CAJ | 학술논문

实验设计,尤其是混合水平实验设计,已经在科研领域和制造业中得到了广泛使用.虽然很多学者已经提出了许多实验设计理论以及实验设计用表,但是对于含有混合水平的大样本体系的实验设计方法仍然是相当困难的.本文根据Yong Guo提出的实验设计方法,并依据正交设计理论和遗传算法,把此方法改进为适用于大样本体系的实验设计方法.为了配合遗传算法,而改进了交叉与变异操作:为了从复杂体系的全实验设计中高效地获取某一次实验设计,引进了数学方法中的求余运算.最后,当把此实验设计方法应用于一个真实的、多因子多水平的体系时,获得了比较满意的实验方案.因此,对于大样本体系,此实验设计方法所得到的实验方案是可以被接受的.
실험설계,우기시혼합수평실험설계,이경재과연영역화제조업중득도료엄범사용.수연흔다학자이경제출료허다실험설계이론이급실험설계용표,단시대우함유혼합수평적대양본체계적실험설계방법잉연시상당곤난적.본문근거Yong Guo제출적실험설계방법,병의거정교설계이론화유전산법,파차방법개진위괄용우대양본체계적실험설계방법.위료배합유전산법,이개진료교차여변이조작:위료종복잡체계적전실험설계중고효지획취모일차실험설계,인진료수학방법중적구여운산.최후,당파차실험설계방법응용우일개진실적、다인자다수평적체계시,획득료비교만의적실험방안.인차,대우대양본체계,차실험설계방법소득도적실험방안시가이피접수적.
Experimental designs,especially mixed-level experimental designs,are widely used in science researches and manufactures.Although many experimental design theories and tables were proposed,it was still difficult for researchers to design experimental strategy for large sample systems with mixed-levels In this paper,an intuitionistic and easy programmable experimental design strategy proposed by Yong Guo.based on orthogonal design(OD)theory and genetic algorithm,Was improved for large sample systems.Crossover and mutation operators ofgenetic algorithm were improved to solve this kind of problems.In order to obtain each experimental run from full experimental space and also have high efficiency,experimental runs were derived through the mathematical operation aiming at arithmetical compliment remainder so as to meet experimental designs of complex system.At last,the algorithm was applied to a real multi-factor and mixed-level system with constraints.The result showed that the proposed method indeed gave out a reasonable experimental scheme.Therefore,for large sample systems,this experimental design strategy can provide an acceptable scheme.