Model-Free Predictive Control of Grid-Forming Inverters Based on RLS-DCD Algorithm

Traditional model predictive control (MPC) is highly reliant on the system's mathematical model. When there are uncertainties in the model or parameter variations, the control performance is vulnerable to being affected. In view of this, model-free predictive control (MFPC) has gradually become the research focus in this field. Against this backdrop, this paper conducts in-depth research. The system's mathematical model is replaced with the auto regressive with extra inputs (ARX) model. This study deeply analyzes the internal reasons for the high computational complexity of the traditional recursive least square (RLS) algorithm during the parameter identification of the ARX model. Furthermore, it innovatively proposes the recursive least square-dichotomous coordinate descent (RLSDCD) identification algorithm as an advanced solution. The study combines rigorous theoretical analysis with extensive simulations and experimental validation. Results demonstrate that the proposed algorithm reduces computational complexity significantly while maintaining control robustness. It has been successfully applied to the control of GFI, providing a more efficient and reliable solution for the optimization of GFI control strategies.