This paper presents a new procedure for signal coordination along oversaturated arterials. It is based on queue management and efficient green time utilization concepts. The main idea is to dynamically manage queue formation and dissipation on system links such that traffic flow is maximized and reasonable through bands are attained. This is achieved by efficiently using green time, preventing intersection blockage, accounting for the non-steady state conditions and providing dynamic, time-dependent control. The problem is formulated as a throughput maximization subject to state and control variables. A solution using Genetic Algorithms (GAs) is then presented. Results show that offsets and green times were dynamically changed as a function of demand and queue lengths. With long queues present, the algorithm utilized appropriate (negative) offsets to reduce or eliminate these queues, and then gradually moved to positive offsets to provide forward green bands, hence normal traffic progression. The algorithm has good potential for on-line real time implementation in an Intelligent Transportation Systems (ITS) environment. (A)
Abstract