Soft Computing Tools to Predict Progression of Percent Embedment of Aggregates in Chip Seals

Abstract

In this study, soft computing and multilinear regression techniques were employed to develop models for prediction of progression of chip seal percent embedment depth (P_e). The model uses inputs such as cumulative equivalent traffic volume, Vialit test results, dust content of aggregates, and initial embedment depth. Multilinear regression, adaptive neuro-fuzzy system, and artificial neural network techniques were used to estimate the P_e. The contribution of the variables affecting P_e was evaluated through a sensitivity analysis. The results indicate that while most of the proposed models were able to predict the P_e reasonably, the artificial neural network model performed the best.