An Information-distribution Joint Matching Method for Motor Intelligent Diagnosis under Sharp Varying Working Conditions
Yifu Ren, Dayong Zheng, Yanyong Yang, Geye Lu, Fengtao Gao, Qinghao Zhang, Zhihuang Ruan, Yatai Ji, Dongbo Guo, Dacheng Song, Pinjia Zhang · 2025
The modeling performance of motor intelligent diagnosis under sharp varying working conditions is limited. The core reason lies in the sharp varying working conditions, which inevitably cause significant differences in the information and distribution of motor faults, leading to the failure of the established diagnostic model. To address the above problem, this paper proposes an information-distribution joint matching method for motor intelligent diagnosis under sharp varying working conditions. Specifically, the information features and distribution features of multi-condition fault samples are extracted to provide the feature basis for the information-distribution joint matching; the information-distribution joint matching method is designed, in which fault samples under sharp varying working conditions are matched in both distribution dimension and information dimension, while fault information under different working conditions is reliably preserved, so as to improve the accuracy and generalization of motor intelligent diagnosis modeling under sharp varying working conditions. Experiment results show the proposed method outperforms the existing methods for motor intelligent diagnosis modeling under sharp varying working conditions.