Predictive Control Strategies for Multilevel Power Converters and Electric Drives

This thesis investigates advanced predictive control strategies for power electronic converters and electric drive systems, with a focus on multilevel architectures and open-end winding configurations. The work follows a progressive approach, starting from two-level voltage source inverters used as a benchmark to evaluate and compare model-based and model-free predictive control techniques. Cascaded H-Bridge converters are then addressed, where the exponential growth of switching states is mitigated through reduced-complexity strategies such as tree-based optimization and adaptive look-up tables, enabling real-time implementation. In parallel, model-free predictive control approaches based on ARX models and recursive identification are developed to reduce dependency on accurate system models and improve robustness. Finally, the study focuses on open-end winding synchronous reluctance motor drives, demonstrating how predictive control can simultaneously regulate currents, floating capacitor voltage, and power factor, extending the torque–speed operating range. Both simulation and experimental results validate the effectiveness of the proposed methods, highlighting the trade-off between performance, robustness, and computational complexity, and confirming the potential of predictive control for next-generation electric powertrain and industrial drive applications.