Keywords: track monitoring, suspension diagnoses, inverse problems.

The present work proposes methods to monitor the railway system using dynamic reactions measured during commercial journeys. More precisely we propose to combine railway dynamic simulations and data models to detect damage of vehicle suspensions on one side and track geometry on the other side. On a scientific point of view, the monitoring is based on the solving of a complex inverse problem. Indeed, the problem is non-linear (wheel/rail contact, suspensions, aerodynamics forces...) and contains many sources of uncertainties (wheel/rail friction coefficient, wind loads, …). The problem cannot therefore be posed in a deterministic way and a probabilistic framework must be adopted. The statistical dimension of the problem is moreover very large because it involves functional inputs and outputs (they depend on time and/or space). Dedicated methods have thus to be adopted. The developments are first applied to the monitoring of high-speed train suspension elements over several years. The results obtained are encouraging and make it possible to monitor the various damage and maintenance. The second application concerns the monitoring of lateral offset defects. These are more difficult to reconstruct than the levelling defects due to the play in the track. A method for detecting large defects is proposed here.

download the full-text of this paper (PDF, 603 Kb)

go to the previous paper
go to the next paper
return to the table of contents
return to the volume description