GB2406423A

GB2406423A - Modelling future railway performance of a railway network

Info

Publication number: GB2406423A
Application number: GB0322798A
Authority: GB
Inventors: Roland Timothy Cornah; Karl Strickland
Original assignee: AEA Technology PLC
Current assignee: Ricardo AEA Ltd
Priority date: 2003-09-27
Filing date: 2003-09-27
Publication date: 2005-03-30
Also published as: GB0322798D0

Abstract

A method and system for modelling the future performance of a railway network comprises the steps of obtaining data from a plurality of sensors, analysing the data and on a first computer modelling future performance of the network. Preferably the data analysis comprises deducing the times at which trains arrive at specific locations, the speed and acceleration of trains and times where trains are stationary. In use the model is used to determine the effect of changes in infrastructure or timetables will have on the network. The data from the sensors may also be stored on a second computer.

Description

Railway Computer System This invention relates to a process for modelling

a railway network, and a computer system for use in modelling a railway network.

Many aspects of railway system operation are nowadays controlled and monitored by computer systems.

For example a computer system may be used to store measurements made on the track, for example to monitor rail wear; a computer system may be used to model the operations of trains on a network, for example for generating a timetable; a computer system may be used to control signalling, and so control the movements of trains; and a computer system may be used to monitor use of each vehicle in a train fleet, so that decisions on maintenance of each vehicle can be made. Such computer- operated systems can be usefully classified according to three different criteria: the computer process itself; the aspect of the railway network; and the geographical breadth. Considering the computer process, this may involve "measuring" i.e. the obtaining and storing of measured data, which may be real-time; it may involve "managing", that is to say making decisions, for example on servicing; or it may involve "modelling", predicting how a system will work in the future. The aspect of the railway network may be the infrastructure (e.g. stations, platforms, and tracks), or the operations (that is to say the planning or control of the movements of trains); or the trains themselves. And the geographical breadth may be local, regional, or supra-regional (i.e. covering several regions, for example national or international).

According to the present invention there is provided a process for modelling performance of a railway network to predict future performance, the modelling being - 2 performed by a first computer, the process involving the steps of: obtaining data from a plurality of sensors indicating passage of trains on a network during a period of time; deducing from this data the times at which the trains arrive at specific locations on the network, the speeds and accelerations of the trains, and the times for which any passenger-carrying trains are stationary at stations; and the first computer then using the deduced information on arrival times, speeds, accelerations, and times spent at stations in modelling future performance of the railway network.

It will be appreciated that real-time data is generally not essential in such a modelling process, so that the data from the sensors may be recorded and then transferred to the first computer. For example it may be recorded on a portable storage medium, and then downloaded from that storage medium to the first computer. More preferably the data from the sensors is recorded by a second computer, and is then transferred to the first computer. This may for example involve data transmission cables linking the computers, or a network wherein each computer is connected to a hub computer through which data is transmitted. This would enable transmission of real-time information, if that is desired.

The invention also provides a computer system for performing this process.

The invention will now be further and more particularly described, by way of example only. - 3

Example 1

This example relates to modelling the effect of new infrastructure. Where the future performance of a part of a railway network is to be modelled, this has hitherto been carried out on the supposition that the trains will enter this part of the railway network and pass through it in accordance with timetable data. It has now been appreciated that this may give misleading results.

A first computer is used to model train operations, and is to be used to model the overall effect on train reliability of construction of a proposed new platform, or of changes to the signal layout. Although a model might be based on the train timetable, in practice the trains running on a particular day may differ quite significantly from those predicted from the timetable.

For example changes may have to be made because of unforeseen circumstances, such as an engine fault so a train cannot travel at full speed, or lack of staff so that a train is cancelled. In addition to trains running slow or being cancelled, there may be additional trains (for example freight trains) that are not in the timetable, and the train timings may differ from those in the timetable owing to passengers taking longer to get on or off the train than predicted. Maintenance work may cause trains on one line to run slow, and this can have consequential effects on other trains. If a passenger train is delayed, and is late in arriving at a station, the railway operator may delay the departure of other trains from that station to ensure connections. It will be appreciated that there are many different reasons why trains may not run according to the timetable.

A second computer records train movement information - 4 on a continuous basis. This records, for each train, the time at which it passes certain locations in the rail network such as stations and signals, at which there are train-detecting sensors. From this second computer data can therefore be obtained, ultimately from these sensors, providing an indication of every train movement over a period of say ten days. This data is then transmitted to the first computer, either through an electronic link or recorded on a data storage medium such as a floppy disk.

The first computer firstly works out, from the sequence of locations that an individual train passed, the route that it had followed, checking from a database of locations that this is a permissible and unambiguous route. From the times at which the train passed those locations the first computer can estimate a train speed and a traction performance so that the model or simulation corresponds to the observed times; furthermore if the discrepancy between a simulated and an actual journey time between successive measuring points is significant (say more than 2 minutes) then the simulation can introduce into the model the train coasting to lose time, or indeed stopping.

In this way the first computer can generate a model of all the train operations during the time period for which observations were available, this simulation accurately representing the route, traction performance, and any stops, of each and every train. The proposed infrastructure changes (such as the proposed new platform) can then be introduced into the simulation, and the effect on overall train performance can be modelled.

The resulting simulation is considerably more realistic than one based on the planned timetable information; indeed, in one specific application of this Example the realistic simulation made it clear that it was impossible for the trains to keep to the timetable unless a new platform was provided.

Example 2

This example relates to the preparation of train timetables. New timetables usually start from the previous timetable as a starting point. This assumes that the old timetable worked, and the intention is that the new timetable will be an improvement on this. In reality, as discussed above, trains do not always run as planned, and there may be additional trains.

A new timetable is therefore generated by modifying the previous timetable and taking into account the data obtained, as described in relation to the previous Example, from the second computer. In the same way as described above, the first computer first generates from this historic data a model of all the train operations during the time period for which observations were available, this simulation accurately representing the route, traction performance, and any stops, of each and every train. The new timetable can use such data. For example if one service (due at say 8:00) is observed to arrive a few minutes late at a particular station over the period for which observations were taken, the timetable may be modified so that train is supposed to be at that station for example five minutes later (i.e. 8:05). The effect of this modification can then be modelled, applying the traction performance and station- time information obtained on the 8:00 train to the train now scheduled at 8:05. This modelling may show that it is necessary to change the scheduled time of another train; and this modification to the timetable can be made and modelled.

In addition to modifications made in this manner, it may be apparent from the observed data and the modelling that there are some times of day at which the section of the network is overloaded. The effect of removing one train, or changing its scheduled departure time by say 5 or 10 minutes, can then be modelled, to find if the risk of such congestion can be reduced. The modelling may also show up periods of time at which it would be possible to run additional trains.

Modelling the new timetable will highlight any problems. In this way a timetable can be developed which will be robust enough that it is not unduly affected by perturbations in train times and performance. At this stage certain operational decisions may be made, in order to enable the timetable to work effectively. These may for example include decisions or rules about priority (e.g. that a local stopping service should be held up if it would otherwise be ahead of a high-speed service, if the latter is running no more than 2 minutes late).

These rules are preferably highlighted at this stage, so that when the finalised new timetable is provided to the train controllers and signallers, these priority rules can also be made explicit.

Example 3

This example relates to providing guidance in the event of an accident. If an accident occurs, such as a derailment, this will close at least part of a line for a period of time. Various different courses of action may be adopted to minimise the disruption to other train services. For example it may be possible to divert some services via another railway line. By performing modelling as described in the preceding examples, though 7 - in this case using substantially real-time data (for example to determine traction performance), each of these different courses of action can be modelled, and hence the least-disruptive course of action identified.

Claims

Claims 1. A process for modelling performance of a railway network to

predict future performance, the modelling being performed by a first computer, the process involving the steps of: obtaining data from a plurality of sensors indicating passage of trains on a network during a period of time; deducing from this data the times at which the trains arrive at specific locations on the network, the speeds and accelerations of the trains, and the times for which any passenger-carrying trains are stationary at stations; and the first computer then using the deduced information on arrival times, speeds, accelerations, and times spent at stations in modelling future performance of the railway network.
2. A process as claimed in claim 1 wherein the modelling is carried out so as to determine the effect of changes in infrastructure.
3. A process as claimed in claim 1 wherein the modelling is carried out so as to determine the effect of changes

in timetable.
4. A process as claimed in any one of the preceding claims wherein the data from the sensors is recorded by a second computer, and is then transferred to the first computer.
5. A computer system for modelling performance of a railway network to predict future performance, the modelling being performed by a first computer, the 9 - computer system comprising: a plurality of sensors for indicating passage of trains on a network; means to transmit data obtained by the sensors over a period of time to the first computer; the first computer being arranged to deduce from this data the times at which the trains arrived at specific locations on the network, the speeds and accelerations of the trains, and the times for which any passengercarrying trains were stationary at stations; and the first computer then being arranged to use the deduced information on arrival times, speeds, accelerations, and times spent at stations in modelling future performance of the railway network.
6. A computer system as claimed in claim 5 wherein the computer system incorporates a second computer for storing the data obtained by the sensors.
7. A computer system as claimed in claim 6 wherein the first computer and the second computer are linked by an electronic data transmission means.
8. A process for modelling performance of a railway network to predict future performance, substantially as hereinbefore described with reference to any one of the

Examples.