冲压成形仿真数据的主成分与模糊聚类分析

Principal component and fuzzy C-means clustering analysis of stamping simulation results

冲压成形数值仿真结果中隐含着大量的领域知识。文章将主成分分析与模糊聚类方法应用于基于仿真模型与模拟结果数据的冲压件相似性判别与成形性能判别。通过对油箱冲压成形有限元仿真结果数据进行处理,分析了压边力、拉延筋设置参数、摩擦系数等工艺参数对成形性能的相对重要程度;构造了油箱成形性能的模糊概念,描述其破裂、起皱程度。通过对汽车覆盖件有限元模型数据的分析,对汽车覆盖件进行模糊分类及相似性判别。结果表明,面向有限元仿真结果的数据挖掘技术,可以为冲压成形领域知识发现提供一种有效的新途径。

Large amount of implicit and useful knowledge is embedded in the simulation results of stamping. In this paper, the principal component analysis and fuzzy C-means clustering is introduced into the similarity and formability assessment of stamping components based on data processing of simulation data. The relative importance degree of various technique parameters, including blankholder force, drawbead setting and friction coefficient, are obtained for an oil pan stamping process. The fuzzy concepts are developed by fuzzy C-means clustering method to describe crack and wrinkling of the oil pan. With these two methods, the similarities of the auto panels are also analyzed based on the finite element data. The analysis results show that data mining technology oriented to FEM results provides an effectively novel method of knowledge acquisition in stamping field.