计算机集成制造系统 ›› 2019, Vol. 25 ›› Issue (第11): 2913-2922.DOI: 10.13196/j.cims.2019.11.020

• 当期目次 • 上一篇    下一篇

基于熵权—广义线性模型与满意度的柔顺精密产品非正态质量特性稳健设计

吴佳伟1,宋华明1+,万良琪2,黄甫1,杨慧1,杨加猛3   

  1. 1.南京理工大学经济管理学院
    2.南京航空航天大学经济与管理学院
    3.南京林业大学经济管理学院
  • 出版日期:2019-11-30 发布日期:2019-11-30
  • 基金资助:
    国家自然科学基金资助项目(71172105,71571102,51665017,51365015)。

Robust design for precision compliant product with non-normal distribution of quality characteristics based on entropy weight-GLM and desirability function

  • Online:2019-11-30 Published:2019-11-30
  • Supported by:
    Project suported by the National Natural Science Foundation,China(No.71172105,71571102,51665017,51365015).

摘要: 针对柔顺精密产品质量特性非正态情形下,质量模型预测不准确、均值和方差权重分配困难的问题,提出一种熵权—广义线性模型与满意度的柔顺精密产品稳健设计方法。采用全因子设计和有限元分析方法获得初始样本数据,利用广义线性模型构建柔顺精密产品非正态质量特性模型。运用熵权法建立均值和方差的加权函数,在此基础上结合满意度函数将柔顺精密产品非正态质量特性稳健设计问题归结为满意度最大的目标优化问题,来解决优化结果受方差波动影响的难题。通过算例验证分析表明,所提方法能够在考虑稳健性的同时解决权重分配问题,并降低优化函数方差波动对结果的影响。

关键词: 柔顺精密产品, 熵权, 广义线性模型, 满意度函数, 稳健设计

Abstract: Under the circumstance that the quality characteristics with non-normal distribution of precision compliant product,for the low predictability of quality model and the weight assignment difficulty of mean and variance problem,a robust design method based on entropy weight-Generalized Linear Model (GLM) and desirability was established.The initial sampling data were obtained with full factorial experiment and finite element method,and the quality characteristic with non-normal distribution model was built through GLM approach.The entropy weight method was used to build the mean and variance weight function.On this basis,the robust design problem was transformed into target function maximum satisfaction problem combining with desirability function to deal with variance influence on result.A numerical example showed that the robust design variables obtained by the proposed approach could solve weight distribution problem and the effect of variance volatility could be mitigated.

Key words: precision compliant product, entropy weight, generalized linear model, desirability function, robust design

中图分类号: