AI-Driven Optimization of Project Cost and Duration in Infrastructure Development Projects

Abstract

The rapid increase in global infrastructure costs is causing delays, and it is difficult for policy-makers and engineers to generate sustainable and consistent outcomes. In this research study, we provide a framework of artificial intelligence to improve cost and schedule in large-scale infrastructure projects, through aggregated holistic data acquisition, gradient boosted prediction models, and particle-swarms multi-objective optimisation. Using historical project data, current sensor IoT data, digital twin simulation, and drone surveys, we develop a quality dataset for validation and training. The models generated are highly predictive in performance compared to traditional scheduling methods, and robust when material prices vary and/ or labour disruptions apply. In addition we conduct scenario testing to confirm that the framework is able to provide realizable recommendations and allow for adaptive scheduling modifications through the use of interactive dashboards. Being able to provide actual costing estimation and adaptive scheduling in real-time provides construction professionals and PMP's the opportunity to better management site performance thereby reducing overruns and simplifying resource allocation, and also very soon capable of responding to site change. The artificial intelligence approach is a promising route to intelligent, data-driven project positioning artificial intelligence as a viable long-term approach towards saving costs and planning timelines strategically, and sustainable construction project scheduling.