Automated real-time railway traffic control: an experimental analysis of reliability,resilience and robustness

Railway transportation provides sustainable, fast and safe transport. Its attractiveness is linked to a broad concept of service reliability: the capability to adhere to a timetable in the presence of delays perturbing traffic. To counter these phenomena, real-time rescheduling can be used, changing train orders and times, according to rules of thumb, or mathematical optimization models, minimizing delays or maximizing punctuality. In the literature, different indices of robustness, reliability and resilience are defined for railway traffic. We review and evaluate these indices applied to railway traffic control, comparing optimal rescheduling approaches such as Open Loop and Closed Loop control, to a typical First-Come-First-Served dispatching rule, and following the timetable (no-action). This experimental analysis clarifies the benefits of automated traffic control for infrastructure managers, railway operators and passengers. The timetable order, normally used in assessing a-priori reliability, systematically overestimates unreliability of operations that can be reduced by real-time control.     

Reinforcement learning approach for train rescheduling on a single-track railway

Optimal rail network infrastructure and rolling stock utilization can be achieved with use of different scheduling tools by extensive planning a long time before actual operations. The initial train timetable takes into account possible smaller disturbances, which can be compensated within the schedule. Bigger disruptions, such as accidents, rolling stock breakdown, prolonged passenger boarding, and changed speed limit cause delays that require train rescheduling. In this paper, we introduce a train rescheduling method based on reinforcement learning, and more specifically, Q-learning. We present here the Q-learning principles for train rescheduling, which consist of a learning agent and its actions, environment and its states, as well as rewards. The use of the proposed approach is first illustrated on a simple rescheduling problem comprising a single-lane track with three trains. The evaluation of the approach is performed on extensive set of experiments carried out on a real-world railway network in Slovenia. The empirical results show that Q-learning lead to rescheduling solutions that are at least equivalent and often superior to those of several basic rescheduling methods that do not rely on learning agents. The solutions are learned within reasonable computational time, a crucial factor for real-time applications.     

An improved car‐following model for railway traffic

In this paper, we propose an improved traffic model for simulating train movement in railway traffic. The proposed model is based on optimal velocity car‐following model. In order to test the proposed model, we use it to simulate the train movement with fixed‐block system. In simulations, we analyze and discuss the space–time diagram of railway traffic flow and the trajectories of train movement. Simulation results demonstrate that the proposed model can be successfully used for simulating the train movement in railway traffic. From the space–time diagram, we find some complex phenomena of train flow, which are observed in real railway traffic, such as train delays. By analyzing the trajectories of train movement, some dynamic characteristics of trains can be reproduced. Copyright © 2011 John Wiley & Sons, Ltd.     

Railway traffic disturbance management—An experimental analysis of disturbance complexity,management objectives and limitations in planning horizon

With the increasing traffic volumes in European railway networks and reports on capacity deficiencies that cause reliability problems, the need for efficient disturbance management becomes evident. This paper presents a heuristic approach for railway traffic re-scheduling during disturbances and a performance evaluation for various disturbance settings using data for a large part of the Swedish railway network that currently experiences capacity deficiencies. The significance of applying certain re-scheduling objectives and their correlation with performance measures are also investigated. The analysis shows e.g. that a minimisation of accumulated delays has a tendency to delay more trains than a minimisation of total final delay or total delay costs. An experimental study of how the choice of planning horizon in the re-scheduling process affects the network on longer-term is finally presented. The results indicate that solutions which are good on longer-term can be achieved despite the use of a limited planning horizon. A 60 min long planning horizon was sufficient for the scenarios in the experiments.     

Multiple-phase train trajectory optimization with signalling and operational constraints

The train trajectory optimization problem aims at finding the optimal speed profiles and control regimes for a safe, punctual, comfortable, and energy-efficient train operation. This paper studies the train trajectory optimization problem with consideration of general operational constraints as well as signalling constraints. Operational constraints refer to time and speed restrictions from the actual timetable, while signalling constraints refer to the influences of signal aspects and automatic train protection on train operation. A railway timetable provides each train with a train path envelope, which consists of a set of positions on the route with a specified target time and speed point or window. The train trajectory optimization problem is formulated as a multiple-phase optimal control model and solved by a pseudospectral method. This model is able to capture varying gradients and speed limits, as well as time and speed constraints from the train path envelope. Train trajectory calculation methods under delay and no-delay situations are discussed. When the train follows the planned timetable, the train trajectory calculation aims at minimizing energy consumption, whereas in the case of delays the train trajectory is re-calculated to track the possibly adjusted timetable with the aim of minimizing delays as well as energy consumption. Moreover, the train operation could be affected by yellow or red signals, which is taken into account in the train speed regulation. For this purpose, two optimization policies are developed with either limited or full information of the train ahead. A local signal response policy ensures that the train makes correct and quick responses to different signalling aspects, while a global green wave policy aims at avoiding yellow signals and thus proceed with all green signals. The method is applied in a case study of two successive trains running on a corridor with various delays showing the benefit of accurate predictive information of the leading train on energy consumption and train delay of the following train.     

Modeling and simulation of vehicle projection arrival–discharge process in adaptive traffic signal controls

Real‐time signal control operates as a function of the vehicular arrival and discharge process to satisfy a pre‐specified operational performance. This process is often predicted based on loop detectors placed upstream of the signal. In our newly developed signal control for diamond interchanges, a microscopic model is proposed to estimate traffic flows at the stop‐line. The model considers the traffic dynamics of vehicular detection, arrivals, and departures, by taking into account varying speeds, length of queues, and signal control. As the signal control is optimized over a rolling horizon that is divided into intervals, the vehicular detection for and projection into the corresponding horizon intervals are also modeled. The signal control algorithm is based on dynamic programming and the optimization of signal policy is performed using a certain performance measure involving delays, queue lengths, and queue storage ratios. The arrival–discharge model is embedded in the optimization algorithm and both are programmed into AIMSUN, a microscopic stochastic simulation program. AIMSUN is then used to simulate the traffic flow and implement the optimal signal control by accessing internal data including detected traffic demand and vehicle speeds. Sensitivity analysis is conducted to study the effect of selecting different optimization criteria on the signal control performance. It is concluded that the queue length and queue storage ratio are the most appropriate performance measures in real‐time signal control of interchanges. Copyright © 2010 John Wiley & Sons, Ltd.     

A delay propagation algorithm for large-scale railway traffic networks

In scheduled railway traffic networks a single delayed train may cause a domino effect of secondary delays over the entire network, which is a main concern to planners and dispatchers. This paper presents a model and an algorithm to compute the propagation of initial delays over a periodic railway timetable. The railway system is modelled as a linear system in max-plus algebra including zero-order dynamics corresponding to delay propagation within a timetable period. A timed event graph representation is exploited in an effective graph algorithm that computes the propagation of train delays using a bucket implementation to store the propagated delays. The behaviour of the delay propagation and the convergence of the algorithm is analysed depending on timetable properties such as realisability and stability. Different types of delays and delay behaviour are discussed, including primary and secondary delays, structural delays, periodic delay regimes, and delay explosion. A decomposition method based on linearity is introduced to deal with structural and initial delays separately. The algorithm can be applied to large-scale scheduled railway traffic networks in real-time applications such as interactive timetable stability analysis and decision support systems to assist train dispatchers.     

Automatic identification of risky weather objects in line of flight (AIRWOLF)

Adverse weather conditions are hazardous to flight and contribute to re-routes and delays. This has a negative impact on the National Airspace System (NAS) due to reduced capacity and increased cost. In today’s air traffic control (ATC) system there is no automated weather information for air traffic management decision-support systems. There are also no automatic weather decision-support tools at the air traffic controller workstation. As a result, air traffic operators must integrate weather information and traffic information manually while making decisions. The vision in the Next Generation Air Transportation System (NextGen) includes new automation concepts with an integration of weather information and decision-making tools. Weather-sensitive traffic flow algorithms could automatically handle re-routes around weather affected areas; this would optimize the capacity during adverse conditions. In this paper, we outline a weather probe concept called automatic identification of risky weather objects in line of flight (AIRWOLF). The AIRWOLF operates in two steps: (a) derivation of polygons and weather objects from grid-based weather data and (b) subsequent identification of risky weather objects that conflict with an aircraft’s line of flight. We discuss how the AIRWOLF concept could increase capacity and safety while reducing pilot and air traffic operator workload. This could translate to reduced weather-related delays and reduced operating costs in the future NAS.     

Rescheduling through stop-skipping in dense railway systems

Based on the analysis of the railway system in the Paris region in France, this paper presents a rescheduling problem in which stops on train lines can be skipped and services are retimed to recover when limited disturbances occur. Indeed, in such mass transit systems, minor disturbances tend to propagate and generate larger delays through the shared use of resources, if no action is quickly taken. An integrated Integer Linear Programming model is presented whose objective function minimizes both the recovery time and the waiting time of passengers. Additional criteria related to the weighted number of train stops that are skipped are included in the objective function. Rolling-stock constraints are also taken into account to propose a feasible plan. Computational experiments on real data are conducted to show the impact of rescheduling decisions depending on key parameters such as the duration of the disturbances and the minimal turning time between trains. The trade-off between the different criteria in the objective function is also illustrated and discussed.     

Development of hybrid optimization of train schedules model for N-track rail corridors

From a capacity perspective, efficient utilization of a railway corridor has two main objectives; avoidance of schedule conflicts, and finding a proper balance between capacity utilization and level of service (LOS). There are several timetable tools and commercial rail simulation packages available to assist in reaching these objectives, but few of them offer both automatic train conflict resolution and automatic timetable management features for the different types of corridor configurations. This research presents a new rescheduling model to address some of the current limitations. The multi-objective linear programming (LP) model is called “Hybrid Optimization of Train Schedules” (HOTS), and it works together with commercial rail simulation tools to improve capacity utilization or LOS metrics. The HOTS model uses both conflict resolution and timetable compression techniques and is applicable to single-, double-, and multiple-track corridors (N-track networks), using both directional and bi-directional operations. This paper presents the approach, formulation and data requirements for the HOTS model. Single and multi-track case studies test and demonstrate the model’s train conflict resolution and timetable compression capabilities, and the model’s results are validated by using RailSys simulation package. The HOTS model performs well in each tested scenario, providing comparable results (either improved or similar) to the commercial packages.     

Prediction of ground vibration amplitudes due to urban railway traffic using quantitative and qualitative field data

The growth of railway transport in urban areas has lead to an increase in ground vibrations enhancing their negative environmental impact. Therefore is mandatory to predict and control ground vibrations. This work presents a methodology for the determination of prediction models of ground vibration amplitudes due to railway train circulation in urban environments. Using quantitative predictors (train speed and distance) and qualitative predictors (railway track type, dominant geology and building type), being the use of the latter predictors justified by the fact that, most frequently, quantitative parameters are very difficult to obtain in the urban environment due to their characterization. Thus, a detailed statistical study based on the proposal and validation of multiple linear regression models, is successfully applied in order to predict vibration amplitudes produced by railway train circulation, in the considered domain, as function of quantitative and qualitative predictors, easily obtained in field work. A multiple linear regression model for ground vibration prediction due to underground railway traffic has been presented for the Lisbon area.     

Distributed model predictive control for railway traffic management

Every day small delays occur in almost all railway networks. Such small delays are often called “disturbances” in literature. In order to deal with disturbances dispatchers reschedule and reroute trains, or break connections. We call this the railway management problem. In this paper we describe how the railway management problem can be solved using centralized model predictive control (MPC) and we propose several distributed model predictive control (DMPC) methods to solve the railway management problem for entire (national) railway networks. Furthermore, we propose an optimization method to determine a good partitioning of the network in an arbitrary number of sub-networks that is used for the DMPC methods. The DMPC methods are extensively tested in a case study using a model of the Dutch railway network and the trains of the Nederlandse Spoorwegen. From the case study it is clear that the DMPC methods can solve the railway traffic management problem, with the same reduction in delays, much faster than the centralized MPC method.     

A delay root cause discovery and timetable adjustment model for enhancing the punctuality of railway services

Knock-on delay, which is the key factor in punctuality of railway service, is mainly related to two factors including the quality of timetable in the planning phase and disturbances which may result in unscheduled trains’ waiting or meeting in operation phase. If the delay root cause and the interactions among the factors responsible for these can be clearly clarified, then the punctuality of railway operations can be enhanced by taking reactions such as timetable adjustment, rescheduling or rerouting of railway traffic in case of disturbances. These delay reasons can be used to predict the lengths of railway disruptions and effective reactions can be applied in disruption management. In this work, a delay root cause discovery model is proposed, which integrates heterogeneous railway operation data sources to reconstruct the details of the railway operations. A supervised decision tree method following the machine learning and data mining techniques is designed to estimate the key factors in knock-on delays. It discovers the root cause delay factor by logically analyzing the scheduled or un-scheduled trains meetings and overtaking behaviors, and the subsequent delay propagations. Experiment results show that the proposed decision tree can predict the delay reason with the accuracy of 83%, and it can be further enhance to 90% if the delay cause is only considered “prolonged passengers boarding” and “meeting or overtaking” factors. The delay root cause can be discovered by the proposed model, verified by frequency filtering in operation records, and resolved by the adjustment of timetable which is an important reference for the next timetable rescheduling. The results of this study can be applied to railway operation decision support and disruption management, especially with regard to timetable rescheduling, trains resequencing or rerouting, system reliability analysis, and service quality improvements.     

Robust single-track train dispatching model under a dynamic and stochastic environment: A scenario-based rolling horizon solution approach

After a major service disruption on a single-track rail line, dispatchers need to generate a series of train meet-pass plans at different decision times of the rescheduling stage. The task is to recover the impacted train schedule from the current and future disturbances and minimize the expected additional delay under different forecasted operational conditions. Based on a stochastic programming with recourse framework, this paper incorporates different probabilistic scenarios in the rolling horizon decision process to recognize (1) the input data uncertainty associated with predicted segment running times and segment recovery times and (2) the possibilities of rescheduling decisions after receiving status updates. The proposed model periodically optimizes schedules for a relatively long rolling horizon, while selecting and disseminating a robust meet-pass plan for every roll period. A multi-layer branching solution procedure is developed to systematically generate and select meet-pass plans under different stochastic scenarios. Illustrative examples and numerical experiments are used to demonstrate the importance of robust disruption handling under a dynamic and stochastic environment. In terms of expected total train delay time, our experimental results show that the robust solutions are better than the expected value-based solutions by a range of 10-30%.     

Real-time automatic rescheduling strategy for an urban rail line by integrating the information of fault handling

In the operation of urban rails, faults are inevitable, which leads to deviation between the actual timetable and the planned timetable. In nowadays, timetable rescheduling strategies rarely integrate the information of fault handling. In this paper, we develop a real-time automatic rescheduling strategy, which integrates the dynamic information of fault handling. The rescheduled timetable is obtained by a mathematical optimization model, the constraints set of which is automatically generated and adjusted as more information of fault handling is feedback. Compared with the experience-based rescheduling methods, the automatic rescheduling strategy reacts more quickly, and uses the information of fault handling more efficiently. A simulation system for testing the automatic rescheduling strategy is built, which uses the data of the Beijing Yizhuang metro line. Via testing on the simulation system, the effectiveness and efficiency of the automatic rescheduling strategy are validated.     

Characterization and prediction of air traffic delays

This paper presents a new class of models for predicting air traffic delays. The proposed models consider both temporal and spatial (that is, network) delay states as explanatory variables, and use Random Forest algorithms to predict departure delays 2–24 h in the future. In addition to local delay variables that describe the arrival or departure delay states of the most influential airports and links (origin–destination pairs) in the network, new network delay variables that characterize the global delay state of the entire National Airspace System at the time of prediction are proposed. The paper analyzes the performance of the proposed prediction models in both classifying delays as above or below a certain threshold, as well as predicting delay values. The models are trained and validated on operational data from 2007 and 2008, and are evaluated using the 100 most-delayed links in the system. The results show that for a 2-h forecast horizon, the average test error over these 100 links is 19% when classifying delays as above or below 60 min. Similarly, the average over these 100 links of the median test error is found to be 21 min when predicting departure delays for a 2-h forecast horizon. The effects of changes in the classification threshold and forecast horizon on prediction performance are studied.     

Analyzing passenger train arrival delays with support vector regression

We propose machine learning models that capture the relation between passenger train arrival delays and various characteristics of a railway system. Such models can be used at the tactical level to evaluate effects of various changes in a railway system on train delays. We present the first application of support vector regression in the analysis of train delays and compare its performance with the artificial neural networks which have been commonly used for such problems. Statistical comparison of the two models indicates that the support vector regression outperforms the artificial neural networks. Data for this analysis are collected from Serbian Railways and include expert opinions about the influence of infrastructure along different routes on train arrival delays.     

Ant colony optimization for the real-time train routing selection problem

This paper deals with the real-time problem of scheduling and routing trains in a railway network. In the related literature, this problem is usually solved starting from a subset of routing alternatives and computing the near-optimal solution of the simplified routing problem. We study how to select the best subset of routing alternatives for each train among all possible alternatives. The real-time train routing selection problem is formulated as an integer linear programming formulation and solved via an algorithm inspired by the ant colonies’ behavior. The real-time railway traffic management problem takes as input the best subset of routing alternatives and is solved as a mixed-integer linear program. The proposed methodology is tested on two practical case studies of the French railway infrastructure: the Lille terminal station area and the Rouen line. The computational experiments are based on several practical disturbed scenarios. Our methodology allows the improvement of the state of the art in terms of the minimization of train consecutive delays. The improvement is around 22% for the Rouen instances and around 56% for the Lille instances.     

Within day rescheduling microsimulation combined with macrosimulated traffic

The concept of rescheduling is essential to activity-based modeling in order to calculate effects of both unexpected incidents and adaptation of individuals to traffic demand management measures. When collaboration between individuals is involved or timetable based public transportation modes are chosen, rescheduling becomes complex. This paper describes a new framework to investigate algorithms for rescheduling at a large scale. The framework allows to explicitly model the information flow between traffic information services and travelers. It combines macroscopic traffic assignment with microscopic simulation of agents adapting their schedules. Perception filtering is introduced to allow for traveler specific interpretation of perceived macroscopic data and for information going unnoticed; perception filters feed person specific short term predictions about the environment required for schedule adaptation. Individuals are assumed to maximize schedule utility. Initial agendas are created by the FEATHERS activity-based schedule generator for mutually independent individuals using an undisturbed loaded transportation network. The new framework allows both actor behavior and external phenomena to influence the transportation network state; individuals interpret the state changes via perception filtering and start adapting their schedules, again affecting the network via updated traffic demand. The first rescheduling mechanism that has been investigated uses marginal utility that monotonically decreases with activity duration and a monotonically converging relaxation algorithm to efficiently determine the new activity timing. The current framework implementation is aimed to support re-timing, re-location and activity re-sequencing; re-routing at the level of the individual however, requires microscopic travel simulation.