灰色神经网络仿真及其在MCMQC中的应用  

Simulation of Grey Neural Network in MCMQC

作　　者：樊树海[1] 肖田元[2] 曹霞[1] 倪卫红[1] 

机构地区：[1]南京工业大学工业工程系,江苏南京210009 [2]清华大学,北京100084

出　　处：《计算机仿真》2008年第12期258-262,共5页Computer Simulation

基　　金：国家自然科学基金(70801036);中国博士后科学基金(2005037725);江苏省高校自然科学基金(07KJB460045)

摘　　要：为了解决大规模定制生产质量控制中质量数据缺乏的灰色不确定性问题,将灰色理论引入到了神经网络建模中。以灰色理论的核心思想(信息递阶、时变和多维特性)为基础,构造了基于灰色体系的建模方法,并对其进行了简化,给出了经过简化的一般性建模方法——横向、纵向和混合建模法,并将其与神经网络建模相结合。以一个典型的灰色BP神经网络为例,以横向建模方式对其进行了实现,并将其应用到大规模定制生产过程质量控制(MCMQC)中。仿真实验结果表明,灰色神经网络解决大规模定制生产质量控制中的灰色不确定问题是有效可行的。To solve the problem of grey uncertainty of quality data in mass customization manufacturing quality control, the grey theory was integrated into the model building of neural network. Based on the kernel of grey theory （ hierarchical, poly - dimensional and time - variant）, the model building methods of grey system were built. To simplify the computation, the paper also gave out three simplified model building methods （ crosswise, lengthwise and hybrid modeling method）, which can be integrated into neural network modeling. Finally, the paper gave out an example for a typical grey BP neural network in which the crosswise modeling method was used and applied to mass customization manufacturing quality control. The final simulation result showed the feasibility and effectiveness of the grey neural network.

关 键 词：灰色理论 神经网络 大规模定制生产质量控制 仿真 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]