Neural network models for analysis and prediction of raveling

The most unacceptable damage on porous asphalt is raveling. Therefore it is important to predict when porous asphalt will achieve a critical level of raveling. In this paper Artificial Neural Network (ANN) was employed to predict raveling having input parameters related to time-series data of raveling, climate, construction and traffic factors obtained from SHRP-NL database. For raveling low, Moderate and High correlation factors were R-2=0.986, R-2=0.926 and R-2=0.976. Another ANN model provided sensitivity analysis indicating relative contribution percentage of input parameters. Finally another model analyzed the relationship between materials and raveling. ANN proved to be powerful technique to predict and analyze raveling.