A Stator Current-Based Method for Diagnosing Broken Rotor Bars in Induction Motors
Abstract
This paper addresses the critical challenge of early diagnostics for broken rotor bars in high-voltage squirrel-cage induction motors (SCIMs), which are essential for auxiliary systems of thermal power plants. Existing vibration and current analysis methods, often based on Fourier analysis, can be costly or lack sensitivity for incipient fault detection. This study proposes a novel diagnostic method based on regression analysis of the stator current signal modulus. An experimental setup with modified rotors enabled controlled simulation of one to four broken bars. Acquired three-phase stator currents were transformed into a single signal representing the modulus of the resulting stator current vector. This signal was then approximated using a Fourier basis regression model. A diagnostic feature was introduced, calculated as the normalized difference between the coefficient vectors of the undamaged and damaged motor states. Results demonstrate that the algorithm detects a single broken bar with high accuracy, causing a significant shift in the diagnostic feature of 5 percent for a 3 kW motor and over 6 percent for a 0.16 kW motor. The method achieved a 97 percent accuracy rate in identifying rotor circuit damage, showing significant promise for integration into microprocessor-based relay protection and automation systems as a cost-effective and reliable diagnostic solution.