the requirements, the existing unqualified production line transformation becomes very urgent. A control scheme of thickness control system of cold rolling mill based on Siemens S7-400PLC is presented. How to improve in a stable in the process of rolling steel strip thickness control precision, and improve the yield of plate and strip, is today a most popular courses in the field of the strip thickness control, aluminum plate is in hot rolling process of cold rolling mill or coil as wool roll casting process, a multiple passes, the rolling rolling out the qualified finished product coil, A rolling mill that provides billets for sheet shearing, stretching and bending straightening and aluminum foil deep processing. Artificial neural network technology has been successfully applied in many fields of automation system, such as system modeling and identification, robot control, system pattern discrimination, system fault diagnosis and so on. In the lag link of AGC system, the self-learning ability of the neural network is used to adjust the integral controller parameter value online. The weight of the neural network is corresponding to the integral parameter value, and the integral parameter can be adjusted according to the dynamic characteristics of the controlled system, thus improving the adaptive ability of the regulator.

For special systems such as nonlinear large time delay, there is a problem that the control effect of conventional PID controller is not ideal. Therefore, a fuzzy-PID control method is proposed by combining the advantages of fuzzy control and conventional PID control. By combining different basic features, this method makes the meaning of alarm clearer, and has better scalability and success. At the same time, it provides a good platform for information interaction of anomaly detection. The electronic devices in the computer control system mostly work in high-voltage environment. The working intensity of the sensor is very high, but its high precision determines that its fault characteristic pole signal is easily interfered. In the hydraulic control system, the technical state of the sensor in the detection link directly affects the working performance of the system. Fuzzy controller can control complex and unclear systems simply and effectively. Therefore, combining the advantages of traditional PID and controller, and considering the characteristics of fuzzy control, the idea of fuzzy PID and controller is put forward. The simulation experiment on the actual power load data proves that this method can accurately detect the abnormal data online and provide correct abnormal data correction.

In the rapid development of social economy, the development level of China's aviation industry is also improving, which plays a positive role in promoting the development of aviation high-tech and domestic and foreign transportation and other fields. The design of aircraft control system is a difficult point in the study of aircraft flight management system. Therefore, on the basis of understanding the current development status of China's aircraft control system, this paper carries out empirical analysis according to the design of wireless sensor network and its application in the control system. The final results show that the air vehicle control system based on wireless sensor network has unique advantages, both track tracking effect and anti-interference performance have been improved, and more effective countermeasures to solve practical design problems have been mastered.