Research on AlexNet Model-Based Partial Discharge Diagnosis of Cable Terminals

The main cause of cable accidents is the insulation degradation of cables or terminal joints, and the detection and identification of partial discharges in cables can effectively avoid the safety hazards caused by insulation degradation. In order to improve the speed and accuracy of recognition, this paper proposes an improved convolutional Neuronal Network (CNN) algorithm for partial discharge image recognition method based on variational mode decomposition and wigner-ville distribution (VMD-WVD) time–frequency spectrum gray image combined with AlexNet network model, which collects partial discharge signals by building a cable partial discharge test platform, determines the image size by bilinear interpolation after joint VMD-WVD analysis, and uses the grey-scale processed image as the network input to construct different defect time–frequency spectra. Finally, the feature samples were used to train the improved convolutional neural network to identify the three types of T-type terminal partial discharge defect types tested. The results show that the partial discharge image recognition method based on the AlexNet network model can extract image features at a deeper level than the traditional principal component analysis and support vector machine (PCA-SVM) recognition method, and has a better recognition effect and faster training speed in the recognition of partial discharge defect types.