Travel time estimation based on historical congestion maps and identification of consensual days

Abstract

In this paper, a new method for real-time estimation of traffic conditions and travel times on freeways is introduced. Using a combination of a Principal Component Analysis and a Gaussian Mixture Model, observation days of historical data are first clustered. Then, a consensus day is identified in each group as the most representative day of the community according to the congestion maps. Such a map is binary visualisation of the congestion propagation on the freeway giving more important to the traffic dynamics. Then, the first measurements of a new day are then used to determine in real-time which consensual day is closest to this new day. The past observations recorded for that consensus day are then used to predict future traffic conditions and travel times. This method is tested using two years of data collected on a French freeway and shows very encouraging results.