Detection of Bearing Fault in Induction Motors Based on Noise-Assisted Adaptive Local Iterative Filtering

In response to the challenge that weak fault signatures in current signals are susceptible to strong background noise interference, complicating feature extraction, this paper proposes a noise-assisted adaptive local iterative filtering. Firstly, to address the incomplete separation of feature scale information in traditional Adaptive Local Iterative Filtering methods, the noise-assisted technique is introduced to optimize the distribution of extreme points in the signal. Secondly, a new method based on correlation analysis and relative error analysis is proposed to overcome the shortcomings of traditional noise-assisted algorithm, such as lack of theoretical basis for empirical parameter setting and inability to achieve optimal decomposition. Experimental results validate the effectiveness and superiority of the proposed method in induction motor bearing fault detection.