常用正交表的构造原理及SAS实现  

Structure Principle of Orthogonal Array and Implementation of SAS Software

作　　者：陈远方[1] 林曦晨[1] 徐利华[1] 汤洪秀[1] 汪宏晶[1] 尹平[1] 

机构地区：[1]华中科技大学同济医学院公共卫生学院流行病与卫生统计学系,430030

出　　处：《中国卫生统计》2012年第4期470-474,共5页Chinese Journal of Health Statistics

摘　　要：目的介绍常用正交表的构造原理,并应用SAS宏语句编制通用程序模块生成各种常用的正交表。方法应用哈达玛矩阵法、正交拉丁方法和并列法构造不同类型的常用正交表。结果采用哈达玛矩阵直积法、特征函数法、正交拉丁方法和并列法可以分别构造试验次数N=2s型、N≠2s且N=2(p+1)型(p为素数且p≡1(mod4))的2水平正交表、Lt2(tm)型正交表和混合型正交表。结论不同类型的正交表构造原理方法不同,应用SAS软件可生成各种常用的正交表,方便实用。Objective The structures principle of common or- thogonal arrays were analyzed, and their common computer models were implemented by macro statements of SAS softwale. Methods Construc- tion of Harlamard Matrix, Design of Orthogonal Latin Squares and Parallel principle were used to construct different kinds of orthogonal arrays. Re- sults The methods of Kronecker product Harlamard Matrix, Eigenvalue Function, orthogonal Latin squares and Parallel principle constructed the model of N = 2^s of second level orthogonal arrays ,N≠2^s and N = 2 （ p ＋ 1 ） （p is prime number and p =-1 （mod4）） of second level orthogonal arrays, Lt2 （ t^m ） of orthogonal arrays, and Mixed orthogonal arrays respectively. Conclusion Different kinds of orthogonal arrays have different structure principle. Common program modules of SAS software were used conven- ient, simple and practical to construct different kinds of orthogonal arrays.

关 键 词：正交表 原理 SAS实现 

分 类 号：R195[医药卫生—卫生统计学]