High-speed train axle fault diagnosis method based on bimodal coding and pruning

The invention discloses a high-speed train axle fault diagnosis method based on bimodal coding and pruning. The high-speed train axle fault diagnosis method specifically comprises the steps that 1, high-speed train axle vibration signals are collected; step 2, carrying out equalization processing on the collected axle vibration signal set of the high-speed train; 3, performing signal decomposition and denoising on the balanced high-speed train axle vibration signal; 4, performing compression and dimension reduction on the decomposed and denoised high-speed train axle vibration signal, and performing sliding window interception on the dimension-reduced high-speed train axle vibration signal; 5, integrating the intercepted axle vibration signals of the high-speed train by adopting a bimodal fusion coding method to obtain a spliced image; step 6, carrying out image enhancement on the spliced image; and step 7, classifying the enhanced images. According to the method, the problem that the diagnosis accuracy is af