Prototype of a stochastic optimal control based approach to incident-responsive coordinated ramp control

This paper presents a real-time coordinated ramp control approach in response to incident-induced traffic congestion on freeways. The proposed control methodology includes two major functions to alleviate in real time incident impact on multi-segment traffic flows of freeways, i.e. (1) identification of control zone-based congestion patterns, and (2) stochastic optimization of group-based ramp metering control. To reduce congestion, the modified generalized sequential probability ratio testing (M-GSPRT) technology together with stochastic optimal control methods are used, respectively, in the above two mechanisms. With the aid of the Paramics microscopic traffic simulator, numerical results obtained under various simulated incident-induced congestion conditions indicate the feasibility of the proposed method. Numerical test results indicate both the feasibility of the proposed ramp control method in responding to diverse lane-blocking incident conditions, and its efficiency, particularly in improving the averaged hourly throughput by 67.5% in medium-volume incident cases, compared with the no-control strategy.