基于变系数非参数回归的稳健性参数设计  

Robust Parameter Design Based on Varying Coefficient Non-Parametric Regression Model

作　　者：姜英[1] 袁代林[1] 李裕奇[1] 刘妙妙[1] 

机构地区：[1]西南交通大学数学学院,四川成都610031

出　　处：《吉首大学学报（自然科学版）》2014年第5期22-25,共4页Journal of Jishou University(Natural Sciences Edition)

基　　金：中央高校基本科研业务费专项基金项目(2682013CX037)

摘　　要：利用变系数的非参数回归理论讨论多因子试验过程中因子间相互性对质量特性的影响,通过调节系数变量改变其他因子对过程的影响,找到最优的因子水平组合,进行稳健试验.提出了在稳健性参数设计中应用变系数的非参数回归模型理论,建立质量特性的均值和方差模型,得到最优的因子水平组合进行试验分析.最后,通过油墨打印机的实例说明此理论的可行性.Varying coefficient Regression theory is applied to the discussion on quality characteristic influenced by factors＇ mutuality during the process of multifactor test. In order to find the optimal combination of factor levels,the effect of other factors is changed by adjusting the coefficient variable. The paper studies varying coefficient regression theory in robust design, builds quality characteristic mean model and variance model and gets an optimal combination of factor levels. The feasibility of the theory is illustrated by the ink printing.

关 键 词：变系数模型 非参数回归 稳健性设计 交叉验证 

分 类 号：O212.6[理学—概率论与数理统计]