Estimation of urban link travel time plays a vital role on real-time trafficinformation. Both on-line analysis and decision-making process need accurate linktravel time to figure out bottlenecks. This paper reviews contemporary applications ofprobe vehicle data and models about path travel time estimation. Based on past studies,this paper adopts a Bayesian method to make estimation on path travel time. To solveposterior probability of Bayesian model, this paper uses a reversible jump Markov chainto explore possible solutions. Based on Metropolis-Hasting sampling, suitable updatestep size is chosen to iterate through link parameters and map-matching results of GPSdata. In comparison with high-frequency sampling vehicle’s trajectory, results fromBayesian method outperform estimates from other benchmark methods. |