基于改进残差网络的XLPE电缆局部放电声纹诊断方法OA北大核心CSTPCD

XLPE cable is an important equipment in the power system.Aiming at the problems of large amount of calculation and low accuracy in the cable fault diagnosis based on traditional residual network(ResNet)model,this paper proposes a cable partial discharge fault diagnosis method based on improved residual convolution network.Firstly,the time-frequency spectrum of three typical partial discharge faults is collected and preprocessed through the test platform.Then,the paper uses the Sigmoid weighted liner unit(Silu)as the activation function,and introduces the efficient channel attention(ECA)mechanism module into the residual block to obtain an improved residual network model.Finally,the trained model is used to identify the time-frequency spectrum of partial discharge fault.The results show that the recognition rate of the improved residual network can reach 97%,which is better than other classical deep learning networks,and is significantly better than the traditional machine learning algorithms.