Fault Diagnosis of Vehicle Continuously Variable Transmission Based on Mutual Information Entropy and Support Vector Machine

The fault inspection and diagnosis of vehicle transmission plays a special role of safely steering and knocking the traffic accidents down. For the fault character of vehicle continuously variable transmission(CVT), the multi-resolution singular-spectrum entropy method, based on the mutual information entropy, to effectively extract the signal character of fault, and the multiclass classification support vector machine(SVM) algorithm to be easily implemented to deal with the problems of classification for fault state are proposed. Experiment results have shown the feasibility and efficiency of the combination mutual information entropy with multiclass classification SVM in fault diagnosis of vehicle CVT, additionally, the algorithms are very reliable to be implemented with fault inspection.