SNCF Uses Neural Networks to Model the Dynamic Response of a Railway Vehicle

Track geometry monitoring is necessary to ensure the safety and comfort of trains. To take into account the dynamic responses of trains, SNCF, France's national state-owned railway, wants to model the behavior of the vehicles according to the parameters of the track. Developing multibody physical models is possible, but is complex and costly. SNCF therefore decided to explore developing "black box" or "gray box" models using neural networks and MATLAB^®. This approach allows easy integration into the measurement chain.