Research on Fault Diagnosis Method Based on Improved CNN

Table 1

Structure parameters of improved CNN.

Parameters of the model
Length of convolution kernel in the first CNN layer group: 30	Learning rate: 0.0004
Number of convolution kernels in the first CNN layer group: 8	Batch size: 100
Length of convolution kernel in the second CNN layer group: 15	Loss: crossentropy
Number of convolution kernels in the second CNN layer group: 16	Optimizer: Adam optimizer
Training steps: 500	Number of neurons in the middle layer: 1024