基于主成分分析与TOPSIS模型相结合的函数型产品质量特性的优化方法研究
发布时间：2018-07-29 09:47
【摘要】：基于稳健参数设计理论提出了一种将主成分分析法与逼近理想点决策方法(TOPSIS)相结合的非线性轮廓图(Non-Linear profile)优化方法。首先利用两步建模法拟合响应模型,计算模型参数的满意度函数值,其次对模型参数的满意度函数值进行主成分分析,消除参数之间的相关性,并构建模型参数变异模式图,确定选定主成分的优化方向。最后利用TOPSIS模型求得选定主成分的最优贴近度(OPI),将其作为最终的优化指标。传统的优化方法都忽略了模型参数之间的相关性及优化过程的稳健性,并且需要复杂的数学计算,而本文所提方法可以有效解决这些问题。最后利用该方法对文献中的实例进行了分析研究,证明本文方法切实可行,优化结果令人满意。
[Abstract]:Based on the robust parameter design theory, a nonlinear contour map (Non-Linear profile) optimization method is proposed, which combines principal component analysis (PCA) with approach to ideal point decision method (TOPSIS). First, the response model is fitted with two-step modeling method, and the satisfaction function value of model parameters is calculated. Secondly, the principal component analysis of satisfaction function value of model parameters is carried out to eliminate the correlation between parameters, and the model parameter variation pattern diagram is constructed. Determine the optimization direction of the selected principal components. Finally, the optimal closeness degree (OPI), of selected principal components is obtained by using TOPSIS model as the final optimization index. The traditional optimization methods ignore the correlation between the model parameters and the robustness of the optimization process, and need complex mathematical calculation. The proposed method can effectively solve these problems. Finally, the method is used to analyze and study the examples in the literature, and it is proved that the proposed method is feasible and the optimization results are satisfactory.
【作者单位】： 天津大学管理与经济学部;
【基金】：国家自然科学基金杰出青年基金资助项目(71225006)
【分类号】：F273.2;F224
[Abstract]:Based on the robust parameter design theory, a nonlinear contour map (Non-Linear profile) optimization method is proposed, which combines principal component analysis (PCA) with approach to ideal point decision method (TOPSIS). First, the response model is fitted with two-step modeling method, and the satisfaction function value of model parameters is calculated. Secondly, the principal component analysis of satisfaction function value of model parameters is carried out to eliminate the correlation between parameters, and the model parameter variation pattern diagram is constructed. Determine the optimization direction of the selected principal components. Finally, the optimal closeness degree (OPI), of selected principal components is obtained by using TOPSIS model as the final optimization index. The traditional optimization methods ignore the correlation between the model parameters and the robustness of the optimization process, and need complex mathematical calculation. The proposed method can effectively solve these problems. Finally, the method is used to analyze and study the examples in the literature, and it is proved that the proposed method is feasible and the optimization results are satisfactory.
【作者单位】： 天津大学管理与经济学部;
【基金】：国家自然科学基金杰出青年基金资助项目(71225006)
【分类号】：F273.2;F224
【参考文献】
相关期刊论文 前2条
1 许静;何桢;;基于逼近理想点的渴求函数法在Linear-Profile优化中的应用[J];工业工程与管理;2015年05期
2 何桢;张迎冬;;基于主成分分析的多响应稳健性优化方法研究[J];工业工程与管理;2012年06期
【共引文献】
相关期刊论文 前4条
1 许静;何桢;袁荣;陈U喼，
本文编号：2152288
本文链接：https://www.wllwen.com/jingjilunwen/hongguanjingjilunwen/2152288.html
