Estimating Long-Term Delay Risk with Generalized Linear Models

Transportation Research Record: Journal of the Transportation Research Board, 2018

Train operations are subject to stochastic variations, reducing service punctuality and thus the quality of service (QoS). Models of such variations are needed to evaluate and predict the potential impact of disturbances and to avoid service punctuality reduction in train service management and timetabling. In this paper, through a case study of the Wuhan–Guangzhou (WH–GZ) high-speed rail (HSR), we show how a wealth of train operation records can be used to model the stochastic nature of train operations at each level, section and station. Specifically, we examine different distribution models for running times of individual sections and show that the Log-logistic probability density function is the best distributional form to approximate the empirical distribution of running times on the specified line. Next, we show that the distribution of running times in each section can be used to accurately infer arrival delays. Consequently, we construct the underlying analytical model and d...

2018

The purpose of this paper is to provide a foundation for modelling train delays as multivariate time series. A pertinent issue with this kind of analysis is that the individual time series do not follow the Gaussian normal distribution. Since the normal distribution constitutes an assumption of classical time series methods, these cannot be applied blindly to the data. The solution explored by the paper is to identify the distributions of the time series and subsequently transforming the data into normal distributions. To this end, a dataset from the Swedish Transport Administration was used containing delays for every train that either departed from the region Skane as a first departure, or arrived in Skane as a final destination. The dataset spans from January 1st, 2014 to December 31st, 2016. Time series were constructed for every station with a significant amount of data points by computing the mean daily delays for these stations. Using software for distribution fitting it was ...

2014

The analysis of random variables-train delays due to various reasons (infrastructure, carrier, external influences) is based on processing of statisticsdata available on the number and duration of failures. The statistics data for these reasons are available in an aggregate form, making it necessary to develop a model for the evaluation of the failure duration. The probability dependencies (well-known in the Theory of Probabilities) of central moments of a random variable, which is a sum of random number of independently distributed random variables, are used. This approach makes it possible to find the mean value and variance of the duration of failure /delay/ and these two features can be used to approximate the unknown distribution of the duration of failure by Gamma distribution.

Congrès Lambda Mu 19 de Maîtrise des Risques et Sûreté de Fonctionnement, 2015

On présente ici un modèle probabiliste de prévision du retard total cumulé induit par les défaillances techniques dans un réseau ferroviaire urbain. Cette prévision est donnée par l'estimation de la distribution de probabilité du retard total cumulé. L'estimateur est calculé à partir d'un échantillon de retards générés par des simulations de Monte Carlo d'un modèle de propagation de retards. On utilise ici cette estimation pour construire un intervalle de confiance autour de la valeur moyenne, mais on pourrait tout aussi bien l'appliquer également à des questions d'analyses de risques. L'accent est mis sur les défaillances techniques du système de signalisation, mais la méthodologie est applicable aux défaillances d'autres éléments du système de transport, tels que, par exemple, le matériel roulant.

Transportation Research Part A: Policy and Practice, 2011

We investigate how passengers on long-distance trains value unexpected delays relative to scheduled travel time and travel cost. For scheduled services with high reliability and long headways, the value of delays is most commonly assumed to be proportional to the average delay. By exploring how the valuation of train delays depends on delay risk and delay length, using three different stated choice data sets, we find that the ''average delay'' approach does not hold: the disutility increases slower than linearly in the delay risk. This means that using the average delay as a performance indicator, a guide for operations planning or for investment appraisal will underestimate the value of small risks of long delays relative to large risks for short delays. It also means that estimated valuations of ''average delay'' will depend on the delay risk level: valuations will be higher the lower the risk levels in the study are.

2019

Delay prediction is an important issue associated with train timetabling and dispatching. Based on real-world operation records, accurate estimation of delays is of immense significance in train operation and decisions of dispatchers. In the study, we establish a model that illustrates the interaction between and the factors affecting the same via train operation records from a Dutch railway system. Based on the main factors that affect train delay and the time series trend, we identify the independent and dependent variables. A long short-term memory (LSTM) prediction model in which the actual delay time corresponded to the dependent variable is established via Python3.6. Finally, the prediction accuracy of the random forest model and artificial neural network model is compared. The results indicate that the LSTM model outperforms other models.

Transportation Research Record: Journal of the Transportation Research Board, 2009

The impact of delays on passenger railways, with specific reference to the national rail network in Great Britain, is examined. A key distinction is made between punctuality (trains running late) and reliability (trains canceled). In Britain these have been combined into a public performance measure, which deteriorated markedly as a consequence of the Hatfield accident in 2000 but has since gradually improved. Causes of delay can be associated with train operators, infrastructure authorities, and external factors. In Great Britain, train operators apparently have been better able to manage delay than the infrastructure authority. Traditional means of valuing delay have used stated preference methods that incorporate a measure of mean lateness and have shown that a minute of late time is typically valued as equivalent to 3 min of timetabled journey time. This is referred to as the reliability multiplier. Studies have also focused on the value placed on the standard deviation of journ...

2018 21st International Conference on Intelligent Transportation Systems (ITSC), 2018

India runs the fourth largest railway transport network size carrying over 8 billion passengers per year. However, the travel experience of passengers is frequently marked by delays, i.e., late arrival of trains at stations, causing inconvenience. In a first, we study the systemic delays in train arrivals using norder Markov frameworks and experiment with two regressionbased models. Using train running-status data collected for two years, we report on an efficient algorithm for estimating delays at railway stations with near accurate results. This work can help railways to manage their resources, while also helping passengers and businesses served by them to efficiently plan their activities.

Journal of Rail Transport Planning & Management

This study investigated the climate effect under consecutive winters on the arrival delay of high-speed passenger trains in northern Sweden. Novel statistical learning approaches, including inhomogeneous Markov chain model and stratified Cox model, were adopted to account for the time-varying risks of train delays. The inhomogeneous Markov chain modelling for the arrival delays has used several covariates, including weather variables, train operational direction, and findings from the primary delay analysis through stratified Cox model. The results showed that the weather variables, such as temperature, snow depth, ice/snow precipitation, and train operational direction, significantly impact the arrival delay. The performance of the fitted inhomogeneous Markov chain model was evaluated by the walk-forward validation method. The averaged mean absolute errors between the expected rates and the observed rates of the arrival delay over the train line was obtained at the level of 0.088, which implies that approximately 9% of trains may be misclassified as having arrival delays by the fitted model at a measuring point on the train line.

“TOPICAL ISSUES OF THERMOPHYSICS, ENERGETICS AND HYDROGASDYNAMICS IN THE ARCTIC CONDITIONS”: Dedicated to the 85th Birthday Anniversary of Professor E. A. Bondarev

Modelling public transportation delays is crucial for building efficient and modern city infrastructure. It is directly related to passenger behavior and satisfaction, as well as to pollution, traffic jams and sustainable development. In this paper we focus on use of jump diffusion processes in analyzing delays. We believe that such an approach is better suited for handling all factors that affect transportation and can also be used to successfully model not only the absolute time spans of waiting times, but also disproportionally large economic impact of huge delays. Our model is tested with real world data and application of obtained results is discussed.