ADAPTIVE INCIDENT DETECTION ALGORITHM FOR SUBURBAN EXPRESSWAY

Incident is one of the major causes of non-recurring congestion on the expressways. Effective incident detection and management would greatly reduce non-recurring congestion delays caused by incidents. This paper aims to implement an incident detection algorithm for a suburban expressway route in Greater Bangkok, Thailand. The overly long sensor spacing, i.e., 5 km apart presents a challenge in developing an effective detection algorithm. Two classic algorithms, namely, California and McMaster algorithms were experimented. Based on the results, it was found that the comparative approach such as California algorithm is not suitable for a long span section due to its low detection rate and extremely long detection time. A single station algorithm like McMaster algorithm showed a more promising result and was then modified. By introducing a simple rule-based algorithm with an adaptive threshold, the Rule-Based McMaster algorithm yielded an exceptionally high detection rate and low false alarm rate.