Diagnosis by pattern recognition for PMSM used in more electric aircraft

Presently, condition monitoring and fault diagnostics in electric drives are essential to optimize maintenance operations and increase reliability levels. This paper presents a diagnosis method for electrical and mechanical faults detection. This method combines a detection method based on expertise with a pattern recognition approach so as to detect different faults appearing on the system but also to classify their origins and their severity by reference to an initial data base. In order to prove reliability and efficiency of this method, experimental results are presented using a permanent magnet synchronous motor (PMSM) drive.