Neurobrake

Various tests show that the coefficient of friction from a brake pad can change up to ±40% and more depending on the history of brake applications. In the design of the brake system of a train unrealistic assumptions of the friction behaviour could lead to an over- or underestimation of the stopping distance resulting in possible critical events or increased brake pad wear. Due to the high number of parameters influencing the frictional behaviour of a brake pad an analytical calculation of the coefficient of friction does not lead to satisfying results.

The target of Neurobrake is to develop a procedure which allows the estimation of the frictional behaviour for each single brake event based on artifical neural networks and fuzzy logic. The training of the neural network is based on dynamometer tests which were made available by our industrial partner Federal-Mogul. The Neurobrake module is directly linked to route profile simulation which allows the estimation of the change in friction during a complete train run.

Bericht in Semester@HM

Railway Research

Railway Research

Research

Kopieren bestätigen

Sicherung wiederherstellen