Artificial neural network-based control for phase-shifted full-bridge DC–DC converter under constant power loads
摘要
Phase-shifted full-bridge DC–DC converters are widely employed for high-voltage conversion ratio in many applications such as telecommunication power supplies, industrial motor drives, renewable energy systems, electric vehicles, and battery charging systems. When fed with the tightly regulated converter, these loads act as constant power loads (CPLs). Further, constant current load, pulse power load, and CPL affect the performance and stability of the system leading to control loop instability and dynamic load response with substantial fluctuations or transients in the output. For regulated power output, the system stability should be maintained. The proportional-integral (PI) controller, commonly used for its simplicity, is evaluated against the artificial neural network (ANN) controller, which provides nonlinear and adaptive capabilities. Both strategies are assessed for voltage regulation, transient performance, and robustness to load variations. Simulation results show that the ANN controller substantially reduces overshoot, improves steady-state stability, and maintains better performance under varying load profiles compared with the PI controller. These findings are validated using OPAL-RT OP4610-XG field programmable gate array-based real-time simulator. These findings confirm the ANN controller’s potential for high-performance converter applications.