CAJ | 학술논문

稳健设计旨在选择可控因子的水平组合来降低产品或过程对噪声因子的敏感性,从而能够有效地提高和改进产品质量.针对稳健设计中如何确定权重和建立优化模型的问题,本文提出了基于熵权理论和双响应曲面的稳健设计方法;并通过结合两个实例对所提方法进行了实证研究.研究结果表明,结合熵权理论与双响应曲面法能够有效地实现产品的稳健设计.
은건설계지재선택가공인자적수평조합래강저산품혹과정대조성인자적민감성,종이능구유효지제고화개진산품질량.침대은건설계중여하학정권중화건립우화모형적문제,본문제출료기우적권이론화쌍향응곡면적은건설계방법;병통과결합량개실례대소제방법진행료실증연구.연구결과표명,결합적권이론여쌍향응곡면법능구유효지실현산품적은건설계.
Robust design is based on the concept of building quality into product or process design.It aims to minimize performance variation and deviation from the target value of interest via determining the optimum operating conditions,which play an important role in optimal design.As increasing number of studies are emerging in this field,companies need to optimize product design in order to successfully grab market share.This study combines entropy weight theory with dual response surface approach in order to improve the product or process design.We construct an optimization strategy to minimize weighted mean squared error （MSE）,which is a simple and effective method in dual response surface optimization.If the mean and the variance of MSE are not equally important,they need to be weighted properly.The determination of the weight coefficient is significant in optimal design since it has a far-reaching impact on optimization results.However,in reality decision-makers usually obtain the weight of responses based on their individual preference and experience,which cannot reflect the degree of importance of responses owing to the strong subjectivity.To date,a few methods have been proposed to obtain the weight of responses.Nevertheless,the calculations of weight in these approaches are either too complicated or inefficient.Therefore,the concept of entropy is introduced to the optimization process to avoid information loss.A method of calculating the true weight coefficient is obtained by considering the information of responses.In addition,entropy weight is very convenient for engineering operators who don＇t have much mathematical and statistical knowledge.Meanwhile,dual response surface optimization is an important tool in modern quality engineering,which takes into account the mean and the variance simultaneously.In the modeling process,the dual response surface optimization firstly utilizes the response surface to build regression functions for both the mean and the variance of the response,and construct optimization strategies to obtain the best setting of factors in the feasible region.Therefore,we combine entropy weight theory and mean squared error approach to optimize the product or process design,which is called the weighted mean squared error approach.In summary,the design optimization method of product or process design based on entropy weight theory and dual response surface methodology （RSM） considers not only the relative importance of different responses but also the optimization of the mean and the variance.In this paper,we firstly adopt entropy weight theory which takes response into consideration in order to obtain the weight of response,and then fit the quadratic models of the mean and variance responses.Next,we construct the optimization strategy,that is,weighted MSE,followed by optimizing the objective function and getting the combination of control factors.Finally,two examples are given to illustrate the efficiency of the proposed approach,compared with the optimization solutions obtained by other methods.