Electrical machine fault detection using adaptive neuro-fuzzy inference

This paper proposes a new integrated diagnostic system for induction machine electrical fault diagnosis by means of a neurofuzzy approach. New features that are of multiple frequency resolutions are extracted by wavelet packet decomposition of the stator current. These features can then clearly differentiate the healthy and faulty conditions. Features with different frequency resolutions together with the slip speed of the induction motor am used as the input sets for a neuro -fuzzy inference system. Two common electrical faults, the rotor bar breakage and the air gap eccentricity are considered. The system is validated on a 5 HP three-phase induction motor. Successful implementation of the proposed diagnostic system has been demonstrated