Detection and classification of broken rotor bars faults in induction machine using K-means classifier

Abstract

Diagnostics of broken rotor bars fault in induction machine can be accomplished by analyzing the anomalies of machine local variable such as magnetic flux, stator current and neutral voltage signature analysis. This paper presents an accurate approach for modeling and simulating the dynamic behavior of the induction machines (IM), including rotor bar fault, and presents an advanced technique to detect and classify the broken rotor bars faults in (IM). The techniques used here are based on vector quantization method (k-means). This detection and classification are based on motor current signal analysis (MCSA). The presented structure manifests promising results using experimental and simulation data of stator current; under variable load, using the FFT approach for identification of the predominant frequency components and using the (K-means) approach for classification and class detection.