基于偏最小二乘回归法的大坝渗漏分析与预测

渗漏是水库大坝主要病害形式之一，进行大坝渗漏分析预测对了解大坝渗流性态和提高工程管理具有重大意义。由于坝体内部渗漏表现出的灰色甚至黑色特征增加了渗流性态的分析难度，基于常规的最小二乘法无法解决渗漏变量间多重共线、样本较少等问题。针对此类分析的难点，本文通过偏最小二乘回归法，依据监测数据建立了渗流统计回归模型，探讨了主要影响因素与渗漏量的关联程度。采用该模型对吉林台面板堆石坝工程运行条件下渗漏量进行了预测，分析结果表明了该方法的可行性。

Study on analysis and prediction of reservoir leakage by partial least squares

Leakage is one of the main problems about dams.Analysis and prediction about dam leakage help significantly to understand dam seepage and improve project management.But the gray and even black characteristics shown by the dam internal leakage increased the difficulty in analyzing seepage.Conventional least squares method can't be used because of the multiple collinear variables and small sample.As to such problems,this paper has established a statistical regression model and has analyzed the main factors as...