ABSTRACT
Local, state, and federal highway agencies run some form of maintenance and rehabilitation program during the design life of highways. Due to budgetary restrictions, maintenance and rehabilitation actions must be prioritized based on the future condition of the highway section. There are important factors that affect the performance of highways. To properly assess the condition of the pavement and operate maintenance, prediction models with significant condition variables are essential. Mississippi Department of Transportation (MDOT) utilizes probability-based prediction models to determine which sections of the highway and when they need rehabilitation. The current probability models predict the performance without the section-specific parameters. The goal of this study is to develop a new set of performance prediction models for the composite Pavements in Mississippi by using machine learning. The best-performing model can be used as a simple and user-friendly tool to allow the user to visualize the future projections of the pavement section. MDOT personnel can employ this application to predict the condition of the composite pavement section and prioritize the maintenance and rehabilitation schedule.
