Data-Driven Particle Filter for Travel Time Prediction
The research presented in this paper develops a data-driven particle filter to predict travel times by sampling from historical data. In the proposed method, each particle corresponds to a travel time sequence from a database of historical data. The particle weight is calculated using a dissimilarity measure between measurement and particle sequences. A resampling method is developed in the data-driven particle filter to eliminate particles with low weights and re-select samples according to the probability of each track. Travel time predictions are computed by aggregating the weighted travel times of each particle. A freeway stretch from Newport News to Virginia Beach is selected to test the proposed algorithm using five-minute aggregated traffic data in 2010 provided by INRIX. The travel time prediction results during the summer season demonstrate that the proposed method outperforms two Kalman filter methods by reducing the prediction error by 30% and 57%.
-
Supplemental Notes:
- This paper was sponsored by TRB committee ABJ30(3) Travel Time, Speed and Reliability
-
Corporate Authors:
500 Fifth Street, NW
Washington, DC United States 20001 -
Authors:
- Chen, Hao
- Rakha, Hesham
-
Conference:
- Transportation Research Board 92nd Annual Meeting
- Location: Washington DC, United States
- Date: 2013-1-13 to 2013-1-17
- Date: 2013
Language
- English
Media Info
- Media Type: Digital/other
- Features: Figures; Maps; References; Tables;
- Pagination: 16p
- Monograph Title: TRB 92nd Annual Meeting Compendium of Papers
Subject/Index Terms
- TRT Terms: Algorithms; Data collection; Freeways; Information processing; Kalman filtering; Mathematical prediction; Particles; Probability; Travel time
- Geographic Terms: Virginia
- Subject Areas: Data and Information Technology; Highways; Operations and Traffic Management; I73: Traffic Control;
Filing Info
- Accession Number: 01473664
- Record Type: Publication
- Report/Paper Numbers: 13-4392
- Files: TRIS, TRB, ATRI
- Created Date: Feb 25 2013 8:52AM