The Experimental Planning Of Blocked Regular Designs

Experimental design,as an important branch of statistics,has been widely used in industry,agriculture,biomedicine,aerospace etc almost all natural science fields.It is a problem that the experimenter pays more attention to select the efficient and favorable design from the numerous experimental designs.Based on the effect hierarchy principle,statisticians put forward a variety of criteria to measure the optimal normal design from different angles,and obtained the optimal design under the corresponding criteria.In order to achieve the experiment of optimization,not only to select the optimal design among all kinds of designs,but also to arrange the experimental factors into the design columns effectively.Especially when the experimenter has a prior knowledge of the importance order of factors,how to arrange the important factors on the less mixed column is related to the estimation of the factor effect directly.In the blocked normal design,mainly studies the factor arrangement problem under the B~2-GMC criterion and the MA-Type criterion.The specific contents are as follows:1.Under B~2-GMC criterion,first put forward the blocked factor aliased effect number pattern(B~2-F-AENP)and hierarchy principle of factors in blocked designs,it can measure the quality of the columns in the design and ranks the design columns intuitively according to the degree of confusion.Then decomposed blocked design was to analyze the effect of block structure on the treatment factor,and get the formula for calculating B~2-F-AENP of the normal design under a certain parameter range.It can calculate the confounders for each column and give the design columns sorted by B~2-F-AENP.Finally,give the specific strategy of arranging test factors and lists the design table for the optimal column selection when the number of horizontal combinations is 16,32 and 64and the number of blocked factor is 2 or 3 in the B~2-GMC design.2.Under MA-Type criterion,first put forward the blocked individual word length(B-I-WLP),and then give the definition and algorithm steps to measure the quality of the columns in the optimal design.It can write the order of all the columns in the design.Then analyze the relationship between B-I-WLP and clear effect to judge whether the factor effect is clear or not.Finally,according to the important order of the factors,the arrangement strategy of the factor was written for experimenters so that the important factors can be arranged on the important columns,the main effect and the second order interaction effect of the factor can be better estimated.Figure 1;Table 5;Reference 58...