Many trains operated by Deutsche Bahn are equipped with the global positioning system (GPS) for position tracking. Fundamentally, however, GPS is not very precise. At some locations, for example, in tunnels or stations, the satellites are not accessible. In addition, there is the matter of the high speed at which the trains are moving, and their considerable length. For more precise automatic vehicle location, other methods have to be used.
“We have therefore considered how we can make the automatic vehicle location more accurate”, says Konrad Winkler, who is responsible for the RailPos project at DB Systel. First there was an attempt to address the inaccuracy of the global positioning system. With “differential GPS”, correction data is transmitted by radio and compared with the satellite data. This enables the discrepancies to be corrected. “In addition, the vehicle knows the precise route and can check the measured data against the track centre line”, says Winkler.
Determination of the exact position of the train during the journey can also help to pinpoint sources of damage. On some trains, problems with the tilting system can cause overloading, leading to failures. To be able to determine reliably where possible causes of overloading may be located on the routes, it is necessary to record not only the exact position but also the spatial orientation or tilting of the train. It soon became clear that GPS alone was not up to this job.
Sensors measure tilting and loading
For this reason, DB Regio and DB Systel now rely on sensors that measure the precise position and inclination of the trains. The sensor-based measuring system, in connection with the differential GPS, records the vehicle movement and the position and records the tilting processes and loading torques. “We rely on the use of commercially available sensors”, says Winkler. These have steadily fallen in price over recent years, enabling the measuring processes to be performed on a relatively small budget.
DB Systel provides the complete IT ecosystem for the measurements. There are still some challenges to face: for example, the volume of data must remain manageable. The installed sensors produce a veritable flood of data. One additional problem is that the exchange of data must also continue to function even under difficult conditions. For example, a wireless communication network is not universally available.
The supplied on-board computer prepares and transmits the data. “This performs a preliminary selection of data during the journey in order to reduce the total volume”, reports Winkler. The information should be processed before it leaves the train. From the total volume of data recorded, further information can be generated at a later stage in a Big Data approach.
Development specially geared towards rail transport
In collaboration with DB Regio, DB Systel is developing the corresponding software for this solution for the analysis, as well as the interfaces to the sensors. “The RailPos solution that is tailored specifically to rail transport and the algorithms we have developed represent a unique selling point”, says Winkler. It means that the data can be sampled while the train is running, rather than waiting for the next maintenance stop.
The algorithms that we have developed represent a unique selling point.
Trial recording started recently on a shunting locomotive. The measured values are used for calibrating the system and improving the algorithms. On completion of the test phase, it will then be possible to tell precisely which sensors are to be used and how the antennas are to be positioned.
Once the pilot project has been completed, the system can then prove its effectiveness in productive service on a fleet of test vehicles. The possibility of recognising the need for maintenance in trains and, in addition, possible locations of damage on the tracks will continue to be developed: with RailPos, it will be possible to evaluate operating hours and other data from the trains in real-time, which can then be used for early warning, fault analysis and predictive maintenance planning within the workshops. The pinpoint location and use of sensors not only creates significant potential for savings, but also opens up further fields of application for the future, such as forecasts regarding connections to other means of transport.
Locating trains in stations could also be improved with the RailPos system and simplify navigation for passengers, according to Winkler. For example, passengers could use an application in stations to see exactly where the compartment with their reserved seat is located. Even “live” deviations from the timetable could be calculated using automatic vehicle location and taken into consideration in recommendations for passengers.
Indoor vehicle location for navigation in workshops and multi-storey car parks
But RailPos is not limited to the positional location of trains. For indoor navigation, where satellite location cannot be used, there are further possible applications that DB Systel is currently working on. For example, the precise position of certain tools or containers can be determined in a repair shop. This is done by radio positioning using hotspots installed in the roof of the building and receivers mounted on the tools or containers. This considerably shortens the journeys that personnel have to make in the workshops, as they always know exactly where each tool is to be found.
The same would apply, for example, to the Flinkster Carsharing fleet, according to Winkler. The precise position of vehicles in multi-storey or underground car parks could be determined in this way, thereby enhancing the travel experience for customers. Winkler can even imagine an app in which all information about the current locations of different modes of transport – such as buses, trains and Call-a-Bike cycles – would be available. With the aid of such an integrated application, planning a journey would be made much easier. Thanks to such innovations, not only customer satisfaction is increased, but considerable savings are also achieved, as in the case of the tilt sensors.