A new approach to optimisation through intelligent integration of railway systems: the InteGRail project

Paolo Umiliacchi1, Roger Shingler2, Gerhard Langer3, Uwe Henning3

1CNC Centro Nuova Comunicazione, Bologna, Italy, 2Bombardier Transportation, Derby, UK, 3Siemens AG, Erlangen, Germany

Abstract

The European Rail Research Advisory Council (ERRAC) has proposed a target for the year 2020 to double passenger traffic and triple freight traffic by rail. Such a goal should be achieved reducing costs and enhancing environmental sustainability, while at the same time keeping the present good safety level of railways, compared to other modalities. Starting in January 2005, InteGRail is the answer of railway research to ERRAC’s challenge, to achieve which it will mobilise significant resources during the next 4 years, involving 39 participants from 10 EU countries plus one participant from non-EU countries (Chile). The project total cost is in excess of 20 million Euro and will be funded (about 50%) by the European Commission, within the 6th Framework Programme for Community Research The paper will present in detail its objectives and work-plan, focusing on the preliminary results achieved so far. The InteGRail project aims to create a holistic, coherent information system, integrating the major railway sub-systems, in order to achieve higher levels of performance of the railway system in terms of capacity, average speed and punctuality, safety and the optimised usage of resources. Building on results achieved by previous projects, InteGRail will propose new intelligent structures and procedures and will contribute to the definition of new standards, in accordance with EC directives and TSI’s. To achieve its core objectives, InteGRail addresses for the first time all railway subsystems simultaneously. In doing so it is striving for better integration of sub-systems through a global approach, with the objective of achieving a higher level of efficiency at global system level. The results can be measured by means of suitable Key Performance Indicators. InteGRail starts from the current situation of railways and improves and extends possibilities by, identifying standard ways to define their performance and gathering data to measure it. Monitoring data are elaborated and transformed into information so as to support the decision process at all levels (operational and strategic). The support for decision making can bring improvement in procedures which can be applied to the railway system in order to improve its performance. This improvement process can be iterated in a continuous optimisation loop. To handle different aspects of the project, a wide project structure was needed, so the project has been divided into seven subprojects. The project’s time-phased activities will involve all subprojects independently, but in a closely coordinated framework. At the core of the system, SP3D will ensure a smooth communication flow between all the other parts: SP3A, dealing with monitoring and control of rolling-stock and infrastructure, SP3B dealing with maintenance objectives and SP3C, which is in charge of the decision support systems. The expected results will be able to contribute, gradually but significantly, to the ERRAC objectives for year 2020. The railway network capacity will be increased, while improving reliability, availability and punctuality and at the same time reducing irregularities and maintenance costs.

Introduction

An efficient and well integrated European railway system is the only answer to the growing demand for mobility in Europe which can cope with all technical, logistic and environmental constraints, to enable sustainable growth of the enlarged Union. To keep up with such expectations, railways need to improve in terms of capacity, average speed and punctuality, safety and the optimised usage of resources. Railway research is challenged to provide new solutions, answering to present and future needs of the railway system.

Building on results coming from previous projects [3, 1, 2], InteGRail aims to create a holistic, coherent information system, integrating the major railway sub-systems, in order to achieve higher levels of performance of the global railway system.

The project

The formal launch of the InteGRail project was a significant step forward in the history of Europe’s railways. If rail is to remain competitive with other transport modes in Europe, further work is needed on development of fully-interoperable technology, so that railway operations are not obstructed by technical or operational barriers while crossing borders between different states. In common with MODTRAIN and other EC funded projects, InteGRail will contribute to forming the basis for the next generation railway system. In the MODTRAIN project, which was started in 2004 as the first of its kind in the area of joint European railway research, the necessary functional, electrical and mechanical interfaces and validation procedures for rolling stock will be defined and proved, InteGRail will create a holistic, coherent information system, integrating the major railway sub-systems. If MODTRAIN, InteGRail and other new projects like MODBRAKE, MODURBAN and RAILENERGY have success in achieving their targets, people may look back in the future and compare these project launches to other historic dates like for example the demonstration of Richard Trevithick’s steam locomotive almost 200 years earlier. The InteGRail project economic advantages, together with the technical solutions, fulfil the objectives of increased railway competitiveness and interoperability defined in the agenda of the European Rail Research Advisory Council (ERRAC), which has proposed as a target for the year 2020 to double passenger traffic and triple freight traffic by rail (as a percentage of market share). As an Integrated Project, InteGRail has a total budget of about Euro 20 million, of which Euro 10 million are financed by the European Union’s 6th Framework Programme for Research and Technological development. The second share of about Euro 10 million is financed by the 39 project partners. Together with the European Commission, the contracting partners of InteGRail include the system integrators Alstom, Ansaldobreda, Bombardier and Siemens, important railway operators, various railway subsystem suppliers as well as research centres and universities. InteGRail is a key enabler of the integration of different Transport Domains, to improve the Railway Network Capacity and Performance and to reduce ownership cost by using Cooperative and Predictive Maintenance. The project provides proposals for definition and improvement of TSI’s (Technical Specifications for Interoperability) and standards. With regard to the system architecture approach, the project result will be a Specification of Service Oriented and Process Centric Railway solution. The service oriented architecture will facilitate integration for the end users. The project provides the specification of a new holistic information concept applicable to the railway system: • Specification of functions/services for Railway Operation, Rolling Stock, Infrastructure and Traffic

Management domains • Specification of an Information Model for Railway Operation, Rolling Stock, Infrastructure and Traffic

Management domains • This specification will include the AEIF Reference Architecture (Functional and Product Breakdown

structures) Railway operators and industry would benefit from two directions of project results, the first direction are technology oriented developments: • Specification and the architecture of a generic monitoring device applicable for wayside and on board

elements • Specification and the architecture of a generic Integration device ensuring the integration of IGR

services in existing systems (Flexible Core Adapter) • Regarding Telecommunication Framework, standard services for information transfer over

heterogeneous networks with advanced features, rich application environment supporting application models, transparency and networks management, standard Railway addressing schema.

The second direction are process and support oriented developments: • Implementation of Maintenance and Decision Support System • Ontology (which includes in its model the candidates to interoperability), to demonstrate the benefit

for the management of Interoperability.

InteGRail will demonstrate that operational improvements are possible in the management of an open Railway Infrastructure, Rolling stock availability and maintenance cost effort as well as in the coordination of Traffic management with Rolling Stock – Infrastructure– Operation domains. The project structure was designed in such a way that the different aspects of the project could be handled effectively. The project has been divided into seven subprojects (Fig. 1). The implementation process will involve all subprojects independently, but in a closely coordinated framework. At the core of the system, SP3D will ensure a smooth communication flow between all the other parts: SP3A, dealing with monitoring and control of rolling-stock and infrastructure, SP3B, dealing with maintenance issues, and SP3C, which is in charge of the decision support systems.

Fig. 1: Project breakdown into subprojects

The optimisation cycle

The first goal of InteGRail is to measure the performance of railways by adequate monitoring of a number of Key Performance Indicators (KPI). Considerable work on the technical and consensus building level involving all stakeholders made it possible to define an agreed set of KPI’s, which can possibly become a standardisation proposal, in order to define a standard way to monitor and measure the performance of a railway system. The impact on KPI’s due to the new solutions developed can then be analysed in order to take appropriate decisions about the future improvements. This will bring to an optimisation cycle (fig. 2) which can be iterated until the defined level of performance in the overall system has been reached.

SP2 – System requirements , architecture and continuous assessment

SP 3AIntelligent

SystemMonitoring and

Control

SP 3BIntelligent

SystemMaintenance

SP 3DAdvanced

SystemCommunication

SP 3CIntelligent

SystemManagement

SP4 – System Integration, test and Validation

SP1ManagementIntegration

DisseminationTraining

Academic coord.

SP2 – System requirements , architecture and continuous assessment

SP 3AIntelligent

SystemMonitoring and

Control

SP 3BIntelligent

SystemMaintenance

SP 3DAdvanced

SystemCommunication

SP 3CIntelligent

SystemManagement

SP4 – System Integration, test and Validation

SP1ManagementIntegration

DisseminationTraining

Academic coord.

Fig. 2: The optimisation cycle concept

It can be distinguished (fig. 3) between real-time optimisation (when decisions are taken in order to influence the process in progress itself) and long-term optimisation (when decisions involve a longer implementation time, so as to have a later impact on the process).

Fig. 3: Impact of decisions on the different railway processes

The former has an immediate or almost immediate impact on the railway system behaviour, while the latter has an impact in the range of weeks or years, but in this case the impact is broader and more enduring. The process to define the InteGRail system involved several steps, trying to match the railway issues with the instruments offered by current state-of-the-art technologies, as depicted in the following diagram (fig.4):

Fleet

Infrastructure

Operations

Traffic management

apply/remove speed restrictions

rolling stock rosters

timetable planning, including dangerous goods

10 years 1 year 1 monthTrain performance specs

Passenger facilities on train

major renewals planning

driverless train operation

off-line strategic and tactical

periodic maintenance planning

periodic maintenance planning

crew rosters

staff recruitment and training

1 week 1 dayservicing fault repair

on-line/real-time operational

condition based maintenancefault repair

condition based maintenancestaff allocation to rosters

emergency action

1 hour 1 minute

quick repair

re-scheduling

incident managementre-scheduling

3C Strategic 3C Operational

3B3B or 3CFleet

Infrastructure

Operations

Traffic management

apply/remove speed restrictions

rolling stock rosters

timetable planning, including dangerous goods

10 years 1 year 1 monthTrain performance specs

Passenger facilities on train

major renewals planning

driverless train operation

off-line strategic and tactical

periodic maintenance planning

periodic maintenance planning

crew rosters

staff recruitment and training

1 week 1 dayservicing fault repair

on-line/real-time operational

condition based maintenancefault repair

condition based maintenancestaff allocation to rosters

emergency action

1 hour 1 minute

quick repair

re-scheduling

incident managementre-scheduling

3C Strategic 3C Operational

3B3B or 3C

Fig. 4 – Methodology followed to define the InteGRail system The railway system can be split into 4 main sub-systems: Rolling Stock, Infrastructure, Operations and Traffic Management. The Top-Down approach defined a high-level description of the railway system in terms of Key Performance Indicators (KPI). The result was a KPI-tree for each of the four subsystems. Links between KPI’s in the different trees have been identified and mapped. Further analysis allowed derivation of user needs, which were expressed in terms of requirements. Finally, requirements have been prioritised, according to the evaluation carried out by railway stakeholders. These lists of items represent the Needs and Requirements Baseline (NRB) for the new InteGRail solutions, based on a common system architecture. The NRB has been analysed in order to identify scenarios and map them to the relevant processes, which define the structure of the InteGRail system: monitoring and control, maintenance, system management and communication. The related Subprojects charged with deriving functional specifications for each of them. This started in parallel with a Bottom-Up approach, which analysed the current practices and requirements starting from suitable questionnaires. Based on a state-of-the-art analysis, the InteGRail teams identified the items which should be monitored by diagnostics facilities in the future for rolling stock and infrastructure (including infrastructure-rolling stock cooperation and interaction). This was the starting point to define how to improve the processes and satisfy the requirements.

From data to information

Transforming data into information by use of ontology This new approach leads us from a data-centric approach to a knowledge and information based one, while all knowledge about the actual state of Railway assets from Rolling-Stock parts to Infrastructure together with the relations of components and physical dependencies etc. are represented in a corresponding Ontology. In this way, formal modelling (fig. 5) is used to represent monitoring information enriched with its context and formalised in this ontology while the basic idea is to represent measured values together with its context directly in the ontology language, e.g. OWL (Ontology Web Language). This ontology captures what is commonly understood about a specific railway monitoring topic by domain experts and then is instantly related to the actual sensor value as it is perceived. Within the perception layer, monitored data from the real world can be enriched on-line with enough semantic assertions to form information to be shared, this includes information about disjoined subclasses, the property of transitivity etc. that can be expressed in a formal and machine processable way.

Railway Domain

Technology Domain

KPI

NeedsRequirements

Priorities Scenarios

Monitoreddata

InformationOntology Knowledge Services

InteGRailSystem

Definition

Consolidation ofState of the Art

Railway Domain

Technology Domain

KPI

NeedsRequirements

Priorities Scenarios

Monitoreddata

InformationOntology Knowledge Services

InteGRailSystem

Definition

Consolidation ofState of the Art

From XML Taxonomies to XML Ontology InteGRail starts from results of previous projects like EuRoMain [5] in which a domain specific XML-language was used to represent Railway taxonomies. While taxonomies enable us to formulate an accurate hierarchical classification to support the discovery of data and enables a first step of aggregation and combination, the Ontology is an enabling technology that allows new information to be inferred from existing one. We will now embed the data into its proper formal relationships that allows a very fine grained on-line analysis of monitoring information with automated reasoning since the knowledge representation together with the measured values from the sensor level is directly semantically interpretable by software agents.

Fig. 5: modelling the system

Machine processable monitoring information for improving the intelligent integration approach of InteGRail OWL, the Web Ontology Language, uses XML-based encoding and is compatible with RDF-Schema, the Resource Description Framework of W3C standard conceptual modelling that enables us to capture relationships in more detail. The big advantage of OWL is that it has a well founded formal semantics on the basis of description logics that provides automatic reasoning support for the given ontology. OWL offers the possibility for execution semantics in the form of on-line distributed intelligent processing to answer specific management and maintenance queries as they arise, because it represents domain concepts, attributes, values and relations, while the actual values are mapped automatically via the perception layer. Improving the perception layer of instruments and sensors At sensor level to observe all assets of Rolling-Stock, Infrastructure and to take into account Operations and Traffic-Management, will achieve a spatial distributed virtualisation of instruments and sensors. Dynamic adaptation together with better self-description of the measured data and proper sensor and system relations is used. This can be done to describe several views of the system such as electrical properties e.g. power-supply, mechanical dependencies e.g. pneumatics, and also spatial and topological relationships that will influence the system behaviour and maintenance process. In this way, information is added and enriched with its relevant context, and the Ontology approach is used as enabling technology for better integration. As a result it is possible to make content meaningful to

computers. So adding the ontological characterisation will enable the knowledge representation to be exploited by a description logic approach for inferencing other information on higher levels such as maintenance and management aspects based on actual terms detected by sensing the real world. Furthermore it is possible to detect inconsistencies in the model and system behaviour by machine processable inference analysis that opens the door for a semantic driven condition based monitoring and maintenance approach. The project aims to build a Railway Ontology as a basis for further standardisation work.

Interoperability of information systems

An initial common understanding is established among different stakeholders on the basis of a common ontology. This is the first step for integration, because it ensures the basis for interoperability and clear semantic inference in the required distributed generic architecture. A holistic coherent information system with specific views The perspectives from which the different railway domains like Rolling-Stock, Infrastructure, Operations and Traffic-Management are requesting information in order to fulfil their management and maintenance optimisation are different in some respects, but should be also be consistently related to each other. Therefore InteGRail aims to build specific views on a holistic coherent information system (fig. 6). Furthermore they are distributed in locations across Europe, which is also true for the mobility aspect of Rolling-Stock equipment.

Fig. 6: Integration by means of a common model

Scalability of the intelligent information system The sensor and other domain knowledge is evolving and therefore should be extensible in the future. This is the reason why content will have continuously scalable characteristics to allow evolution for future needs. This implies that scalable interfaces, with contexts, which can be composed in different ways, are used as a continuum for modularisation and scalability. For these generic services, mediation within a network of collaborating agents is needed. This leads to a distributed adaptive service grid for ubiquity and pervasive computing. Intelligent services allow the analysis from different views and therefore will support specific queries, route and delegate those to other analysis services e.g. services for predictive

analysis that enable condition based monitoring because the context is conveyed and related within the ontology. This can only be done, if semantic interoperability is achieved, preserving formal properties such as logical consequence within the proposed distributed railway ontology. In this way we are able to open shared information spaces to integrate the major railway subsystems knowledge.. From composed services to semantic integrated services This leads to flexible information sharing mechanisms based on Semantic Web Services that are well described in their semantic behaviour, to solve service discovery, mapping and composition based on the underlying execution semantics for the services. In addition, the challenge for InteGRail is to achieve inference as distributed reasoning on the given spatial distributed ontology and corresponding inference engines. Routing and delegation from one service to the next one within the service grid is needed and we have to take dynamic characteristic and scalability of the system into account. The integration development: from data exchange to information integration Instead of sending thousands of messages describing events that are classified to hundreds of strong predefined and fixed states that lead to pre-planned decisions, there is an on-line opportunity to users to explore and deduce new information based on the actual information set. Instead of using a closed and dedicated approach that leads to a stove-pipe system, open, flexible and generic inference mechanisms can be applied to the ontology. The intelligent query mechanism will find the corresponding information, its relations and is able to compute variables to the set of possible values under the given restrictions e.g. to explore all possible root causes for a failure. This intelligent inference approach integrated with condition monitoring supports Condition Based Maintenance and incipient fault detection to improve proactive maintenance procedures because the context is conveyed. Trust determination by adding semantics It is also important to take security aspects into account, for InteGRail we expect these to be based on cryptographic credentials. For many stakeholders it is important to allow or deny access to information based on access rights and agreed service level for a trusted party, which is authenticated via digital signatures. Even more important is the requirement that, smart services only make logical statements about monitoring resources that they can trust. Therefore security aspects will be integrated into the ontology and services of InteGRail. InteGRail will allow on the one hand modelling of machine understandable situations and on the other hand universally interoperable services within an agent based adaptive service grid.

Adding intelligence

Current Practice As demonstrated in many other projects, the rail industry is now very able in the areas of monitoring and information collection with large numbers of parameters being monitored and recorded. The increase in telecommunications bandwidth and ubiquity combined with reduction in cost paves the way for systems which routinely transfer the recorded information from monitored assets to wayside information repositories. The result of this progress is that there are many different repositories of information which can be accessed with bespoke tools for further manual and automated analysis. As the needs or new technologies become evident, new repositories and tools are being created to automate additional monitoring and analysis functions. Value of monitoring and analysis Monitoring and analysis can provide information to reduce costs in a number of areas: Two examples are Maintenance and Operation. Maintenance costs are reduced by automation of inspection activities and also by increasing the period between maintenance tasks. Operation costs are reduced by early detection of faults before they impact operation and improved information on asset condition. Enabling this requires greater intelligence in the sub-systems, additional sensors, better connectivity and new storage and analysis tools.

So far the development efforts have produced workable systems for supporting reduction in costs, by monitoring and analysis of information from railway assets. However further development in this area has revealed some difficulties in analysing and sharing information that are now becoming evident. Problems encountered There are several main problems that restrict the development of more sophisticated monitoring and analysis tools:

• Rail systems, sub-systems and devices are intrinsically similar but have been developed in isolation and so their monitoring and storage systems do not use a standard schema for representation of information .In some cases where the same schema is used there may be differences in context of the information due to differences in application of the monitored device and the environment it operates in

• Information required by one system is actually already monitored and analysed by another system, which means that adaptors and system integration is necessary or in the worst case parallel monitoring systems are created

• Systems produced are specialised for a particular product or family of products which leads to the need for multiple tools to interface to sub-systems and analyse data.

• Direct access to sub-systems is not always provided and the proxy or gateway functionality used changes the structure of the information provided, so standardisation in sub-systems is lost when information is transmitted (for example to wayside)

InteGRail Efforts to Solve These Problems The InteGRail project is working on a number of developments to support systems that can solve the problems mentioned. The consistency and utility of collected information will be improved by the development of an improved representation of information through development of the railway ontology. This structured method will allow us to classify the railway assets or ‘objects’, adding additional attributes and describing the relationships between objects. This has the effect of adding information and putting it into context. This is part of the tacit skill of (for example) a maintenance technician when assessing information about failures and problems. The systems that operate on this information should have less need for specialisation and can therefore be applied more widely, for example an analysis tool that considers door problems can be applied to all instances of doors, and does not need to be specially adapted for each different door type and installation in different vehicles. Once this improved information layer is added we should be able to improve the continuous analysis of large volumes of information through development of automated reasoning systems that operate on the railway ontology and need less manual work in both their development and operation. Using the example of door analysis again, a possible example is the automation of the analysis of blocked doors across different fleets with different doors, running on the same route. This information layer will also enable easier development of solutions for creation of interoperable systems, such as systems that exchange information between infrastructure monitoring systems and rolling-stock monitoring systems. Areas of benefit It is expected that activities to integrate the following areas can benefit from this development:

• Short and long-term maintenance planning (Rolling-stock and Infrastructure) • Vehicle condition monitoring • Condition based maintenance • Depot automation/ rolling-stock interaction • Operations centres • Traffic management • Infrastructure/ rolling-stock interaction • Service level management • Process management (of Maintenance, operation, depot, etc)

Test scenarios We will test scenarios that need analysis of complex distributed and interacting systems such as:

• Wheel/ infrastructure interface, where information can be analysed coming from both the vehicle and ground systems

• Operations/ maintenance depot interface, where effects of competing decisions can be analysed • Traffic management/ vehicle condition interface, where the effects of vehicles in different

conditions of functionality on the operating railway can be analysed. Integration with processes We will also investigate the effects of this information layer on business process operating systems which are being used in service industries to enforce and automate workflows and manage service level agreements.

Expected final results

The final outcome of InteGRail will be: • Specifications for InteGRail sub-systems and architecture for

o Intelligent On-board systems o Intelligent Maintenance Systems o Intelligent Management Systems o Intelligent Communication

• Proposals for standards From the above specifications the project will submit proposals to a number of different bodies and committees for standardisation. In particular, working groups in CENELEC and ERA will be addressed

• A number of integrated demonstrations It is planned that demonstrations will be made in a number of countries, where actual rolling stock, infrastructure and systems will be used (although setup will be in demonstration conditions)

• An example working system The project will also deploy an example system on the public internet (subject to a registration process) to allow further more limited demonstrations to be made (although this will be constrained by limits to access to ‘live’ railway systems)

Conclusions

The project started in January 2005 with the collection of initial requirements and will be finalised at the end of year 2008 following the demonstrations and final report. Presently, the specification phase is in progress and by the end of this year the implementation work will start. A demonstration involving real trains and infrastructure is envisaged, which will show the real benefits of the InteGRail system, coming from interoperability [4] of railway information systems, smooth information flow and ubiquitous availability of relevant information. It is important to underline that the InteGRail holistic approach addresses for the first time all parts of the railway system, in order to achieve a global optimisation concept. This required to define a new methodology for inter-disciplinary cooperation, using state of the art technologies, and to activate adequate resources, in terms of funding and expertise. First results are encouraging and show a big potential. Final results will be delivered in 2008, so they will be available and proposed for presentation at the next WCRR event.

Acknowledgements

This InteGRail project is partially funded by the European Commission DG RESEARCH under the “Sustainable Surface Transport” thematic area. The content of this publication is the sole responsibility of the authors and in no way represents the view of the European Commission or its services. The authors thank UNIFE, the Union of European Railway Industries (www.unife.org), coordinating partner of the InteGRail project, for its valuable support.

References

[1] E.Renner, P. Umiliacchi: “TrainCom: an Integrated Communication System for Intelligent Train Applications” – WCRR ‘03 – World Conference on Railway Research – Edinburgh, Scotland – 28 September - 1 October 2003

[2] R. Shingler, P. Umiliacchi : “Advances in railways maintenance: the EuRoMain project” – WCRR ‘03

– World Conference on Railway Research – Edinburgh, Scotland – 28 September - 1 October 2003 [3] P. Umiliacchi : “The role of European Research in the railways modernisation process : the ROSIN

(Railway Open System Interconnection Network) project” – WCRR ‘97 - World Congress on Railway Research – Florence, Italy - 16-19 November 1997

[4] P. Umiliacchi “EC projects for railway interoperability”, ITS Hannover 2005, Strategy Session 7:

ITS@Rail - Hannover, Germany – 1-3 June 2005 [5] U. Zahner, “Information Technologies in support of Maintenance: EuRoMain – European Diagnostic

Data Network EDDN”, EuroMaintenance 2004, Barcelona, Maintenance and Sustainability Session – 11-13 May 2004