基于质量损失的交叉质量结构参数设计方法

摘要

本文根据复杂产品的复杂质量特性构成,提出了具有交叉性的质量结构,将田口方法中参教设计方法引入到具有交叉质量结构的产品质量设计中来.在一般田口参数设计基础上,将信噪比转化为标准质量损失,以多元回归分析为工具,以使复杂产品系统综合质量损失最小为目标,确定各设计参数的设计水平,并以实例验证了所提方法的可行性和有效性.%Complex products have sophisticated quality characteristics, composition, and parameters in the parts design. Hence, any attempt of optimizing single quality characteristics in the design of complex products is prone for failure. The traditional Taguchi method mainly resolves problems related to single-layer and simple design. This paper applies Taguchi method to the design of products with crossed structure, and transforms SN ratio into standard quality loss. Taguchi method offers a solution to resolving quality improvement problems for complex products.Several quality characteristics of the designed products are firstly determined. Controllable factors are then selected as parameters to help determine quality structure for complex products. Engineers help configure parameter levels according to their expertise. Taguchi method is then employed to design parameters and calculate SN ratio for each quality characteristics. However, Taguchi parameter design method can' t resolve issues related to crossed quality structure because parameters designed separately to measure quality have conflicts. In order to overcome the shortcoming of Taguchi method, we propose the following process in setting parameter levels:First, we conduct the ANOVA analysis to calculate SN ratio in order to design parameters, and determine the effect of parts on quality characteristics. Second, we transform S/N ratio into a standardized quality loss. Quadratic regression analysis is conducted by taking significant factors of each quality characteristic as an independent variable. Each independent variable has values in three levels ( 1, 2 or 3 ). Dependent variable is standardized quality loss. Third, integer programming is constructed to minimize the comprehensive quality loss of the production system. Optimal combination of various parameters is set to resolve programming problems.In summary, SN ratio is transformed into quality loss based on the traditional Taguchi parameter design. Multiple regression analysis is used to determine the levels of different parameters and minimize comprehensive quality loss of complex product systems. Last, an example is used to verify the feasibility and effectiveness of the proposed method.