PLS在状态检测多元线性回归中的应用  

The Model of Multivariate and Linear Regress in State Detection Based on PLS

作　　者：尚伟[1] 梁玉英[1] 巴宁[2] 

机构地区：[1]军械工程学院,河北石家庄050003 [2]武汉军械士官学校,湖北武汉430075

出　　处：《计算技术与自动化》2011年第1期30-32,共3页Computing Technology and Automation

基　　金：国家自然科学基金项目(60472009)

摘　　要：系统的状态检测过程中,各影响因子之间往往存在着多重相关性,给回归建模分析带来许多负面影响,采用偏最小二乘(PLS)回归建模分析可以很好地解决这个问题。在简述PLS原理的基础上,结合系统状态检测仿真数据,建立系统的状态检测的PLS模型,分别验证PLS回归预测模型用于多个自变量对单个因变量和多个自变量对多个因变量的回归模型的用于系统状态检测的可行性。Multiple correlations often exist among the different impact factors in systematic state detection course.And will bring about a series of negative effects on regression modeling and analysis.However,the partial least—squares regression method can solve this problem in a reasonable manner.Based on the principle of PLS,and also combined with system State detection emulation data,a partial least—squares regression model are established.validate of these two models include more to one and more to more model ware made,and good results was achieved.

关 键 词：偏最小二乘 回归分析 状态检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]