Artificial neural networks as a tool for selecting the parameters of prototypical under sleeper pads produced from recycled rubber granulate

Abstract

This study investigates the possibility of predicting, by means of artificial neural networks (ANNs), the values of static and dynamic bedding moduli of prototypical under sleeper pads (USPs) based on recycled SBR (styrene-butadiene rubber) granulate, in order to adapt their technical parameters (density and thickness) at the production stage to meet the requirements of the railway infrastructure manager or the specifics of a particular investment project. USPs are usually produced from elastomeric materials such as rubber or polyurethane. Here, the authors propose a sustainable approach, where the use of raw materials is reduced and a recycled rubber from end-of-life tires is reused. It is proved that the parameters of a vibration isolator, such as its thickness and density, can be designed using the proposed ANN-based models, which are aimed at the prediction of static and low frequency dynamic bedding moduli of the pad.