Multi-Stage Model Predictive Control with Enhanced Discrete-Time Models for Multilevel Inverters

Model accuracy, computational complexity, and weighting factor dependency are three primary challenges associated with multi-stage model predictive control (MS-MPC) of multilevel inverters (MLIs). Typically, existing MS-MPC employs the forward Euler integration method to discretize continuous-time models, due to the ease of implementation in real-time controllers. This approach faces a significant deterioration in computational accuracy as sampling period increases. To address this issue, this paper proposes a new formulation of MS-MPC for MLIs. Also, an enhanced discrete-time model of MLI by using Heun's integration method is proposed to implement the proposed MS-MPC, resulting in a more precise formulation and indirect minimization of common-mode voltage (CMV). Unlike the existing methods, the proposed method does not employ a cost function with weighting factor or offline selection of voltage vectors (VVs) to minimize the CMV. Furthermore, it does not increase computational complexity compared to the existing MS-MPCs. Simulation studies are conducted on a four-level MLI system to validate the efficacy of the proposed MS-MPC, and its performance is compared with the existing methods.