This report develops a modelling scheme within an urban traffic control (UTC) system so that its performance can be monitored on line as part of its daily traffic operation. A review is given of the modelling techniques employed for off line signal optimisation, on line adaptive traffic control, and on line performance monitoring. It suggests that a fruitful approach to on line performance monitoring would be to estimate platoon delays with a TRANSYT like model within the framework of a control system such as SCATS. The accuracy of TRANSYT was then assessed by validation with floating car survey data, and by a comparative study with the microscopic model NETSIM. The assessment confirms that TRANSYT modelling techniques would be appropriate for system monitoring. A scheme that makes use of the traffic flow profiles retrievable from stop line detectors in SCATS is proposed. It consists of a recursive least squares estimator for link flow estimation, and a queue length/ delay estimation procedure. This procedure models queue formation and dissipation at a stop line from the predicted arrival profile and the actual stop line departure profiles. The scheme was tested with simulated data, and with SCATS data collected in a field study. Both the queue lengths at the start of a green period, and the maximum back of queue that was formed as the head of a platoon joined the stationary queue at a stop line, were recorded. The results confirm the validity of the proposed scheme. The stop line flow profiles were also successfully utilised for automatic incident detection. Lane by lane queue length estimation was also found to be feasible. It is recommended that the scheme be employed in a control system for performance monitoring. A scenario of the future of adaptive UTC in the context of dynamic traffic management is also provided (A). The ISBN of the microfiche version is 0-86910-396-2.
Abstract