AI-Driven Digital Twins for Smart Urban Infrastructure: A Comparative Survey of Monitoring and Predictive Maintenance Methodologies

Abstract

The increasing complexity, ageing, and operational demands of urban infrastructure have created a need for intelligent approaches to continuous monitoring and predictive maintenance. Digital Twins (DTs) have emerged as a promising technology for creating dynamic virtual representations of physical infrastructure, while Artificial Intelligence (AI) enables automated analysis, anomaly detection, structural condition assessment, deterioration prediction, and maintenance planning. This paper presents a comparative survey of AI-driven Digital Twin methodologies for smart urban infrastructure, with particular emphasis on monitoring and predictive maintenance. Existing approaches are examined across key technological dimensions, including Digital Twin architecture, IoT-based sensing, machine learning and deep learning, Building Information Modelling (BIM), Geographic Information Systems (GIS), numerical simulation, Remaining Useful Life (RUL) prediction, and Explainable AI (XAI). The survey compares the capabilities, strengths, limitations, and application domains of existing methodologies to identify the extent to which these technologies are integrated for infrastructure management. The comparison indicates that existing studies have demonstrated significant advances in individual areas such as structural health monitoring, damage detection, predictive maintenance, and Digital Twin development; however, many approaches remain domain-specific or focus on limited combinations of technologies. In particular, integration of BIM, GIS, IoT, AI, simulation, explainability, and maintenance decision support within a scalable framework remains insufficiently addressed. The findings highlight the need for more integrated and generalizable Digital Twin methodologies capable of progressing from real-time infrastructure monitoring to predictive and decision-oriented maintenance. The survey provides a structured basis for identifying current research gaps and guiding the development of future AI-driven Digital Twin solutions for smart urban infrastructure.