摘要
针对制造单元构建的问题特征,提出基于模糊C均值逻辑的制造单元构建算法。采用改进的减法聚类法产生初始聚类中心,防止陷入局部优化;提出制造单元构建专用的距离函数测定零件的相似性,避免欧几里德距离函数造成的对零件的错误聚类;采用最优迭代方案选择算法,避免因停止参数选择不当造成算法最终迭代方案不可行。通过对10组文献数据和90组随机产生数据的大规模测试和比较分析,表明了算法性能的优越性。
A new fuzzy clustering algorithm was proposed for manufacture cell formation to avoid infeasible and non-optimal solutions with Fuzzy C-Means. An improved subtract algorithm was used for creation to avoid local optimization, a new distance function was developed to avoid clustering errors among components caused by Euclidean distance, and a best-solution-selection procedure was advocated to avoid final infeasible solution caused by unsuitable algorithm stop-parameter selection. Large-scale comparative tests with 10 literature data sets and 90 random-generated data sets have verified superiority of the proposed algorithm.
出处
《计算机集成制造系统》
EI
CSCD
北大核心
2004年第12期1561-1566,共6页
Computer Integrated Manufacturing Systems
基金
河北省自然科学基金资助项目(603091)。~~
关键词
单元制造
制造单元构建
模糊C均值算法
模糊聚类
cell manufacturing
manufacturing cell formation
fuzzy C-means
fuzzy clustering
