面向多尺寸链的计算机辅助选择装配模型研究  

Computer aided selective assembly model for multi-dimension chains

作　　者：刘建东[1] 常智勇[1] 莫蓉[1] 张栋[1] 魏江峰[1] 

机构地区：[1]西北工业大学现代设计与集成制造技术教育部重点实验室,陕西西安710072

出　　处：《计算机集成制造系统》2008年第5期855-860,共6页Computer Integrated Manufacturing Systems

基　　金：国家863计划资助项目(2006AA04Z159);航空科学基金资助项目(05H53080)~~

摘　　要：针对现有计算机辅助选择装配的研究仅限于单一尺寸链,且质量损失模型只能用于对称公差带这一问题,提出一种新的面向多尺寸链计算机辅助选择装配的模型。该模型基于田口理论,建立了面向非对称公差带的质量损失评价规则。以遗传算法为基础,设计了可保持种群多样性的快速多目标优化算法,并通过加权巴莱托方法来描述偏好信息。将该模型应用于企业信息系统,并采用统计方法来验证其优化效率。计算实例表明,在相对短的时间内,随着迭代次数的增加,每代的非受控点逐渐收敛于巴莱托前沿,而且解的分布较为均匀,符合一个多目标优化算法的核心要求。Studies of selective assembly problem were criticized for two reasons： most models were confined to single dimension chain; quality loss could be used only on symmetrical tolerance. To deal with these problems, a new selective assembly model for multi-dimension chains was proposed. Based on Taguchi method, evaluation rules for quality loss oriented to asymmetrical tolerance was established. According to Genetic Algorithm （GA）, a new multi- objective GA was designed to remain population diversity and weighted Pareto method was used to represent preference information. This model was applied in enterprise information system, and its optimization efficiency was proved by statistics method. Simulation results revealed that the proposed model was able to find much better spread of solutions and better convergence near the true Pareto-optimal front.

关 键 词：选择装配 多目标规划 遗传算法 偏好信息 质量损失成本 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]