Diagnosis of stator faults of the single-phase induction motor using acoustic signals

Abstract

An early diagnosis of faults prevents financial loss and downtimes in the industry. In this paper the authors presented the early fault diagnostic technique of stator faults of the single-phase induction motor. The proposed technique was based on recognition of acoustic signals. The authors measured and analysed 3 states of the single-phase induction motor: a healthy single-phase induction motor, a single-phase induction motor with shorted coils of auxiliary winding, a single-phase induction motor with shorted coils of auxiliary winding and main winding. In this paper an original method of feature extraction called MSAF-RATIO30-MULTIEXPANDED (Method of Selection of Amplitudes of Frequency - Ratio 30% of maximum of amplitude Multiexpanded) was described. This method was used to form feature vectors. A classification of obtained vectors was performed by the KNN (K-Nearest Neighbour classifier), the K-Means clustering and the Linear Perceptron. The early fault diagnostic technique can find application for protection of the single-phase induction motors. It can be also used for other rotating electrical machines.