A Novel Methodology for Predicting Roadway Deterioration in Iraq

Abstract

The accurate prediction of roadway conditions is challenging for infrastructure services, especially when considering an increase in traffic volume. This is the first study conducted in Iraq that focuses on predicting roadway condition deterioration and its relation to yearly traffic volume, using surveying data collected between 2019 and 2021. The main purose of the conducted study was to inspect the accuracy, reliability, and ability of a combination of predictive techniques, this combination including Markovian Chains (MCs) and Artificial Neural Networks (ANNs), known as (MC-ANN), accurately to forecast mid-term to long-term (yearly) roadway condition. The principal findings of this research are as follows: a) MCs is a powerful method applied to predict future condition depending on previous one; b) ANNs modelling was performed that be able to produce a more reliable model of roadway condition based on selected road traffic volume change, climate circumstances and road age. The study reached a correlation coefficient of 0.94 between inspected and predicted roadway conditions using a valid collected dataset and a slight mean square error of 0.0195.