Research On Protection Status Monitoring And Early Warning Of Substation Electronic Fence

Electronic fence has strong blocking and deterrence function to substation.Through the processing and analysis of the collected parameters such as voltage,current and field strength of the electronic fence of the substation,improve the condition monitoring and early warning efficiency of the electronic fence,ensure the more reliable and safe operation of the substation,and then ensure the reliable and stable operation of the power system.At present,most domestic substations focus on data collection and simple processing for electronic fence status monitoring,and can not timely and effectively provide reliable judgment and early warning and targeted maintenance suggestions for special abnormal conditions,resulting in the problems of low intelligence and high missing rate of electronic fence in unattended substations.The accuracy and response speed of condition monitoring still need further research.This paper mainly studies the condition monitoring and early warning of substation electronic fence based on deep learning algorithm.Firstly,a set of electronic fence simulation test system is designed to facilitate the acquisition of data.After preprocessing and feature extraction of electronic fence data,different convolution neural network structures are used to classify and analyze the extracted features.Finally,a set of signal analysis software for monitoring and early warning of substation electronic fence is designed and applied in practical engineering projects.The main research contents are as follows:(1)A set of electronic fence simulation test system is designed,which mainly includes signal generator,signal acquisition device and signal analysis software.The maximum output voltage of the signal generator can reach 20 k V,and the coincidence with the real electronic fence voltage is as high as 95%.The signal measuring equipment integrates Pico oscilloscope,industrial computer,wireless transmission and other modules.The device is equipped with a shield for anti-interference protection to ensure the accuracy of collected data and equipment safety.The signal analysis software adopts deep learning algorithm to provide theoretical basis for real-time monitoring and early warning of the state of electronic fence in intelligent substation.(2)Two convolutional neural networks VGG Net 16 and Google net are used to process and analyze the electronic fence data.The results show that the recognition accuracy of VGG Net 16 and Google net is 96.13% and 94.35% respectively,and the recognition speed is 0.02 s.Compared with Google net,the accuracy of VGG Net 16 is improved by 1.78%,which is more in line with the requirements of electronic fence condition monitoring and early warning.(3)According to the functional requirements of substation electronic fence monitoring and early warning,the interface of signal analysis software is visually designed by using c#language,and a set of human-computer interactive signal analysis software is developed,mainly including the design and development of login interface and operation interface,which has the functions of online analysis,feature display,anomaly monitoring,user management and so on.Starting from the actual project,this paper studies the existing problems of electronic fence in unattended substation.At present,experiments have been carried out on the electronic fence device of a 330 k V substation in the eastern and northern suburbs of Xi’an.Everything is normal during operation,which better meets the functional requirements of daily monitoring and early warning of the electronic fence of the substation.