Hybridizing four wise neural-metaheuristic paradigms in predicting soil shear strength

Abstract

Due to the vital significance of precise determination of soil shear strength (SSS) in many civil engineering projects, this study is dedicated to proposing novel intelligent models for estimating this parameter. To this end, elephant herding optimization (EHO), shuffled frog leaping algorithm (SFLA), salp swarm algorithm (SSA), and wind-driven optimization (WDO) are synthesized with artificial neural network (ANN) to create neural ensembles. The results indicated the efficiency of metaheuristic science for dealing with the non-linear analysis of the SSS and influential soil parameters. Also, a comparison between the models revealed that the SSA-MLP (Error = 0.0386 and Correlation = 0.8219) presents the most efficient prediction, followed by WDO-MLP (Error = 0.0403 and Correlation = 0.8025), SFLA-MLP (Error = 0.0408 and Correlation = 0.7559), and EHO-MLP (Error = 0.0436 and Correlation = 0.7195). Therefore, the proposed SSA-MLP can function as a reliable substitute for traditional approaches in prediction of the SSS. (C) 2020 Elsevier Ltd. All rights reserved.