Large-scale hydro-steel structures (LS-HSS) are pivotal in hydraulic engineering, boasting high lift, expansive apertures, and discharge capacities. Conventional methods of operation and maintenance for LS-HSS require a digital revolution to ensure safety, reliability, and durability. This study introduces a novel approach, harnessing the power of a digital twin (DT) to integrate physical and virtual elements in LS-HSS operation and maintenance seamlessly. A digital twin modeling framework of LS-HSS operation and maintenance is first presented, incorporating five levels of a physical entity, a virtual entity, DT data, services, and connections. Subsequently, a DT-driven intelligent operation and maintenance (DT-IOM) platform is constructed and applied, which features real-time mapping, closed-loop control, dynamic health assessment, and intelligent decision-making. This platform provides a pathway for digitally transforming conventional LS-HSS operation and maintenance and paves the way for a new era of intelligent infrastructure management. The proposed approach has been successfully implemented and validated with multiple engineering projects, including radial gates of reservoir spillways, plain gates of flood discharge orifices, and emergency gates of pumped storage power stations. The impressive results demonstrate a significant anomaly response and processing efficiency improvement of over 60%. Moreover, the platform has substantially reduced field inspection time and equipment failure rate by approximately 50% and 20%, respectively. These tangible benefits further highlight the platform’s functional advantages, including its capability for IoT connection, modeling and integration, virtual-real interaction, 2D/3D visualization, and service expansion. The results demonstrate the unique features in the closed loop of perception, analysis, decision-making, and optimization for DT-IOM, especially for integrating data, models, knowledge, and services in LS-HSS intelligent operation and maintenance.