Sie können Operatoren mit Ihrer Suchanfrage kombinieren, um diese noch präziser einzugrenzen. Klicken Sie auf den Suchoperator, um eine Erklärung seiner Funktionsweise anzuzeigen.
Findet Dokumente, in denen beide Begriffe in beliebiger Reihenfolge innerhalb von maximal n Worten zueinander stehen. Empfehlung: Wählen Sie zwischen 15 und 30 als maximale Wortanzahl (z.B. NEAR(hybrid, antrieb, 20)).
Findet Dokumente, in denen der Begriff in Wortvarianten vorkommt, wobei diese VOR, HINTER oder VOR und HINTER dem Suchbegriff anschließen können (z.B., leichtbau*, *leichtbau, *leichtbau*).
Dieses Kapitel geht den erheblichen Auswirkungen von Wetterbedingungen auf Zugverspätungen und Pünktlichkeit nach und betont die Notwendigkeit, Wetterdaten in Vorhersagemodelle einzubeziehen, um die Genauigkeit zu erhöhen. Die Studie konzentriert sich auf die Strecke Oslo-Trondheim und nutzt maschinelle Lernmodelle, insbesondere den LSTM-Algorithmus, um Verzögerungen auf Grundlage von Wettermerkmalen wie Temperatur und Niederschlag vorherzusagen. Wichtige Ergebnisse zeigen, dass Wettereinflüsse Zugverspätungen erheblich beeinflussen, wobei extreme Bedingungen zu längeren Verspätungen führen. Die Forschung unterstreicht auch die Bedeutung von Clustern und Gradation als Merkmale, die die Anwendung von Wetterdaten in Vorhersagemodellen unterstützen. Darüber hinaus identifiziert die Studie Verzögerungsspitzen im Zusammenhang mit Staus und Wartungsarbeiten aufgrund extremen Wetters und liefert wertvolle Erkenntnisse für die Verkehrsplanung und -steuerung. Die Schlussfolgerungen aus dieser Studie unterstreichen die Bedeutung der Integration von Wetterelementen in Vorhersagemodelle für Verspätungen, um die allgemeine Widerstandsfähigkeit und Zuverlässigkeit von Eisenbahnsystemen zu verbessern.
KI-Generiert
Diese Zusammenfassung des Fachinhalts wurde mit Hilfe von KI generiert.
Abstract
Estimation of train delays is crucial for customer information. One cause of the train delays can be easily blamed on the weather. The effect of weather on signaling and dispatching can be indirectly articulated from the arrival and departure delays at the stations. The study uses Norwegian delay data from 2021, 2022 and parts of 2023. This data contains scheduled and actual departure and arrival times of trains on the Dovre line between Oslo and Trondheim. This article talks about acquiring freely available weather data using APIs at the stopping station and checking the effect of weather elements on the departure delay. Weather elements correlated with the departure delays were rainfall (precipitation) and temperature. This study attempts to articulate the quantitative nature of the effect of these weather elements on the departure delays of the trains. The delay prediction model uses different neural network algorithms. The prediction results from different algorithms are compared to provide a deeper insight into the effect of weather on the delay characteristics.
1 Introduction
Due to the unpredictability of weather conditions, transportation networks are experiencing unprecedented challenges. The railway system is thought to be the least weather-sensitive of the transport modes, often expected to run when other forms of transport are disrupted [1]. However, weather conditions can influence the railway traffic and cause delays. Additionally, extreme weather can affect the train operator’s performance directly or indirectly through disturbances in the railway infrastructure.
It is hard to identify the best implementation options to lessen the negative effects of disruptions. However, it is widely acknowledged that transportation planning and management organizations must increase system resilience to minimize losses and disruptions [2]. This is partly caused by an inadequate knowledge of how various weather phenomena and specific extreme weather occurrences affect the functionality of transportation systems. Uncertainty exists over how effects vary across various event types and geographically and temporally for various means of transportation within the same event. Therefore, providing a comprehensive picture of their effects is crucial, mapping all essential steps toward increased network robustness and resilience and finding chances for mode substitution [3].
Anzeige
As stated by [4], weather and climate are frequently treated as synonyms even though they differ. Weather is the short-term state of the atmosphere at a specific time and place. On the other hand, climate is the long-term manifestations of weather and other atmospheric conditions in a given area during a period long enough to ensure that representative values are obtained. Weather refers to the daily variation in the atmosphere, including the temperature, relative humidity, cloud cover, precipitation, and wind. In contrast, climate refers to the general conditions over a long period in a given location. More studies on climate change and its impact on the transport sector have been published in the 21st century, hence the railway system [5].
There are several ways to examine the impact of the weather conditions on the transport mode, according to [2]. The differences in performance can compare transport systems between regions with very different weather conditions. According to [4], the weather influences individual travel demand. Travelers have availability to information on the exact weather conditions and advanced weather forecasts, which seems to have an impact on their choice of transportation mode.
Each railway system has its particular challenges related to the local weather. While the cold weather is unfamiliar in the southern nations, it isn’t easy in the Northern countries. The weather-related issues can be solved in a variety of ways. The British railway system has its own Leaf Fall Timetable Changes [6]. Thousands of tonnes of leaves fall across the railway network during the autumn. The leaves on the track are compressed by passing trains, producing a thin, black coating that affects train braking and acceleration. Timetables are modified throughout the autumn to prevent unforeseen delays to journeys.
There has been some recent research on how weather conditions affect disturbance in the railway industry. However, the examination of published literature has a heavier emphasis on other modes of transportation [7, 8]. Yet, as [9] pointed out, the literature has begun to pay more and more attention to the need to comprehend how weather and climate change affect railway delays and punctuality.
Anzeige
Including weather variables in the train delay prediction models has been shown to improve the overall accuracy of the models. [10] developed a dynamic train delay prediction system that uses state-of-the-art tools and techniques to combine heterogeneous data sources and interact with dynamically changing systems to provide meaningful information to traffic management and dispatching operations. These models are further enhanced by including exogenous meteorological data [10, 11]. [12] attempted to comprehend the trends in weather and railway delays. According to the study, train delays in severe weather are most influenced by poor weather. Still, delays in fair weather are primarily influenced by the duration and frequency of previous train delays. Further, [12] found that infrequently occurring weather, such as snow in southern cities, has a stronger influence and causes longer train delays.
The impacts of weather phenomena, including wind, temperature, and precipitation, on railway operators’ performance of passenger train services are estimated by [1]. The study concluded that 4–8% of all train disturbances were caused by bad weather conditions. [13] discovered that trains frequently display the same delay patterns as in the past and that adverse weather frequently results in longer delays. [14] found that 60 % of late arrivals in Finland between 2008 and 2010 were connected to winter weather by modeling the co-variation between harsh weather and freight train delays. The weather factors that influence the punctuality of trains on the Norwegian railway Nordland Line were examined by [8]. The study demonstrates that extreme winter cold is a significant influencing factor that leads to delays and poor punctuality. Moreover, snow depth is the meteorological factor that most explains changes in passenger train punctuality daily and weekly. [15] investigated the impact of weather on railway punctuality. As the temperature drops below 0\(^\circ \), punctuality reduces exponentially; at –5\(^\circ \), it drops by 7.5%, and at –30\(^\circ \), it drops by 50%. The temperature variation was highly correlated with the travel distance. [16] introduced a train delay prediction model that included a fully connected neural network with two LSTM components. The study highlighted the importance of considering the interactions between trains, stations and weather-related factors regarding prediction accuracy.
The present study tries to provide data-driven insights into delay modeling by including weather elements. This research specifically targets the existing delays and the effect of different weather elements on the departure delays of the trains.
This research is a part of an EU-funded project “FutuRe”. The flagship project FP6 – FutuRe (GA 101101962) [17] under Europe’s Rail Joint Undertaking) aims at providing new innovative technical requirements, methods, solutions, developments and services based on the latest leading-edge technologies to make regional rail cost-efficient while meeting safety standards and improving the reliability, availability and capacity of the railway system. The work presented here is related to the FutuRe project area Regional Rail Customer Services, focusing on customer service and aiming to develop highly accurate multimodal passenger information on-board and/or at stations for passenger and freight management.
2 Methodology
The data was acquired from Bane NOR for the arrival and departure delays of long-distance trains running on the Oslo-Trondheim line from 1 January 2021 to 28 February 2023 at the stopping and non-stopping stations. The data was acquired from different sources into TIOS database as shown in Fig. 1. The data includes trains starting from Oslo and trains starting from Trondheim. Machine learning model, specifically the long short-term memory (LSTM) algorithm of the neural network domain, was used for prediction modeling. Three weather features (temperature and precipitation) were acquired from the Norwegian Meteorological Institute for the corresponding dates. Correlations were established between the effects of different aspects of weather. These results were compared to the predictions without considering the weather effect. The data was cleaned and analyzed using Python modules.
Fig. 1.
Illustration of the data flow of the train record information (reproduced from [18])
The delay data was selected between the interquartile range (IQR), a statistical measure of the spread of a data set, as shown in Eq. 1. It is calculated as the difference between the third (Q3) and the first quartile (Q1) of the data distribution, which is the range between the 25th and 75th percentiles of the data. The IQR method for identifying outliers is used by setting up a “fence" outside of Q1 and Q3, or minimum and maximum values. Values outside of these boundaries are considered outliers.
The delays, temperature, and rainfall values (parameters) were graded using a formula in Eq. 2 to derive additional features for the prediction model. The value of variable grad ranges from 1 to 4 while the values v1, v2 and v3 are changed as per the parameter.
Chemical structure diagram showing a hexagonal benzene ring with alternating double bonds. Attached to the ring is a hydroxyl group (OH) and a carboxyl group (COOH).
(2)
There were negative delays at some non-boarding stations in the dataset, which were ignored in this analysis. The Entur Geocoder API was used for getting the latitude and longitude of the stations to connect them with the nearest weather stations. The FrostMET API was used to get historical data on temperature and rainfall from weather stations nearest to the train stations.
Entur Geocoder API is free to use and requires no credentials. However, bulk requests are blocked by the server. The application of FrostMET API requires duly formed credentials, which can be overcome by web scraping. However, there are some unresolved issues regarding bulk requests and multiple weather elements. These issues will be resolved in the future research.
3 Analysis and Results
The effect of gradation as a modeling feature is shown in the form of Pearson coefficient values (\(R^2\)), mean absolute error (MAE, in minutes) and root mean square error (RMSE, in minutes) of the LSTM prediction model (as per Fig. 2) in Table 1.
The general gradation of delay values was present in all the datasets by default unless stated in the table. It can be seen that gradation increases the \(R^2\) values while keeping the error values in check. It can be seen that without gradation, the weather data correlation becomes much weaker [19].
Table 1.
Statistical inferences from different prediction models
Models
MAE
RMSE
\(R^2\)
Weather without gradation (delay gradation absent)
2.0231
2.5375
0.0109
temperature and rainfall data without gradation
0.9358
1.4162
0.6919
temperature and rainfall data with gradation
0.9813
1.2976
0.7414
Only temperature data with gradation
0.9528
1.2730
0.7500
No weather data at all
0.9712
1.3205
0.7312
Fig. 3.
Delay averaged over the weeks of the year (1 is the first week and 53 is the last week on the x-axis)
The delay averaged over the weeks of the year is shown in Fig. 3. The delays can be related to two factors, i.e., congestion-based delays due to increased demand and weather-based delays. The peak delay is in the middle of the year, specifically the 26th week. This coincides with the peak of the holiday season in Norway, and people travel during this time. The second-highest peak is at 37th week, around the arrival of the fall season in Norway. This requires more maintenance at the railway tracks, thus causing delays on single-line tracks.
4 Conclusions
The major conclusion that can be drawn from the study and the literature studied in the present study are:
1.
The application of weather elements to the prediction model surely increases the quality of the results. However, clustering and gradation as a feature are more helpful in bolstering their application.
2.
The peak delays can be related to two major factors: congestion due to increased demand and maintenance due to extreme weather.
Acknowledgements
We are delighted to acknowledge the help of Bane NOR for providing the delay data for this study.
Funding
This research activity is part of the FP6 FutuRe project, partially funded by the European Commission through Europe’s Rail Joint Undertaking under the Horizon Europe Programme with the grant agreement number 101101962.
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Xia, Y., Ommeren, J.N., Rietveld, P., Verhagen, W.: Railway infrastructure disturbances and train operator performance: the role of weather. Transport. Res. Part D: Transp. Environ. 18, 97–102 (2013). https://doi.org/10.1016/j.trd.2012.09.008CrossRef
2.
Koetse, M.J., Rietveld, P.: The impact of climate change and weather on transport: an overview of empirical findings. Transport. Res. Part D: Transp. Environ. 14(3), 205–221 (2009). https://doi.org/10.1016/j.trd.2008.12.004CrossRef
3.
Stamos, I., Mitsakis, E., Salanova, J.M., Aifadopoulou, G.: Impact assessment of extreme weather events on transport networks: a data-driven approach. Transport. Res. Part D: Transp. Environ. 34, 168–178 (2015). https://doi.org/10.1016/j.trd.2014.11.002CrossRef
Baker, C.J., Chapman, L., Quinn, A., Dobney, K.: Climate change and the railway industry: a review. Proc. Inst. Mech. Engineers Part C: J. Mech. Eng. Sci. 224(3), 519–528 (2010). https://doi.org/10.1243/09544062JMES1558CrossRef
Nagy, E., Csiszar, C.: Analysis of delay causes in railway passenger transportation. Period. Polytech. Transp. Eng. 43(2), 73–80 (2015). https://doi.org/10.3311/PPtr.7539CrossRef
8.
Zakeri, G., Olsson, N.O.E.: Investigating the effect of weather on punctuality of Norwegian railways: a case study of the nordland line. J. Mod. Transport. 26(4), 255–267 (2018). https://doi.org/10.1007/s40534-018-0169-7CrossRef
Oneto, L., et al.: Advanced analytics for train delay prediction systems by including exogenous weather data. In: 2016 IEEE International Conference on Data Science and Advanced Analytics (DSAA), pp. 458–467 (2016). https://doi.org/10.1109/DSAA.2016.57
11.
Oneto, L., et al.: Dynamic delay predictions for large-scale railway networks: deep and shallow extreme learning machines tuned via thresholdout. IEEE Trans. Syst. Man Cybern. Syst. 47(10), 2754–2767 (2017). https://doi.org/10.1109/TSMC.2017.2693209CrossRef
12.
Wang, P., Zhang, Q.: Train delay analysis and prediction based on big data fusion. Transport. Saf. Environ. 1(1), 79–88 (2019). https://doi.org/10.1093/tse/tdy001CrossRef
13.
Ling, X., Peng, Y., Sun, S., Li, P., Wang, P.: Uncovering correlation between train delay and train exposure to bad weather. Physica A: Stat. Mech. Appl. 512, 1152–1159 (2018). https://doi.org/10.1016/j.physa.2018.07.057CrossRef
Palmqvist, C.W., Olsson, N.O.E., Hiselius, L.W.: Some influencing factors for passenger train punctuality in Sweden. Int. J. Prognost. Health Manag. 8(3), 1–14 (2017). https://doi.org/10.36001/ijphm.2017.v8i3.2649
16.
Huang, P., et al.: Modeling train operation as sequences: a study of delay prediction with operation and weather data. Transport. Res. Part E: Logist. Transport. Rev. 141, 102022 (2020). https://doi.org/10.1016/j.tre.2020.102022CrossRef
Skjøren, C., Larsen, A.S.: Predictions for railway traveller information. Master’s thesis, Norwegian University of Science and Technology (2023). https://hdl.handle.net/11250/3088477
19.
Mandhaniya, P., Olsson, N.O.E., Larsen, A.S., Skjøren, C.: How complex systems sometimes follow murphy’s law: train delay prediction at a station using delays at previous stops as the features. Procedia Comput. Sci. 239, 1778–1783 (2024). https://doi.org/10.1016/j.procs.2024.06.357CrossRef
There were negative delays at some non-boarding stations in the dataset, which were ignored in this analysis. The Entur Geocoder API was used for getting the latitude and longitude of the stations to connect them with the nearest weather stations. The FrostMET API was used to get historical data on temperature and rainfall from weather stations nearest to the train stations.
