GD-Based Robust Model Predictive Control for DC-DC Converters with Inductance Identification

The voltage performance of conventional model predictive control (MPC) depends on the accuracy of leakage inductance parameters in dual active bridge (DAB) converters. To address this issue, a gradient-descent-based robust model predictive control (GD-RMPC) is proposed. By integrating the mathematical model of the DAB converter, a gradient-descent equation is established to achieve real-time online identification of the leakage inductance parameter, ensuring robust output voltage control for the DAB converter. The proposed method allows for rapid, accurate, and online identification of the leakage inductance parameter, suppressing the adverse effects of parameter mismatches on conventional MPC. Finally, a DAB converter experimental platform is established, and the effectiveness of the proposed method is validated.