Culvert Inspection Framework Using Hybrid XGBoost and Risk-Based Prioritization: Utah Case Study

Abstract

Resource constraints often prevent Departments of Transportation (DOTs) from performing routine culvert inspections. This study focused on the Utah DOT (UDOT), which lacks a comprehensive culvert management system and faces budgetary constraints for culvert maintenance. These limitations hinder strategic inspection planning and may lead to missed opportunities for preventative maintenance. Without regular inspection, minor defects can escalate into major issues, necessitating costly and extensive repair or replacement that could have been avoided. UDOT’s current approach to culvert inspections is reactive, resulting in an incomplete inventory and a failure to safeguard this critical infrastructure effectively. To address this challenge, an intelligent culvert inspection framework was proposed for Utah by integrating a culvert condition prediction model and a risk-based prioritization approach. An eXtreme Gradient Boosting (XGBoost) model was developed as the foundation of the culvert condition prediction model, and five optimization algorithms were employed, namely the gray wolf optimizer (GWO), whale optimization algorithm, moth–flame optimization algorithm, genetic algorithm, and Bayesian optimization algorithm, to tune its hyperparameters and improve its predictive performance. Based on the results, the GWO-XGBoost model outperformed the others. Subsequently, a risk-based strategy was developed using UDOT’s maintenance data and the GWO-XGBoost model’s output for prioritizing culvert inspections. The case study was conducted on 272 Utah culverts as a validation test and showed the effectiveness of the proposed method for culvert inspection planning by decreasing the cost of culvert failures based on the Monte Carlo simulation results. With this innovative approach, UDOT can allocate resources more efficiently while prioritizing the inspection and maintenance of critical culverts. Furthermore, it optimizes maintenance budgets and resource utilization while improving transportation infrastructure safety and reliability.