关键词: 重力坝, 安全监测, 经验模态分解, 响应面代理模型, 遗传算法, 反演分析

Abstract: It is an important means for dam safety monitoring to establish a numerical model which can accurately simulate the behavior of dam structure through back analysis on material parameters. Back analysis on concrete dam parameters based on water pressure component in monitoring data is a common method at present. Empirical Mode Decomposition (EMD) is an intrinsic mode function that can eliminate the time effect component in monitoring data and separate water pressure component. It facilitates accurate water pressure component separation. At the same time, the response surface surrogate model and genetic algorithm are integrated into the analysis to improve the parameter back analysis efficiency. Taking a YL Gravity Dam as an example, the elastic moduli of the dam body and dam foundation were back analyzed based on the measured dam crest displacement along river and the strain of dam body. The results show that the water pressure component of displacement simulated based on inversion parameters is more consistent with the measured value than that simulated based on design parameters, thus the working behavior of the gravity dam can be analyzed and interpreted more accurately.

Key words: gravity dam, safety monitoring, empirical mode decomposition(EMD), response surface surrogate model, genetic algorithm, back analysis