Fault Diagnosis of Motor Bearing based on Multi-Feature Fusion
Ali Hui, Peng Wei · 2024
Aiming at the loss of single sensor signal and low fault identification rate in mining motor bearings, a fault diagnosis algorithm based on the fusion of multiple sensor signals is proposed. First, multiple time domain and frequency domain fault features are extracted for each vibration signal. Secondly, the maximum relevance and minimum redundancy ($m$RMR) algorithm is used to screen the key features and reduce the feature redundancy. Finally, the golden jackal optimization algorithm (GJO) improved by Cauchy variation and weight decision is used to optimize the penalty factor and kernel function parameters in the support vector machine (SVM) model, so that it can be better applied in the fault diagnosis of motor bearings. Experimental results show that the proposed mRMR-IGJO-SVM model has higher accuracy, stability and faster convergence.