Experimental Study on Damage Identification of Dynamically Excited Structures

摘要在地震、风和其他荷载作用下，基于振动数据的土木工程结构损伤检测、定位和量化越来越受到人们的关注。这种做法的早期动机是1994年北岭地震和1995年阪神大地震对钢结构的潜在损害的发现，以及结构构件在施工后通常无法进入的事实。文献中已经提出了几种分析方法，并且还在为此开发更多的分析方法。目前正在使用模拟数据来检查这些方法的有效性；但是，它们还没有在实际实验数据上进行验证和比较。在这个探索性项目中，提出了实验研究和验证外生输入自回归模型、小波变换、特征系统实现算法（ERA）和子空间识别（SSI）方法在土木工程结构综合模态和损伤识别中的有效性。对于损伤定位和量化，将使用具有和不具有损伤定位向量的柔性矩阵和旋转柔性矩阵。为了更可靠的损伤量化，还将对这些方法进行进一步的分析改进。测试将在一个按比例设计的三层九自由度结构模型上进行，该模型的刚度特性可以通过计划改变一个或多个支撑，通过使用磁流变阻尼器实时突然移除一个或两个支撑，以及通过松开几个选定关节的连接硬件来改变。制作这种可变比例模型的主要费用将由该机构承担。该模型的基座将由弗吉尼亚理工大学的5英尺乘5英尺的振动台进行振动。该模型还具有被安装在三层中的任何一层的激振器激励的能力。加速度响应的动态测量将使用十个加速度计，三个在两个正交的方向上放置在每层楼，另一个放在桌子上。正在寻求SGER资助，以确定更全面的后续研究所需的有效的损害定位和量化方法。随后将提出的后续研究将开发一个集成系统，该系统由经过验证的损伤识别技术组成，具有在线后果预测，然后在必要时在线驱动控制系统，以将结构的性能保持在可接受的水平。这项探索性工作的成功成果也将成为结构诊断和健康监测直接实际实施的非常有用的工具。本研究的智力价值在于开发了经过实验验证的损伤识别（检测、定位和量化）方案，用于全面的结构健康监测。本研究的广泛影响将发生在两个层面：研究生和本科生的教育和培训，以及结构健康监测研究人员和从业人员急需的工具和数据的可用性。研究生和一名本科生将参与这个项目。本科学生将通过REU计划获得支持。她/他将通过弗吉尼亚理工大学工程多样性增强中心（CEED）从可用的非裔美国人、西班牙裔、美洲印第安人和女性学生中选出。在本研究中产生的信息和数据将通过课堂教学、网站创建、会议演讲以及在档案期刊和会议论文集上发表来传播给其他学生、研究人员和从业者。

AbstractThere is an increasing interest in the vibration data-based detection, localization, andquantification of damage in civil engineering structures exposed to earthquake, wind, and otherloads. Early motivation for this was provided by the discovery of hidden damages in steelstructures caused by the 1994 Northridge and 1995 Great Hanshin (Kobe) Earthquakes, and alsoby the fact that structural element are usually inaccessible after construction. Several analyticalmethods have been proposed in the literature and still more are being developed for this.Currently simulated data are being used to check the effectiveness of these methods; however,their validation and comparison on actual experimentally obtained data is yet to be done.In this exploratory project, it is proposed to experimentally study and validate theeffectives of the autoregressive with exogenous input model, wavelet transform, eigensystemrealization algorithms (ERA), and subspace identification (SSI) approaches for comprehensivemodal and damage identification of civil engineering structures. For damage localization andquantification the flexibility and rotational flexibility matrices with and without and damage locatingvectors will be used. Further analytical improvements in these methods will also be made formore reliable damage quantification. The tests will be conducted on a scaled three-story, ninedegrees-of-freedom structure model prepared such that its stiffness properties can be changedby planned changes in one or more braces, by sudden removal of one or two braces in real-timeusing magneto-rheological dampers, and by loosening the connecting hardware at few selectedjoints. The major cost of preparing this changeable scaled model will be borne by the PIsinstitution. The scaled model will be excited at its base by the 5' by 5' shaking table at VirginiaTech. The model would also have the capability of being excited by shakers installed at any ofthe three levels. The dynamic measurements of the acceleration response will be made usingten accelerometers, three placed at each floor level in two orthogonal directions plus one on thetable.The SGER funding is being sought to identify validated damage localization andquantification methods which will be needed in a more comprehensive follow-up study. Thisfollow-up study to be proposed later will develop an integrated system consisting of the validateddamage identification techniques with online consequence prognosis followed by onlineactuation, if necessary, of a control system to maintain the performance of the structure at anacceptable level. The successful outcome of this exploratory work will also be a very useful toolfor direct practical implementation in structural diagnostics and health monitoring.The Intellectual merit of this study is in the development of experimentally verifieddamage identification (detection, localization, and quantification) schemes for a comprehensivestructural health monitoring.Broader impacts of this research will occur at two levels: theeducation and training of graduate and undergraduate students and the availability of a muchneeded tool and data to researchers and practitioners involved in structural health monitoring. Agraduate student and an undergraduate student will work on this project. The undergraduatestudent will be supported through the REU program. She/he will be selected through the Centerfor the Enhancement of Engineering Diversity (CEED) at Virginia Tech from the pool of availableAfrican American, Hispanic, American Indian, and women students. The dissemination ofinformation and data generated in this research to other students, researchers and practitionerswill be done through class room teaching, website creation, presentations in conferences, andpublications in archival journals and conference proceedings.