Sensorless Capacitor Voltage Balancing of a Grid-Tied, Single-Phase Hybrid Multilevel Converter with Asymmetric Capacitor Voltages using Dynamic Programming

This paper shows a sensorless capacitor voltage balancing control approach for a grid-connected, single-phase hybrid multilevel inverter based on an NPC main stage with a voltage stiff DC-link and an arbitrary number of H-Bridge modules (capacitor modules) with asymmetric capacitor voltages. Using nearest-level control, a model predictive control (MPC) approach with a prediction horizon of one time step is chosen to find an optimal switching-state combination among the redundant switching combinations to balance the capacitor voltages as quick as possible. Using the Lyapunov stability criterion, it is shown that an offline calculated optimal switching-state sequence for each discrete output voltage level can be used to operate the inverter without using any voltage sensors for the capacitor voltages. To validate the stability of the approach, a laboratory inverter with a resistive load is operated with the offline calculated optimal switching-state sequences and it is shown that the capacitor voltages converge to their desired reference voltages.