Induction motor fault diagnosis based on the machine temperature, vibration analysis and sensors fusion

Abstract

The most common motor used for industrial, residential and commercial applications is the induction motor (three or single phase). This motor is very reliable, but faults still may occur. The present paper focuses on the diagnosis of induction motor faults based on its temperature and vibration behaviors on steady-state operation. The proposed method is based on the Extended Park Transform, enabling sensor fusion which reduces the amount of data required for fault identification to 1/3 and allows the usage of a shallow artificial neural network. To validate the proposed method, experiments have been carried using a single phase induction motor operating under normal and fault conditions (short-circuit between main winding turns, auxiliary turns, main-auxiliary windings and with contaminated bearing lubrication). The results proves the efficacy of the proposed method, which has reached an accuracy over 99.5% in the process of fault identification using low cost sensors/equipment.