Research on regenerative braking strategy for pure electric vehicles based on deep reinforcement learning algorithms

Electric vehicles represent a significant direction for the future development of new energy vehicles. This study aims to further extend the battery life of pure electric vehicles (PEVs) and ensure optimal regenerative braking efficiency under varying conditions such as speed, acceleration, vehicle mass, and road conditions. This research proposes a regenerative braking strategy based on a deep reinforcement learning optimization algorithm, specifically the Deterministic Policy Gradient (DDPG) algorithm within the framework of deep reinforcement learning, for energy management. This strategy considers the diverse characteristics of actual driving conditions and basic vehicle attributes. By analyzing the regenerative braking mechanism of PEVs, a reward function based on the State of Charge(SOC) of the battery is constructed to evaluate battery status. Simulation experiments demonstrate that the dynamic programming control strategy based on deep reinforcement learning achieves a 57.69% faster convergence rate compared to the traditional SAC (Soft Actor-Critic) electric energy control strategy for automotive applications. By comparing the reward curves of the two control algorithms, it is evident that the control strategy proposed in this study, which is based on the deep reinforcement learning algorithm, exhibits superior performance in terms of battery energy regenerative braking for PEVs.