学术论文

      基于图论聚类的随机子空间模态参数自动识别

      Automatic stochastic subspace identification of modal parameters based on graph clustering

      摘要:
      为提高随机子空间法模态参数识别过程中的自动化程度,减少人为干预,提出了基于图论聚类的桥梁结构模态参数自动识别方法.首先,初步剔除由于数据精度以及噪声等引起的虚假模态;其次,采用图论聚类法,对结构模态结果依次根据基于结构频率和模态保证准则(MAC)指标定义的距离进行聚类,以自动识别出结构的真实模态.随后基于灌河大桥0.5 h的加速度数据,采用所提方法实现了结构模态参数的自动识别,并通过结构的有限元模型对识别结果进行验证.最后,将所提出的方法应用到基于灌河大桥健康监测系统采集的一年加速度数据的模态参数识别过程中,表明了该方法在桥梁结构海量加速度数据的结构模态参数自动识别中是可行的.
      Abstract:
      In order to improve the degree of automation in the process of modal parameter identification for bridge structures based on the stochastic subspace identification method, and reduce human intervention, an automatic modal parameter identification method based on graph clustering for bridge structure is proposed.First, some methods are adopted to initially weed out the false modes caused by the data accuracy, noise and so on.Secondly, the graph clustering theory is used to identify the structural modal parameters according to the distances defined by structural frequency and modal assurance criterion (MAC) index, respectively, so as to finish the automatic modal parameters identification.The automation modal parameters identification of the structure is realized by the proposed method based on 0.5 h acceleration data of Guanhe bridge, and the identification results are verified by the corresponding finite element model.Then, the proposed method is used to identify the modal parameters of Guanhe bridge based on one-year acceleration data from its structural health monitoring system, which indicates that the method is feasible for the modal parameter automatic identification of the bridge structure with massive acceleration data.
      作者: 郑沛娟 [1] 林迪南 [2] 宗周红 [1] 余道兴 [1]
      Author: Zheng Peijuan [1] Lin Dinan [2] Zong Zhouhong [1] Yu Daoxing [1]
      作者单位: 东南大学土木工程学院,南京,210096 福建省建筑科学研究院福建省绿色建筑技术重点实验室,福州,350025
      年,卷(期): 2017, 47(4)
      分类号: TB122 U441.3
      在线出版日期: 2017年8月15日
      基金项目: 国家自然科学基金资助项目,江苏高校优势学科建设工程资助项目