Detection of wheel faults with smart algorithms

AI

Wheel faults lead to high loads on the railway infrastructure and generate noise. SCS has developed algorithms for recognising wheel faults. This means that affected vehicles can be repaired accordingly.

  • Initial situation: Wheel fault

    Wheel faults (e.g. flat spots, chipping) lead to high stress on the infrastructure and generate noise. Various data must be linked and analysed to record different wheel defects.

  • SCS solution

    SCS has tested and implemented various algorithms for recognising wheel faults. The procedures are based on the signals supplied by the RLC.

  • Benefit SBB

    The presence of wheel faults is recognised. An estimate of the severity of the wheel defect is made for certain wheel defects. Affected vehicles can be reported to the railway undertaking so that it can repair the vehicles.

RLC systems (wheel load checkpoints) measure the weight of passing vehicles as part of the train control equipment (ZKE). The signal curves are influenced by wheel faults. The signal curves can therefore not only be used to determine weight, but also indicate whether wheel faults are present.

In the first approach for determining a wheel error, an easily determined parameter (dynamic coefficient) is calculated from the signal curve of each wheel. If the dynamic coefficient is small, there is no wheel error and if the dynamic coefficient is large, there is a pronounced wheel error.

In the second approach, an algorithm developed by SCS detects flat spots and estimates their length. It utilises the fact that a typical signal curve results from a flat spot that occurs exactly above a sensor. The algorithm searches for this typical signal curve on the measurements. If one is found, it is analysed in more detail by the algorithm in order to estimate the length of the flat spot. To estimate the length, characteristic points must be determined on the signal curve.

In the third approach, the “unwinding” of the wheel is determined, i.e. the outside of the wheel and its condition are shown on a graph. This provides a condition of the wheel over the entire circumference and the individual defects on the tread are visualised. As the sensors on the RLC system do not cover the entire circumference of the wheel, additional sensors are used for processing. The algorithms developed by SCS process the measured data from the individual sensors and then combine them to produce the processing.

The three approaches allow existing wheel defects to be detected and analysed in more detail. Based on the analysis, a recommendation can then be issued for the refurbishment of vehicles with wheel defects.

Cover picture: Heitersberg [CC-BY-SA-4.0] via Wikimedia Commons (cropped)

Related projects

Visual inspection of railway wagons

Is the block brake worn? How thick is the pantograph contact strip? Is an incorrectly fitted screw coupling hanging down? In future, SBB's "Visual ... More

Visual innovation through AI in ophthalmology

Optical coherence tomography (OCT) is a gentle laser microscope and is used by ophthalmologists to examine the retina for abnormalities. Trained AI ... More

Reliable AI for the measuring wheel set

SBB uses so-called measuring wheel sets to measure the dynamic contact forces between rail and wheel. Sensors on the wheels measure the forces. It is ... More

Predictive Alarm Detection

SBB ZKE measures the bearing and wheel temperatures of passing trains on the open track. SCS has trained an AI model that can predict applied brakes ... More

Blockchain for safety-critical interlocking logic

Blockchain makes it possible to run several computers in a network and thus increase reliability. Particularly in safety-critical areas such as ... More
Show all projects
Pascal Iselin Algorithms, Analytics & AI How can I help you?