Predicting Road Condition using Linear Hierarchical Modelling

Abstract

Roads are vital elements of societal connectivity, facilitating the seamless flow of goods and people. Increased use of roads because of population growth and urbanization, combined with increased intensity and frequency of extreme events due to climate change have been accelerating road pavement deterioration. This leads to a cascade of adverse consequences, increasing safety risks, traffic congestion, and maintenance costs. Predictive maintenance can reduce failure risks, optimize resource allocation, and extend road lifespan, but it requires accurate deterioration models. Presented herein is an application of linear hierarchical modeling for predicting road condition. Hierarchical modelling is a state-of-the-art technique in industrial health management for data representing a population. The proposed method quantifies the uncertainty of prediction and considers the heterogeneity across different subpopulations of pavements caused by different design, traffic and environmental conditions. The results suggest that this technique can be effectively used for avoiding high variance caused by sparse health inspection data for some of the road sections. This method has the potential to significantly contribute to enhance asset predictive maintenance and consequently to operation and safety of roads and other infrastructure assets.