Estimation of Pavement Roughness Based on Surface Distresses Using Artificial Neural Network (case study: Iran’s arterial roads)

Today, roads are considered among the main assets of each country so there is a need for a specific mechanism for their preservation and maintenance. Hence, the pavement management system, as an effective tool for decision-making and identifying effective and economical strategies, is used in pavement evaluation and treatment and also in maintaining roads in acceptable levels. In order to implement this system, having access to accurate measurement of different pavement indices, is vital. The goal of this study is identifying the effect of pavement distresses on the roughness and establishing a correlation between the two parameters to be used for evaluating the International Roughness Index (IRI) and roughness growth rate. In order to do this, using the Laser Crack Measurement System (LCMS), the roughness index and pavement distresses are measured in 10-meter length sections with lateral resolution of 1mm in several arterial roads of Iran. After the preliminary analyze of the LCMS output, pavement distresses with higher impact on roughness were identified and then, using artificial neural network (ANN), a correlation was established between IRI and pavement distresses. The relationship showed a correlation coefficient of 0.70. Putting this model into use, is a low-cost approach for road agencies to evaluate the roughness index as well as the roughness growth rate based on pavement distresses in network level. This in part would lead to better policy making and more efficient maintenance and treatment activities.