论文部分内容阅读
运用连续型Ｈｏｐｆｉｌｅｄ网络以及连续双向联想存储器（ＢＡＭ）详述结构分析的两种新方法。新的方法是建立在比较上述两种方法的能量函数与结构位移法的能量函数（也就是所谓的刚度矩阵法）的关系基础上的，并且利用了人工神经网络所具备且优于经典数值方法的计算特点。几个相关结构分析的例子说明了以ＢＡＭ为基础的神经网络优于其他传统的数值方法及Ｈｏｐｆｉｅｌｄ模型。
Two new methods of structural analysis are described using continuous Hopfiled network and continuous bidirectional associative memory (BAM). The new method is based on a comparison between the energy functions of the above two methods and the energy function of the structure displacement method (so-called stiffness matrix method), and makes use of the advantages of the artificial neural network The calculation features. Several examples of related structural analysis illustrate that BAM-based neural networks outperform other traditional numerical methods and Hopfield models.