A New Algorithm for the Detection of Faults in Permanent Magnet Machines

Most condition monitoring techniques are based on steady state analysis. However, machines typically operate under transient conditions thus prompting the development of non-stationary fault detection techniques. In this paper, the steady state analysis technique (motor current signature analysis) is extended to a new non-stationary fault detection technique to detect several different faults in a low-voltage high current PMSM. It will be shown that the motor current signature analysis (MCSA) technique cannot be applied directly to machines operating under transient conditions. The new technique, which is based on an adaptive algorithm capable of extracting non-stationary sinusoids, is able to extract fault information during transient operation of the machine