数值仿真结果中知识发现的模糊-粗糙集方法  

Fuzzy-Rough Method of Knowledge Discovery on Numerical Simulation Results

作　　者：尹纪龙 李大永[1] 彭颖红[1] 

机构地区：[1]上海交通大学机械与动力工程学院,上海200030

出　　处：《上海交通大学学报》2004年第9期1448-1452,共5页Journal of Shanghai Jiaotong University

基　　金：上海市科技启明星计划跟踪项目(01QMH1411)

摘　　要：分析了塑性成形过程数值仿真结果的特点和采用离散化作为粗糙集数据预处理方法的不足,提出用模糊-粗糙集算法对仿真结果进行数据挖掘,并以NUMISHEET'93中盒形件拉伸为例进行了数值仿真,从中提炼出对加工工艺和模具设计有指导意义的产生式规则,且每条规则都有可靠性度量,有利于知识推理中的规则调度.研究结果表明,该算法是一种适用于本领域特点的更有效的算法.The characteristics of simulation results in metal forming were analyzed. In light of the deficiency of the discretization method, as a pre-process for rough set method, a fuzzy-rough algorithm was proposed for data mining on numerical simulation results. Taking the square box's drawing process in NUMISHEET'93 as an example, this paper did numerical simulation. Through knowledge discovery from the simulation results using fuzzy-rough method, many instructive 'what if' rules were induced. Every rule has a reliability measure which is helpful to rule scheduling in knowledge reasoning. The results show that the fuzzy-rough method is a more efficient way to knowledge discovery from simulation results.

关 键 词：模糊集 粗糙集 知识发现 数值仿真 

分 类 号：TG302[金属学及工艺—金属压力加工]