桁架结构拓扑形态数字化信息提取

Digital abstraction of information for topology configuration of truss

为利用神经网络实现桁架结构的形态映射机制 ,采用形态基因方法对桁架结构拓扑形态进行了数字化信息提取 .以桁架拓扑结构形态基因的方式表述桁架的拓扑结构 ,应用预先布置节点图、桁架特征基因串、基因分片、压缩等方法 ,将桁架的拓扑结构表述成一串 0～ 1之间的小数形式 ,使其代表桁架拓扑结构的形态特征 .结果将形态化的桁架结构转化成为数字化的信息 ,形成了规范化的神经网络数据输入方法 ,为进一步实现桁架结构拓扑形态推理作好了准备 .算例证明此方法有效 。

The information of topology configuration of truss structure is digitally extracted to form a mapping mechanism of configurations of the truss structure through neural network. The truss configuration is described as a topology configuration germ. After laying a node map, forming characteristic germ cluster of truss, cutting germ and compressing germ, the truss configuration is described as a series of decimal fraction between 0 and 1, which represents the character of the truss configuration. As a result, the visual truss structure is transformed into digital information. The whole program produces standardized input data for neural network, and provides a foundation for configuration inference of truss structure by neural networks. Numerical simulation shows that this method can efficiently extract and compress the topology configuration information of truss structure.