Predge Rolling Stock™️ - Wheel Profile Prediction (WPP)

Predge Rolling Stock – Wheel Profile Prediction (WPP)

Wheel Profile Prediction, or WPP, aims at delivering refined information from wheel profile measurements. Predicting the wear and introducing remaining useful life estimates is one step towards optimizing the wheelsets’ life length.


Wheel profile measurements, in general, provide a good insight into reactive measures related to maintenance of the wheels. Relying on individual measurements however is not enough to enable a preventive and fully optimized maintenance approach. It requires insights about future wear and reliable information. To enhance the measurements’ data, handle uncertainties, and varying conditions, further analytics is needed.


The wheel wear prediction solution, WPP, replicates every wheel as a digital twin tracking the condition of the individual wheel, self-correcting itself, and performing analytics as new data comes in but also predicts what next measurement should be expected from the field. The feature manages measurements provided handheld devices, one or many wayside devices, or a combination of both. It can synchronize the data with distance information to gain results in distance rather than time when needed. Since it treats every wheelset individually, it will account for different vehicle dynamics and contextual differences

The analytics behind the feature is based on our AI and Machine learning principles with hybrid models. The hybrid models are not purely data-driven and take the physical aspects behind these damages. Still, it is continuously and automatically updated as new data is collected.


The WPP introduces a high level of reliable information about the wheelsets, created in the context of its operation. It gives the users the information about the current profile and its predicted future and information about the wear rate and when to perform which action. It offers excellent flexibility by treating every single wheel side as an individual, and in combination with its maintenance and operational data, this builds up a digital twin of wheelsets. We enable high confidence in prognostic and prescriptive forecasts to generate automatic work orders. Users can overview the entire fleet or compare individuals. Combined with integration to maintenance systems, individual wheelsets can be monitored. The digital twin then supports reliability, maintainability, supportability, availability, and safety evaluation and analysis of the wheelsets.

What’s next

In our continuous development process, we always strive to improve and develop all our features. Besides the overall improvements, we want to broaden the value provided by the analytics. By integrating data from the wheel lathe, we will predict the need for turning on an individual level and the entire wagon fleet. Having this type of integration will also allow for better optimization of the wheelsets’ life-length and a better understanding of the wear throughout the wheelset’s whole life cycle. This would help to extend the life-length.

Success story

In the north of Scandinavia, Europe’s heavy haul operator and mining company LKAB, are dependent on reliable railway operations to efficiently transport their iron-ore products to the harbors in Norway and Sweden. The heavy haul railway is a vital key factor for the LKAB supply chain, linking the mine and the port. Hence, keeping the preventive maintenance activities dominant and reactive maintenance as low as possible is essential in their maintenance strategy.

The Wheel Profile Prediction (WPP) is a key feature for us to enable preventive maintenance activities. It gives us individual RUL estimates and wears rates in terms of wheel profile for our entire fleet. Since it has been tailored to our operations the deviations are based on our defined maintenance actions. The deviations with high confidence have been automated to generate work orders in our maintenance system.

Robert Pallari, Senior Engineer Heavy Haul Wagons at LKAB Malmtrafik AB

Predge Rolling Stock – Wheel Damage Prediction