Development of intelligent models for ravelling using neural network

The most unacceptable damage observed on porous asphalt is raveling. Therefore it is important to predict this detriment accurately and understand it deeply. Artificial neural network (ANN) was employed to predict raveling using time-series raveling and climate, construction and traffic factors. The necessary data was obtained from SHRP-NL database. Model I is able to forecast raveling low, moderate and high with correlation factor of R²=0.986, 0.926 and 0.976. Model II provided sensitivity analysis indicating the relative contribution of factors related to climate, traffic factor, thickness, roughness and age. Color contours illustrated lots of facts such as heavy traffic and low thickness cause raveling on old asphalt at cold rainy days. Model III and its optimized version were developed to analyze relation between material properties and raveling. ANN proved to be a powerful technique to predict and analyze raveling opening great opportunities for development of ANN models for other detriments.