Enhancement of Urban Drainage System Resilience by Artificial Intelligence: A Comprehensive Review

Urban infrastructure resilience is crucial for developing urban drainage systems (UDSs) in response to uncertainties from climate change and urbanization. Artificial intelligence (AI) techniques have been employed to enhance the resilience of urban drainage systems. This review examines the state-of-the-art AI applications to urban drainage systems, focusing on aspects including AI-based urban flooding modeling, system diagnosis, and real-time control. It explores potential improvements in light of challenges such as climate change, aging infrastructure, and rapid urbanization. This review suggests that deep learning algorithms (e.g., long short-term memory, deep reinforcement learning, graph neural network) possess considerable potential for enhancing resilience of UDSs, but most studies are still in early stages and not widely implemented in engineering practice. Five key research areas─data enhancement, algorithms and models, interpretability, system safety, and digital twins─are identified as promising for advancing practical applications of AI. Meanwhile, data acquisition and efficient application, the cultivation of professional talents, the support of regulations, and economic sustainability remain essential prerequisites beyond technology for the application of AI in existing UDSs. This review offers insights for innovative AI applications in urban drainage systems, contributing to resilient and sustainable urban water systems.