桁架结构拓扑形态数字化信息提取

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成进行了数字化信息提取，以桁架拓扑结构形成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，对桁架结构拓扑形态信息进行了有效的提取和压缩。

Digital abstraction of information for topology configuration of truss

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 configuration of truss is described as a topology configuration germ. After laying a node map, forming characteristic germ cluster of truss, cutting germ, and compressing germ, the configuration of truss is described as a series of decimal fraction between 0 and 1, which represents the character of the configuration of truss. 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.