TAM DERİNLİKLİ ESNEK ÜSTYAPILARIN KATMAN ÖZELLİKLERİNİN TAHMİNİ İÇİN LİG ŞAMPİYONASI ALGORİTMASI

Abstract

This study proposes a backcalculation tool, based on the hybrid use of League Championship Algorithm (LCA) and Artificial Neural Network (ANN), in order to predict the stiffness related layer properties of full-depth asphalt pavements. The proposed algorithm, namely LCA-ANN, is composed of two main parts; (i) an ANN forward response model, which is developed with the nonlinear finite element solution, for computing the surface deflections, and (ii) LCA search algorithm which is employed to search and provide the best set of layer moduli to the ANN model. In order to evaluate the performance of the proposed method, a synthetically generated dataset and real field data are utilized. Moreover, to assess the searching ability of LCA, well-accepted metaheuristic algorithms; Simple Genetic Algorithm (SGA) and Particle Swarm Optimization (PSO) are employed for comparison purposes. Obtained results reveal that the proposed algorithm can predict the layer properties with a low order of error values and enables fast and reliable tool for backcalculation studies.