【作者】 赵延明；

【导师】 刘德顺；

【作者基本信息】 中南大学 ， 机械工程， 2013， 博士

【摘要】 机械产品质量和成本是影响其竞争力的重要因素。本文从产品级的角度出发,面向具有多元性、动态性、复杂性的质量特征,深入分析产品质量特征与其零部件质量特征之间的关系,提出产品质量损失建模和公差优化设计方法,提高产品质量,降低产品成本,快速响应市场需求变化。(一)提出了面向多质量特征的产品质量损失三层次建模方法。充分利用产品生产全过程历史检测数据,应用神经网络、层次分析方法,确定产品零件质量、产品质量特征、产品质量之间的权重；考虑零部件加工生产过程中的返修成本和报废成本,采用分段函数理论改进了田口质量损失函数,并根据质量特征服从不同的分布,导出了相应的零件质量损失；建立了多元产品质量损失模型。以微型电动机为例,对具有多质量特征的产品质量损失进行了计算与分析,验证了该方法的有效性,为多质量特征产品质量综合评价提供了解决途径。(二)提出了基于服役寿命分布的质量损失建模方法。面向产品全寿命周期,定义产品质量损失为其投入服役后因产品质量特征超出规定范围而导致报废所造成的损失现值；在此质量观基础上,应用资金折现原理建立了基于服役寿命分布的产品质量损失模型；针对服从不同分布的质量特征,提出了基于质量特征公差要求的产品服役寿命分布密度函数计算方法。结合三个实例,分别对基于服役寿命分布的望目特征、望小特征和望大特征的产品质量损失进行计算与分析,深入讨论了产品特征分布参数和质量要求对质量损失的影响,并与田口质量损失模型、Teran质量损失模型进行了比较分析。结果表明：该产品质量损失模型描述了一种产品投入服役后因质量特征超出其技术规范要求而报废给社会带来的实际损失,能够反映不同产品在不同服役条件和环境下产生的质量损失,将产品质量从设计制造阶段扩展到运行服役阶段,为产品质量在全寿命周期的变化规律研究提供了新的视角。(三)提出了基于数据挖掘的产品公差体系优化设计方法。从产品广义公差出发,分析产品各层次质量特征的相互影响关系,建立产品公差体系模型；利用产品制造检验过程中的历史数据,建立了基于支持向量非线性回归机的产品质量特征偏差与零部件质量特征偏差关系模型,并以微电机为例,构建了产品质量特征偏差与零部件质量特征偏差关系支持向量非线性回归机(SVNR)模型,并与RegressFunction方法进行了比较分析；分析了非尺寸公差与产品成本的关系,建立了基于产品广义公差的产品制造成本模型；以产品成本最小为目标,建立了产品公差体系优化设计模型。结果表明：所提出的产品公差体系优化方法行之有效,为实现产品公差设计提供了新途径。(四)提出了面向用户性能需求的产品公差体系再设计方法。针对用户对产品性能需求不断变化或增加新的用户需求,在基于数据挖掘的产品公差体系优化设计方法的基础上,建立面向产品用户性能需求的产品公差体系再设计优化模型；并以微电机为实例,分析了微电机的装配关系以及性能影响因素,建立微电机公差体系模型,再以微电机制造成本最低为目标,用户性能需求为约束,建立微电机公差体系再设计优化模型,根据用户对微电机产品性能的不同需求,进行了微电机公差体系再设计优化,分析探讨了微电机公差和成本的变化规律。所提出的公差体系再设计优化方法能够根据用户性能需求变化直接对产品公差体系进行改进优化,缩短了产品设计与制造周期,快速响应市场变化,生产企业只需在原产品生产系统的基础上适当调整生产制造系统和质量检验体系,就能使产品既能满足用户需求而制造成本又最低,实现了产品公差设计从产品开发阶段到产品全市场周期的扩展。更多还原

【Abstract】 Mechanical product quality and manufacturing cost are the important factors that affect its competitiveness. From the perspective of product level, the relationships between product quality characteristics and parts quality characteristics are analysed, and then the methods of product quality loss model and tolerance system optimization design is proposed for the product quality characteristics with multiplicity, dynamic nature and complexity, which improve product quality, reduce product cost, and quickly respond to market demand change.Firstly, three hierarchical modeling method of product quality loss for multi-quality characteristics is proposed. Making full use of the historical test data recorded in the whole production process, the weights among parts quality, product quality characteristic and product quality are determined with neural network method and analytic hierarchy process. Considering the rework cost and scrap cost during parts manufacturing process, the piecewise function theory is used to improve Taguchi quality loss function. The corresponding parts quality loss is deduced with the different distribution of parts quality characteristic, and product quality loss for multi-quality characteristics is established. A micro-motor is used as an example to calculate and analyse product quality loss with multi-quality characteristics, which show that the developed method is effective, and provides a comprehensive evaluation approach solution of product quality with multi-quality characteristics.Secondly, modeling method of product quality loss based on service life distribution is proposed. For product whole life cycle, Product quality loss is defined as present worth of a loss caused by product obsolescence because its quality characteristics are out of specification after it is put into service. On the basis of this view on quality, the present worth of quality loss model based on service life distribution is established with the principle of discounted cash flow, and the calculation methods of product service life distribution density function based on the technical requirements for tolerance of quality characteristic are proposed for the quality characteristics with different distribution. Combining with three cases, the fatigue strength in the dangerous cross-section of rotation shaft, the product quality loss of nominal-is-better, smaller-the-better and larger-the-better quality characteristics are calculated and analysed, respectively. The impacts of various parameters of the new model on product quality loss are discussed, and the new model is contrasted with Taguchi quality loss model and Teran present worth of quality loss model. The results show that the new model describes an actual loss to society, which is caused by product obsolescence because its quality characteristics are out of specification after it is put into service, and can reflect quality loss of different products in the different service conditions and environment, and is more realistic than Taguchi quality loss model and Teran quality loss model. The new model provides new perspective for researching product quality variation in product whole life cycle.Thirdly, product tolerance system optimization design method based on data mining is proposed. From the perspective of general tolerance, the product tolerance system model is established after analysing the interaction relationships among the quality characteristics of product levels. The support vector non-linear regression relational model between parts quality characteristic deviations and product quality characteristic deviations is established by using the historical test data in the product manufacturing process, and a micro-motor is used as an example to structure the support vector non-linear regression (SVNR) model between parts quality characteristic deviations and product quality characteristic deviations, and compare with the Regress Function method. The relationship between non-dimension tolerance and product cost is analysed, and the product manufacturing cost model based on product general tolerance is set up. Then, the optimal design model of product tolerance system is established by minimizing product manufacturing costs as target. The results show that the developed method is effective and provides a new way for the product tolerance design.Finally, tolerance system redesign method for user performance requirements is proposed. For the users often change product performance requirements or the new user requirements are added, the product tolerance system redesign model for user performance requirements is established on the basis of optimal design method of product tolerance system based on data mining. A micro-motor is used as an example, its assembly relationship and influence factors of performance are analyzed, and micro-motor tolerance system model is established. The micro-motor tolerance system redesign model for user performance requirements is established by minimizing product manufacturing costs as target and user performance requirements as constraint. The micro-motor tolerance system redesign is optimized according to the different user performance requirements. The change rule between product tolerance and cost is analysed. The proposed tolerance system redesign optimization method can directly improve and optimize product tolerance system according to user performance requirements change, so that the manufacturers can shorten product design and manufacturing cycle, and respond quickly to market change. The manufacturing enterprises can manufacture the products that can not only meet the user requirements, but also minimize its manufacturing total costs because they only need to appropriately adjust manufacturing system and quality inspection system on the basis of the original product manufacturing system. The new method expands product tolerance design from the product development stage to the product whole market cycle.更多还原