Adaptive Energy Management in Dual-Motor Electric Vehicles Using Deep Deterministic Policy Gradient

Authors:

Abstract
With dual-motor systems emerging as a key trend in electric vehicles, energy management strategies (EMSs) for such architectures remain underexplored. This study presents an EMS based on deep deterministic policy gradient (DDPG) for dual-motor electric vehicle, comparing an average-efficiency reward function (DDPG-AE) and a novel efficiency-differential reward function (DDPG-DE), the latter engineered to exploit front–rear motor efficiency differentials without relying on empirical parameters, thereby markedly enhancing energy utilization and overall performance. Compared with the dynamic programming (DP) benchmark, under the new European driving cycle (NEDC), DDPG-AE and DDPG-DE reached 91.01% and 93.12% of the optimal benchmark energy consumption, respectively. Under random driving conditions, these algorithms achieve the benchmark energy consumption of 90.61% and 92.77%, respectively. These results confirm the proposed strategy’s superior energy efficiency and adaptability compared with traditional average-torque distribution and DP-based approaches.
Funding Information
  • Science Fund for Distinguished Young Scholars of Guangxi Province (2023GXNSFFA026013)
  • National Natural Science Foundation of China (2021NSFC52175081, 2024NSFC52475097)

This publication has 40 references indexed in Scilit: