| Summary |
Chih-Liang Chien, 2008.02
Graduate Institute of Transportation Science, Tamkang University
This research addresses the relationship between the needed amount and spatial distributions of Probe Vehicle, and the travel time estimation errors, when practically using Probe Vehicles on signalized arterials to estimate travel time.
Tools of travel time’s data-collection can be categorized as Vehicle Detector and Probe Vehicle. Probe Vehicle is becoming widely used in Taiwan because of its lower setup cost. However, Probe Vehicle researches at present all focus on highway, non-signalized arterials, resulting in the lack of discussion in how traffic signals affect travel time in urban districts.
Therefore, the purpose of this research contains two main points: first, under limited amount of probe vehicles, how to efficiently use their positions to correct estimation errors of travel time on signalized arterials; second, if the number of Probe Vehicle becomes very popular in the future, how to select Probe Vehicles’ data in which locations to narrow down the errors within limited sampling.
This research uses experimental design and ANOVA (Analysis of Variance) to explore the relationship between Probe Vehicles’ spatial distribution and Space Mean Speed, and furthermore, uses Probe Vehicles’ spatial distribution as Artificial Neural Network’s variables to correct estimation errors. The research results show that the relationship between Probe Vehicles’ spatial distribution and Space Mean Speed is not obvious for the short link under 200 meters, but it is significant for the link longer than 200 meters. When Probe Vehicles’ spatial distribution is used as variables in Artificial Neural Network, travel time estimation errors can be efficiently corrected and more accurate signalized arterials travel time can be obtained. |
