R. Mussa, M.F. Selekwa
Pages: 57-70
Abstract
The continued growth of traffic on urban streets demands that coordinated traffic signals operate optimally even during the transition period between timing plans. Inefficient handling of the transition process can cause traffic perturbations that may take hours to recover. The need for better transition algorithms is heightened by the growing use of traffic responsive systems as well as signal preemption systems, which cause transition between timing plans to occur more frequently than in the past when the only transition was between time-of-day timing plans used in closed loop systems. This paper reports on the development of a transition procedure based on the quadratic optimization method. The procedure is aimed at reducing disutility measures to motorists during the transition period. The transition is modeled as a linear dynamic process, and the disutility measures are modeled as the sum of squares of the deviations of the coordination parameters—that is, cycle length, phase split, and offset—from the optimal values during the transition. Optimal control techniques are used to determine the step size and optimal number of steps necessary to complete the transition with minimum disruption to traffic flow. The proposed transition procedure does not go beyond the current and the target cycle lengths thus eliminating the need for the user to specify minimum and maximum values of splits and cycle lengths to accommodate pedestrians and other local intersection needs. A simulation study showed that the proposed transition method reduced the queue delay for minor streets in one scenario by an average of 5.88 percent, compared to the immediate transition method embedded in CORSIM. However, the method slightly increased delay on the major street by approximately 1.43% compared to CORSIM’s immediate method. The paper further discusses techniques that could improve the performance of the proposed algorithm under various geometric and traffic conditions.
Keywords: optimization; traffic signals; traffic modeling; signal transition