Disturbance-rejection adjacent vector model predictive control strategy based on extended state observer for EV converter

Abstract

Conventional single-vector model predictive control (MPC) can suffer from low control accuracy, while multi-vector MPC is often criticized for its complexity and heavy computational burden. In order to address these issues, an adjacent vector-based MPC is investigated in this paper for an electric vehicle battery charging and discharging converter. The voltage vector selection table based on the principle of using adjacent vectors has been designed and this reduces the number of iterations and thus the computational burden. A threshold is used in the adjacent vector-based MPC to coordinate the use of the single and multi-vector MPCs considering a balance between the control accuracy and computational burden. In addition, to enhance the robustness of MPC to parameter changes, an extended state observer for active disturbance rejection control has been used to derive the predictive model, and an adjacent vector-based MPC using extended state observer is studied. The method does not need accurate system parameters. Instead, it only requires the system input and output measurements to calculate the predicted current. The robustness of the controller against the parameter mismatch is enhanced compared to alternative approaches and the experimental results verify the feasibility and effectiveness of the proposed strategy.