摘要
为了有效地提取环境影响下结构损伤的异常信息,提出一种基于离散小波变换(DWT)和快速独立分量分析(FastICA)相结合的无监督损伤识别方法。首先,通过离散小波变换预处理结构响应,并将处理后的混合信号作为FastICA的输入信号,提取独立的损伤特征信号;然后,根据分离出的含有损伤突变的特征分量信号及其对应的混合矩阵识别结构损伤时间和位置;最后,通过地震激励下三层框架的数值模拟结果验证了该算法的有效性和可行性。
In order to extract more effectively the abnormal damage information of the structure under the affect- ing environment, this paper proposes an unsupervised damage identification method based on a combination of Discrete Wavelet Transform（DWT） and Fast Independent Component Analysis （FastICA）. First, the structural responses are preprocessed by Discrete Wavelet Transform （DWT）, and the wavelet-domain mixtures are fed into the FastICA model to extract independent damage novelty signals; then, the damage instant and damage location are identified respectively according to the feature component which contains damage information and the corre- sponding recovered mixing matrix separated by FastICA. Finally, the simulation results of the three-story frame under seismic excitation are used to prove the effectiveness and feasibility of the algorithm.
出处
《苏州科技学院学报（工程技术版）》
CAS
2016年第3期35-40,共6页
Journal of Suzhou University of Science and Technology （Engineering and Technology）
基金
江苏省自然科学基金项目(BK20141180)
江苏省结构工程重点实验室开放课题(Z1405)
江苏省建设系统科技项目(2015ZD77)
关键词
离散小波变换
快速独立分量分析
损伤识别
特征分量
discrete wavelet transform, fast independent component analysis, damage identification, feature component
作者简介
刘文波（1989-），男，河南驻马店人，硕士研究生。
通信联系人：常军（1973-），男，教授，博士，主要从事健康监测与振动控制的研究，Email：changjun21@mail．usts．edu．cn。
