A comparative analysis of short-range travel time prediction methods

Giovanni Huisken, Eric C. van Berkum

Research output: Contribution to conferencePaperAcademicpeer-review

48 Downloads (Pure)

Abstract

Increasing car mobility has lead to an increasing demand for traffic information. This contribution deals with information about travel times. When car drivers are provided with this type of information, the travel times should ideally be the times that they will encounter. As a result travel times must be predicted, often on a short-term basis. Available data for such a prediction are spot measurements of speed and flow from dual induction loop detectors. In this contribution a prediction method that uses a neural network is described. The performance of the neural network approach is compared with two naïve methods that are currently in operation, using data from a short-range motorway site: the A13 motorway from The Hague to Rotterdam.
In order to be able to assess the performance of these methods it is imperative to use data on travel times. Since this data is not readily available, an estimation algorithm was selected where travel time is determined using speed and flow data from loop detectors. Five algorithms to estimate travel times were assessed using a data set with actually measured travel times through license plate recognition.
Results of the assessment of short-range travel time predictions show that the Artificial Neural Network (ANN) method significantly outperforms the Dynamic Travel Time Estimation (DTTE) method, which in turn outperforms the Static Travel Time Estimation (STTE) method.
Original languageEnglish
Number of pages21
Publication statusPublished - 12 Jan 2003
Event82nd Transportation Research Board (TRB) Annual Meeting 2003 - Washington, United States
Duration: 12 Jan 200316 Jan 2003
Conference number: 82

Conference

Conference82nd Transportation Research Board (TRB) Annual Meeting 2003
CountryUnited States
CityWashington
Period12/01/0316/01/03

Fingerprint

Travel time
Neural networks
Railroad cars
Detectors

Keywords

  • METIS-209286

Cite this

Huisken, G., & van Berkum, E. C. (2003). A comparative analysis of short-range travel time prediction methods. Paper presented at 82nd Transportation Research Board (TRB) Annual Meeting 2003, Washington, United States.
Huisken, Giovanni ; van Berkum, Eric C. / A comparative analysis of short-range travel time prediction methods. Paper presented at 82nd Transportation Research Board (TRB) Annual Meeting 2003, Washington, United States.21 p.
@conference{e61ad66dbb424af397f7a84fd5223a45,
title = "A comparative analysis of short-range travel time prediction methods",
abstract = "Increasing car mobility has lead to an increasing demand for traffic information. This contribution deals with information about travel times. When car drivers are provided with this type of information, the travel times should ideally be the times that they will encounter. As a result travel times must be predicted, often on a short-term basis. Available data for such a prediction are spot measurements of speed and flow from dual induction loop detectors. In this contribution a prediction method that uses a neural network is described. The performance of the neural network approach is compared with two na{\"i}ve methods that are currently in operation, using data from a short-range motorway site: the A13 motorway from The Hague to Rotterdam.In order to be able to assess the performance of these methods it is imperative to use data on travel times. Since this data is not readily available, an estimation algorithm was selected where travel time is determined using speed and flow data from loop detectors. Five algorithms to estimate travel times were assessed using a data set with actually measured travel times through license plate recognition.Results of the assessment of short-range travel time predictions show that the Artificial Neural Network (ANN) method significantly outperforms the Dynamic Travel Time Estimation (DTTE) method, which in turn outperforms the Static Travel Time Estimation (STTE) method.",
keywords = "METIS-209286",
author = "Giovanni Huisken and {van Berkum}, {Eric C.}",
year = "2003",
month = "1",
day = "12",
language = "English",
note = "82nd Transportation Research Board (TRB) Annual Meeting 2003 ; Conference date: 12-01-2003 Through 16-01-2003",

}

Huisken, G & van Berkum, EC 2003, 'A comparative analysis of short-range travel time prediction methods' Paper presented at 82nd Transportation Research Board (TRB) Annual Meeting 2003, Washington, United States, 12/01/03 - 16/01/03, .

A comparative analysis of short-range travel time prediction methods. / Huisken, Giovanni; van Berkum, Eric C.

2003. Paper presented at 82nd Transportation Research Board (TRB) Annual Meeting 2003, Washington, United States.

Research output: Contribution to conferencePaperAcademicpeer-review

TY - CONF

T1 - A comparative analysis of short-range travel time prediction methods

AU - Huisken, Giovanni

AU - van Berkum, Eric C.

PY - 2003/1/12

Y1 - 2003/1/12

N2 - Increasing car mobility has lead to an increasing demand for traffic information. This contribution deals with information about travel times. When car drivers are provided with this type of information, the travel times should ideally be the times that they will encounter. As a result travel times must be predicted, often on a short-term basis. Available data for such a prediction are spot measurements of speed and flow from dual induction loop detectors. In this contribution a prediction method that uses a neural network is described. The performance of the neural network approach is compared with two naïve methods that are currently in operation, using data from a short-range motorway site: the A13 motorway from The Hague to Rotterdam.In order to be able to assess the performance of these methods it is imperative to use data on travel times. Since this data is not readily available, an estimation algorithm was selected where travel time is determined using speed and flow data from loop detectors. Five algorithms to estimate travel times were assessed using a data set with actually measured travel times through license plate recognition.Results of the assessment of short-range travel time predictions show that the Artificial Neural Network (ANN) method significantly outperforms the Dynamic Travel Time Estimation (DTTE) method, which in turn outperforms the Static Travel Time Estimation (STTE) method.

AB - Increasing car mobility has lead to an increasing demand for traffic information. This contribution deals with information about travel times. When car drivers are provided with this type of information, the travel times should ideally be the times that they will encounter. As a result travel times must be predicted, often on a short-term basis. Available data for such a prediction are spot measurements of speed and flow from dual induction loop detectors. In this contribution a prediction method that uses a neural network is described. The performance of the neural network approach is compared with two naïve methods that are currently in operation, using data from a short-range motorway site: the A13 motorway from The Hague to Rotterdam.In order to be able to assess the performance of these methods it is imperative to use data on travel times. Since this data is not readily available, an estimation algorithm was selected where travel time is determined using speed and flow data from loop detectors. Five algorithms to estimate travel times were assessed using a data set with actually measured travel times through license plate recognition.Results of the assessment of short-range travel time predictions show that the Artificial Neural Network (ANN) method significantly outperforms the Dynamic Travel Time Estimation (DTTE) method, which in turn outperforms the Static Travel Time Estimation (STTE) method.

KW - METIS-209286

M3 - Paper

ER -

Huisken G, van Berkum EC. A comparative analysis of short-range travel time prediction methods. 2003. Paper presented at 82nd Transportation Research Board (TRB) Annual Meeting 2003, Washington, United States.