Together with our partners at the University of Paderborn, we recently published an article in IEEE Transactions on Power Electronics entitled “Universal Direct Torque Controller for Permanent Magnet Synchronous Motors via Meta-Reinforcement Learning.” The article addresses highly automated controller synthesis for any permanently excited synchronous drives and demonstrates how a control strategy learned using data-driven algorithms can be transferred from simulative training to real-world laboratory applications. The methodology and empirical investigation presented represent an important step toward the autonomous commissioning of power electronic systems in heterogeneous application contexts in order to achieve optimal system performance with minimal (human) effort.
Link to open access publication: https://ieeexplore.ieee.org/document/11263992
DOI: 10.1109/TPEL.2025.3635741
