Improved Quantile Regression Analysis on Small Sample Multicollinear Time Series Measured Data

Abstract

To solve the problems of little sample, multi-collinearity and bad robust ability of normal model remaining in measured dam data in process of analysis, this paper analyzed the monitoring data of measured dam crest crown cantilever and both sides of 1/4 arch of a gravity dam in 2013 using circannual monitoring data. The research shows that quantile regression analysis method based on POD can conquer the problems above when analyzing measured dam data and excavate more safety dam information.