Model Predictive Control of an Active Capacitor for Ripple Energy Compensation in Single-Phase DC-to-AC Converters

Single-phase dc-to-ac converters produce second-order harmonic currents on the dc side. To reduce this low-frequency ripple current, normally a bulky electrolytic capacitor is used, which contributes significantly to the increase in size and weight of the overall system. In order to reduce the ripple current and simultaneously reduce the size of the electrolytic capacitor, a dc-to-dc boost converter is employed. The capacitor of the dc-to-dc boost converter, the active capacitor, is used for ripple energy compensation. This paper presents a model predictive control (MPC) strategy of the active capacitor. The dc-to-dc boost converter switches are directly manipulated in order to divert the ripple current to the active capacitor. To improve the system performance, long prediction horizons are required. However, it becomes computationally challenging to solve the underlying optimization problem for long prediction horizons. Therefore, a branch-and-bound method is employed along with a move blocking strategy to solve the optimization problem in a practically realizable time. Simulation results are provided to verify the effectiveness of the proposed control strategy.