摘要
利用偏最小二乘回归建模,由于自变量中含有大量与因变量无关的信息,这些信息被提取成具有大方差和小相关系数的成分,使入选的成分虽具有大的协方差,但是仍对因变量缺乏解释能力。为了克服这一缺点,应用改进的偏最小二乘法进行回归建模,介绍了该方法的改进原理及步骤,并将该方法应用于谐波源识别分析中,仿真示例说明了该方法的有效性及优越性。
In the partial least squares regression modeling, due to a large number ot irrelevant reformation which is exrractea into large variance and small correlation coefficient of the ingredients in the variables and the dependent variable. The irrelevant informa- tion makes the selected components with a large covariance, but lacks the explaining ability of the variable. An improved partial least -squares regression is used for modeling in order to overcome this shortcoming. The fundamental principle of the improved method as well as its modeling steps is introduced in the paper. Furthermore, the improved method is applied to the quantitative analysis of har- monic. Simulation example illustrates the superiority of this method.
出处
《中国农村水利水电》
北大核心
2013年第5期164-167,共4页
China Rural Water and Hydropower
基金
福建省教育厅2011年A类科技项目(JA11333)
关键词
偏最小二乘回归
方差
相关系数
解释能力
改进的偏最小二乘
the partial least squares regression
variance
correlation
explaining ability
improved partial least-squares regression
