A Methodology Utilizing Artificial Intelligence to Optimize Road Maintenance Scheduling Programs

Globally, the need for proper maintenance of highway assets is gaining more significance due to the reduction in budgets for highway agencies. In order to fulfil this requirement, it is essential for the highway agencies to analyze and prioritize the assets requiring maintenance to prevent economic loss. An artificial intelligence approach based on Bayesian Belief Networks (BBN) is presented in this study to optimize road maintenance scheduling programs. Data from road networks collected from the northern emirates of the United Arab Emirates are evaluated. Performance indices such as cracking, rutting, International Roughness Index (IRI) and deflection of the road networks are considered in the analysis. The correlations among these performance indices obtained through BBN approach is compared with a deterministic approach, Principal Component Analysis (PCA). Understanding the relations among the road distress parameters aid in determining the variables of which decision makers could consider as key deterioration indicators, where prioritization policies are drawn for better road infrastructure management. The results obtained based on the two approaches are presented and it was observed that both approaches produce similar inferences. In addition, the strength of BBN while compared to PCA methods are discussed. Further, the long-term pavement performance (LTPP) datasets of road distresses of pavements with asphalt concrete surfaces are employed to demonstrate the capability of the BBN approach in analyzing big data.