Detection of Broken Rotor Bars in Induction Machines using Machine Learning Methods

Induction machines are used in a wide range of industrial applications due to their simplicity, ruggedness, and low cost. Despite their robustness, induction machines eventually fail at one point due to a variety of failure mechanisms. Many faults exhibit specific fault frequencies in the motor current spectrum, which allows for fault detection. Many classical fault detection methods have been developed for grid-connected machines with quasi-static operating points. In inverter-driven machines with a wide operating range, these methods cannot reliably detect and classify faults. Machine learning methods have been successfully used for various classification tasks. This work applies a combination of classical fault detection approaches with different fault classification algorithms to reliably detect induction machine faults over a wide operating range.The developed fault classification method is evaluated using steady-state measurements on an inverter-fed 5.5kW induction machine. The algorithm shows promising fault detection and classification capabilities and achieves an accuracy of 97.4%.