严寒地区混凝土重力坝变形行为分析与预测模型

Displacement behavior analysis and prediction model of concrete gravity dams in cold region

变形是环境荷载动态变化与结构性能演化耦合作用下大坝服役性态的直观表征,合理的变形行为分析与预测模型是科学诊断大坝健康态势并预测其未来运行行为的重要科学手段。考虑到严寒地区混凝土坝变形行为受环境温度变化影响显著,为有效解译环境温度动态波动导致的热变形特征,构建了基于实测边界温度的严寒地区混凝土重力坝变形行为分析模型。同时,为深入挖掘变形及其解释变量间复杂的因果函数关系,引入具有优良非线性训练能力的孪生支持向量回归(Twin Support Vector Regression,TSVR),并结合鲸鱼算法(Whale Optimization Algorithm,WOA)对TSVR参数优化求解,据此提出了基于优化TSVR的混凝土重力坝变形预测模型。以严寒地区某混凝土重力坝为例,利用所建变形行为分析模型剖析了该坝某表孔溢流坝段坝顶水平位移变幅大且与其它测点水平位移变化规律相反的不协调变形行为的成因,研究结果对深入认识严寒地区混凝土坝变形行为具有重要价值;同时,基于优化TSVR的变形预测模型拥有出色的非线性信息挖掘与建模预测能力,为高精度预测大坝变形提供了一种新方法。

Displacement is the intuitive reflection of comprehensive operating behavior of dams under the coupling effect of dynamic change of environmental loads and structural performance evolution.A reasonable displacement behavior analysis and prediction model is crucial to scientifically diagnose the health status and to predict the future operating behavior of dams.Considering that the displacement behavior of concrete dams in cold region is significantly affected by ambient temperature variation,to effectively interpret the thermal displacement characteristics caused by the dynamic fluctuation of ambient temperature,a measured boundary temperature-based displacement behavior analysis model of concrete gravity dams in cold region is established.Meanwhile,to further explore the complex causal function relationship between displacement and its explanatory variables,twin support vector regression(TSVR)with excellent nonlinear training ability is introduced to construct the displacement prediction model.And TSVR parameters are simultaneously optimized by whale optimization algorithm(WOA).Taking a concrete gravity dam in cold region as an example,the changing amplitude of horizontal displacement of an overflow dam section crest is quite large and the variation law of this displacement is different to that of other monitoring points.The cause of this disharmonious displacement behavior is interpreted with the proposed displacement behavior analysis model.The result is of great significance for further understanding the displacement behavior of concrete dams in cold region.In addition,engineering examples indicate that the proposed optimized TSVR-based displacement prediction model has excellent nonlinear information mining and predictive performance.A new approach is provided for accurately predicting dam displacement.