| 引用本文: | 顾晓光,马义中,张延静,蒋兆.基于置信水平和熵权法的参数和容差经济性设计[J].哈尔滨工业大学学报,2017,49(11):73.DOI:10.11918/j.issn.0367-6234.20171084 |
| GU Xiaoguang,MA Yizhong,ZHANG Yanjing,JIANG Zhao.Parameter andtolerance economic design based on confidence level and entropy method[J].Journal of Harbin Institute of Technology,2017,49(11):73.DOI:10.11918/j.issn.0367-6234.20171084 |
|
| 摘要: |
| 为降低生产运作成本并提高产品稳健性，需要对产品系统进行参数和容差的经济性设计.本文考虑模型参数不确定性和质量损失函数系数对优化结果的影响，从质量损失成本和容差成本的角度，提出了基于置信水平和熵权法的参数和容差经济性设计方法.该方法分为两个阶段：在参数设计阶段，根据因子响应试验数据构建质量特性的响应曲面模型，进而得到质量特性规格限为约束区间的置信水平计算公式，再通过最大化置信水平得到最优设计变量；在容差设计阶段，构建质量特性的均值方差关于设计变量和容差的模型，同时根据容差成本试验数据构建容差成本模型，再采用熵权法计算质量损失函数系数进而构建总成本函数，最终通过最小化总成本计算最优容差值.本文以树脂生产过程的胺添加试验为案例，系统地研究了温度、搅拌速度和添加速度这三个设计变量对树脂粘度的影响.结果表明，与已有方法相比，通过本文方法所确定的最优设计变量及其容差值，减少了模型参数不确定性对优化结果影响，同时降低了质量损失和容差成本.
|
| 关键词: 参数 容差 置信水平 熵权法 经济性设计 |
| DOI：10.11918/j.issn.0367-6234.20171084 |
| 分类号:O213.1 |
| 文献标识码:A |
| 基金项目:国家自然科学基金(71471088；61603347) |
|
| Parameter andtolerance economic design based on confidence level and entropy method |
|
GU Xiaoguang,MA Yizhong,ZHANG Yanjing,JIANG Zhao
|
|
(School of Economics and Management, Nanjing University of Science and Technology, Nanjing 210094, China)
|
| Abstract: |
| Parameter and tolerance design are used to reduce the cost of production and improve the robustness of product. We consider the effects of model parameter uncertainty and quality loss function coefficient, and propose a parameter and tolerance economic design approach based on confidence level and entropy method. The proposed approach consists of parameter design stage and tolerance design stage. In parameter design stage, we obtain the best design variables through maximizing the confidence level of quality characteristics. In tolerance design stage, we get the quality loss function coefficient through entropy method, and then obtain the best tolerances through minimizing the total cost function which is composed of quality loss cost and tolerance cost. In this paper, we have analyzed the amine addition experiment of resin production process and studied the influence of temperature, agitation and rate of addition on the viscosity of resin systematically. Experimental results indicate that the optimal design variables and their tolerance values obtained by the proposed approach minimizes the effects of model parameter uncertainty while reducing the quality lost and tolerance cost.
|
| Key words: parameter tolerance confidence level entropy method economic design |
