Hoppa till innehållet

The digital railway switch- Digitalized railway switches for the future (DigiSwitch)

Publicerad: 18 maj 2020

The goal is to investigate if and how it is in principle possible to, by measuring vibrations at the point machine and using AI, achieve an automatic correlation between the vibration levels and degree of wear/ flaws.

Researchers: Jan Lundberg (PL), Taoufik Najeh

Sponsor: Formas, Bombardier, Trafikverket, Infranord

Duration: 2019-2021

Goal: To investigate if and how it is in principle possible to, by measuring vibrations at the point machine and using AI, achieve an automatic correlation between the vibration levels and degree of wear/ flaws anywhere on the turnout, so that this information can help TRV to perform automatic preventive maintenance actions to increase punctuality and traffic capacity. Also the aim is to build a prototype, showing that this is possible.

Project status and results: An experimental set up consisting of a full scale railway switch with included point machine and a  movable rail wagon boogie has been  build up by us and studied for a number of different relevant wear and flaws on the switch blades, support rail, middle rail, and crossing. Also tamping error, wheel plates and trailing switch has been studied. Although the speed of the wagon was less than 0, 8 m/s it was possible to detect and separate level of the different type of wear and flaws at most of the positions. These promising results is indicating that the suggested measurement solution should be developed further in additional projects, discussions are right now with Alstom, Vossloh, Infranord and TRV. The resulting vibration data from the measurements was also successfully used as input to neural network in order to create automatic correlation between measured vibration features and the degree of actual wear and flaws. The applied system developed in the present study can also easily be assembled into a cheap and robust device that ca be delivered by the manufacturing company of point machinery, this type of system is also applied as patent proposal. Fig 1 shows the experimental set up.

D1
D2
D3