A new method for analysis of rotor broken bar fault in induction machines

Abstract

The transition of automotive industry from traditional combustion engines to hybrid vehicles and also research on renewable energies show a rapid need for more electrical machineries and consequently a huge demand for their condition monitoring and diagnostic systems. Prodigious improvement in signal processing techniques and availability of cost effective DSP hardware provide a road map for developing such systems. Primarily, these techniques depend upon locating specific features such as harmonics in line current, vibration signals, or thermal measurements. These features are then processed using modern classifiers and classification techniques. By removing correlations among features and harmonic contents, the classification result will be an indication of the health of system and or whether there is a need for maintenance. This paper provides a brief review of past and presents methods, and presents new analytical method to model and analyze broken bar fault of the IM. The method detects the existence and severity of broken bar fault in the line currents of an IM that yields a reliable feature for its fault classification. Theoretical results are verified by experiments and shown at the end of the chapter. © 2005 IEEE.