Development of Roughness Index Model of Low Volume Roads Using Machine Learning Techniques

Abstract

In the present scenario, one of the most common challenges in urban areas is road damage. The effect of these damaged roads will cause delay in the travel time, traffic congestion, and road accidents. The main cause for these distresses are due to the improper planning of road, overloading of the vehicles, water on the road surface and the age or serviceability of the road. Improper maintenance of the road will also contribute to the primary function of damage caused to the road. The factors affecting the maintenance are delays in financing, improper handling and accelerating climate change. Hence, it is very important for incorporating a proper maintenance to the existing roads beside the development of new highways for the growth in country’s economy. Pavement condition and roughness data are collected on the selected study stretch and were used to develop the relationships of distresses and UI values using machine learning techniques, artificial neural network and support vector machines. It has been observed that, the roughness is usually manifested as a combined effect of different individual pavement deterioration parameters such as cracking, potholes, raveling, patching, and rutting.

Keywords: Pavement distress, roughness, low volume roads, pavement condition, support vector regression, artificial neural network, Matlab, performance measures