Optimized ANN Algorithm for Analyzing the Road Rigid Pavements

The prediction of pavement responses on the basis of Advanced Finite Element Programs provides endless opportunities for the analysis of complex problems in the pavement engineering. Reducing considerable time in the analysis of such problems with the model of artificial neural network analysis is possible. Artificial neural networks are functionally very practical models whose calculational speed is entirely independent of the complexity of mathematical algorithms or the method used for providing training samples. In this paper, the analytical results of 624 jointed plain concrete specimens is used in order to choose an appropriate neural network algorithm as a reliable tool for road rigid-pavements response. The suggested analytical model is based on results of artificial neural network in the form of a 6-18-2 back propagation ANN network with sigmoid transfer function which provides ability to find critical stress and deflection in PCC in short time for generating model and analysis process of several pavement sections simultaneously.