Intelligent Fault Diagnosis Method of Motor Gear Based on Transfer Learning Under Variable Working Conditions

Abstract

A transfer learning (TL) based intelligent diagnosis method for motor gear faults under variable operating conditions is proposed. Firstly, an improved threshold function is used to eliminate the signal noise caused by gear gap oscillation. Secondly, a feature TL model is developed. The model uses the transfer component analysis (TCA) method to preserve the maximum embedment space of the source and target domain data characteristics. Finally, similar features are mapped to the new space of labeled data transmission. The simulation results show that the method solves the problem of large data volume and computational complexity. At the same time, it improves the diagnosis efficiency of motor gear faults under variable working conditions.