Hybrid data-based/model-based inter-turn fault detection methods for PM drives with manufacturing faults

Abstract

In this paper, improved online inter-turn fault detection methods are proposed in order to take into account some design specifications related to recent Permanent Magnet Synchronous Machine (PMSM) topologies having modular stator combined with Fractionnal Slot Concentrated Winding (FSCW). This winding allows compact PM motor mainly by reducing end windings. It is often associated with segmented stator in order to simplify the construction of the motor as in the case of Electro-Mechanical Actuators (EMA). This technique may lead to the presence of manufacturing faults such as additional air gaps between the various parts of the modular stator. This urges users to consider complex models of PMSM far removed from classical dq models. Consequently, undesirable harmonics, torque ripples can interfere with fault detection such as inter-turn fault detection algorithms. This paper proposes invasive methods interacting with PMSM control using both model-based and data-based methods in order to take into account these aspects for control and online inter-turn fault detection. Various experimental tests on an industrial EMA prototype validate the effectiveness of the proposed solution.