Single-Loop Model Predictive Control for Fast Electric Vehicle Chargers

This paper proposes a single loop model predictive control (MPC) approach for DC fast electric vehicle chargers (DCFCs). The proposed controller simultaneously regulates the DC-link voltage, grid current, and charging power through a unified cost function, while incorporating an intelligent prediction of DC-link voltage dynamics and power delivery behavior to the battery. The prediction stage aims to forecast the DC-link voltage behavior and generate an optimal grid current reference for the rectifier and phase shift reference for the dual active bridge (DAB) DC-DC converter. The forecasted DC-link voltage is compared with the measured value in the subsequent control cycle, and the resulting deviation is utilized to adapt and refine the forecasting rules. The proposed controller includes four-control intervals, DC side and EV side capacitors charging for smooth startup and inrush current elimination, DC-link voltage forming for pre-charging initialization, and DC-link voltage regulation during the charging process. The main objectives of the proposed controller are to eliminate the need for an outer loop and cascaded voltage controller, mitigate losses and current reference generation errors, and ensure compatibility with various grid voltage levels (e.g., 110 V and 240 V) as well as a wide charging voltage range (200 V-1000 V), without requiring any modification to the controller structure or control gains. The proposed controller is experimentally implemented on an industry-scale 30kW charging module prototype.