Pattern identification method for partial discharge signal of electrical primary equipment

A mode recognition method for a partial discharge signal of electrical primary equipment comprises the following steps: S1, acquiring a partial discharge sample of the electrical primary equipment, performing VMD decomposition and CWD time-frequency analysis on the discharge sample to obtain a VMD-CWD time-frequency spectrogram of a discharge type, and outputting the VMD-CWD time-frequency spectrogram in a picture form; s2, performing image preprocessing on the time-frequency spectrum images obtained in the step S1, and constructing a training sample set and a test sample set; s3, introducing a cross-layer connection and optimization algorithm to construct a cross-layer feature fusion optimization convolutional neural network CFFO-CNN; and S4, training the CFFO-CNN network and identifying the partial discharge type of the test sample set. The method overcomes the defect that a traditional shallow layer recognition method is poor in high-dimensional data processing capacity, image pixel features can be autonom