摘要
为提高基于随机子空间(stochastic subspace identification, SSI)法模态参数识别的自动化程度,提出了基于博弈K均值(game-based K-means, GBK)聚类的两阶段模态参数自动识别算法。首先,对采集到的数据通过SSI识别出大量待分析的极点并根据所提出的硬性指标对虚假模态进行初步剔除;其次,将前后两阶模态的多类型偏差指标构造成特征矩阵进行GBK聚类,进一步将剩余虚假模态进行剔除;之后,将提取的结构模态进行层次聚类得到各阶模态;随后将提出的方法通过六自由度质量-弹簧结构数值模型进行验证,最后,将此算法应用于某大跨悬索桥的实际监测数据中进行模态参数自动识别,进一步验证了该方法在实际工程中对采集到的海量数据进行自动化识别的可行性和适用性。
In order to improve the automation of modal parameter identification based on stochastic subspace identification(SSI),a two-stage automatic modal parameter identification algorithm based on game-based K-means(GBK)was proposed.Firstly,mode candidates were identified from a large number of system orders through SSI for the collected data and certainly spurious modes were removed using hard validation criteria.Secondly,a multi type deviation indicator of two adjacent poles was constructed into a feature vector matrix for GBK clustering,and the spurious modes were further eliminated.Then,the extracted structural modes were hierarchically clustered to obtain each order of modes;a six-degree-of-freedom mass-spring structure numerical model was then used to validate the suggested approach.Finally,it was applied to the actual monitoring data of a long-span suspension bridge for automatic modal parameter identification,which further verifies the feasibility and applicability of the proposed method for automatic identification of massive data collected in practical engineering.
作者
郭鹏
李东升
郭鑫
GUO Peng;LI Dongsheng;GUO Xin(Guangdong Engineering Center for Structure Safety and Health Monitoring,MOE Key Laboratory of Intelligent Manufacturing Technology,Shantou University,Shantou 515063,China;School of Transportation Institute,Inner Mongolia University,Huhehot 010030,China)
出处
《振动与冲击》
EI
CSCD
北大核心
2023年第24期92-100,共9页
Journal of Vibration and Shock
基金
国家自然科学基金(52078284)
广东省自然科学基金(2021A1515011770)
汕头大学科研启动基金(NTF18012)
桥梁结构健康与安全国家重点实验室开放课题(BHSKL20-10-KF)。
关键词
结构健康监测
运行模态分析
模态自动识别
议价博弈
随机子空间
structural health monitoring
operational modal analysis
automated modal identification
bargaining game
stochastic subspace identification