Induction Motor Fault Diagnostic and Monitoring Methods

Induction motors are used worldwide as the workhorse in industrial applications. Although, these electromechanical devices are highly reliable, they are susceptible to many types of faults. Such fault can become catastrophic and cause production shutdowns, personal injuries, and waste of raw material. However, induction motor faults can be detected in an initial stage in order to prevent the complete failure of an induction motor and unexpected production costs. Accordingly, this thesis presents two methods to detect induction motor faults. The first method is a motor fault diagnostic method that identifies two types of motor faults: broken rotor bars and inter-turn short circuits in stator windings. These two types of faults represent 40 to 50% of all reported faults. Moreover, this method identifies the motor fault\u27s severity through the identification of the number of broken bars and the number of turns involved in an inter-turn short. The second method is a motor fault monitoring method that classifies the operating condition of an induction motor as healthy or faulty. The faulty condition represents any number of broken bars. This method has two major advantages. First, this is a robust technique, which is trained with datasets generated by time-stepping finite element methods in order to monitor faults of real induction motors in operation. Thus, the high cost associated with destructive tests to generate the training sets is not required. Second, it will be demonstrated here that this method, which is trained with simulated data of only one motor, can be used to monitor faults of real motors even with different design specifications. This establishes the scalability of this method. Both methods are validated through experimental tests