Model Predictive Control of Single-Phase Three-Level Inverters with Wide Inductance Variation

Model predictive control (MPC) has been widely considered as a design strategy for grid-interactive applications due to its simplicity and ease of implementation. However, the control performance of the inverter depends heavily on the accuracy of the system model when using the MPC-based method. To design cost-effective and high-efficiency inverters, the magnetic powder core-based filter inductor is a potential candidate since it involves the advantages of moderate cost and low losses. In the inverter operation, inductor current can vary from 0 to rated value in one line cycle of the grid. This will result in wide inductance variation, which may inevitably increase the current ripple and even deteriorate system stability. In this paper, an inductance estimation algorithm was introduced to the MPC method for a single-phase three-level inverter. By replacing the nominal value of the inductor with the predicted value obtained by the estimation algorithm, better performances can be achieved for the grid current tracking and neutral point potentials (NPPs) balance control in both steady and dynamic state. Finally, all-digital simulation results are presented to verify the theoretical expectations and the effectiveness of the proposed scheme.