Pavement analysis is an important field in pavement engineering, because of its causes on pavement behavior and obtaining more accurate equations for it. By science developing, many pavement analysis softwares have been developed which most of them are based on Multi-Layer Theory and a few work based on Finite Elements Method (FEM). Lots of input, spending much time in modeling process and to be unable to running more than one pavement modeling in a time are their main deficiencies. Artificial Neural Networks (ANN) as one of artificial intelligence algorithms has a lot of advantages such as low input numbers, reducing considerable time in the modeling process, multi pavement modeling at a time. In this article, after modeling verification, simulated an artificial neural network based on 384 models which were modeled by finite elements software and found a back propagation ANN with 8-10-2 combination with sigmoid transfer function which was the optimum network for the network.