對(duì)稱化偏差及其改進(jìn)
發(fā)布時(shí)間:2018-06-01 02:59
本文選題:試驗(yàn)設(shè)計(jì) + 均勻設(shè)計(jì) ; 參考:《華中師范大學(xué)》2017年碩士論文
【摘要】:均勻設(shè)計(jì)是一種穩(wěn)健設(shè)計(jì),是最常用的空間填充設(shè)計(jì)之一.在沒(méi)有充分的先驗(yàn)信息時(shí),試驗(yàn)區(qū)域中的每一個(gè)試驗(yàn)點(diǎn)被認(rèn)為同等重要,因此均勻設(shè)計(jì)主張將試驗(yàn)點(diǎn)均勻地分布在目標(biāo)區(qū)域中.為了度量設(shè)計(jì)點(diǎn)在試驗(yàn)區(qū)域分布的均勻性,許多偏差被定義和發(fā)展.在文獻(xiàn)中,最常用的偏差有:星偏差、中心化偏差(CD)、可卷偏差(WD)、對(duì)稱化偏差(SD)和混合偏差(MD)等.為了比較不同的偏差,最近覆蓋率的定義被提出,本文根據(jù)覆蓋率的定義和再生核希爾伯特空間中核函數(shù)的定義,推導(dǎo)出各種偏差的覆蓋率公式,并針對(duì)公式研究了現(xiàn)有的各種不同偏差在覆蓋率意義下的優(yōu)良性,發(fā)現(xiàn)對(duì)稱化偏差具有常值覆蓋率,具有突出的優(yōu)良性.利用對(duì)稱化偏差這樣的優(yōu)良性,我們借鑒了混合偏差的思想構(gòu)造了新的混合偏差.最后,我們從不同的角度研究了對(duì)稱化偏差,及新的混合偏差,如:覆蓋率、GMA準(zhǔn)則等.
[Abstract]:Uniform design is a robust design. It is one of the most commonly used space filling designs. In the absence of sufficient prior information, each test point in the test area is considered equally important, so the uniform design advocates that the test points are evenly distributed in the target area. In order to measure the uniformity of the distribution of the design points in the test area, many of the design points are distributed. Deviation is defined and developed. In the literature, the most commonly used deviations are: Star deviation, centralization deviation (CD), coiling deviation (WD), symmetric deviation (SD) and mixed deviation (MD). In order to compare different deviations, the definition of the recent coverage rate is proposed. In this paper, the definition of the coverage rate and the definition of the kernel function in the regenerated kernel Hilbert space are derived. The formula of the coverage rate of various deviations is given, and the excellent properties of the existing different deviations in the sense of coverage are studied. It is found that the symmetry deviation has the constant coverage rate and has the outstanding quality. By using the good of symmetry deviation, we use the idea of mixing deviation to construct a new mixing deviation. Finally, We have studied the symmetry deviation and new mixing bias from different angles, such as coverage, GMA criterion and so on.
【學(xué)位授予單位】:華中師范大學(xué)
【學(xué)位級(jí)別】:碩士
【學(xué)位授予年份】:2017
【分類號(hào)】:O212.6
【參考文獻(xiàn)】
相關(guān)期刊論文 前1條
1 王元,方開泰;A NOTE ON UNIFORM DISTRIBUTION AND EXPERIMENTAL DESIGN[J];A Monthly Journal of Science;1981年06期
相關(guān)碩士學(xué)位論文 前1條
1 于美玲;對(duì)稱化L_2-偏差優(yōu)良性研究[D];華中師范大學(xué);2016年
,本文編號(hào):1962598
本文鏈接:http://sikaile.net/kejilunwen/yysx/1962598.html
最近更新
教材專著