Real-time energy management strategy for shore power hybrid energy supply system based on PG-MPC

Growing energy demands at ports and mounting environmental pressures have driven interest in hybrid shore power systems that integrate photovoltaic (PV) systems, energy storage systems (ESS), and the national grid. To realize stable and economical operation of the shore power hybrid energy supply system, a real-time particle swarm optimization (PSO) and generalized regression neural network (GRNN) integrated multi-level model predictive control strategy, designated as PG-MPC, is introduced for energy management. PG-MPC is composed of three distinct tiers: prediction, economic, and control. In the prediction tier, a ship shore power demand prediction method based on PSO-GRNN is developed. The economic tier accounts for the costs associated with electricity purchased from the grid and the losses in the PV and ESS. The control tier employs an enhanced MPC framework to correct system deviations in real time. Simulation analysis shows that the PSO-GRNN power prediction model outperforms traditional methods in predicting ship shore power demand. Compared to fault-tolerant MPC (FT-MPC) and PSO-MPC, the operating cost of PG-MPC is decreased by 4.33% and 5.09%, in turn. The results indicate that the developed PG-MPC approach has excellent control performance and costeffectiveness, providing theoretical support and practical references for optimizing energy management by integrating shore power technology and distributed power supply.