An innovative algorithm for the evaluation of wheel - rail contact points based on neural networks

The multibody simulation of a railway vehicle requires a reliable efficient method to solve the wheel/rail contact problem. The paper describes a series of studies concerning the first part of the problem: the research of the contact points. The aim of the work is then the development of a method for the evaluation of the positions of all contact points between wheel and rail, and its implementation in a multibody simulator of the railway vehicle dynamics. An application of neural networks to the wheel-rail contact problem is proposed in order to reduce the time of evaluation of contact points. The paper will summarize the formulation of the analytic methods and will describe the details of the implementation based on the neural network. The results obtained with the different solutions will be shown and compared.