Time-optimal finite control set model predictive control of non-isolated DC-DC converters

The authors propose a time-optimal finite control-set model predictive control (FCS-MPC) formulation, generalised to the three most common non-isolated DC-DC converters (buck, boost, buck-boost) tracking a constant switching frequency. The generalised switching model is used to formulate natural trajectories and the internal dynamic model for all three converters. The proposed FCS-MPC also allows the control designer to implement current and voltage constraints to limit current spikes and voltage deviations, respectively. The proposed FCS-MPC is compared to classical FCS-MPC and boundary controllers that also use natural trajectories for time optimality but at the cost of large voltage deviations. Classical FCS-MPC, time-optimal boundary control and the proposed FCS-MPC have been implemented in PLECS for all three converters. The current constraint does not impact control performance while the voltage constraint improves voltage deviation performances without significantly impacting the control speed compared to time-optimal boundary control. Finally, a hardware implementation of the proposed FCS-MPC on a buck converter proves that the control scheme is time optimal and mitigates current spikes while operating at a constant switching frequency at steady-state. A proposed finite control set model predictive control is compared to boundary controllers that also use natural trajectories for time optimality but at the cost of large voltage deviations. In this work, the voltage constraint improves voltage deviation performances without significantly impacting the control speed compared to time-optimal boundary control. Results are validated in simulation and a low power hardware test bed. image