Development of Bayesian Multi-State Travel Time Reliability Models

The objective of this project is to develop a Bayesian multi-state travel time reliability approach for modeling travel time uncertainty under various traffic conditions. The reliability of travel time is a key performance index of transportation system and has been a major transportation research area. Reliability is one of the four key focus areas of the Strategic Highway Research Plan (SHRP2). Travel time is affected by multiple factors such as traffic condition, weather, incidents etc. Many of these factors are random in nature and stochastic modes should be used in modeling the uncertainty associated with travel time. Traditionally, uni-mode distributions have been adopted for travel time reliability modeling and the log-normal distribution has been the most popular model. In recent years, the multi-state travel time reliability model has been proven to be a superior alternative by providing substantial improved data fitting, scientifically sound interpretation, as well as close relationship with the underline traffic flow characteristics. Majority of the current stochastic models, however, focus primarily on provide the best fitting for the travel time data. Limited researches have been conducted to link travel time uncertainty with traffic conditions and other external factors. Part of the reason is that traditionally used uni-mode distributions lack the flexibility to accommodate variation in travel time, let alone the complex interaction with external factors. On contrast, the multi-state model used a two-level structure to represent 1) the probability of encountering a traffic delay, and 2) the distribution characteristics of travel time in both delay or non-delay conditions. Previous studies have shown that the parameters of the multi-state model are directly related to the time of day. The results fit the intuition that the probability of encountering traffic delay is much higher during peak hours. However, previous studies are exploratory in nature and had not quantitatively evaluated the relationship between traffic conditions and parameters of the multi-state models. To establish quantitative relationship between traffic condition and the key parameters of mult-state models, i.e., the probability of encountering delay and the distribution parameters for each travel time state, will significantly increase our understanding of the relationship between traffic condition and travel time. Due to the complexity of the problem, traditional Expectation-Maximization (EM) algorithm for model fitting is not computationally attractive. A full Bayesian approach will be more appealing for both research and practical purposes. The results of this study will benefit congestion management for traffic management authorities as well travel time prediction for individual travelers.

    Language

    • English

    Project

    • Status: Active
    • Funding: $90560.00
    • Contract Numbers:

      DTRT12-G-UTC03

    • Sponsor Organizations:

      Mid-Atlantic Universities Transportation Center

      Pennsylvania State University
      201 Transportation Research Building
      University Park, PA  United States  16802-4710

      Research and Innovative Technology Administration

      University Transportation Centers Program
      1200 New Jersey Avenue
      Washington, DC  United States  20590
    • Performing Organizations:

      Virginia Polytechnic Institute and State University, Blacksburg

      Virginia Tech Transportation Institute
      3500 Transportation Research Plaza
      Blacksburg, VA  United States  24061
    • Principal Investigators:

      Guo, Feng

    • Start Date: 20120901
    • Expected Completion Date: 0
    • Actual Completion Date: 20131231
    • Source Data: RiP Project 33630

    Subject/Index Terms

    Filing Info

    • Accession Number: 01472290
    • Record Type: Research project
    • Source Agency: Mid-Atlantic Universities Transportation Center
    • Contract Numbers: DTRT12-G-UTC03
    • Files: UTC, RiP
    • Created Date: Feb 14 2013 1:00AM