A Multi-Objective Reinforcement Learning Based Energy Management Strategy for Electric Vehicles With Battery and Supercapacitor Integration

This paper proposes a Soft Actor-Critic (SAC)-based Energy Management Strategy (EMS) for All Electric Vehicles (AEVs) equipped with a Hybrid Energy Storage System (HESS). A custom-designed reward function is formulated to explicitly drive the SAC agent toward achieving three key objectives: extending battery lifetime, increasing vehicle mileage, and enhancing energy efficiency. The SAC agent, developed in Python, communicates with a Simulink-modeled environment over TCP/IP using the Simulink_gym wrapper. Simulation results demonstrate that the SAC-based EMS outperforms a conventional rule-based strategy in terms of State of Charge (SoC) preservation, power smoothing, and loss reduction. Future work includes real-time validation through Hardware-in-the-Loop (HIL) testing.