Dam early warning model based on structural anomaly identification and dynamic effect variables selection

Developing monitoring and warning indicators for dams is essential for facilitating the rapid assessment of dam safety. However, current methodologies for establishing these warning indicators often fail to incorporate the actual structural responses and engineering reliability. This limitation may result in indicators that do not accurately represent abnormal behaviors in dam structures. This article presents a data-driven approach for assessing dam health and introduces a model for developing warning indicators, intending to enhance the intelligence and accuracy of dam safety monitoring. This approach effectively identifies abnormal behavior in dam structures by integrating Variational Mode Decomposition with the Cloud Model. The model can detect and classify non-sudden and sudden changes in monitoring data. Furthermore, a formulation method for warning indicators is proposed, which is based on a dynamic sliding window and the small-probability method. This method aims to improve the accuracy and robustness of dam safety monitoring and warning systems. By applying this methodology to a typical high-inclined core rockfill dam equipped with a comprehensive monitoring system, we validated the accuracy, reliability, and advantages of the proposed approach. The findings demonstrate its suitability for formulating warning indicators related to both trend analysis and periodic monitoring data.