Enhanced Sampled-Data Models for Multistage Predictive Current Control of Four-Level Inverters

The use of forward Euler-based sampled-data models in multi-stage predictive current control (MS-PCC) for multilevel inverters (MLIs) results in a poor prediction accuracy with the rise in sampling time. These models also lead to higher harmonic distortion and capacitor voltage ripple in MLIs. To address these problems, modified Euler-based models are proposed for an MS-PCC, and they are applied to a four-level MLI. Also, the proposed MS-PCC is formulated to reduce the common-mode voltage indirectly, thereby eliminating the need for pre-selection of voltage vectors and weighting factors. The performance of MS-PCC with the proposed modified Euler and the existing Euler-based models is investigated experimentally on a dSPACE-controlled laboratory prototype under identical operating scenarios.