Adaptive Signal Control for Overflow Prevention at Isolated Intersections Based on Fuzzy Control
Tianyu Yao, Cheng Zhang, Jing Zhao, Ankit Gupta, Satyajit Mondal
Abstract
Nonrecurrent congestion on urban roads may cause overflow, possibly leading to gridlock at intersections. Signal control is one of the most important measures to prevent such phenomena. For nonrecurrent congestion with uncertain bottleneck capacity, however, existing signal control methods have mainly adopted the strategy of cutting off the green signal of the overflow movement while the release priority of green time has not been considered. Consequently, the green time allocated to overflow movements is too long, aggravating overflow and congestion. Therefore, a signal control method is proposed to prevent overflow caused by uncertain bottleneck capacity. The proposed method extends existing real-time adaptive signal control to explicitly consider the risk of queue spillover. When an overflow risk is detected, the green times of overflow movements will be reassigned using fuzzy control to determine the release priority depending on the queue length and red time. Numerical studies were conducted for two scenarios with different downstream bottleneck capacities. The results demonstrate that the proposed method not only outperforms existing methods in respect of intersection efficiency but also reduces the risk of overflow. The proposed method achieves a 27.7% to 33.8% reduction in the maximum queue length of overflow movements compared with a traditional adaptive signal control method and a 10.9% to 11.0% decrease compared with a modified signal control method with a cutting-off strategy. The sensitivity analysis indicates that the average delay resulting from the proposed method is 51.1% and 18.0% less than the delays resulting from the traditional and modified methods, respectively.