DESIGN OF NEURAL NETWORKS FOR PAVEMENT RUTTING

Rutting is one of the major distresses of asphalt pavements. Currently, an Asphalt Pavement Analyzer (APA) can be used to evaluate the rut potential of asphalt pavements in the laboratory. Although it is preferable to conduct APA il tests to predict the rutting potential of an asphalt pavement, such tests are not i always feasible for a project due to economic reasons. A rut prediction model I can be a useful tool in such situations. Prediction of rutting using a model is a I I rather challenging task. Traditional statistical models have often exhibited weaknesses in predicting reliable rut values (Tarefder et al. 2002). This study proposes Neural Networks (NNs) to predict rutting of asphalt pavements.