基于神经网络的大跨钢结构缺陷损伤的定位研究

PDF(1163 KB)

振动与冲击 ›› 2011, Vol. 30 ›› Issue (9) : 191-196.

论文

基于神经网络的大跨钢结构缺陷损伤的定位研究

熊仲明1;王超1;林涛2

作者信息 +

Research on the defect determination of large-span steel structure based on neural network

Xiong Zhong-ming1; Wang Chao1; Lin Tao2

Author information +

文章历史 +

摘要

大跨钢结构目前被广泛应用于体育馆等大型公共建筑中。但其缺陷损伤位置的确定，至今没有得到很好的解决，这必将影响其使用过程中的安全性。本文利用神经网络技术，以某高校体育场馆的大跨钢结构为工程背景进行模拟损伤定位研究，通过ANSYS计算软件对该大跨钢结构建模分析，得出了该结构在损伤前后的模态参数，并将其结果作为网络的输入参数。为了提高神经网络模型对该结构缺陷损伤判定的收敛速度及诊断精度，在进行损伤识别时，将该大跨结构细分成许多子结构，缩小损伤的范围，同时将高阶频率引入到不同的神经网络训练样本中进行网络训练，检验其对该结构及构件损伤识别的影响。分析结果表明，本文采用神经网络技术对大型复杂结构进行损伤定位是可行的，并通过该方法的改进，将识别精度大大的提高，所得结论为今后进行网络改进，提高网络的准确性、抗干扰性和泛化能力提供了有意义的参考

Abstract

Large-span steel structure has been widely used in large-scale public buildings such as stadiums in recent years. But the location of defects has not been solved well which will affect the safety in the using process. In the background of the large span steel structure of a college stadium, a simulating study on the location of the defect based on neural network was put forward, the modal parameter before and after the structural damage was obtained from ANSYS model analysis of the large span steel structure, and which is taken as input parameter of the neural network. In order to improve the convergent speed and the diagnostic accuracy of the network model for the structural defects, when locating the structural damage, on one hand, the structure subdivided into a large amount of sub-structure and narrow the damage scope; on the other hand, input high frequency into different experimental samples of the neural network to conduct experiments to detect the influence of the damage location and damage degree caused by it. The results showed that the technology of neural network for the defects location of large complex structure was feasible, and the identification accuracy was greatly enhanced by improving the method, it will provide a meaningful reference for network improvements and improving the accuracy, interference immunity and generalization ability of neural network

导出引用

熊仲明;王超;林涛. 基于神经网络的大跨钢结构缺陷损伤的定位研究[J]. 振动与冲击, 2011, 30(9): 191-196

Xiong Zhong-ming;Wang Chao;Lin Tao. Research on the defect determination of large-span steel structure based on neural network[J]. Journal of Vibration and Shock, 2011, 30(9): 191-196