Travel time is a vital performance index in assessing transportation network performance.Vehicle speeds along any network route fluctuate and route travel times are essentiallyrandom. This technical note first examines travel time modelling and estimation via therandom modelling of speeds, and a general approach is presented for travel time estimationbased on speed distributions. Since normal and log-normal distributions are commonlyemployed for speed modelling in traffic engineering, travel time estimation is furtherdiscussed in the paper with regard to both distributions. Most probability distributions(including normal and log-normal) assume that a modelled random variable spreads over thewhole or half range of the real number axis, but, in practice, any traffic quantity of interestmakes sense only within a limited value range. In view of this, the concept of truncateddistributions is introduced, and, specifically, the probability features of truncated normal andlog-normal distributions are explored in the paper with regard to the discussed travel timeestimation issue.
展开▼