Investigating the Two Optimization Algorithms (GWO and ACO) Coupling with Radial Basis Neural Network to Estimate the Pile Settlement

Document Type : Original Article


1 School of Chemical Engineering, Yeungnam University, Gyeongsan 712-749, Rep. of Korea

2 Department of Mechanical Engineering, Dezful Branch, Islamic Azad University, Dezful, Iran


Immunizing projects such as piled bridges entails considerations that ensure safety over the operation period. Pile Settlement (PS) which seems one of the most critical matters in constructional project failure, has attracted experts’ attention to be predicted before starting projects with piles. The variables to appraise the pile movement would help us determine the perspectives after and during loading. Theoretical ways to calculate the pile movement mathematically have been adopted to model the PS, mostly by using artificial intelligence (AI). This paper has aimed to estimate the pile settlement rates based on pile samples. For this reason, a new hybrid model containing a Radial Basis Function Neural Network (RBFNN) joining with Grey Wolf Optimization (GWO) and Ant Colony Optimization (ACO) were used in a framework. In fact, optimizers utilized for calculating the neuron number of hidden layer in RBFNN at optimal level. In Malaysia, the Kuala Lumpur transportation network was investigated to examine the pile movement based on ground conditions and properties through the developed hybrid RBF-GWO and RBF-ACO algorithm. Evaluating each framework’s performance was done via the indices. So, the RMSEs of RBF-GWO and RBF-ACO reached values 0.5176 and 0.6562, respectively, and the MAE showed the rates 0.2583 and 0.3386, respectively. The correlation R-value also showed the RBF-GWO suitable accuracy with 1.23 percent higher than another model. Therefore, results have implied the RBF-GWO desirable performance to estimate PS.