simultaneous bearing faults detection in three phase induction motor based on feature fusion method and random forest algorithm

Fault detection and classification (fdc) is a vital area in the health monitoring of three-phase induction machines. according to the failure survey of three three-phase induction machines, bearing-related faults cause a percentage of motor failures in the range of almost 41-50% which is very significant. these faults may occur one or multiple at a time in the bearing. with a well-designed fault detection method, failure of the motor can be reduced and productivity can also be increased. this paper proposes the simultaneous bearing fault detection and classification in three three-phase induction machine using the combination of feature fusion method and intelligent random forest (rf) algorithm. the paper contributes in two folds. in the first part of the paper, the performance of traditional methods such as vibration and current analysis is tested in which statistical parameters obtained from current and vibration signals are passed separately to the intelligent random forest classifier. in the second part of the paper, statistical parameters obtained from current and vibration signals are fused together and used as inputs to the rf classifier. the accuracy and various other performance measures are calculated and based on experimental results; a remarkably high detection/classification performance is achieved.