【作者】 王颖；

【导师】 何桢；

【作者基本信息】 天津大学 ， 工业工程， 2007， 博士

【摘要】 产品/过程设计中的多元质量特性优化方法研究是当今质量工程领域面向六西格玛设计中的热点研究问题。本文从面向普通设计人员、易于推广应用的角度出发,基于传统的TOPSIS方法对产品/过程设计中的多元质量特性优化方法进行了深入研究,具体内容包括以下几个方面:首先,对产品/过程设计中多元质量特性优化问题的特点与难点进行深入分析,指出了传统设计模式的弊端,强调了并行质量工程、六西格玛设计对提高产品/过程质量的重要作用。其次,针对望目类型多元质量特性优化问题的自身特点进行深入研究,指出传统TOPSIS方法存在的不足。改进的TOPSIS方法中给出了适合望目质量特性的正理想解,并根据公差限的实际需求确定合理的负理想解,然后给出优化的步骤、构建评价指标。改进方法不仅可以对备选方案排序,还可以确定最佳因子水平组合。再次,研究了考虑稳健性需求的优化方法。传统的TOPSIS方法没有考虑到实际生产对稳健性的要求,改进方法分别构建基于均值与标准差的决策矩阵,考虑了决策者的偏好与实际情况的需求。此外,改进方法中的相对贴近度使决策者可以在质量特性的均值与标准差之间进行权衡,增强了优化的稳健性。最后,根据同时考虑偏差与稳健性的需求进行深入研究。针对传统TOPSIS方法存在的不足构建基于均方误差平方根的决策矩阵,改进方法使决策者可以在靠近目标值与减小方差之间进行衡量,增加了决策过程的灵活性。本文提出的几种改进方法简单有效,对操作人员的统计与数学知识水平要求不是很高,模型求解过程通过Minitab或Excel等常规软件即可完成,因此改进方法易于在实际中得到推广应用。此外,改进方法考虑了决策人员的偏好及实际情况对稳健性与偏差的要求,使优化结果更符合实际情况,也使决策者能够得到更满意的优化方案。更多还原

【Abstract】 The optimization methodology research for multiple quality characteristics in products or process design is considered as hot topic research issue on Design for Six Sigma in quality engineering. In the consideration of the demands of ordinary technical persons with not much statistical knowledge and the requirements of easy to be implemented and popularized, based on the conventional TOPSIS methodology, this paper brings you an advanced research on optimization methodology for multiple characteristics in the design of products or process. The main content is described as follow:First of all, you will find, in this paper, through the analysis of the problems and difficulties on optimization issue research for multiple quality characteristics in the design of product or process, the author has indicated the conventional design’s defaults and shown you the importance of concurrent quality engineering and Design for Six Sigma in the quality improvement of product or process.After that, the author focuses her research on the self-particularity of the nominal-the-best type optimization with multiple quality characteristics and then puts out the shortcomings of conventional TOPSIS method. The positive ideal solution suitable for nominal-the-best type quality characteristic is presented in the proposed method, and the negative ideal solution is determined reasonably by the requirement of tolerance limits in reality. Thus, the order of alternatives and the optimal factor level combination can be determined by the improved optimization process and assessing index.Thirdly, the optimization method to meet the demand of robustness is studied. After the limitation analysis of conventional TOPSIS method, an improved methodology taking robustness into consideration is proposed, and then the decision matrix based on mean value and standard deviation are established, and the preference of decision maker and requirement of reality are both considered in the proposed method. Besides, the balance between mean value and standard deviation of characteristics can be done by using combined relative closeness which enhances the robustness of optimization.Finally, the author addresses her research on the requirements of deviation and robustness. An improved method considering both deviation and robustness is proposed to make up the defects of conventional TOPSIS method, and the decision matrix based on root of mean square error is established, thus the tradeoff between closing to the target and decreasing variation can be done by the decision maker which increases the flexibility of decision process.On conclusion, this paper proposed some improved methods which are simple, practical, and effective, also these methods require technical persons to know not much knowledge of statistics and mathematics, and the model optimization can be done through any normal soft wares such as Minitab or Excel. Moreover, it makes the decision making process more flexible and reasonable after taking into consideration of the decision maker’s preference and practical requirements of robustness and deviation, which makes the result of optimization correspond more with the reality and decision maker get more satisfactions about the optimization resolution.更多还原