Date of Completion

Spring 5-1-2017

Thesis Advisor(s)

Ali M. Bazzi, Helena Silva

Honors Major

Electrical Engineering


Controls and Control Theory | Other Electrical and Computer Engineering | Power and Energy


This paper introduces a novel time-domain method for detecting the four major types of induction motor faults without requiring complex signal processing or additional sensors other than those already in place for closed-loop motor control. This method artificially modulates the motor current feedback signal with a perturbed frequency that oscillates about the characteristic frequency of the fault in question. After filtering out the unwanted frequency components that result from the modulation, the selected fault-indicative component is a very slow sinusoid that is only present when the fault is present. The method looks for this fault-indicative component by monitoring the time since the last zero crossing, while also adapting the modulating signal in real-time based on the motor speed and torque. The perturbations of the modulation frequency are employed to accentuate the differences between fault and no-fault conditions and increase detection speed. Simulations are conducted in Simulink to validate the accuracy and speed of this detection method for each fault type under a variety of operating conditions and commanded speeds. The proposed method is capable of correctly detecting all the fault types, while offering exceptional detection speed and the ability to detect multiple faults concurrently.