Detect damage to tracks and points early

Railways detect damage to tracks and trains from sensor data

Needs setupAI Act: no extra obligationsMonitoring sensor and machine dataPassenger transport & mobility

Fits: Production, engineering & maintenance

Wear on points, tracks and vehicles becomes visible before it leads to disruptions and speed restrictions. Maintenance plans work in scheduled possessions instead of after a failure.

Is this for you?

  • Points failures, track faults or damage to wheelsets cause noticeable delays or closures for you.
  • Point machines, track recording runs or trackside measuring points already deliver data that is hardly analysed today.
  • Track or rolling stock maintenance can sustain a project over months with IT, operations and safety officers.
First step today

For five sets of points at one station, export the daily throw times and motor currents over three weeks, plus the outside temperature. Replace the points numbers with codes such as W1 and have the values checked with the template below. This shows you whether the data can distinguish stiffness from the effect of frost.

How to go about it

  1. Define assets and types of damage
  2. Collect measurement data from point machines, recording runs and vehicles
  3. Have the normal state of each asset learned
  4. Define warning levels
  5. Pass warnings to maintenance and operations management
  6. Report findings back and evaluate them

Templates to copy

Try it now
in the training
Analyse
in the training
Plan the rollout
in the training
Included in the training

The complete guide

  • 6 steps explaining exactly what to do
  • 3 proven templates to copy into your own AI
  • Safe use: data, checking, deciding, informing
  • Pitfalls, effort and how to recognise success

With the training you get all 430 guides for as long as your access runs.

See the trainingOne-off per person, no subscription

Same approach, other use cases

8 more use cases follow the pattern “Monitoring sensor and machine data”.

Detect battery degradation in your EV fleet earlyDetect falls in residents’ rooms without a cameraHave AI monitor process data and suggest setpointsMonitor cows with sensors and barn camerasPredict machine failures from vibration and temperatureDetect leaks in the water network earlyPilot electronic noses for quality testingPilot network fault prediction and automatic control
‹ PreviousSee all use casesNext ›

PV-02 · Legal status 08.10.2026