智能微调的混合动力汽车能量管理策略研究

Research on intelligent fine-tuning of hybrid electric vehicle energy management strategy

为提高混合动力汽车的燃油经济性,利用深度确定性策略梯度(deep deterministic policy gradient,DDPG)算法优化自适应等效能耗最小策略(adaptive equivalent consumption minimization strategy,A-ECMS)。在基于SOC反馈的基础上考虑上一时刻的需求功率、发动机和电机扭矩来调整等效因子,优化发动机与电机的扭矩输出,达到保持SOC和减少油耗的目的。由FTP75工况训练可知,提出的控制策略可以在原有的强化学习的能量管理策略的基础上用更短的时间学习到更好的控制动作,动作优化了6.05%,训练效率提升了60%,百公里等效油耗同基于规则相比减少了7.07%。在测试工况NEDC上的百公里等效油耗同基于规则能量管理策略相比油耗减少了2.27%,燃油经济性有明显改善。

Energy and environmental issues restrict the development of the traditional automobile industry.Hybrid vehicles with the advantages of traditional fuel vehicles and pure electric vehicles have become a research hotspot.The energy management strategy is the core of the hybrid vehicle control system.The quality of the control strategy directly determines the fuel economy and emission performance of the vehicle.This paper is inspired by the Adaptive Equivalent Consumption Minimization Strategy(A-ECMS).Using reinforcement learning algorithm to optimize the control performance of A-ECMS based on SOC feedback,an intelligently fine-tuned hybrid electric vehicle energy management strategy is proposed.The main idea is to adjust the equivalent factor of the output by considering the required power,engine and motor output torque at the previous moment based on the SOC feedback,and optimize the torque output of the engine and the motor to achieve the purpose of maintaining SOC and reducing fuel consumption.It can be seen from the training results on the FTP75 drive cycle that the proposed energy management strategy can learn better control actions in a shorter time on the basis of the original reinforcement learning energy management strategy,the action is optimized by 6.05%,the training efficiency is increased by 60%.The equivalent fuel consumption per 100 kilometers is 7.62 L,which is 7.07%lower than the rule-based energy management strategy,and only 0.01 L more than the dynamic programming energy management strategy.In order to verify the superior performance of the proposed energy management strategy,the trained control strategy is used for comparison and verification on NEDC drive cycle.The equivalent fuel consumption per 100 kilometers under the NEDC drive cycle is 7.74 L,which is 2.27%lower than that of the rule-based energy management strategy,and 0.02 L more than the dynamic programming energy management strategy.The fuel economy is significantly improved.From the simulation results,it can be seen that the proposed intelligent fine-tuning hybrid electric vehicle energy management strategy can not only improve the fuel economy of the vehicle,but also has good applicability.