An iterative-learning-based compensation for PWM VSI nonlinearities

The nonlinearities of pulse width modulation voltage source inverter (PWM-VSI) lead to output voltage and current distortion, fundamental wave amplitude decrease, torque ripple that may cause instability in the low-speed range of permanent magnet synchronous machine (PMSM) drives. To eliminate the influence on PMSM drives due to the nonlinearities of the VSI, models of the above nonlinearities were established and their influence was treated as a total disturbance voltage. Through coordinate transformation and time domain to angle domain transformation, the voltage disturbance is comprised of a periodic disturbance with regard to the rotor position and a constant disturbance in the synchronous rotating reference frame. A composite compensation strategy, composed of the iterative learning control and feed-forward, was proposed to eliminate the periodic disturbance and constant disturbance respectively in the angle-domain. The proposed method is easy to implement without the knowledge of the system accurate model, attaining effective compensation by only using current sensors. Simulation and experimental results demonstrate that the proposed composite compensation can reduce the current harmonic effectively, improve the stability in low-speed range and is applicable to the different speed operations. ©2015 Chin. Soc. for Elec. Eng.