Fault Detection and Diagnosis of Mine Hoist Machine Based on RS and SVM

Abstract

A fault diagnosis method is proposed for the mine hoist machine fault diversity and redundancy of fault data based on Rough Sets (RS) and Support Vector Machine (SVM). RS theory is used to analyze the stator current fault data of mine hoist machine in order to exclude uncertain, duplicate information. For getting the optimal decision table, the equivalence relationship of positive domains of between decision attributes and different condition attributes is analyzed in the decision tables to simplify condition attributes. The optimal decision table is as the SVM input samples to establish the SVM training model. And the mapping model which reflects the relation of the characteristics between condition attribute and decision attribute is obtained by SVM training model in order to realize the fault diagnosis of the mine hoist machine. The simulation results show that the fault diagnosis method based on RS and SVM can improve the accuracy of fault diagnosis.