An intelligent automated method to diagnose and segregate induction motor faults

In the last few decades, various methods and alternative techniques have been proposed and implemented to diagnose induction motor faults. In an induction motor, bearing faults account the largest percentage of motor failure. Moreover, the existing techniques related to current and instantaneous power analysis are incompatible to diagnose the distributed bearing faults (race roughness), due to the fact that there does not exist any fault characteristics frequency model for these type of faults. In such a condition to diagnose and segregate the severity of fault is a challenging task. Thus, to overcome existing problem an alternative solution based on artificial neural network (ANN) is proposed. The proposed technique is harmonious because it does not oblige any mathematical models and the distributed faults are diagnosed and classified at incipient stage based on the extracted features from Park vector analysis (PVA). Moreover, the experimental results obtained through features of PVA and statistical evaluation of automated method shows the capability of proposed method that it is not only capable enough to diagnose fault but also can segregate bearing distributed defects.