A Signal-End Data Augmentation Method for Mechanical Fault Diagnosis Based on Self-Sensing Motor Driver

The performance of diagnosis model highly depends on the amount and quality of data used in training. However, it is difficult to collect sufficient fault data in practical applications, especially under different operation conditions. To overcome this problem, a signal-end data augmentation method is proposed to provide adequate fault data for mechanical fault diagnosis under different speeds. Firstly, the self-sensing motor driver is used to collect monitoring signals, which not only contains faulty information but also operation conditions. Secondly, based on the dynamics of mechatronic system and mechanism of fault, the monitoring signals are decomposed into three components, and utilized to augment synthetic data. Finally, the proposed method is validated by signals obtained from a test rig with self-sensing motor driver under different speeds. Experimental results show that the proposed method can effectively bridge the gap between different working conditions and solve the few-shot problem.