Detection of Bearing Fault in Induction Motor Based on Improved Adaptive Chirp Mode Decomposition

Bearing fault detection based on stator current signals has the characteristics of non-invasive and easy to implement, but weak fault features are submerged by strong background noise, posing a great challenge. Although adaptive chirp-mode decomposition (ACMD) has achieved good results in processing non-stationary signals in many fields, it requires some prior information to initiate, which limits its widespread application. Therefore, an improved ACMD method is proposed. First, the instantaneous frequency of initial estimation is obtained based on general linear chirplet transform. Then, the instantaneous frequency is used as the iterative condition, and the motor current signal will be decomposed into several modes. Finally, use power spectrum analysis to determine if it is faulty. The experimental results indicate that the improved method proposed in this paper is feasible for bearing fault detection.