主成分回归分析在大坝安全监测中的应用  

Application of Principal Component Regression to Dam Safety Monitoring

作　　者：黄伟朵[1] 许昌[1] 段永刚 邓成发[2] 

机构地区：[1]浙江水利水电专科学校水利工程系,浙江省杭州市310018 [2]浙江省水利河口研究院,浙江省杭州市310020

出　　处：《水电自动化与大坝监测》2009年第1期57-59,81,共4页HYDROPOWER AUTOMATION AND DAM MONITORING

摘　　要：介绍主成分回归的原理,对某大坝的位移预测进行分析研究。利用SPSS软件进行多重共线性诊断后进行主成分分析,确定主成分的个数,将原自变量的主成分代替原自变量进行回归分析,建立主成分回归模型,并对该大坝位移进行预报。对影响大坝位移的各因子进行有效分离,既保留了原指标的绝大部分信息,又有主成分之间不相关的特点,弥补了最小二乘法回归无法有效识别和消除因子间多重相关性影响的不足。Principal component regression is introduced,and displacement prediction of a dam is analyzed.Multi-collinearlity is diagnosed and the principal component is analyzed by SPSS software.Then the number of principal components is determined,and regression is analyzed by principal components instead of the original independent variables.A principal component regression model of the dam is established and its displacement is predicted.The factors affecting the dam displacement are separated effectively.Most of the original information is retained,and the characteristics in the main components is avoided,and thus,the problem that multi-collinearlity in factors can not be effectively identified and eliminated by regression with the least square method is solved.

关 键 词：主成分回归 多重共线性 大坝 安全监测 预报 

分 类 号：TV698.1[水利工程—水利水电工程]