Plug-in Hybrid Electric Vehicles (PHEVs) play a vital role in advancing sustainable transport, offering flexibility by switching between electric and internal combustion modes. However, managing energy flow and power distribution in real-world driving remains complex. This paper presents a data-driven control framework that integrates machine learning (ML) and fuzzy logic to optimize PHEV power management. The ML model predicts battery state of charge (SOC) using real-time driving data, while fuzzy logic determines the optimal distribution of power across four modes: full electric, series hybrid, parallel hybrid, and full internal combustion. The framework dynamically adapts to diverse driving conditions, consistently minimizing fuel use and emissions through electric propulsion. Simulation using the Worldwide Harmonized Light Vehicles Test Cycle demonstrates improved fuel economy, reduced emissions, and an 84 km all-electric range with 80% battery utilization. The approach also enhances gasoline-equivalent fuel efficiency by 20%. These results underscore the framework’s potential for future PHEV applications

Plug-in Hybrid Electric Vehicles (PHEVs), Fuzzy Logic Control, Machine Learning (ML), Energy Management System (EMS), Real-World Driving Cycles, Fuel Efficiency, Emission Reduction, Battery State of Charge (SOC) Prediction