A Self-tuning Signal Control Algorithm for Isolated Intersections Based on Time-space Diagrams

Abstract

In this study, a self-tuning signal control algorithm is developed for isolated intersections. Traffic dynamics at signalized intersections are represented on time-space diagrams using the shockwave theory and information from detectors installed upstream of intersections. Splits are incrementally adjusted so that the delay per cycle is gradually diminished. Cycles are modified to have an efficient use of the provided green times without causing the residual queues. Unlike most algorithms, the proposed method uses occupancy information and can manage traffic even when queues extend beyond detector locations. Simulation experiments on a four-phase intersection with different demand scenarios are performed to demonstrate efficiency of the developed algorithm. Hypothesis tests are conducted to statistically verify the efficiency comparison between the proposed method and the Webster formula. It is found that in case of fixed demand, the proposed method can optimize splits and cycle lengths with no worse performance measures than the optimal fixed-time signal settings. For the variable demand case, the result indicates that the algorithm can adjust splits and cycle lengths in response to the change of demand and provides better performance measures than the Webster formula. The proposed algorithm has demonstrated itself to be a potential split and cycle optimization for an adaptive signal control system.