Event-Triggered Neural Predictor-Based FCS-MPC for MMC

Traditional predictive control schemes dominated the research field of numerous power electronic applications over the past years, since they usually lead to control solutions with good dynamic and steady results. However, these solutions strongly depend on the available knowledge of the control system (e.g., accurate modeling information), which often results in the lack of robustness in the presence of parametric uncertainties. Furthermore, unnecessary energy loss heavily correlates with high switching frequency, which directly affect efficiency. Aiming to cumbersome the aforementioned scarcities, this letter is concerned with a novel approach to the controller design for modular multilevel converter, which makes use of event-triggered neural predictor-based finite control-set model predictive control methodology under low switching frequency operation. The salient feature of the proposed controller is that the uncertainties and unnecessary energy loss in practical systems can be explicitly dealt with, while keeping the switching frequency in a low value. Finally, steady-state and transient-state performance tests together with the analysis of the experimental results confirm the interest of the proposal, and the results found are promising and motivate further research in this field.