乘员约束系统耐撞性数据挖掘  

Data mining on crashworthiness data of occupant restraint system

作　　者：赵志杰[1] 金先龙[1,2] 

机构地区：[1]上海交通大学机械与动力工程学院,上海200240 [2]上海交通大学振动、冲击、噪声国家重点实验室,上海200240

出　　处：《振动与冲击》2010年第3期13-17,共5页Journal of Vibration and Shock

基　　金：国家自然基金(90612017);国家自然基金(50705058);国家自然基金(60970049、50875166);上海市科技攻关计划项目(061111013);教育部博士点基金(20070248110)

摘　　要：把数据挖掘方法应用到乘员约束系统FE(Finite Element)数据及其碰撞仿真结果数据上。针对CAE数据的特点,基于面向对象思想对乘员约束系统数据进行数据挖掘前的数据准备。应用属性选择、决策树分类算法对98例碰撞仿真实验数据进行挖掘,得到乘员约束系统关键参数与乘员伤害WIC值的关系,并应用交叉验证方法对所得决策树进行了评估。通过实例验证,所得知识符合实际、对乘员约束系统的设计具有一定参考价值,是对海量CAE仿真数据进行知识发现的成功尝试。The applicability of data mining on the finite element data and the crashworthiness simulation data of occupant restraint system was demonstrated.According to the characteristics of CAE data,a framework for data preparation was developed based on object-oriented ideology.Decision tree and attribute selection algorithms were used to perform data mining on the crash simulation data of 98 instances.The results of the data mining give us a deeper insight into the interrelations between the key parameters of occupant restrain system in design and the performances in crash simulation.Finally,the learned rules have been tested with the real data sets.The results of the testing show that the learned rules are proper,and can be used in design of occupant restraint systems.

关 键 词：乘员约束系统 数据挖掘 有限元分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] U49[自动化与计算机技术—计算机科学与技术]