摘要
将有限元法与人工神经网络相结合 ,利用人工神经网络的非线性映射能力 ,训练一个可以预测不同输入参数时的结构应力或位移等信息的人工神经网络 ,使其可以方便地代入到优化设计程序中 ,其中重点介绍的是人工神经网络的教师样本构造问题。用一实例详细说明了方法的使用过程 ,结果表明了这种方法的有效性。
Letting the finite element method be combined with artificial neural net, using the nonlinear mapping ability of artificial neural net, an artificial neural net, which could predict the information of structural stress or displacement while different parameters were being inputted, has been trained and let it be conveniently substituted into the optimized designing program. The structural issues of teacher′s sample of artificial neural net were being introduced emphatically. The operational process of this method was explained in detail by a living example.
出处
《机械设计》
CSCD
北大核心
2004年第12期18-20,共3页
Journal of Machine Design
关键词
结构优化
有限元法
人工神经网络
structural optimization
finite element method
artificial neural net
