Machine learning for natural fibre-reinforced compressed earth blocks

Abstract

For centuries, natural fibres have found wide application in traditional natural building materials, such as earth. However, the great variety of fibres in nature and the extreme variability between them make the design of modern blends difficult. The aim of this study is to continue to investigate the possibility of using artificial neural networks to develop predictive models that support the design of compressed earth blocks reinforced with randomly distributed natural fibres. Real and synthetic data from literature mining form the database used to train the network. The overall results show a very high prediction accuracy. For example, a correlation coefficient (R-value) of 0.97 was found for the prediction of the compressive strength value.