Finite-states model predictive control of a four-level diode-clamped inverter

This paper presents a new and simple finite-states model predictive control strategy for a four-level three-phase diode-clamped inverter. This strategy allows for fast load current control while keeping the balance of the dc-link capacitor voltages. A discrete-time model of the load and of the dc-link capacitors is used to predict the behavior of the load current and the capacitor voltages for each possible switching state. A cost function that considers the load current error and the capacitor voltages error is used to evaluate each prediction. The switching state that minimizes the cost function is selected and applied during a whole sampling period. Simulation results are shown verifying the good performance of the proposed predictive controller.