Particle Swarm Optimization based Multi-Step Prediction for Start-up Switching Sequence Optimization of NPC-DBSRC

Direct minimum current trajectory (MCT) control can induce beat-frequency oscillation during start-up process in dual-bridge series-resonant converter (DBSRC). To address this issue, this paper proposes a particle swarm optimization (PSO)-based multi-step predictive framework for neutral-point-clamped DBSRC (NPC-DBSRC) start-up process. The external phase-shift ratio and NPC duty cycle over two consecutive switching periods are jointly optimized offline under the designed fitness function with a continuity term, and the resulting switching sequence is downloaded as a lookup table (LUT) for real-time execution on a DSP. The method achieves dual optimization in both start-up duration and eliminating beat frequency oscillation. Simulation and prototype experiments validate the effectiveness and robustness of the proposed approach.