均匀试验设计子程序  

A SUBROUTINE FOR UNIFORMLY EXPERIMENTAL DESIGN

作　　者：刘范[1] 

机构地区：[1]华中师范大学化学系

出　　处：《华中师范大学学报（自然科学版）》1992年第1期63-68,共6页Journal of Central China Normal University：Natural Sciences

摘　　要：在多因素、多水平的试验中,常采用正交设计法。当水平数q较多时,需要安排的试验次数常常很大,甚至于是不允许的,因而1980年中科院应用数学所方开泰等提出了均匀设计法,它的理论基础是数论方法,特点是在试验范围内安排的试验点充分地均匀分散,因而具有代表性,只要进行很少次数(等于水平数)的试验就可找出基本的规律,曾在我国飞航式导弹的设计中得到有效的应用,是一种有效的设计安排试验的方法。其缺点是对试验结果的分析处理比较麻烦,但在计算机已被广泛应用的今天,这个缺点就不算是个什么问题了。The uniformly experimental design proposed by Fang etal in 1980 [Acta Mathematicae Applicatae Sinica, 3 (1980) 363; Kexue Tongbao, 26(1981) 485] is an effective and successful method for the arrangement of experiment containing a considerable number of levels q (integers from 4 to 31) and number of factors s (s≤q). Its theoretical foundation is the theory of numbers,especially the relation of congruence. The resultant schemes of the experimental arrangement are: (A) P.(k)=(ka_1,ka_2,ka_3,…,ka,) (mod q), k=1,2,3,…,q where α_i and q are integers, 1≤α_i≤q (1<i<s),α_i≠α_i and (α_i, q)=1. (B) If the number of levels q equals a prime number p,the suggested scheme is P_(?)(k)=(k,ka,ka^2,…, ka^((?)-1))(mod P), k=1, 2, 3,…,p where 1<α<p and α~i(?) a^(?) (rood p). The subroutine offered here can be employed for all the situations discussed in the cited references.

关 键 词：均匀设计 试验设计 BASIC程序 

分 类 号：O212.6[理学—概率论与数理统计]