International Roughness Index (IRI) prediction using various machine learning techniques on Flexible Pavements

Pavement performance is an essential factor for companies to shape their strategies for maintenance, repair, and sometimes reconstruction of damaged roads. Pavement roughness is a crucial indicator for performance measurement. The international roughness index (IRI) represents pavement roughness. Over the last decade, researchers have been working on linking pavement roughness to other performance indicators. In India, no IRI (International Roughness Index) research has been conducted that considers the characteristics and parameters of Indian road conditions. This paper explores the possibility of using different models to determine the characteristic of traffic and the structural parameters of the road. Using the data set collected from the Central Road Research Institute of India (CRRII) database, different machine learning models were applied to effectively predict the IRI to determine the pavement condition. IRI using ANN (Artificial Neural Network) and XGB-Regressor gave the lowest mean absolute percentage error in terms of data analysis compared to the other models used.