Improved Performance With Dual-Model Predictive Control for Cascaded H-Bridge Multilevel Converter

There has been increasing interest in solid-state transformers (SSTs) as a future replacement for conventional distribution transformers as it is able to provide ancillary services in addition to step-up/step-down of voltages. In SSTs, a common topology used in the ac-dc stage is the cascaded H-bridge (CHB) multilevel converter. This paper proposes a novel dual-model predictive control method for the CHB multilevel converter with faster dynamic response, improved harmonic performance, and voltage balancing of the CHB cells. The primary model enables conventional control over source current and output voltages, whereas the secondary model is able to generate the reference source current through power modeling, which eliminates the need for a proportion-integral (PI) controller, thus resulting in faster dynamic response. A Luenberger-based observer is also implemented to estimate the dc load current in each CHB cell. The second-order harmonic notch filter and the CHB cell dc voltage balancing cost function aim at improving the performance of the source current and CHB cell dc voltages, respectively. The proposed control techniques are then validated under steady-state and transient conditions through simulation and experiment with a laboratory prototype.