Predictive Maintenance Optimization Framework for Pavement Management

The preservation of highway infrastructure is essential for maintaining its capacity, safety, and efficiency for commerce and defense. Pavements are among the most important elements of highway systems that deteriorate over time. Hence, the goal of pavement asset management is to seek efficient investments where the methods applied will aid in identifying the most appropriate allocation of the resources available to the highway agencies. In the absence of unlimited resources, such decisions will always result in trade-offs in which funding certain assets will be needed at the expense of the other. Decision-makers need data-driven information regarding trade-offs to avoid the reactive solutions that are far from optimum and may be counter-productive over the long run. This paper proposes using a multi-objective predictive maintenance optimization framework using a non-dominated sorting multi-objective evolutionary algorithm (MOEA), for the optimum upkeep of pavements. The algorithm aims to find a spread of Pareto-optimal solutions by concurrently minimizing the life cycle cost and maximizing the level of service (LOS). A case study was developed to compare the model's effectiveness based on the maintenance data from the asset management plan of the California department of transportation. The results from the study will help develop promising techniques for the application of various multi-objective optimization systems and thus pave the way for efficient decision-making tools for the maintenance of highway infrastructure projects.