基于随机灰靶模型的多质量特性产品参数选择

Parameter Choice for Multi-Quality-Characteristic Product Based on Stochastic Grey Target Model

针对传统田口质量损失函数的一些不足,提出一组改进型田口质量损失函数,并给出了质量特性取值"中靶"和"脱靶"的范围,为文中剔除"脱靶"对策、确定满意区域提供依据;以损失期望作为灰靶模型效果测度的基本工具,在此基础上,提出一系列灰靶效果测度方法;以正交试验设计为基础,针对多质量特性产品参数选择的具体问题,考虑噪声因素的概率分布,构建了基于改进型田口质量损失函数的多质量特性产品参数设计随机灰靶模型,并给出最优对策的判断标准。案例研究表明该模型可以确定相对最优参数水平组合,解决多质量特性产品参数选择问题,该方法新颖清晰、易于操作,可辅助工程设计者进行决策。

To cover the shortage of traditional Taguchi quality loss functions, a group of improved Taguchi quality loss functions are developed to decide the range of ＂on-target＂ and ＂off-target＂ for a quality characteristic to help both excluding ＂off-target＂ strategies and determining the satisfactory domain. Moreover,the expectation of quality loss is used as the basic measure for the effect in the grey model,based on which a series of other methods for effect measure are developed. Based on orthogonal experiment design, a stochastic grey target model for parameter design of multi-quality-characteristic product is proposed, in which the probability distribution of events is considered and a judgment criterion of choosing a relative optimal strategy is provided. This model can finally determine the optimal parameter combination for multi-quality-characteristic product. This method is clear and easy to operate, which can aid engineering designers to make decisions.