The flow of the proposed spiral bevel gear fault diagnosis method is as follows:
Step 1: Collect the vibration data of spiral bevel gear transmission system, and divide the data into several equal length data segments.
Step 2: Carry out bispectral analysis pretreatment for each data segment, generate two-dimensional bispectral image, intercept the 1/4 local bispectral image in the upper left corner to construct a feature map sample set, and divide it into training set and test set.
Step 3: Initialize CNN network structure and parameters, such as learning rate, iteration number and minimum training amount.
Step 4: Input the training set into the CNN model for model training, and select key parameters such as the number of iterations and learning rate through multiple tests to complete the training of building the CNN model.
Step 5: Input the test sample set into the trained CNN model for identification and verify the validity of the model, so as to realize the fault diagnosis of spiral bevel gears.