Application of Artificial Neural Networks in Predicting Abrasion Resistance of Solution Polymerized Styrene-Butadiene Rubber Based Composites

TL;DR

This paper demonstrates that Artificial Neural Networks can accurately predict the abrasion resistance of SSBR-based composites, offering a robust alternative to traditional linear models.

Contribution

The study introduces a neural network model with 3 nodes for predicting abrasion resistance, capturing non-linear relationships more effectively than previous linear models.

Findings

ANN model achieved RMS error of 0.07

Neural network outperforms linear regression in prediction accuracy

Model validated with 23 data groups

Abstract

Abrasion resistance of solution polymerized styrene-butadiene rubber (SSBR) based composites is a typical and crucial property in practical applications. Previous studies show that the abrasion resistance can be calculated by the multiple linear regression model. In our study, considering this relationship can also be described into the non-linear conditions, a Multilayer Feed-forward Neural Networks model with 3 nodes (MLFN-3) was successfully established to describe the relationship between the abrasion resistance and other properties, using 23 groups of data, with the RMS error 0.07. Our studies have proved that Artificial Neural Networks (ANN) model can be used to predict the SSBR-based composites, which is an accurate and robust process.