Development of Artificial Neural Network Model for Prediction of Marshall Parameters of Stone Mastic Asphalt

Stone Mastic Asphalt (SMA), firstly introduced in the 1960s, is a durable and rut-resistant asphalt mixture that uses stone-on-stone contact to improve strength, and a rich mortar binder to provide durability. Marshall parameters, such as Marshall Stability (MS) and Marshall Flow (MF) are critical mechanical properties of SMA, representing the performance of asphalt concrete. The two Marshall parameters are widely used for the evaluation of resistance to displacement, distortion, rutting, and shearing stresses of SMA. As the pavement is frequently subjected to traffic loads, it is highly required to find out an optimum manner to determine these Marshall parameters. However, such a procedure is complicated, costly, and time-consuming. The primary aim of the present work is to develop an alternative numerical tool using artificial neural network (ANN) to predict the MS and MF of SMA mixtures. The results show that the ANN algorithm is an excellent predictor based on the excellent values of statistical criteria such as root mean square error, and the Pearson correlation coefficient. This study's results pave the way towards selecting a suitable machine learning approach to accurately determine the Marshall parameters of SMA mixtures. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.