Rotor fault detection in induction motors using the fast orthogonal search algorithm

This paper presents a method of detecting rotor faults in induction motors using the fast orthogonal search (FOS). Proper online condition monitoring of induction machines is very important to ensure safe operation, timely maintenance, and efficiency. It has been shown that when a fault occurs in the rotor, it will exhibit itself as a series of sidebands around the fundamental frequency in the stator current, which can be detected using a spectrum analyzer. Conventional methods based on the fast Fourier transform (FFT) are inadequate for motors operating under light load because the fault signatures will be close to the fundamental. Therefore a balance between resolution and sampling time must be achieved, which is difficult with the FFT. The higher degree of resolution of FOS makes it a promising choice for broken bar detection in motors operating under light load, and the reduction in sampling time is beneficial for motors that are prone to transient conditions. Experimental results using a 1/4 horsepower motor with a rotor winding fault are presented to verify this approach.