Negative Freeway Feedback On-ramp Metering Based On Q-learning
Traffic congestion has caused severe problems in the normal operation of freeways,and various methods have been employed to relieve the congestion.In this paper,we propose a negative feedback model based on density control(NFMD)method to control the traffic inflow from ramp merging areas and keep the traffic density around the critical value.It can maintain an optimal operation to the two on-ramps metering thus maximize the utilization of mainstream.The method including traffic state,action space as well as the reward function is designed based on Markov Decision Process and we utilize the Q-learning algorithm in our model.We demonstrate our method through a benchmark network with two on-ramps.The results show that the NFMD method can keep mainstream traffic flow close to mainstream freeway capacity and successfully regulate traffic congestion.
Jianguo Zhang Xianjun Yang Lingfeng Lin Chunyang Zhang
College of Mathematics and Computer science,Fuzhou University,China;Fujian Key Lab for Systems and Applications on Intelligence
国内会议
福州
英文
1-6
2015-12-11（万方平台首次上网日期，不代表论文的发表时间）