Visual control of railroad wagons
Is the block brake worn? How thick is the collector shoe of the pantograph? Is an improperly mounted screw coupling hanging down? In the future, SBB's "Visual Inspection System" (VIS) will monitor passing trains with cameras. Vehicle owners and rail transport companies can install their own algorithms on the platform to evaluate the images.
-
Initial situation
Rail vehicles for goods and passengers must be maintained regularly. But how often? Short intervals are expensive. If, on the other hand, you wait too long, you increase the risk of an accident.
-
Solution SCS
The "Visual Inspection System" (VIS) uses cameras to monitor trains passing through. Railroad companies, vehicle owners and infrastructure managers can evaluate the images of their vehicles using their own algorithms.
-
Added value
The "Wayside Intelligence" (WIN) platform provides operators with the condition of their vehicles, or the calculated values from the algorithms. From this, they can plan maintenance and keep safety high - despite longer intervals.
Project insights
Today, around 250 measuring systems already check trains passing through Switzerland's rail network: they check whether brakes or axles are overheating, a wheel is running out of round, a wagon's load has slipped or truck covers and antennas are infringing the loading gauge. If a hazard is detected, the train control unit (CCU) alerts the control center and the train is immediately stopped, slowed down or directed to an intervention track at the next station. The train control facilities have been continuously expanded since 2008 and supplemented, for example, with natural hazard alarm systems that stop train operations in the event of rockfalls and avalanches.
The next step is now to monitor the trains with cameras. Modern image processing algorithms are able to detect many damages and dangers. A typical example is the screw couplings used to couple freight cars or even older passenger cars. Since each wagon has a screw coupling on each side, one remains unused. This must be correctly suspended, otherwise it dangles down, can hit the ground and destroy railroad infrastructure. Today, dedicated detectors are used only for dangling parts. In the future, cameras will detect various things simultaneously, for example also the thickness of the collector shoe on the pantograph, the wear of the block brakes or the remaining mileage of a wheel. This not only detects hazards, but also makes it easier to plan the maintenance of the cars.
However, the system cannot send all the images to every vehicle owner and every rail transport company that runs trains on the SBB infrastructure. The data volumes would be too large. The WIN platform is therefore set up in such a way that fast computers directly in the vicinity of the cameras evaluate the images and send only the results to the customers of SBB Infrastruktur AG (the values from the algorithms). Of course, in case of a striking value, images can also be consulted. These are stored on the computers on site.
The rail transport companies and vehicle owners can install their own image processing algorithms on the WIN platform. Via an RFID tag on the vehicle, the system recognizes who the vehicle owner is. Now the images are evaluated by the algorithms and the results are sent to the authorized users via a machine-readable interface - usually the deviation from the standard value. Based on this data, the companies can plan maintenance. SBB Infrastructure's customers therefore decide what they want to measure and at what point they want to take action.
The existing measurement systems of the train control equipment will also be integrated into the WIN platform. For example, weight measurement on the track: If a wheel has ever jammed for some reason, there may be a flat spot because it has slipped on the rail. This spot now hits the track with every rotation. Weight sensors on the track (several strain gauges on the rail over a short distance) detect when the weight is not passing through evenly. For a short time, the wheel may even be in the air, only to hit the rail all the harder afterwards. A dynamic coefficient is calculated from these weight values and sent to the authorized users.
The WIN platform is already operational today and is being successively expanded with additional functions. The first camera measuring points have been installed and algorithms are being integrated.
