摘要：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.
关键词：bearing failure; diagnosis; distributed bearing faults; induction motor; park transform;;
and neural networks.
