Multi-scale feature fusion gearbox fault diagnosis method based on self-attention mechanism

The invention discloses a multi-scale feature fusion gearbox fault diagnosis method based on a self-attention mechanism, and relates to the technical field of fault diagnosis, and the method comprises the steps: carrying out the random window drawing of a one-dimensional original vibration signal of a gearbox, and obtaining a fixed-length diagnosis sample; a multi-scale feature fusion fault diagnosis model based on a self-attention mechanism is constructed, and a Softmax function is used as a classifier for training; carrying out model training through a back propagation method by utilizing a dynamic cutting Adam optimizer; and storing the trained fault diagnosis model for online diagnosis. According to the method, low-frequency features and local time-domain features of original vibration signals of the gearbox are extracted through convolution kernels of different scales respectively, then an improved self-attention mechanism is introduced to construct a multi-scale feature fusion network to replace a tradi