Abstract
Road sign support systems are not usually well managed because bridges and pavement have budget and maintenance priority while the sign boards and sign supports are considered as miscellaneous items. The authors of this paper suggested the implementation of simplified machine learning algorithms for asset risk management in highway sign support systems. By harnessing historical and real-time data, machine learning models can forecast potential vulnerabilities, enabling early intervention and proactive maintenance protocols. The raw data were collected from the Connecticut Department of Transportation (CTDOT) asset management database that includes asset ages, repair history, installation and repair costs, and other administrative information. While there are many advanced and complicated structural deterioration prediction models, a simple deterioration curve is assumed, and prediction model has been developed using machine learning algorithm to determine the risk assessment and prediction. The integration of simplified machine learning in asset risk management for highway sign support systems not only enables predictive maintenance but also optimizes resource allocation. This approach ensures that decision-makers are not inundated with excessive detailed information, making it particularly practical for industry application.