Partial Discharge Pattern Recognition Of Mining Cable Based On Convolutional Neural Network

Mine cable is important equipment for transmitting and distributing electrical energy in the underground power supply system of coal mine.The harsh environment in coal mines may cause cable failure and endanger the safety of equipment and worker.The statistical results show that the insulation defects existing in the production or operation of mining cable is the main cause of their failure.At the beginning,these defect is partially discharged under live conditions.Eventually,the development of partial discharges lead to cable failure.Study the partial discharge pattern recognition of mine cables,to quickly and accurately identify the type of partial discharge,and lay the foundation for online diagnosis of partial discharge of mine cables.Firstly,this paper starting starting from the structure of mine cable,analyze the cause of partial discharge of mine cable,Ansoft Maxwell software is used to analyze the electric field of the mine cable with partial defect and the physical model of partial discharge,and find that the defect will cause the local electric field change of the mine cable.The regularity of the electric field change of the mine cable and its corresponding discharge model.Secondly,build a partial discharge detection system for mine cable and analyze the PRPD spectrum characteristics of the test data;To solve the problem of low recognition rate of the existing mining cable pattern method,based on convolutional neural network design a partial discharge pattern recognition of mining cable.Thirdly,optimizing the convolutional neural network for partial discharge pattern recognition of mining cable,and constructing a two-dimensional grayscale image based on the PRPD spectrum of the physical model of partial discharge of mining cable as the input of pattern recognition,and analyzing the influence of partial discharge pattern recognition of mining cable,the core factor of accuracy is to optimize and train the convolutional neural network for partial discharge pattern recognition of mining cables,and analyze and compare the recognition results.The research results show that the recognition accuracy of the recognition method proposed in the thesis can reach 93.79%in the partial discharge pattern recognition of mining cables,which is more than 3%higher than the recognition accuracy of other traditional partial discharge pattern recognition methods.The identification method proposed in the thesis has the characteristics of simplicity and high accuracy.It can be used in coal mine power supply system to improve the safety of power supply.