Adaptive Ramp Metering Control for Urban Freeway Using Large-Scale Data
Jiming Chen,Weixin Lin,Zidong Yang,Jianyuan Li,Peng Cheng
DOI: https://doi.org/10.1109/tvt.2019.2937880
IF: 6.8
2019-01-01
IEEE Transactions on Vehicular Technology
Abstract:Urban freeway traffic control is of great importance for traffic management and intelligent transportation systems. Various approaches have been proposed to relieve urban freeway traffic jam, among which, ALINEA, a ramp metering strategy, is commonly implemented with fixed triggering threshold and static controller parameter. However, such a strategy may not be able to effectively alleviate the traffic congestion while maintaining certain ramp throughput due to two reasons: i). the congestion threshold can be time-varying due to different factors, such as segment ID, weather condition, time, and etc. ii). The congestion evolution patterns are time-varying even for the same segment. In this paper, based on over 890 million records of vehicles collected on ramps in Hangzhou, China, we established dynamic congestion threshold for each road segment with external factors. Based on such dynamic congestion threshold, we further clustered the congestion evolution patterns, and designed adaptive ramp controller which could switch the controller parameter according to the predicted congestion evolution pattern. Finally, in order to show the performance among different strategies, we introduced three baseline groups, which are without Controller, ALINEA controller, and Direct RBF(radial basis function)-neural network controller, respectively.The evaluation of proposed controller design over real large-scale data indicated that our method achieves 8.4(7.2), 4.62(9.48) efficiency improvement in terms of average speed in km/h (average traffic flow in veh/h) than the performance with normal ALINEA controller and RBF-neural network controller respectively.