摘要
在分析主成分分析PCA和独立分量分析ICA的基础上,建立了基于PCA和ICA的结构损伤识别构架。利用它们对结构损伤信号进行特征提取,并将提取的特征作为3层BP神经网络的输入,以实现对结构损伤的识别。这2个模型通过British Columbia大学IASC-ASCE SHM任务组提供的用于验证分类正确性的结构基准数据集合进行测试。结果显示:PCA和ICA都能降低信号中噪音的影响,并对特征进行有效提取;基于ICA的模型比基于PCA的模型预测更准确。
The architecture of structural damage identification based on PCA and ICA was constructed after analyzing PCA and ICA. It can effectively extract structural features from structural damage signals, and apply them as the input to an artificial neural network, which consists of three layers BP neural network for structural damage identification. Then, it was tested by the benchmark dataset from IASC-ASCE SHM group in British Columbia University. The results show that both PCA model and ICA model can reduce the influence from noise, and correctly extract structural features from the structural damage signals; ICA model predicts more accurately than PCA model.
出处
《武汉理工大学学报》
EI
CAS
CSCD
北大核心
2006年第7期93-96,共4页
Journal of Wuhan University of Technology
基金
教育部高校行动计划项目(2004XD-03)
关键词
主成分分析
独立分量分析
结构损伤识别
特征提取
人工神经网络
principal component analysis （PCA）
independent component analysis （ICA）
structural damage identification
feature extraction
artificial neural network
