Kalman-Cyclisation Dominant Frequency Based PMSM Bearing Fault Detection

Permanent Magnet Synchronous Motors (PMSMs) are becoming increasingly popular in traction applications. PMSMs have gained popularity due to their efficiency, compactness and excellent features related to control. This is mainly because of the permanent magnet placement in the rotor, which negates any additional rotor excitation, hence reducing energy losses considerably. This makes them more efficient compared to induction motors. PMSMs also have a high-power density, enabling them to provide more power relative to their size. These features make them highly feasible in electric vehicles and aerospace applications where size and weight are one of the critical aspects. With the possibility of precise and smooth torque control, they are well adapted to applications that have strict requirements for precise speed and position control. Also, PMSMs provide superior dynamic performance with low maintenance since they do not contain brushes and commutators; hence, they are supposed to have a longer life. But PMSMs have also some challenges, the most critical problem that usually faces PMSMs is fault detection, mainly bearing faults. The bearings in PMSMs are critical components, and any defect in those components creates serious mechanical and electrical problems for the motor. Due to its defective state, the bearing may impose the emergence of vibration, raised friction, and wear, leading to inefficiencies, increased heat generation, and, finally, motor failure. Thus, early detection of bearing faults is indispensable; prolonged operation with faulty bearings results in catastrophic damage to the motor, longer downtime, and more expensive repairs. Thus, the fault diagnosis at the inception stage has become very critical now. In this paper, a technique based on cyclization dominant frequency analysis for bearing fault detection using the phase current of a motor has been proposed. The results have shown that this method can be used for monitoring and detecting bearing faults without the use of any additional sensor.