1. Home
  2. TRID
  3. View Record
https://www.nationalacademies.org/trb/events

Suitability of the travel time prediction model used by the Danish Road Administration in Finland

Tanskan tielaitoksen kayttaman matka-aikaennustemallin soveltuvuus Suomeen

According to the policy of the Finnish Road Administration, the Road Administration produces real time information and forecasts of the road status on road sections with the biggest risk of congestion by the year 2008. The Road Administration of Denmark uses short term forecasts of travel time as a part of the traffic management system on the very busy motorway M3 in the surroundings of Copenhagen. The model used by the Danish Road Administration is a regression model and it is based on the differences between one-minute-average travel times preceding the forecasting moment and corresponding historic averages. The purpose of the study was to evaluate the suitability of the prediction model used by the Danish Road Administration in Finnish conditions. The evaluation was based on test models that were based on travel time information collected by the travel time measurement system of the Finnish Road Administration. The study was based on two sites of a different type: the Ring Road I represented busy urban roads and the road section between Lahti and Heinola on main road 4 represented road sections on which holiday traffic caused congestion. The models were made by adapting the model used by the Danish Road Administration. However, the models were made to cover the whole day and all week days instead of rush hours only. In addition, the predicted quantity was selected to be the median of the travel time instead of the average travel time used in Denmark. Different data was used to make and to test the models. According to the test data, the 15 minute prediction model was able to predict freely flowing traffic 96-99 percent of the time correctly except for the Otaniemi-Perkkaa road section where the proportion was 83 percent. The proportion of correctly predicted travel times reduced as the road status got worse. Two prediction periods were used in the study: 15 and 30 minutes. In congested traffic, the proportion of correctly predicted travel times was 5-9 percent points worse for 30 minute forecasts than for 15 minute forecasts depending on the road section. It seemed that the prediction model was best suited for road sections with regular heavy congestion. A travel time prediction model like the one used by the Danish Road Administration which is based on regression model based on latest measurements and historic averages is recommended to be used on short road sections with regular heavy traffic or alternatively always freely flowing traffic. In addition, it is recommended that the models are made based on 5 minute medians. However, if one-minute-medians are used, it is recommended that the data filtering method is improved. It is recommended that the quality of forecasts is evaluated by the road status classes. This report may be found at http://alk.tiehallinto.fi/julkaisut/pdf/3201002-vtanskan_tielait_kayttama.pdf

Language

  • Finnish

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01038000
  • Record Type: Publication
  • Source Agency: TRL
  • ISBN: 951-803-729-9
  • Files: ITRD
  • Created Date: Dec 19 2006 10:09AM