Determining International Irregularity Index (IRI) Values Through Artificial Neural Network (ANN) Modelling

The quality of a pavement's level of service is generally determined by measuring the combinations of some important factors which affect the speed, travel time, freedom to maneuver, user comfort and convenience. In this study, a feed-forward back-propagation artificial neural network (ANN) algorithm is proposed based on the acquired International Irregularity Index (IRI) data for the highway structures, bridges and culverts, obtained through laser profilometer measurements on the surface irregularity of the bituminous hot mix roads. Analysis of ANN results were carried out through training various hidden number of neural networks for the output prediction, which is the best estimation of the surface irregularity of the roads. Results produced by artificial neural network have been compared with experimental and numerical results through extensive sets of non-training experimental data. As the comparison of results with ANN study having average absolute mean relative errors as 12.68% for bridges and 12.90% for culverts provided very accurate results, the model proposed could be used to obtain the surface irregularity of the roads by avoiding heavy duty of collecting numerous field data. The results obtained through ANN model were found more accurate than the results produced by numerical models.