REAL-TIME ADAPTIVE CONTROL POLICY FOR INTERSECTION WITH DOWNSTREAM DETECTION

摘要

This paper considers the problem of control of time-varying road traffic at signalized intersections. Implementing realtime adaptive control for a road traffic system at a signalized intersection when only departure data is available is a long standing problem for traffic management systems. This study proposes a mathematical model and an algorithm for adaptive traffic control system for an isolated intersection when only downstream detectors are available. The proposed algorithm aims to minimize the average delay faced by vehicles. It defines phase terminating conditions using minimum and maximum green times and a threshold headway value. These minimum and maximum green time bounds for each phase are adjusted based on state of the phase in the immediate previous cycle and history of traffic data obtained from the past few cycles. In this methodology the control system monitors the performance at the end of each phase and if required, the control system re-sets the green time bounds. This paper proposes an approach for setting green time bounds, which involves prediction of queues computed by a linear regression mode. This control policy enables one to adapt to the stochastic and dynamic behavior of traffic flow at the intersection. The proposed algorithm is compared with the vehicle actuated methodology proposed in the Highway Capacity Manual (HCM 2000).