Intelligent fault diagnosis of gear crack based on side frequency feature under different working conditions

Abstract

Aiming at the problem of gear crack fault diagnosis, an intelligent diagnosis method based on side frequency feature is proposed. It enhances the fault information representation ability of the extracted features and the fault identification ability of the model. Firstly, according to the side frequency distribution characteristics of gear crack fault, the side frequency energy features are quantified, and a relatively complete feature set is constructed by combining the time domain features; Secondly, an evaluation method of feature effectiveness is designed to obtain the optimal feature subset; Finally, a three-stage training network is constructed to achieve an increase in fault diagnosis rate. The test results under different working conditions show that the proposed method can more completely represent the fault information and effectively improve the fault diagnosis rate when compared with the machine learning model of a general two-layer network and feature extraction methods based on entropy features.