首页|基于优先贝叶斯与深度Q学习的交通信号工程控制优化研究

基于优先贝叶斯与深度Q学习的交通信号工程控制优化研究

扫码查看
交通信号工程控制一直是城市交通管理中至关重要的组成部分.研究以城市路网中的最小单元—单点交叉口作为研究对象,引入深度Q学习算法,提出了一种新的动作空间设计方法,以解决传统方法难以应对交叉口各方向车流到达率不同的问题.研究通过融合贝叶斯与支持向量机,搭建行程时间预测模型.收集车辆路径数据,预测出行路径和相位需求.结果表明,两种交通状态的仿真实验中,在平峰时段研究所提控制方法在仿真时间超过 400 s后,车辆平均延误基本稳定在32 s~35 s之间.该研究主要针对交通信号控制调控策略进行了分析,希望能提高道路交通效率,为城市居民提供方便及可持续的出行体验.
Traffic signal engineering control optimization based on priority Bayes and deep Q learning
Traffic signal engineering control has always been an important part of urban traffic management.Taking single-point intersection,the smallest unit in the urban network,as the research object,this paper introduces deep Q learning algorithm and puts forward a new motion space design method to solve the problem that the traditional method is difficult to deal with the different arrival rates of traffic in each direction of the intersection.The travel time prediction model is built by integrating Bayes and support vector machine.Collect vehicle routing data to predict travel path and phase requirements.The results show that in the simulation test of the two traffic conditions,when the simulation time of the proposed control method exceeds 400s in the normal peak period,the average delay of the vehicle is basically stable between 32 and 35s.The research is expected to improve the efficiency of road traffic and pro-vide convenient and sustainable travel experiences for city dwellers.

deep Q learningtraffic signalscontrolsbayesian optimizationtraffic condition

张文利、甘新立、邹俊辉

展开 >

贵州理工学院,贵州 贵安 550003

深度Q学习 交通信号 控制 贝叶斯优化 交通状态

2024

自动化与仪器仪表
重庆工业自动化仪表研究所,重庆市自动化与仪器仪表学会

自动化与仪器仪表

CSTPCD
影响因子:0.327
ISSN:1001-9227
年,卷(期):2024.(11)