The impact analysis of traffic incident and prediction model on travel time under incident condition

Abstract

Understanding the impact of traffic incidents, not only can help decision-makers choose a better strategy, but also provide travel recommendations for travelers. This paper uses real GPS data collected from probe vehicle to analyze the impact of incident in urban traffic network. Queuing length and incident duration are used to evaluate impact level incident. Then, a traffic incident pattern classification method based on queuing length variation is proposed to more clearly understand the characteristics of traffic incident. At last, we propose a prediction model on the time that a vehicle takes to pass through the location where traffic incident happens by using the GPS position and velocity information.