Fault Diagnosis of Asynchronous Motor Bearing Based on Time-Shift Filter and Deep Residual Network

Aiming at the problem that data acquisition is easily affected by noise and conventional current bearing fault diagnosis is difficult to accurately identify faults, a bearing fault diagnosis based on time-shifting (TS) and deep residual network (DRN) is proposed. Firstly, TS is used to denoise the original to remove the main noise that interferes with fault classification, and then the denoised input is extracted from the deep residual network of the Residual attention module (Ram) with dense connection and attention mechanism, and finally the fault classification is completed through the Softmax layer. The experimental results show that the proposed method has good superiority in bearing fault diagnosis and is superior to other comparison methods.