论文部分内容阅读
一、正交试验设计的基本方法 在多因素试验时,如果要进行全面试验,则处理数往往太多。例如4因素3水平的试验有3~4=81个处理,10因素3水平的试验有3~(10)=59,049个处理。故全面试验常是不可能的,应在众多的处理中,选出有代表性的进行试验。正交表就是为这个目的而设计的各种现成表格。在试验前,先要选择适用的正交表来安排试验。这种利用正交表来安排多因素试验的方法,叫做正交试验设计。 每个正交表都有一个代号,例如L;(3~4)(见表6—2)的代号中,字母L是表示正交表,L的石下脚码9是表示这个正交表有9行,可用来安排9个试验;括号内的底数3是表示此表安排的各因素都要求是3个水平的,即表中每列恰有1、2、3三种数字;括号内的指数4,是表示此表有4列,最多可考察4个因素。现结合实例来说明。
First, the basic method of orthogonal test design In the multi-factor test, if you want to conduct a comprehensive test, the number of processing is often too much. For example, 4 to 3 levels have 3 to 4 = 81 trials, 10 to 3 levels have 3 to 10 = 59,049. Therefore, a comprehensive test is often impossible and should be handled in a large number of selected representative tests. Orthogonal tables are a variety of off-the-shelf forms designed for this purpose. Before the test, first select the appropriate orthogonal table to arrange the test. This method of using orthogonal tables to schedule multivariate tests is called orthogonal test design. Each orthogonal table has a code, for example, L; (3 ~ 4) (see Table 6-2) in the code, the letter L is the orthogonal table, L stone foot 9 is the orthogonal table 9 lines can be used to arrange 9 tests; the bottom 3 in brackets means that the table arrangement of the various factors are required to be three levels, that is, each column in the table exactly 1,2,3 kinds of numbers; brackets Index 4, is that this table has 4 columns, up to examine four factors. Now combined with examples to illustrate.