基于主成分监控法的混凝土坝变形预警机制研究  

Study of deformation early-warning mechanism of concrete dam based on monitoring method of PCA

作　　者：王丹净[1,2] 王晓琴[3] 

机构地区：[1]江南大学环境与土木工程学院,无锡214000 [2]无锡商业职业技术学院艺术设计学院,无锡214153 [3]西北民族大学土木工程学院,兰州730124

出　　处：《黑龙江大学自然科学学报》2016年第1期135-140,共6页Journal of Natural Science of Heilongjiang University

基　　金：国家自然科学基金资助项目(51179062);中央高校基本科研基金资助项目(31920150020)

摘　　要：针对传统方法处理混凝土坝观测数据中存在多重共线性问题的局限性,引入混凝土坝观测数据分析的主成分分析方法,对相应的原始观测数据进行降维重构,提取了起主要作用的少数主成分,可以对高维数据进行降维处理,且数据信息丢失很小,有效地减少数据冗余及噪声作用。对于混凝土坝平稳观测数据序列引入置信椭圆,提出了一种多效应量的主成分监控法,构建了多效应量的整体观察值控制域。工程实例表明,该方法是一种直观、高效、可行的混凝土坝预警机制。Aiming at the limitation of the traditional analysis method in the multicollinearity of concrete dam observation data processing,the multivariate response variables of concrete dam are reconstructed by introducing the principal component analysis（PCA） method.A few principal components which have major influence on data variance are extracted so as to reduce the dimensionality of high variable space with the minimum loss of information and a low redundancy and realize the reduction of data and noise.For the steady observation series of the concrete dam,a control field for the whole observation values is established based upon the confidence ellipse.The results indicate that the proposed method is highly effective,intuitive and practicable which can be used as the early warning mechanism of the concrete dam.

关 键 词：主成分分析 混凝土坝 置信椭圆 数据缩减 安全监控 

分 类 号：TV698.1[水利工程—水利水电工程]