摘要
当前大型桥梁在线健康监测与损伤识别中主要采用基于结构模态参数的方法,该类方法存在固有缺陷。结构的频响函数包含了更丰富的结构动力信息,但数据量大、冗余度高,不利于工程应用。本文采用主元分析方法,对结构频响函数进行空间变换,得到的主元向量可在最小均方意义下提取数据的主要分布特征,根据主元的累积贡献率,可选取较少的主元可靠重构结构的频响函数,从而使基于频响函数的在线监测在大跨桥梁结构的应用更为可行。具体针对润扬长江大桥,结合实时监测系统的测点布置,对南汊悬索桥进行了动力特性分析,根据降维后的较少主元对结构频响进行重构。误差分析结果表明,采用27或20个主元能提取润扬悬索桥的主要动力特征,重构误差的均方值分别为0.0097和0.0134。
The structural health monitoring and damage detection for long span bridge structure are mainly based on the modal analysis, and the method has inherent defects. Frequency response function （FRF） contains abundant structural dynamic characteristics, but the high redundancy and the huge data confine its application. The structural FRF is mapped into principal component space using principal component analysis, and a few principal components are selected to reconstruct the FRF based on the cumulated contribution. The dynamic charactersistics of Runyang suspension bridge are studied according to the real time monitoring system, and the FRF isreconstructed by a few principal components. Results show that 27 or 20 principal components can extract the dominating features of FRF, and the corresponding covariances of the reconstruction error are 0. 009 7 and 0. 013 4 respectively.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
2007年第3期403-406,共4页
Journal of Nanjing University of Aeronautics & Astronautics
基金
国家自然科学基金(50378017)资助项目
国家863计划(2006AA04Z416)资助项目
南航青年科研基金(Y0513-013)资助项目
关键词
健康监测
特征提取
主元分析
悬索桥
重构
health monitoring
feature extraction
principal component analysis
suspension bridge
re construction
