Model- and Information Theory-Based Diagnostic Method for Induction Motors

Abstract

Introduced is a model-based diagnostic system for motors, that also employs concepts of information theory as a health metric. From an existing bond graph of a squirrel cage induction motor, state equations were extracted and simulations performed. Simulated were various cases, including the response of an ideal motor, which functions perfectly to designer’s specifications, and motors with shorted stator coils, a bad phase capacitor, and broken rotor bars. By constructing an analogy between the motor and a communication channel, Shannon’s theorems of information theory were applied to assess functional health. The principal health metric is the channel capacity, which is based on integrals of signal-to-noise ratios. The channel capacity monotonically reduces with degradation of the system, and appears to be an effective discriminator of motor health and sickness. The method was tested via simulations of a three-phase motor; and for experimental verification, a two-phase induction motor was modeled and tested. The method was able to predict impending functional failure, significantly in advance.