同水平析因设计矩阵的分解结果及其特点

目的介绍同水平析因设计矩阵的分解方法、分解结果及其特点,为多因素试验设计提供一种有效方法.方法以n水平(n≥3)的同水平多因素析因设计矩阵为基本的设计矩阵,不增加任何限定条件,仅考察各试验点对其余试验点的影响,采用"主成分分析、聚类分析和规则归纳"等数据挖掘技术相结合的方法,可将此矩阵分解成一系列彼此互不重叠的独立设计矩阵.结果独立设计矩阵具有一些很好的特点,有些本身就是正交设计,有些本身就是均匀设计,还有些是它们所不包含的特殊设计,这些设计一般都是比较理想的多因素试验设计.结论独立设计不仅涵盖了多因素析因设计、分式析因设计、正交设计、均匀设计,还有一些是具有实用性的特殊设计,它是一类具有很高理论研究价值和实际应用前景的综合性极强的多因素试验设计方法.

The Disassembling Result and Character of the Factorial Design Matrices with the Same Level

Objective To introduce the disassembling methods, disassembling results and the characters of the factorial design matrices with the same levels and to provide an effective method for muhifactor experimental design. Methods Taking the muhifactor factorial design matrices with n levels as the basic design matrices, without adding any additional restrictive condition, we just take on the impacts of each experimental point on the rest of the points and adopt the method of combined data mining techniques, such as principal analysis, cluster analysis and rule-based induction. We can disassemble the matrix into a series of independent design matrices, which are mutually non - overlapped, Results Independent design matrices have considerable fine characters. Some of the designs are orthogonal design, uniform design and special designs, which are generally satisfying muhifactor experimental designs. Conclusion Independent design not only embodies muhifactor factorial design, fractional factorial design, orthogonal design and uniform design, but also it possesses some special designs with better practicability. It is a practically all-inclusive muhifactor experimental design, which has high academic research value and practical application prospect.