Fault Diagnosis Method Based on MED-VMD and Optimized SVM for Rolling Bearings

Abstract

A method of fault feature extraction was proposed based on MED, VMD and fuzzy approximate entropy, and the optimized SVM was used to identify faults. The MED method was used to reduce the noise interferences and to enhance the fault feature informations in the fault signals, and the signals after noise reduction by VMD were decomposed, then, the fuzzy approximation entropy was used to quantify the modal components of fault feature informations after VMD, and the feature vectors were constructed, Finally, the extended particle swarm optimization(EPSO) algorithm was used to optimize the penalty factors and the kernel function parameters of SVM to complete the fault recognition classification. The proposed method was applied to the experimental data of rolling bearings, and the effectiveness of the method was verified. Compared with the feature extraction method based on local mean decomposition(LMD), it is shown that the proposed method may extract the features of rolling bearing faults more accurately and may identify different faults more accurately. Compared with SVM based on grid search algorithm and the least square support vector machines(LSSVM) based on EPSO algorithm, the proposed method has better classification performance and better diagnosis performance.