Fault diagnosis using vibration measurements of a Flux-Switching permanent magnet motor

This paper examines the fault diagnosis (including detection and localization) of a Flux Switching permanent magnet Motor (FSM). Electrical and mechanical models of fault due to a partial short-circuit on one phase is described and used on a numerical simulation in order to detect and localize the fault. Fault detection is realized by a diagnostic observer (Kalman filter) which generates residual functions depending on the kind of fault. This diagnostic observer is designed by the Luenberger method and uses vibration measures to generate these residual functions.