Development of an ECMS-CESO real-time energy management strategy for hybrid electric vehicles

In today’s transportation, hybrid electric vehicles (HEVs) are vital for lowering emissions and fuel consumption in modern transportation. This study modelled and simulated the reconfigured HEV originally designed as a power-split into a pure Series Hybrid Electric Vehicle. In the reconfigured vehicle, the internal combustion engine is completely isolated from the driving wheels and functions only to operate a generator at its optimal brake-specific fuel consumption point. To control the power flow of SHEV, a real-time adaptive energy management strategy called ECMS-CESO was developed. A high-fidelity MATLAB/Simulink model of the reconfigured selected vehicle was built. The developed ECMS-CESO was compared to Model Predictive Control, Instantaneous Optimal Control, Twin Delayed Deep Deterministic Policy Gradient, and Soft Actor–Critic across standard driving cycles. The findings demonstrate that the performance of ECMS-CESO fuel economy has been overall improved by an average of 12.09%, 4.91%, 2.91%, and 0.06%, when compared to MPC, IOC, SAC, and TD3, respectively, while maintaining the battery’s state of charge within the acceptable range. ECMS-CESO integrates a disturbance observer to adapt the equivalence factor in real time. This approach bridges the gap between deterministic and reinforcement learning approaches, offering both robustness to transient load variations and low computational demand.