Fault diagnosis method and system based on wavelet graph convolution

The invention provides a fault diagnosis method and system based on wavelet graph convolution, and the method comprises the steps: extracting the multi-scale time-frequency features of a signal based on a multi-Bezi wavelet, and obtaining the multi-scale wavelet components; performing time-frequency analysis on the wavelet component of each scale, and obtaining a corresponding wavelet coefficient and time-frequency domain information; constructing a multi-scale image convolutional network combined with a similarity attention mechanism, and extracting multi-scale time-frequency image features; fusing the multi-scale time-frequency graph features, and obtaining a multi-scale fusion feature; and obtaining a label of the composite fault. According to the fault diagnosis method based on wavelet graph convolution, the technical problem that the accuracy and rapidity of fault diagnosis of an existing fault diagnosis method of a gearbox are not high due to the fact that complex non-linear and strong coupling relation