基于TD3算法的网联汽车队列控制研究  

Research on Platoon Control of Connected Vehicles Based on TD3 Algorithm

作　　者：张鹏[1] ZHANG Peng(School of Automobile,Chang'an University,Xi'an 710064,China)

机构地区：[1]长安大学汽车学院,陕西西安710064

出　　处：《物流科技》2025年第7期55-59,共5页Logistics Sci Tech

基　　金：西安市科技计划项目(2022JH-GXQY-0074)。

摘　　要：随着汽车智能化和网联化技术的进步,汽车编队行驶逐渐成为缓解城市交通拥堵的有效手段之一。为了提高队列行驶的安全性和稳定性,文章提出了一种基于双延迟深度确定性策略梯度(Twin Delayed Deep Deterministic Policy Gradient,TD3)算法的网联汽车队列控制器。该控制器将队列间距误差和速度误差作为智能体的输入特征,设计了综合考虑队列安全性与稳定性的奖励函数,接着在SUMO仿真平台中搭建训练场景,并进行参数训练。结果表明,与模型预测控制方法相比,提出的TD3算法在安全行驶性能上有显著优势。With the advancement of automotive intelligence and connectivity technologies,vehicle platooning has gradually become one of the effective solutions to alleviate urban traffic congestion.To enhance the safety and stability of platoon driving,this study proposes a connected vehicle platoon controller based on the Twin Delayed Deep Deterministic Policy Gradient(TD3)algorithm.The proposed controller incorporates the inter-vehicle distance error and velocity error as input features for the agent,and a reward function is designed to explicitly account for both safety and stability requirements.Then,a training scenario is built in the SUMO simulation platform for parameter training.Experimental results demonstrate that,compared to the model predictive control approach,the TD3-based controller significantly improves driving safety and overall performance.

关 键 词：网联汽车队列 轨迹优化 深度强化学习 模型预测控制 

分 类 号：U491[交通运输工程—交通运输规划与管理]