Novel approaches to predict the Marshall parameters of basalt fiber asphalt concrete

Experimental approach for evaluating the Marshall Stability (MS) and Marshall Flow (MF) of basalt fiber asphalt concrete (BFAC) is time-consuming and expensive. This study employed the Extreme Gradient Boost (XGB) algorithm in conjunction with two novel optimization algorithms, namely Sailfish Optimizer (SFO) and Aquila Optimizer (AO), to construct novel and enhanced prediction models for MS and MF of BFAC. Two databases were compiled from 18 experimental investigations with 128 and 89 experimental samples, respectively for MS and MF. Ten input parameters covering the mixture components were considered for MS and MF modeling. In addition, cross-validation was applied to evaluate the generalizability of the trained XGB model. The obtained findings demonstrated that the XGB model has an excellent and consistent predictive capacity, thereby demonstrating the model's stability in predicting MS and MF. In addition, the SFO optimization algorithm was applied for a constrained design optimization problem regarding the composition mixture for BFAC. The results were discussed considering existing design standards.