Purpose <p>The rising adoption of Plug-in Hybrid Electric Vehicles(PHEVs) for their fuel-saving and emissions-reducing benefits, existing energy management strategies often overlook engine transients, including performance limits and excess fuel during state transitions. This research aims to develop an advanced energy management system (EMS)that optimizes PHEV operation by integrating renewable energy sources, reducing fuel consumption and operational costs, while improving battery utilization and overall system efficiency.</p> Methods <p>A novel CPO-MCAGCN framework is proposed, combining the Crested Porcupine Optimizer (CPO) and a Multi-Component Attention Graph Convolutional Neural Network (MCAGCN). The CPO optimizes PHEV charging schedules, while the MCAGCN predicts overall power flow and charging efficiency, considering interactions among PV, battery, and fuel cell resources. The approach is validated through MATLAB simulations under varying environmental and load conditions and benchmarked against WOA, PSO, and DDQN methods.</p> Results <p>The results of the simulation prove that the suggested CPO-MCAGCN system is much more effective in comparison to the traditional ones, reaching a lower cost of operations of7,595.37$, a lower fuel consumption of 2.3L/100 km, and an optimal battery State of Charge(SoC) of 97.35%. The framework will guarantee better use of energy, higher operational velocity, and consistent performance in a wide range of situations.</p> Conclusion <p>The proposed framework offers a strong platform to the adaptive, real-time and grid-integrated EMS in PHEVs which offers an effective and affordable solution to the optimization of hybrid vehicle functioning in the city transportation grid in the future.</p>