An Improved Triple‐Vector Model‐Free Predictive Control Method for Grid‐Forming Converters with Runge–Kutta‐Based Data‐Model

The conventional triple‐vector model predictive control (TV‐MPC) method for grid‐forming converters (GFCs) exhibits a strong dependence on precise system parameters, potentially compromising voltage prediction performance. To address this limitation, this paper develops an improved triple‐vector model‐free predictive control method with Runge–Kutta‐based data‐model (TV‐MFPC‐RKDM). First, the impact of system parameters on traditional TV‐MPC for GFCs is analyzed. Subsequently, a Runge–Kutta‐based data‐model (RKDM) for the GFC is established and continuously updated using historical voltage data. To compensate for the inherent one‐step delay in the RKDM, the Lagrange extrapolation method is employed for model estimation. The triple‐vector (TV) principle is then integrated, with vector duration calculation leveraging the updated RKDM to enhance voltage prediction accuracy. The proposed TV‐MFPC‐RKDM facilitates real‐time RKDM updates, eliminating the reliance on precise system knowledge and significantly reducing prediction errors. The effectiveness of the proposed TV‐MFPC‐RKDM method is demonstrated through comprehensive experimental comparisons. © 2026 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC.