368 / 2021-07-19 16:54:37

Application of Residual Network based on Double Threshold Structure in Bearing Fault Diagnosis

residual network, double threshold, one-dimensional vibration data

Special Sessions > Fault Diagnosis of Gears

 Convolution neural network and its derivative network have been widely used in the field of image recognition, but its application in one-dimensional vibration data is not very wide. In this paper, residual network is used to identify the fault of bearing based on one-dimensional vibration data. Due to the actual bearing data will be mixed with noise and invalid data, Therefore, a residual network based on a double threshold is proposed in this paper. The threshold structure of the first layer is mainly used to remove invalid data, and the soft threshold structure of the second layer is mainly used to filter noise data. Compared with the residual network and residual shrinkage network, the combination of residual and double threshold structure has improved the fast convergence of the algorithm and the accuracy of diagnosis in bearing fault diagnosis.