A Novel Model Predictive Control Algorithm with Rolling-Horizon Optimization for Three-Level NPC Rectifiers

Junjie Wei,Yong Wang,Chen Qian,Changxing Shao
DOI: https://doi.org/10.1109/cac59555.2023.10451685
2023-01-01
Abstract:Due to its simple structure and good performance, the model predictive control (MPC) algorithm has attracted a lot of research and applications, making it a promising control strategy. However, the traditional MPC algorithm selects the optimal space voltage vector based solely on one sampling period, which may not be optimal for subsequent periods. In order to enhance the control performance, a novel model predictive control algorithm with rolling-horizon optimization (RO-MPC) for three-level NPC (Neutral Point Clamped) rectifiers has been proposed in this article. The proposed algorithm takes into account the control performance of the next period as well. Based on this, it selects the optimal and suboptimal space voltage vectors at the k+1th time. Using these selected vectors, the algorithm predicts the system state at the k+2th time. And to reduce the computational burden of the system, adjacent candidate space voltage vectors are considered for the next period. Then, the optimal voltage vector that minimizes the cost function is selected. Finally, the effectiveness of the proposed algorithm is verified through simulation experiments, including the steady-state performance, the dynamic-state performance and the robustness under grid voltage fluctuation.
What problem does this paper attempt to address?