Latest revision as of 20:01, 1 February 2021

Abstract

Train dwell time is one of the most unpredictable components of railway operations, mainly because of the varying volumes of alighting and boarding passengers. However, for reliable estimations of train running times and route conflicts on main lines, it is necessary to obtain accurate estimations of dwell times at the intermediate stops on the main line, the so-called short stops. This is a great challenge for a more reliable, efficient and robust train operation. Previous research has shown that the dwell time is highly dependent on the number of boarding and alighting passengers. However, these numbers are usually not available in real time. This paper discusses the possibility of a dwell time estimation model at short stops without passenger demand information by means of a statistical analysis of track occupation data from the Netherlands. The analysis showed that the dwell times are best estimated for peak and off-peak hours separately. The peak-hour dwell times are estimated using a linear regression model of train length, dwell times at previous stops and dwell times of the preceding trains. The off-peak-hour dwell times are estimated using a non-parametric regression model, in particular, the k-nearest neighbor model. There are two major advantages of the proposed estimation models. First, the models do not need passenger flow data, which is usually impossible to obtain in real time in practice. Second, detailed parameters of rolling stock configuration and platform layout are not required, which makes the model more generic and eases implementation. A case study at Dutch railway stations shows that the estimation accuracy is 85.8%-88.5% during peak hours and 80.1% during off-peak hours, which is relatively high. We conclude that the estimation of dwell times at short stop stations without passenger data is possible.

Original document

The different versions of the original document can be found in:

• http://resolver.tudelft.nl/uuid:4b83452e-ef28-4433-8242-57b115da3e82

• http://resolver.tudelft.nl/uuid:8f5c33a5-b6fe-4ec2-a222-629ad729a9a7

• https://onlinelibrary.wiley.com/doi/pdfdirect/10.1002/atr.1380

• http://resolver.tudelft.nl/uuid:4b83452e-ef28-4433-8242-57b115da3e82,

http://dx.doi.org/10.1002/atr.1380

• https://api.wiley.com/onlinelibrary/tdm/v1/articles/10.1002%2Fatr.1380,

https://onlinelibrary.wiley.com/doi/full/10.1002/atr.1380,

http://dx.doi.org/10.1002/atr.1380 under the license http://doi.wiley.com/10.1002/tdm_license_1.1

• https://onlinelibrary.wiley.com/doi/pdf/10.1002/atr.1380,

https://onlinelibrary.wiley.com/doi/full/10.1002/atr.1380,

https://repository.tudelft.nl/islandora/object/uuid%3A4b83452e-ef28-4433-8242-57b115da3e82,

http://resolver.tudelft.nl/uuid:4b83452e-ef28-4433-8242-57b115da3e82,

https://www.narcis.nl/publication/RecordID/oai%3Atudelft.nl%3Auuid%3A4b83452e-ef28-4433-8242-57b115da3e82,

https://repository.tudelft.nl/islandora/object/uuid:4b83452e-ef28-4433-8242-57b115da3e82/datastream/OBJ/download,

http://onlinelibrary.wiley.com/doi/10.1002/atr.1380/full,

https://academic.microsoft.com/#/detail/2344633803

• [ ]