<p>The integration of Wind, Microturbine (MT), Photovoltaic (PV), and Electric Vehicle (EV) systems into microgrids (MGs) presents significant challenges due to renewable energy intermittency, bidirectional power flows, and dynamic load demands. Effective energy management is essential to ensure grid stability, economic operation, and enhanced renewable utilization. This study proposes a novel hybrid energy management framework based on Sea Horse Optimization (SHO) and a Dimension-Augmented Physics-Informed Neural Network (DAPINN), termed SHO-DAPINN. The SHO algorithm is employed to optimize energy dispatch, storage utilization, and power converter operations, while DAPINN forecasts short-term energy generation and demand using physics-guided learning. The proposed SHO-DAPINN approach is implemented in MATLAB and evaluated against benchmark methods, including the Bat Optimization Algorithm (BOA), Modified Super Twisting Algorithm (MSTA), Fuzzy-Sparrow Search Algorithm (FSSA), Spider Monkey Optimization (SMO), and Whale Optimization Algorithm (WOA). Simulation results demonstrate that SHO-DAPINN achieves the lowest total operational cost ($6.26 × 10⁵), outperforming all comparative methods. Additionally, the system exhibits strong adaptability to dynamic load and resource conditions, improves system stability, and achieves high renewable energy utilization and energy efficiency. These findings confirm that integrating intelligent forecasting with metaheuristic optimization significantly enhances MG performance. The proposed framework offers a cost-effective and scalable solution for future smart MGs, particularly in scenarios involving high renewable penetration and EV integration.</p>