Safety monitoring model of gravity dam zoning based on spatial clustering and Bayesian model averaging
In response to the spatiotemporal similarity of multiple measuring points in gravity dams and the uncertainty of statistical model structure,a hierarchical clustering method was used to spatially cluster all deformation measuring points.Principal component analysis was use to extract the comprehensive displacement of multiple measuring points in each zone.A statistical model for the comprehensive displacement of multiple measuring points was established using the Bayesian model averaging(BMA)method.Using the second zone along the vertical monitoring line of the gravity dam of Ankang Hydropower Station as an example,the optimization results of the complex BMA model and the simplified BMA model for comprehensive displacement were discussed,and the optimal factors of the models were given based on the quantitative analysis of the weighted average model.The analysis results indicate that the fluctuations in reservoir water level have no significant lag effect on dam displacement.The impact of downstream water level on dam displacement is very small,and relevant factors can be ignored in the modeling process.The impact of temperature rise and fall on the displacement of the dam shows a significant lag,and it is necessary to maintain high or low temperatures for more than 15 days before significant changes in the displacement of the dam body occur.
gravity damdeformationBayesian model averagingspatial clusteringprincipal component analysisAnkang Hydropower Station
万方数据
维普
