请教HyperStudy中DOE设计中哪个是正交设计、设计变量敏感性分析输出？
请教大家两个问题：1、HyperStudy中DOE设计中哪个是正交设计？
2、设计变量敏感性分析怎么输出，好像只能显示各个设计变量的斜率，
然后斜率陡的说明更敏感，具体数据能否输出呢？ 第二个是正交设计---看纵坐标的值 在计算后的文件夹里卖弄有一个×_runsummary.csv 可以用excel打开 Fractional factorial这个就是正交设计啊？是用Auto select还是选择L什么呢？
L4，L8，L16……这些数字是表示进行的设计计算数吗？好像选择后，提示会error。 正交设计只适合2、3个水平啊？刚看了一下help。
如果是比较多的水平呢？11.0是不是可以啊？ 那个是运算次数--建议看哈实验设计是啊一年水平比较多选拉丁方实验设计--- 5# jinming_02 OK，谢谢。转载一下别人的结论：
Full Factorial适合任何水平；适合因子分析；高精度；设计次数多
Central composite 适合3水平；适合响应面分析；高精度；设计次数多
Fractional factorial 适合2 、3水平；适合与设计变量筛选和因子分析； 正交设计精度高
Box-Behnken适合3水平；适合响应面分析；用很少的设计次数创建2次响应面，比中心复合法设计次数少
Plackett burman 适合2水平设计；适合筛选
Latin Hypercube适合任何水平；高效率；高精度
Hammersley适合任何水平 再转载一下帮助文件
DOE Type
Levels
Properties and Comments
Full Factorial
Any
Requires a high number of points and is therefore unsuitable for most FE studies.
Fractional Factorial
2 or 3
Selected fraction of full factorial; typically a half, quarter, or an eighth of full factorial.
CCD
3
Developed to build a quadratic response surface, typically five levels for each point.Limited to 20 factors and should only be used if the responses are known to be quadratic.
Box-Behnken
3
Three levels for each design variable including central point, economic DoE for building a quadratic response surface.May be used if the responses are known to be quadratic and predictions are not required at the edge of the design space.
Plackett Burman
2
Two levels for each design variable.Used for initial screening, number of points can be as small as N-1.Use when all factors are boolean (eg on/off, 0/1,+/-).Not suitable for an approximation.
Latin Hypercube
Any
Number of points not related to number of design variables.Used when response surface is highly non linear.
Hammersley
Any
Used when response surface is highly non linear, better space filler than Latin Hypercube – generally best for FE work. 这个好象是我传上去的---- 7# jinming_02 呵呵，前几天用记事本记录下来了，没记录下是哪个兄弟发表的。
现在知道了，多谢奉献。
页:
[1]