Model-Free Predictive Control of LCL-Filtered Voltage Source Inverter With Parameter-Free References

The parameter mismatch of conventional model predictive control (MPC) may affect the grid current performance of LCL-filtered voltage source inverter (VSI). To improve the robustness of grid current against parameter mismatch, a model-free predictive control (MFPC) with parameter-free references (PFRs) is presented in this article. First, a model-free prediction method is presented, where the current–voltage differences corresponding to different voltage vectors are updated, stored, and adopted for current–voltage predictions, eliminating the parametric effect on conventional prediction method. In addition, a PFRs method is presented, where the cosine functions between the sampled current–voltage signals of LCL-filtered VSI are established to obtain current–voltage references, which eliminates the parametric effect on conventional reference method. Finally, the performance of the proposed method is evaluated on the LCL-filtered VSI under accurate and mismatched parameters. Compared with conventional MPC and MFPC methods, the proposed MFPC-PFRs achieve comparable current performance and a greater robustness against parameter mismatch.