Condition monitoring of induction motor using negative sequence component and THD of the stator current

Abstract

The Condition Monitoring of Induction Motor (IM) is performed to ensure optimal and reliable operation, as IM has numerous applications spread across varied sectors. Mechanical faults such as Broken Rotor bar fault of IM along with supply PQ disturbances create a high degree of non-linearity in the supply. This non-linearity is examined by stator current signature analysis which involves the computation of the Negative Sequence Components (NSC) and Total Harmonic Distortions (THD) of the stator current. These values are given as inputs to the Artificial Neural Network (ANN), Support Vector Machine (SVM) and k-Nearest Neighbor (kNN) classifiers. The results of the classifiers are obtained and compared. It is seen that the classification accuracy for ANN is found to be 90.63%, while for SVM is found to be 92.71% and that of kNN is found to be 85.41%.