Fault detection on a real three-phase induction motor: simulation and experimental results on residual generation

AC machines have become the industrial standard not only in high power-constant speed applications, but also in the medium and low power range. While most of the problems related to the feedback control of AC motors have been successfully solved, another crucial point emerged with the diffusion of induction motors: the condition monitoring and the detection of their most common faults. In this work, a novel approach to fault diagnosis of AC motors is introduced. Specifically, the architecture of a nonlinear tracking controller made of an observer and a feedback-linearizing module is exploited in order to generate residual signals to be used for fault diagnosis purposes. In the paper, both simulation and experimental results on a real AC motor are provided showing the effectiveness of the proposed approach.