Gear Fault Diagnosis Based on Short-time Fourier Transform and Deep Residual Network under Multiple Operation Conditions

Abstract

To solve the ICPHM 2023 data challenge, a fault diagnosis method is proposed in this paper can accurately predict gear faults under various working conditions. The method is based on the deep learning model and Short-time Fourier Transform with fewer training parameters. The model can learn effective data features without setting too many epochs, which makes the training cost acceptable. In addition, the proposed model only needs to make simple function calls in the fault diagnosis phase, the time cost of the fault diagnosis phase is very low.