基于深度强化学习的入口匝道流量调控方法

打开文本图片集

中图分类号：U491.4 文献标识码：A DOI：10.3969/j.issn.1674-8484.2025.04.009

Abstract：Given that current research on ramp control methods based on reinforcement learning （RL） has notthoroughlyaddressed keyissues suchas learning costand policy transferabilityduringpolicy training，the practicalapplicationof these control strategies remains challenging.Toaddress this issue，this paperproposed a RLapproach aimed at optimizing ramp control strategies and conducted extensive simulation experiments to investigate theportabilityoftheproposed method.Arampcontrolmodel wasconstructed，andamodeltraining method based on deep reinforcement learning was proposed.The bottleneck ina certain convergencearea of Rongwu Expressway in themain external road networkof Xiongan Districtwas selectedas theexperimental scenario.ThedeepRLalgorithmwasusedtotraintherampmetering model，and theperformanceof thecontrol strategyduring the trainingprocess wascompared withtheclassical rampcontrolmethod，therebyquantitatively analyzing the learning cost.Diferent simulation modelsand multiplesetsof model parameters wereselected as the test environment，and the influenceof the diferences between the training environment and the test environment on the control strategy was analyzed.The results show that when the diference between the training environment and the test environment iswithin 20% ，theRL controlmethod is significantly superior to theclassicalramp control method in improving the trafic eficiency.However，when the difference exceeds 20% ，the effects of the two methods are comparable.

Keywords：rampmetering;reinforcement learning;transferability; learning cost

匝道控制是调节高速公路及城市快速路瓶颈路段交通流运行状态的常用控制手段。（剩余13863字）