EGIS (1994), copyright EGIS Foundation.
The operation and maintenance of a railway network can be enhanced with the use of Geographic Information Systems (GIS) and the Global Positioning System (GPS). The development of a rail mapping system and a pilot infrastructure management system in the Department of Geomatics has provided an appropriate platform to investigate the use of these technologies in the railway industry.
The paper focuses on the use of these technologies for rail infrastructure management. A number of issues with the implementation and operation of these technologies in the rail industry is discussed. The main issue identified was the need to establish a homogeneous spatial reference to facilitate the successful implementation of GIS technology.
The Public Transport Corporation (PTC) of the State of Victoria, Australia, was established approximately 140 years ago to manage and operate the State's rail network. During that time numerous surveys of railway network have taken place, with the majority of these surveys referenced to various local datums. This problem was somewhat alleviated in 1966 with the adoption of a national mapping datum across Australia. However, despite the datum being adopted, not all departments within the PTC are using it as a reference or have the need to use it.
A vital part of the PTC's role is the management and maintenance of rail track infrastructure. This is achieved using a track recorder car that measures the geometry of the railway network on a 3 - 4 month cycle. The track recorder measures five attributes to the nearest millimetre and faults are identified when a predefined parameters are exceeded. The fault location is recorded in terms of the nearest whole metre chainage, and the necessary information is then passed on to track maintenance.
The chainages measured by the recorder car are referenced to the kilometre posts (KMP) that are positioned alongside the track. A manual technique is used to reference the system to each KMP that is passed. For example, on passing KMP 63 the adopted chainage would be 63,000. Although a post may seemingly identify the beginning of a kilometre section of track, the distance between two concurrent KMP is rarely one kilometre. The effect of this re-referencing technique is that the track is defined as a series of nominal one kilometre track segments.
[End Page 1384]
The track recorder car also records the position of track furniture. Bridges, track descriptors, crossings, signals, tunnels, and platforms are some of the features that have their positions recorded during each track recording run. These features are used to assist maintenance crews in locating the track faults and the identification of appropriate repair equipment. With an experienced track recorder operator, the accuracy of locating trackside features is in the order of 10 metres. For track infrastructure management this proves to be adequate, and experienced maintenance crews are able to locate work sites with ease using the combination of KMP, track furniture and identified markers.
A fundamental requirement of a GIS is a homogeneous spatial reference system. Thus, it would seem appropriate to use the national datum as the underlying spatial reference for the railway management system. However, the success of the management system has little to do with the choice of reference but rather the ability to represent data to best meet the needs of users. Despite the fact that the national datum will promote the use of data gathered by external and internal parties, it is obviously important to retain a local referencing system to support field operations. For example, the current methodology of using chainages and track furniture to locate work sites is obviously user friendly for track maintenance crews. Therefore, it is vital that users are able to reference features to the national datum and, if required, their own local datum.
The ability to transform between datums means that a large quantity of data from internal and external sources will be available for sharing. However, to ensure the reliability of shared data, accuracy standards relating to data collection must be introduced. With low or unknown data accuracy, the reliability placed upon the information derived from other sources limits the reliability of decision making. Users need to know whether data from additional sources will enhance or inhibit their own analysis, since accuracy of unknown data has implications for the reliability of any analysis performed.
It has been identified that a homogeneous spatial reference must be established for the implementation of a GIS based railway management system. In order to implement the management system a comprehensive mapping programme is needed to establish the homogeneous spatial reference system. This program will also provide a platform on which parameters can be calculated for the transformation from the national datum to any local datum in use. Importantly, the reverse can apply with local datums transformed to the GIS frame of reference to allow field crews to add new or updated data to the system.
The generic task of mapping concerns the real time positioning of objects over ranges and to precisions normally achieved by static field surveying measurements. Recent research into mapping has centred on the use of the satellite based GPS as the primary method of measuring the three dimensional location along a rail track.
On first examination the use of GPS seems to be the most appropriate solution to quickly and efficiently map the railway. The use of this facility in pseudo range or carrier phase measurement modes has the potential to provide the continuous location of a moving platform to accuracies in the range of one centimetre to ten metres (Goad 1988, Schwarz 1990). However, it is recognised that GPS alone is not a complete solution to the mapping task. The pseudo range mode of operation requires additional data sources to improve the local accuracy of positions, whilst the carrier phase operation requires additional data sources to recover from a situation where signal lock is lost on one or more satellites. Experience has shown that an integrated GPS approach using inertial sensors is the most practical (Judd et al 1993).
[End Page 1385]
A number of methods are being investigated by the research community to supplement the GPS data with independent measurements from other sources. Electro-mechanical gyroscopes may be used to give the orientation which, with velocity measurement, will enable continuous tracking of the instrument platform (Chen et al 1990). Video systems may be also used to obtain repeated images of the surrounding detail that will allow the computation of the instrument platform position using real time photogrammetry (Wong et al 1989). Simultaneously, techniques which allow the quick recovery from a loss of lock are being developed.
Mapping systems have been proposed or successfully used for transport studies of both road (Schwarz et al 1990, Novak 1990) and railway (Heij 1989, Judd et al 1993) networks. Research work in the area has been active and it is clear that the optimum solution to the augmentation of GPS positioning data will come from a combination of a number of techniques. Regardless of which technique is adopted in the future, current rapid mapping systems using GPS as the primary data source must utilise supplementary position or dead reckoning information from independent data sources.
Two typical rail operation scenarios are discussed using the GIS and GPS technologies. These scenarios highlight some issues involved with the use of the GIS and GPS technologies in the rail industry.
A pilot GIS was developed using railway track data obtained from a mapping programme utilising an integrated GPS solution (Southby 1993). Using local transformation parameters defined during the mapping programme, the track geometry data was referenced to the homogeneous railway track data. The latter was then combined with the attribute data collected from the track recorder car to form the system. To test the functionality of the system, current maintenance methods were modelled and tested. The basic GIS was then used to investigate extended modelling of the data for rail management.
With the current system, the raw track data collected by the recorder car is processed to yield a variety of tables and reports defining the status of the track. The outputs can be classified as summaries, which are limited to textual information. The limitation of the outputs becomes evident when it is required to summarise and represent the massive amount of data collected by the track recorder car. For example, users are informed of the number of kilometre sections that fall within specified fault ranges, but they are not told where these faults occur. Hence, users are only able to determine the spatial location of faults by manually cross referencing other computer generated reports.
The GIS automatically classifies the faults for each parameter and the results are displayed either graphically or in tabular format. This information can be used to create comprehensive reports which incorporate a balance of textual and graphical information. Digital photographs can also be included in the system to enhance the user' s visualisation of the track.
Queries were developed such that data from different epochs can be analysed. Using one or more fault attributes, the queries allow patterns and trends to be easily detected and monitored. Comparisons between epochs of data are performed quickly and accurately, with results displayed to users in an understandable format. Such analysis can be performed manually, but is far less efficient than using the GIS.
[End Page 1386]
The recorded track geometry data is referenced to time and thus can be used to facilitate the prediction and planning of maintenance requirements and associated budgets. With a wealth of historical data available, current data is able to be combined and averaged over time such that trends can be observed. The rate of deterioration can be monitored and hence maintenance or track renewal can be predicted. The system will require that data be collected for small segments of the track, which will allow for maintenance optimisation and the identification of influencing factors. Through the incorporation of statistical testing, the power of the decision making will be increased.
The system provides users with increased visualisation and analysis of recorded attribute data. With the appropriate management criteria in place, prediction and planning of maintenance projects are additional features that can be included within the system to increase the power of decision making. The ability to analyse and forecast such operations in turn assists management with budget predictions.
GIS for rail infrastructure management can meet the analysis needs associated with the track recording car, provided that all the data is accurately referenced to the underlying spatial system. The repeatability of feature identification is important to support the analysis capabilities of the GIS. Using the current referencing system, a fault which occurs at a chainage of 10,073m may next time be identified by a chainage of 10,062m. It becomes evident that accuracies in the order of 1-2 metres are probably required for spatial referencing of features in a GIS.
In any railway network, careful and detailed planning of train schedules is required for the optimal and safe operation of trains. The scheduling process involves many track parameters including the occupation of track sections by maintenance crews. GIS offers the appropriate level of technology to efficiently deal with these dynamic information requirements. With the implementation of a GIS-based vehicle track system, track and train information can be combined in real-time to facilitate optimal train scheduling in response to changing operational conditions.
A central control office could monitor the position of all trains and be used to monitor the standing of the current train schedules. The GIS might be used to dynamically alter the schedule through a combination of changes in speed and travel route. The system could be used to alert train operators of changes in the schedules through a train communication system interfaced to the GIS. There is also the potential to have a train scheduling system operating in conjunction with an infrastructure management system, that predicts maintenance requirements and alerts the train scheduling system well in advance of the track occupation needs.
This type of system is reliant upon the ability to accurately locate rolling stock throughout the rail network. GPS technology seems to offer an appropriate means of meeting the locational requirement. It has been discussed previously that GPS will not operate in the rail environment continuously due to obstructions and signal masking, and an integrated solution will prove to be necessary.
The use of a single autonomous GPS receiver can only provide location information that is accurate to 100 metres. This may be acceptable for the majority of railways operations, but there will be instances where an increase in accuracy is necessary. For example, the position of trains relative to switches in crossing loops will require location to an accuracy of 1 - 2 metres. This can be obtained using the differential GPS technique and involves the use of a GPS base station.
[End Page 1387]
With trains encountering tunnels, buildings, trees, and bridges, each of these contributes to the effects of multi-pathing and masking of the satellite signal resulting in degraded positional accuracy. In response to this problem an integrated solution is required whereby data from multiple sources is used to locate the position of the trains. The use of GPS in combination with in-track transponders and distance loggers is a solution that may be appropriate, however, the inclusion of on-board inertial positioning sensors is an option that might prove to be more attractive. The outcome is that GPS alone will not be adequate to meet the train location requirements in a rail network.
The pilot project showed that GIS can meet the functional requirements needed for the analysis and management of railway infrastructure. The functionality of the GIS is based on data referenced to a homogeneous and consistent spatial datum. This presents a problem in the rail industry as current work practices do not use a consistent spatial reference datum for the maintenance of the rail infrastructure.
This dichotomy is resolved by establishing transformation parameters between the spatial systems of the GIS and those of the user. This will facilitate the maintenance of current work practices and allow information gathered by field crews to be used in the GIS.
The use of GPS technology can provide the homogeneous spatial reference for the GIS and through transformation parameters maintain the use of local reference datums. In a direct sense, the GPS can be used to spatially reference data as it is collected in the field, and indirectly, the GPS can be used to find the parameters to transform between the various spatial datums.
GPS offers the level of accuracy required to meet the GIS needs, however, to achieve the reliability in spatial reference GPS may need to be operated in combination with other positioning sensors. The outcome is that GIS and GPS have application in the rail industry provided close attention is given to identifying the limitations of using GPS and the associated spatial location requirements and standards.
Betak, J. and Vaidya, A. (1991), On Becoming an Information-Based Company - A Role for GIS, GIS World, 4(8), pp. 44-47.
Chen, W. and Cross, P. A. (1990), Integration of GPS and an Inertial System for Precise Surveying Applications, Survey Review, 30(238), pp. 375-395.
Eissfeller, B. and Spietz, P. (1989), Basic Filter Concepts for the Integration of GPS and an Inertial Ring Laser Gyro Strapdown System, Manuscripta Geodetica, 14, pp. 166-182.
Esvald, C. (1989), Computer- Aided Maintenance and Renewal of Track, The Fourth International Heavy Haul Railway Conference, Brisbane, Australia, pp. 118-123.
Gibbons, G. (1990), On Track with GPS, GPS World, 1(3), pp. 28-37.
Goad, C. C. ( 1988), Potential Real Time Application of GPS Phase Tracking in Construction Surveying, Proceedings of Construction Congress I, pp. 252-257.
[End Page 1388]
Heij, B. (1989), A Dissertation on Track Measuring Systems, The Fourth International Heavy Haul Railway Conference, Brisbane, Australia, pp 541 - 545.
Judd, A.M., Leahy, F.J. and Shortis, M.R. (1993), Rapid Rail Mapping with GPS and Other Sensors, Proceedings Satellite Navigation Conference, Sydney, Australia, 12pp.
Novak, K. (1990), Integration of a GPS Receiver and a Stereo-Vision System in a Vehicle, Proceedings, SPIE Vol. 1395 Close Range Photogrammetry Meets Machine Vision, Zurich, Switzerland, pp 16-23.
Schwarz, K. P. (1990), Kinematic Positioning - Efficient New Tool for Surveying, Journal of Surveying Engineering, 116(4), pp.181-192.
Schwarz, K. P., Lapucha, D., Cannon, M. E. and Martell, H. (1990), The Use of INS/GPS in a Highway Survey System, Proceedings, Commission 5, 19th International Congress of FIG, Helsinki, Finland, pp 238 - 249.
Southby, L. J. (1993), Railway Information System Pilot Project Report, pp.25.
Wong, K. W., Wiley, A. G., and Lew, M. (1989), GPS-Guided Vision Systems for Real-Time Surveying, Journal of Surveying Engineering, 115(2), pp.243-251.
[End Page 1389]