基于深度强化学习的面向无线充电的电动汽车路径规划  

Electric Vehicle Path Planning for Wireless Charging Based on Deep Reinforcement Learning

作　　者：靳勇[1] 陈政超 杨惠珍 JIN Yong;CHEN Zhengchao;YANG Huizhen(School of Computer Science and Engineering,Changshu Institute of Technology,Changshu,Jiangsu 215500,China;Changshu Dawei Transportation Technology Co.,Ltd.,Changshu,Jiangsu 215500,China)

机构地区：[1]常熟理工学院计算机科学与工程学院,江苏常熟215500 [2]常熟大为交通科技有限公司,江苏常熟215500

出　　处：《自动化应用》2025年第2期72-75,共4页Automation Application

基　　金：常熟市科技计划(社会发展)项目(CS202204);苏州市社会发展科技创新项目(SS202155)。

摘　　要：电动汽车的广泛应用使得电动汽车充电面临成本高、效率低和城市电网负荷大等一系列问题。为此,在行车道铺设无线充电线圈以实现电动汽车的无线充电,形式化了电动汽车无线充电调度问题;基于深度强化学习,提出了电动汽车调度算法,以使所有电动汽车在满足截止时间约束和能量约束下的总剩余电量最大。仿真实验分别从电动汽车的数量、充电路段的数量和截止时间的均值等方面分析了所提算法的性能。结果表明,所提算法的总剩余电量性能明显优于对比算法。The widespread application of electric vehicles poses a series of problems such as high cost,low efficiency,and high urban power grid load for electric vehicle charging.To achieve wireless charging of electric vehicles,wireless charging coils are installed on the driving lane,formalizing the scheduling problem of wireless charging for electric vehicles.Based on deep reinforcement learning,an electric vehicle scheduling algorithm is proposed to maximize the total remaining power of all electric vehicles while satisfying deadline and energy constraints.The simulation experiments analyzed the performance of the proposed algorithm from the aspects of the number of electric vehicles,the number of charging sections,and the average cut-off time.The results indicate that the total remaining power performance of the proposed algorithm is significantly better than that of the compared algorithms.

关 键 词：电动汽车 无线充电 路径规划 带限制的最短路径 深度强化学习 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]