关键词: 溃口水动力, 实时监测, 视频智能流场识别, LSPIV, PTV

Abstract: Real-time and accurate monitoring of the hydrodynamic characteristics of river channel flow downstream of a breached dam is fundamental to the safety protecting performance of a hydraulic project and the decision-making of flood control and disaster mitigation. Traditional contact-based monitoring technologies suffer from response delays and low accuracy, failing to meet emergency monitoring demands. By integrating image recognition technology, LSPIV, and PTV techniques, this study develops an efficient HAIT algorithm engine that significantly enhances accuracy and efficiency in the velocity measurement of the flow passing through the breach and its distribution. Aiming at the challenges in the monitoring in complex scenarios, we adopt the L-K pyramid optical flow algorithm and the image pyramid construction method to resolve optical flow calculation issues in the case of large-scale motion discontinuities. And, we develop a comprehensive technology framework－comprising the composition of the system and its workflow－for river video-based intelligent flow velocity monitoring, and validate its reliability and effectiveness through testing its practical application, as a new tool for riverine hydrodynamic characteristics downstream of a breached dam.

Key words: breach hydraulic, real-time monitoring, video intelligent flow field identification, LSPIV, PTV