A Motor Current Signal-Based Fault Diagnosis Method for Harmonic Drive of Industrial Robot Under Time-Varying Speed Conditions

The harmonic drive (HD) of industrial robots often operates under nonstationary and time-varying speed conditions. The fault frequency of HDs under time-varying speed conditions is nonperiodic and random, and it is difficult to extract fault features from the current signal with the interference caused by time-varying conditions. Additionally, the impacts induced by long and short axes alternating of the flexible bearing in HDs can also contaminate the fault feature information. Hence, we propose a fault diagnosis method for HD under time-varying speed conditions. The equal angle displacement signal segmentation (EADSS) is applied to eliminate the effects of time-varying speed on features of the current signal, and an improved nuisance attribute projection (NAP) method is developed to remove the influence of speeds on fault features, in which the cosine distance is utilized to optimize the weight matrix of the NAP. Finally, the principal component analysis (PCAs) is used to extract sensitive features from the feature matrix, and the back propagation neural network (BPNN) is utilized to diagnose faults on different parts of the bearing. Two cases are provided to demonstrate the proposed method, and results show that it is suitable for fault diagnosis of HDs under time-varying working conditions.