Software-as-a-service for digitalisation of switches

Infrastructure managers can just buy IoT sensors to collect data and try to do the rest themselves, but this is not the most optimal way, says Thomas Böhm, Head of Product at KONUX. The company combines Machine Learning algorithms and IoT to deliver software-as-a-service solutions to digitalise railway assets.

Currently, KONUX is working together with Deutsche Bahn (DB) on the digitisation of their switches. The Munich-based AI start-up won a DB tender for the condition monitoring of switches. Initially, KONUX will digitise 650 switches so that passengers can travel more reliably by train on heavily congested lines. DB is investing 15 million euros in this stage of the project. “This was the first ever tender for software-as-a-service in rail. 650 assets are now equipped with the KONUX IIoT devices and monitored in the KONUX system. The installations were performed this year during full traffic, which is quite an achievement in itself”, says Thomas Böhm.

KONUX delivers an end-to-end solution, and also designs the hardware to collect the data that is needed to provide insights about assets. “Many customers think in terms of buying IoT sensors off the shelf, but IoT devices are constantly evolving, they need updates, operation and operation. So do the analytic models built upon the data of those devices. Infrastructure managers are experts in managing things like steel, concrete, and signalling systems, but I believe they are underestimating what it takes to do this with IoT devices and machine learning models with a short lifecycle”, says Böhm.

Predicting degradation

The system already works really well for switches, and in predicting the degradation of crossings and steel surfaces, there are still things to be learned and improved. “As there are quite few assets we currently monitor, there is little failure to observe, which is of course a good thing. However, observing failures can be used to check and improve the predictions.” There are also many variations in traffic depending on the location, and different construction characteristics which need to be taken into account.

An IoT sensor on the tracks

Another challenge is the change in maintenance regime. “Assets usually have a specific inspection cycle. They are checked periodically, for example, every 6 months or 8 weeks. These are set up in a way it is expected that in between the inspection no large failures occur.”

By using data from a system that also monitors how many and what kind of trains run on a section, this can change this regime. “We will be able to see what drives degradation. This means each asset can be looked at individually, and the frequency of inspection is dependent on things like how intensively the track is used.” This requires a system change, says Böhm. However, this cannot be done quickly, as the safety also has to be guaranteed, with national regulation bodies. “Just like with automatic train driving, the technology could be there, but there are also the legal and ethical aspects, which need a longer discussion.”

Data is shared too little

There is a lot of other data available of network usage, such as from other companies working in digitalisation of other aspects of the railway network, as well as measuring trains. “The exchange of that and combining data is happening too little.” Böhm advocates for the integration of many different sources, as a role of the infrastructure manager. “This is not only useful for the railway network as a whole, but it also helps the various systems, which can develop further and become better with more data.”

On the Intelligent Rail Summit on 21-23 September in Bilbao, Thomas Böhm will give examples from Germany, France, and Norway of how artificial intelligence and industrial Internet-of-Things help make maintenance plannable to improve capacity, reliability, and cost-efficiency. Registration for physical and digital tickets is still open.

Author: Esther Geerts

Editor of RailTech.com

Add your comment

characters remaining.

Log in through one of the following social media partners to comment.