The project OnboardEU has shown that trams themselves can serve as mobile measuring systems by using AI to analyse vibrations and noise during the journey, detect damage and thus provide a cost-effective data basis for maintenance and noise reduction.
To this end, trams in several cities were equipped with sensors, large amounts of data were collected and open software and data sets were made available to experts in order to further develop the technology.
How can rails and tracks be monitored efficiently and precisely – without costly measuring journeys? The recently completed research project OnboardEU shows that trams themselves could become mobile measuring systems in the future. With the help of artificial intelligence, they analyse vibrations and noise during the journey, detect damage to tracks and rails – and thus provide a reliable and cost-efficient data basis for maintenance and noise reduction in urban traffic. The German Aerospace Center (DLR), AIT Austrian Institute of Technology GmbH, Vienna, and i4M technologies GmbH, Aachen, worked together to realise an innovative onboard measurement system with edge computing functionality. OnboardEU was funded with around 750,000 euros as part of the mFUND innovation initiative of the Federal Ministry for Digital and Transport (BMDV).
"The aim of the project was to research suitable AI methods for the automatic detection of damage to rails and track superstructures," explains project manager Dr Jörn Groos from the DLR Institute of Transportation Systems. "To do this, we recorded dynamic vehicle reactions over several months during operation and then analysed them using AI methods."
Data collection in regular operation – in four cities
On this basis, the project participants developed various machine learning methods – both supervised and unsupervised – for the detection of damage, noise mapping and low-loss data reduction through edge computing. Particularly relevant: The AI algorithms make it possible to identify stretches of tram tracks with particularly high noise emissions – and to take targeted measures to reduce noise.
Open source for the professional world – and a look into the future
In addition to the public final report, the project team is providing open source software for track-accurate localisation and an anonymised open data set via the Mobilithek. This will allow further AI processes to be trained and put into practice. Several of the onboard systems will remain in operation beyond the end of the project and form the basis for further development of the technologies and increasing their level of maturity.
"Diverse and high-quality data sets are crucial for the successful training of reliable and robust AI processes. By providing open access, we are making an important contribution to the transferability and further use of our results in practice," emphasises Groos.
The project shows how data-driven innovations can help to optimise maintenance, reduce costs and improve the quality of life in urban areas – an approach with great potential for municipal rail transport in Europe.
Vibration and acoustic data is recorded during regular tram operation, then processed on a background system, georeferenced and analysed. This allows conspicuous vehicle reactions and increased noise emissions to be efficiently mapped.
As part of the mFUND funding programme, the BMDS has been supporting research and development projects relating to data-based digital innovations for the mobility of the future since 2016. The project funding is supplemented by active professional networking between stakeholders from politics, business, administration and research and the provision of open data on the Mobilithek. Further information can be found at www.mfund.de.
