Author(s): Wangjiayi Liu; Guanghua Guan
Linked Author(s):
Keywords: Canal; Digital Twin; Prediction intervals; Real-time; State assessment
Abstract: Reliable operation of canal systems requires timely knowledge of their hydraulic state, yet operators rely on sparse observations and deterministic models that poorly reflect uncertainty and evolving risk. This study addresses the problem of quantifying real-time safety within a digital twin framework. A hybrid physics–data model generates probabilistic water-level prediction intervals, from which a unified state index is built to characterise in-interval proximity to safety bounds and out-of-interval exceedance magnitude and duration. Application to a regulated canal demonstrates that the index captures abnormal events, reveals their evolution, and supports risk-aware decision-making.
Year: 2026