This paper presents a new adaptive control strategy for co-coordinating and synchronizing signals in a network using the virtual-fixed-cycle concept. The strategy employs a distributed dynamic programming (DDP) algorithm to determine, at each intersection, phase durations that are constrained by minimum and maximum green times and, when operating in a coordinated mode, by a virtual cycle length and by virtual offsets that are updated based on real-time data. The goal is to provide continuously optimized controls in response to time-dependent variations in demand. The distributed algorithm is executed by means of a three-layer control architecture superimposed on a rolling horizon independent adaptive intersection controller. The strategy was implemented and field-tested successfully in a major U.S. sub-urban arterial corridor. (Author/publisher) For the covering entry of this conference, please see ITRD abstract No. E208120.
Abstract