Transmission Line Foreign Body Fault Detection Using Multi-Feature Fusion Based on Modified YOLOv5Wang, Qianyue; Si, Gangquan; Qu, Kai; Gong, Jiahui; Cui, Lele
doi: 10.1088/1742-6596/2320/1/012028pmid: N/A
The particular work environment of transmission lines makes it liable to malfunction because of the foreign bodies attached to the power equipment. However, traditional detection methods are difficult to identify the exact category of foreign bodies and ignore the relationship between a foreign body and power equipment, which is correlated to determining whether there is a fault or not. In order to detect and distinguish the foreign bodies and give fault warnings automatically in transmission line patrol with high accuracy and efficiency, we improve the original YOLOv5 object detection model and propose a novel fusion detection model using multi-scale appearance and relationship features. The proposed model is capable of transmission line foreign body category detection and fault determination. By comparing with the original YOLOv5, the proposed model shows better accuracy and achieves the strong ability to give the fault determination simultaneously, which makes it convenient for power system operators and protects the safe and stable operation of the power system.
Lossless Compression and Optimization Method of Data Flow Efficiency of Infrared Image Depth Learning Model of Substation EquipmentLiu, Guangwei; Hao, Zhen; Niu, Zheng; Mu, Ka; Ma, Jixian; Zhang, Wenzhe
doi: 10.1088/1742-6596/2320/1/012026pmid: N/A
Infrared detection of substation equipment has become one of the important means for live detection of power grid equipment. The insulation fault and equipment defect of electrical equipment in operation can be found by using infrared thermal imaging technology. Infrared image deep learning model usually needs a very large computational cost and storage space, which greatly limits the application of deep learning model in embedded terminal. In order to apply the deep learning model well in embedded devices with limited resources, this paper proposes a lossless compression method of infrared image deep learning model of substation equipment, which can reduce the size of the deep learning network model by 35 to 49 times without loss of recognition accuracy. A hybrid sparse matrix storage format based on recursion and a cache blocking method based on multi-core/many-core processors are proposed to optimize the data flow efficiency between processors.
Simulation Modeling of Switching over Current of Circuit Breaker in Intercity TrainCao, Weinan; Sun, Ning; Wang, Yong; Li, Meng
doi: 10.1088/1742-6596/2320/1/012018pmid: N/A
The main interference source of axle counter is the over-current caused by the operation of circuit breaker by intercity train. This paper focuses on the simulation modeling technology of switching over-current of intercity train circuit breaker. Firstly, the mechanism of the interference when the circuit breaker is closed is studied by theoretical analysis; Then, the interference source model is established when the circuit breaker is closed, and the equivalent circuit model of interference path including vehicle body, vehicle body grounding system, traction network and high-voltage cable is obtained through calculation and analysis. The current of transformer primary grounding wire is measured on the actual vehicle, and the accuracy of the model is verified by comparing with the simulation model. Finally, the spectrum characteristics of the interference current in the rail are obtained by Fourier transform. In this paper, a simulation modeling method for analyzing the operating overcurrent of intercity vehicle circuit breaker is proposed, which can effectively evaluate the magnitude of the interference current generated by closing the circuit breaker into the rail, and provide a research basis for the follow-up study of the interference characteristics of the radiation field of rail current on the trackside axle counter.
Negative Damping Mechanism Analysis of Power System Stabilizer Based on Damping TorqueZhu, Xueqin; Yang, Yiqiang; Fu, Jiangtao; Bai, Wenrui; Su, Xiaolin
doi: 10.1088/1742-6596/2320/1/012004pmid: N/A
Power system stabilizer plays an important role in suppressing the low frequency oscillations, but a PSS with poorly tuned parameters may produce negative damping and aggravate the system oscillation. The purpose of this paper is to study the mechanism of damping of a PSS in multi machine system and to analyze whether the PSS has produced negative damping. This paper adopts Park model for analysis. According to the concepts of complex frequency and generalized phasor, the correlation between the damping torque phasor and the real part of the complex eigenvalue is studied. With the auxiliary of Park equivalent circuits, the damping torque provided by the excitation system is separated from the electrical torque phasor. Then, the damping torque coefficient of the PSS is defined as an indicator of the damp effect of the stabilizer. Finally, the simulation results of single machine infinite bus system and 4-machine system demonstrate the validation of the proposed method for evaluating the damping effect of PSS on the system.
The Determination of Reward Function in AGV Motion Control Based on DQNChen, Yubin; Li, Dancheng; Zhong, Huagang; Zhu, Ouwen; Zhao, Ziqi
doi: 10.1088/1742-6596/2320/1/012002pmid: N/A
Motion control is a very important part in the field of AGV(Automated Guided Vehicle). A good motion control method can make the movement of AGV more stable. Network models of reinforcement learning is one of the methods to solve the problem of AGV in motion control. This paper introduces the Markov Decision Process and the role of reward function. Besides, it studies and analyzes several classic reinforcement learning cases. DQN(Deep Q-Learning Network) which belongs to deep reinforcement learning network model is adopted. We set up several sets of comparative experiments with different reward functions by using sparse reward setting method, formalized reward setting method and reward coefficient variation reward setting method. Also we adjust the time of training. Through comparison, a reward function suitable for solving the problem of AGV motion control is obtained. In the field of AGV motion control, reinforcement learning model can converge faster and make more correct decisions. The reward function is verified in the simulation environment built by ROS(Robot Operating System).
Design and Implementation of AGV Calling System Based on AndroidZhu, Ouwen; Zhao, Yuli
doi: 10.1088/1742-6596/2320/1/012001pmid: N/A
This paper designs and implements an AGV calling application based on Android. In the actual working environment, administrators need to call automatic guided vehicles to perform tasks and monitor them anytime and anywhere, so the AGV calling application based on Android came into being. Five business function modules of the system are realized: calling module, map monitoring module, task list module, node configuration module and communication configuration module. The application software is developed using Android Studio development tool, with the use of Vue framework and front-end HTML5+CSS+JavaScript technology to achieve interface rendering. TCP and HTTP protocols are used in the connection between the system and the central control system.
Prefacedoi: 10.1088/1742-6596/2320/1/011001pmid: N/A
We are glad to welcome you to the proceedings of 2022 The 15th International Conference on Computer and Electrical Engineering (ICCEE 2022), which was held on June 24-26, 2022. Due to the cases increasing recently in China, in addition to the travel restriction and in respect of the willingness of all participants, ICCEE 2022 was entirely transferred as online conference instead of in Shenzhen University, China as scheduled.ICCEE was held annually and covered a broad range of topics related to Computer and Electrical Engineering. This year, presenters have presented and discussed topics in their respective research areas. The conference was held for 3 days including the online test sessions on ZOOM on the first day, and then followed with a wonderful array of 6 guest speeches along with 5 technical sessions during June 25-26. The authors’ presentations cover topics on Computer Science and Information Engineering; Computer and Control Automation; Machine Vision and Image Processing; Modern Electronic Information Technology and Application; Electrical Engineering and Automation. 15 minutes for each presentation, including 2 minutes for the Q&A. Delegates had a wide range of sessions to choose from and had enough time in deciding which sessions to attend.After the peer-review of all submissions, 40 full papers were accepted and included in the proceedings. All papers in these proceedings were peer-reviewed by conference technical committee members and international reviewers. We’d like to recognize the hard work of the entire committee for their enormous contribution. Our special thanks go to our speakers as well as all the authors for sharing their latest research results. We hope these talks interest and inspire all of you.Computer and Electrical Engineering have had a successful history. This field has pushed the frontiers of fundamental research, led to the emergence of entirely new disciplines, and revolutionized our daily lives. It extended to all walks of life, from the design of a switch to the research of aircraft. We do hope researchers in this field can create innovative, practical solutions as well as invent, design and build technologies and products that matter in the future.The proceedings will be successful only if these papers could motivate some of you. We look forward to meeting all of you next year at the conference in person.Best regards,Conference Program ChairChengwen Luo, Shenzhen University, ChinaList of Committees are available in this pdf.
An Image Encryption Scheme based on Adaptive Bit-Plane DivisionGu, Jun; Xu, Feng; Zeng, Tao; Lyu, Xin
doi: 10.1088/1742-6596/2320/1/012024pmid: N/A
With the application and popularization of digital images, the security of images has been paid more and more attention. The bit-level image encryption method has become one of the commonly used methods. It has the characteristic of changing the pixel value while disturbing the position of the pixel point, but at the same time it is accompanied by a higher computational complexity. In order to improve the efficiency of the encryption algorithm, according to the characteristic that the higher 4 of the 8 bit-planes of the grayscale image contain 94.125% of the information of the entire image, the encryption structure of higher 4 bit-planes and lower 4 bit-planes is widely used. But for different images, this characteristic is not accurately applicable. In response to this problem, we design a method of adaptive bit-plane division, which uses mutual information and Jensen-Shannon divergence to measure the amount of information of bit-planes and then divides bit-planes into two parts. The results of bit-plane division are different for different images. Based on this, we propose an image encryption scheme based on adaptive bit-plane division. Compared with the traditional higher 4-bits and lower 4-bits encryption method, this scheme reduces the encryption cost and improves the encryption efficiency. Experimental results show that the proposed scheme can effectively resist the chosen plain-image attacks and has sufficient security.
Cloud Resource Scheduling Method based on Markov Process and the Cuckoo SearchQi, Bin; Zhang, Pan; Wu, Hong; Yan, Miao
doi: 10.1088/1742-6596/2320/1/012030pmid: N/A
With the rapid development of cloud computing, the computing reliability of cloud servers has become an important research area. To improve the load balancing capability of cloud servers, an autonomous adaptive architecture cloud computing framework based on software-defined networking and network function virtualization were proposed. And a cloud resource scheduling method based on the Markov process model and cuckoo search algorithm to realize the load balancing of cloud nodes was designed. The cloud platform verified by the experiment shows that the method can effectively guarantee the service reliability of cloud computing.
Photovoltaic Output Prediction Method Based on Weather Forecast and Machine LearningZhang, Kaikai; Zou, Guibin
doi: 10.1088/1742-6596/2320/1/012032pmid: N/A
This paper conducts prediction research based on the historical power generation data and meteorological data of a photovoltaic power station. The obtained data is firstly classified by season, and the correlation analysis of meteorological factors is carried out. The data closely related to PV generation is selected for prediction, which improves the prediction efficiency. Then the Kmeans algorithm is used to cluster the historical data, and data are divided into four types: sunny, partly cloudy, cloudy, and rainy, further each type of data is separately trained to obtain four training models. The classified data is decomposed by EEMD, and a set of frequency-distributed IMF components and a monotonic distribution of residual components are obtained. Use support vector machine to predict the dimension-reduced components, and reconstruct the result to obtain the final predicted value. After obtaining the future weather forecast data, select the same type of historical data set for model training. The results show that the prediction accuracy will be significantly improved after signal processing in cloudy and rainy weather.