Periodic learning suppression control of torque ripple utilizing system identification for permanent magnet synchronous motors

This paper introduces two torque ripple suppression control methods for permanent magnet synchronous motor. With a system identification technique, the system model from compensating current input to torque detection output can be expressed as the speed-adjustable 1-dimension complex vector in the coordinate system synchronizing with the torque ripple frequency. The first method introduces a periodic learning I-P control system of compensating current to create the feed-forward compensation tables. The second introduces a speed-adjustable periodic disturbance observer which can improve the transient suppression characteristics. Both methods are suitable for the adjustable-speed drive, so the control parameters and reference model can be adjusted automatically. The most important thing is with these methods excellent performance for mechanical resonance suppression caused by torque ripples can be gotten. And this paper shows the analysis and experimental results to support the methods.