Faults detection on a wound rotor induction machines by principal components analysis

This paper deals with faults detection and localization of wound rotor induction machines based on principal components analysis method. Both, localization and detection approaches consist in analyzing a detection index which is established on principal components. Once the faults are detected, the affected state variables are localized. The EWMA filter is applied to improve the fault detection quality by reducing the rate of false alarms. An accurate analytical modeling of the wound rotor induction machines is proposed and implemented on the software Matlab to obtain the state variables data of both healthy and faulted machines. Several simulation results are presented and analyzed.