Application of the short-time Fourier transform and artificial intelligence to the permanent magnet synchronous motor stator winding fault detection

This paper presents the possibility of using the short-time Fourier transform of the stator phase current and stator current space vector module in the process of permanent magnet synchronous motor stator winding fault symptoms extraction. Additionally, the automatization of the stator winding condition inference process which the use of selected artificial intelligence based algorithms: Support Vector Machine and MultiLayer Perceptron is proposed. The developed diagnostic system has been extended with the functionality of locating the damaged phase. Experimental studies confirmed the high effectiveness of the developed method.