Predictive current control of asynchronous machines by optimizing the switching moments

In this paper a model-based predictive control (MBPC) scheme for the current control of induction machines is presented. The controller directly selects the optimal switch state of the inverter. The proposed scheme uses a longer prediction horizon and a limited amount of optimal switching instants to reduce the average switching frequency. The next iteration of the MBPC-scheme is performed at the established optimal switching instant, as such suppressing the receding horizon property for short time spans. The proposed method is compared to a more conventional MBPC-scheme with a very short prediction horizon. Both simulations and experiments clearly show a significant reduction in average switching frequency. However, with a reduction in switching frequency the torque ripple is increased. To correctly asses the properties of the different schemes, a key performance indicator is proposed that offers a fair and unbiased comparison in terms of switching frequency and torque ripple.