An Iterative Learning Based Compensation in Model Predictive Control for DC/DC Boost Converter

Attributed to the increased processing power of modern microprocessors, model predictive control (MPC) for power converters is gaining more attention. However, the non-minimum phase behavior in DC/DC boost converters complicates the design of model predictive control. When controlling the output voltage directly, it fails to track the reference with short prediction horizons, nevertheless, long prediction horizons cause a heavy computational burden. Although controlling the inductor current is a feasible option with a short prediction horizon, the control accuracy of the output voltage cannot be guaranteed. To address this issue, this work introduces a compensation term into the difference equation of the inductor current. Then the proportion of the compensation term is designed with an iterative learning method to improve the control accuracy. Finally, the results indicate the proposed method can ensure a good control performance with different load occasions.