基于反向传播神经网络的变型设计变异点快速定位  

Fast location of aberrant point for variant design based on the BP neural network

作　　者：赵秀燕[1] 赵婷婷[1] 张强[1] 陈微微[1] 魏小鹏[1] 

机构地区：[1]大连大学信息科学与工程省重点实验室,辽宁大连116622

出　　处：《计算机集成制造系统》2009年第8期1463-1467,1540,共6页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(50575026);辽宁省优秀人才培养基金资助项目(3040014);辽宁省高校科研计划资助项目(2008S007);辽宁省科技计划资助项目(2008219013)~~

摘　　要：为了对产品进行有效变型设计,针对产品模型的最小模块,提出了变异点概念,并在分析产品结构的基础上,将反向传播神经网络引入到变型设计中,建立了快速定位变异点的反向传播人工神经网络模型,较好地解决了人为确定变异点的杂乱性与欠准确性问题。最后进行了实证分析,结果表明,采用反向传播人工神经网络快速定位变异点是可行的。Aiming at the efficiency of variant design,module partition was often carried out till the minimum module after establishing the product model.According to the minimum module,Aberrant Point was defined and the BP neural network model for its fast location was established,which solved well the problem of the disorder and inaccuracy in the earlier man-made location method.Finally,a typical case was given to evaluate the performance of the proposed approach,which demonstrated that the approach was obviously helpful for quick of design variant product.

关 键 词：反向传播神经网络 模块划分 变型设计 变异点 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TH122[自动化与计算机技术—控制科学与工程]