Dam deformation early warning model based on cluster analysis and spatiotemporal data fusion

Abstract

The deformation behavior has a vitally spatial correlation to hydraulic concrete structures. The majority of traditional monitoring model focuses on the displacement law of local monitoring points, in which the single objective method is employed, and results in the ignorance of temporal and spatial diversity of structural behaviors. Thus, the state of the whole structure cannot be truly reflected. In this paper, the correlation between the field data obtained from different measuring points was considered, and a dam deformation early warning model based on spatiotemporal data fusion was proposed. The model was established with respect to temporal and spatial dimensions, and the correlation of monitoring points was explored by utilizing cluster analysis. The cross-sectional series fusion model was used to establish the mapping relationship between environmental factors and cross-sectional clusters for extracting the variation regulation of overall horizontal displacements or the structural deflection curve for dams during long-term operation and identifying the abnormal deformation position. Meanwhile, the time series fusion model was established based on the insight of the Stacked Single-Target strategy, in which the correlation between monitoring points was fully utilized to improve the prediction accuracy. Then, it was used to judge whether the monitoring points conform with the operation regulation in temporal dimension. Based on the early warning model in spatiotemporal dimension, a comprehensive diagnosis method of dam deformation was proposed. In addition, this method can be used for further analyzing the critical environmental factors affecting the local and overall deformation of dams. Finally, a project was selected to verify the effectiveness of the proposed model through cluster analysis and spatiotemporal data fusion.