基于RBF网络的K6型单层球面网壳结构损伤研究  

Damage research of K6 single-layer reticulated shells based on RBF network

作　　者：杨新峰 孙凯 马跃 刘平[1] 徐宁 YANG Xin-feng;SUN Kai;MA Yue;LIU Ping;XU Ning(Civil Engineering and Architecture College,Jiangsu University of Science and Technology,Zhenjiang 212100,China)

机构地区：[1]江苏科技大学土木工程与建筑学院,江苏镇江212100

出　　处：《空间结构》2022年第3期27-32,共6页Spatial Structures

摘　　要：随着越来越多网壳结构的建造,其损伤的测定亟需解决.根据某实际工程设计一个55个节点、126根杆件的K6型单层球面网壳模型.根据穷举法,利用ANSYS建模输出各个节点不同损坏程度的位移响应;将响应数据根据杆件损坏工况进行预分类,并构造杆件损伤信息矩阵;然后将预分类后的响应数据与损伤信息矩阵输入一种改进的RBF网络进行训练;随后使用随机损伤工况数据对已训练的神经网络进行测试.结果表明采用改进的RBF网络可以准确识别出结构的损伤状态,该方法对实际应用有参考价值.With the construction of more and more reticulated shells,the damage measurement is in urgent demand.Based on a real project,a K6 single-layer spherical reticulated shell model with 55 nodes and 126 members is designed.Firstly,the exhaustive method is used to output the displacement response of each node by ANSYS modeling.The response data is pre-classified according to the damage conditions of the members,and the member damage information matrix is constructed.Then the pre-classified response data and damage information matrix is input into an improved RBF network for training and random damage condition data is used to test the trained neural network.The results show that the improved RBF network can accurately identify the damage state of the structure,and the method is valuable for practical application.

关 键 词：单层球面网壳 RBF神经网络 结构损伤 损伤指标 正弦激励 

分 类 号：TU393.3[建筑科学—结构工程]