混料模型D—最优试验设计的改进DE算法研究  

A Modified Differential Evolution Algorithm for D-optimal Experimental Designs of Mixture Models

作　　者：张新风 朱志彬 李光辉[3] 张崇岐 ZHANG Xin-feng;ZHU Zhi-bin;LI Guang-hui;ZHANG Chong-qi(School of Economics and Statistics,Guangzhou University,Guangzhou 510006,China;Undergraduate School of Medical Business,Guangdong Pharmacentical University,Guangzhou 510006,China;School of Science,Kaili University,Kaili 556011,China)

机构地区：[1]广州大学经济与统计学院,广东广州510006 [2]广东药科大学医药商学院,广东广州510006 [3]凯里学院理学院,贵州凯里556011

出　　处：《数理统计与管理》2021年第1期26-35,共10页Journal of Applied Statistics and Management

基　　金：国家自然科学基金(12071096.11671104,11926305.11901260)。

摘　　要：差分进化算法(Differential Evolution Algorithm)有较强的全局收敛能力和稳健性,应用该算法可解决工程学、计算机科学等领域的一些复杂优化问题。本文介绍一个改进DE算法,该算法适用于混料模型的近似最优设计问题,对于具有附加约束试验域的混料问题也可高效求解。最后,本文给出应用改进DE算法求解混料模型D-最优试验设计的例子。DE(Differential Evolution)algorithm holds strong global convergence capability and robustness.which make it popular in solving complex optimization problem of many fields,such as engineering.computer science,etc.This paper shows a modified DE algorithm,which can be used for finding the optimal experimental design for mixture models on mixture region with or without additional constraints.At last,examples of solving D-optimal design for mixture model will be given to show the effectiveness of the modified DE algorithm.

关 键 词：DE算法 改进DE算法 混料试验设计 D—最优设计 

分 类 号：O212[理学—概率论与数理统计] O212.6[理学—数学]