Exploiting a Real-Time Self-Correcting Digital Twin Model for the Middle Route of the South-to-North Water Diversion Project of China

Real-time monitoring and forecasting are essential to ensure an on-time and on-demand supply of water diversion projects. However, water transfer systems currently lack spatiotemporal data in a dense resolution, failing to monitor real-time conditions and test plausible scenarios. To address the problem, this paper proposes a novel digital twin framework. It includes a real-time self-correcting model, which combines (1) a hydraulic solver using the one-dimensional Saint-Venant equations; and (2) a method updating hydraulic states driven by field observed data. This framework consists of four phases: preparation, warming up, tuning, and monitoring and predicting. Particularly in monitoring and predicting, an identification method for diagnosing abnormal events is also proposed as one of the functions of the twin model. The model shows beyond 98% similarity to reality based on the metric similarity (S) proposed in this paper on both of two real-world scenarios: a large flow scenario and a normal one. The deviation is generally lower than 5 cm for water level 2 m(3)/s for discharge. The abnormal situation diagnosis method also provides timely fault detection for daily scheduling. It is anticipated that this framework can be a powerful tool to estimate current canal states and predict change trends, further ensuring the security and efficiency of operations for large-scale water diversion projects.