The use of simulation in the design of a road transport incident detection algorithm

Automatic incident detection is becoming one of the core tools of urban traffic management, enabling more rapid identification and response to traffic incidents and congestion. The existing traffic detection infrastructure within many urban areas (often installed for traffic signal optimisation) provides urban traffic control systems with a near continuous stream of data on the state of traffic within the network. The creation of a computer simulation model to replicate such a data stream can therefore provide a facility for the development of accurate congestion detection and warning algorithms. This paper reports on the development of firstly an urban traffic control simulation platform and secondly a new incident detection system (RAID – Remote Automatic Incident Detection), with the facility to use the simulation platform as an integral part of the design and calibration process. A commercial traffic simulation model was augmented with additional code developed to replicate the traffic status messages generated by a real urban traffic control system using the 250 millisecond data produced from single inductive loop detectors. These messages were then fed directly into an offline version of the RAID detection algorithm. This approach allowed the network managers to both visualise the system in operation and assess the impact on RAID detection accuracy of changing rule settings or traffic detector placements without the need for any alterations to the on-street detection infrastructure .