Success Story: Mobility
Developing Predictive Analytics for Trains
Together with Europe’s largest private lessor of railroad freight cars we build a system of predictive maintenance.
Rail freight is an industry that has traditionally relied on tried and tested technology. But high-tech approaches are also being developed, giving enterprising companies an edge. One such early adopter is Europe’s largest private lessor of freight cars with a fleet of about 100,000 cars that are leased and used by clients across the continent.
Aiming for a Global Novelty
As part of its digitalisation initiative the company is equipping all its freight cars with GPS sensors that provide information about the current location and past routes of the 100,000 cars. Based on that information, we jointly develop a system of predictive maintenance. The application forecasts the dates when certain components of specific cars will need to be replaced – an intelligent tool for freight trains that is probably the first of its kind in the world.
To develop that tool, we sent our data scientist Josef Dukarevic to our client’s department for digitisation. There he applies Data Science technology and proceeds in three steps. First, he uses time series analysis to determine trends in the use of individual freight cars.
Keeping Track
In a second step, Martin uses a combination of statistical algorithms and machine learning to estimate wear of components. Based on the previously determined trends, these algorithms forecast when a certain freight car will likely have travelled the distance to require maintenance. To finally apply these complex algorithms to all of the 100,000 freight cars, Martin relies on data storage solutions and the flexible infrastructure of a cloud service.
“Our client placed a lot of trust in us. Our colleagues had a lot of freedom to experiment with data science and AI and were also involved in strategic discussions about analytics. As a result, they identified all the more with our goals and with our joint success.”
Our work enables our client to progress in several ways. First, concerning expenditures, automatically calculated predictions of freight car usage and their associated wear of components help reduce maintenance cost and downtime. Secondly, on a strategic level, the integration of the predictive models into a cloud service offers flexibility and the possibility to scale the system.
Scope of our Models
Scope of our Models
Materials monitored in component wear
Railcar types factored in
A third advantage for our client is the expansion of its offering. Specifically, our work contributes to a new information service that the company offers to its own customers leveraging intelligent algorithms and geospatial data. The system provides a specially designed interactive analytics dashboard displaying details about freight cars leased by a client.
In particular, the system indicates: Where is a freight car currently located? When is it likely to arrive at its destination? And on which day is each car expected to reach the mileage agreed in its leasing contract?
Workload
Workload
Daily measurements of component wear
Traffic patterns considered
In sum, we support our client in diverse evolutionary steps. Thereby the company gains an edge over its competitors and establishes its position as digitally savvy market leader.
How to Get in Touch
Do reach out to our regional experts who will be glad to assist you or put you in touch with our specialists.
