基于深度确定性策略梯度算法的新能源汽车动力电池充电管理方法

打开文本图片集

中图分类号：U463.63 文献标志码：A DOI：10.20104/j.cnki.1674-6546.20250170

【Abstract】To enhance batery charging eficiency，this studyaddresses the challenges in the fast charging process， includingchargingtime，theralsafety，andbaterylifespanduringfastcharging，alongsidethepooradaptabilityoftraitioal strategiesand insuficient state estimationaccuracyAninteligent chargingstrategy basedon Deep ReinforcementLearning （DRL）isproposed.Athermoelectric-coupledbaterymodel isdeveloped，employinganExtendedKalmanFilter（EKF）for accurate State Of Charge （SOC）estimation.The charging strategy isoptimized usingthe Deep Deterministic Policy Gradient （DDPG）algorithm.Policy training is acomplished through an Ornstein-Uhlenbeck （OU）noiseexploration mechanism，a multi-objectiverewardfunction，combined with experiencereplayand softupdate techniques.Experimentalvalidationonthe MATLAB platform demonstrates that the proposed method achieves acharging time of1625s.While maintaining a comparablechargingtime toconventional methods，itreduces themaximum interal batery temperature byapproximately （204号 2C ，realizing coordinated optimization of charging eficiency and thermal safety.

Key Words：Power battery，Charging management，Deep Reinforcement Learning （DRL）， Thermoelectric coupling model， State Of Charge （SOC） estimation，Deep Deterministic Policy Gradient （DDPG）

1前言

随着新能源汽车的普及，快速充电技术成为缓解用户里程焦虑的关键。（剩余10650字）