• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文OA文献 >Seismic damage identification based on integrated artificial neural networks and wavelet transforms
【2h】

Seismic damage identification based on integrated artificial neural networks and wavelet transforms

机译：基于集成人工神经网络和小波变换的震害识别

页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

In recent years, Structural Health Monitoring (SHM) has been proposed and practiced for condition assessment of structures. SHM covers shortcomings of nondestructive tests and is comprised of a sensory system, data acquisition system, and damage identification system. In this study, numerical and experimental investigations are concentrated on the application of Artificial Neural Networks (ANNs) and Wavelet Transforms (WTs) for damage identification of civil engineering structures. As a major outcome of this research, three novel damage identification methods are developed. The first damage identification method enables the SHM systems to identify damage to cantilever structures through decomposition of mode shapes by integrating WTs and ANNs. The second damage identification method enables SHM systems to identify damage to cantilever structures via decomposition of response accelerations by means of WTs and ANNs. The third damage identification method takes advantage of only ANNs and enables the SHM systems to identify seismic-induced damage to concrete shear walls in real-time by measuring inter-storey drifts. In addition, a novel optimal strain gauge placement method for seismic health monitoring of structures is proposed. This method considers the seismicity of construction site and the importance level of structures. Results from the first method showed that when the imposed damage levels were severe, medium, and light, the proposed method could quantify them with less than 5%, 12%, and 16% errors, respectively. In addition, the second method quantified seismic-induced damage to the studied structure with an averaged error of 8%. Moreover, the third method classified damage levels of the studied concrete shear walls with a success rate of 91%. The proposed optimal strain gauge placement method reduced the number of required sensors for the studied structure from 206 sensors to 73 sensors. The obtained results demonstrated the feasibility, robustness, and efficiency of the proposed methods for damage identification of civil engineering structures.
机译：近年来，已经提出并实践了用于结构状况评估的结构健康监测（SHM）。 SHM涵盖了无损检测的缺点，并由感官系统，数据采集系统和损坏识别系统组成。在这项研究中，数值和实验研究集中在人工神经网络（ANN）和小波变换（WTs）在土木工程结构损伤识别中的应用。作为这项研究的主要成果，开发了三种新颖的损伤识别方法。第一种损伤识别方法使SHM系统能够通过整合WT和ANN，通过模态形状的分解来识别悬臂结构的损伤。第二种损伤识别方法使SHM系统能够通过WT和ANN分解响应加速度来识别悬臂结构的损伤。第三种损伤识别方法仅利用了人工神经网络，并使SHM系统能够通过测量层间漂移来实时识别地震对混凝土剪力墙的损伤。此外，提出了一种用于结构地震健康监测的新型最优应变仪布置方法。该方法考虑了施工现场的抗震性和结构的重要程度。第一种方法的结果表明，当所施加的损害级别为严重，中等和轻度时，所提出的方法可以量化它们的误差分别小于5％，12％和16％。另外，第二种方法量化了地震对研究结构的破坏，平均误差为8％。此外，第三种方法对研究的混凝土剪力墙的破坏程度进行了分类，成功率为91％。所提出的最佳应变仪放置方法将研究结构所需的传感器数量从206个减少到73个。获得的结果证明了所提出的用于土木工程结构的损伤识别的方法的可行性，鲁棒性和效率。

著录项

  • 作者

    Vafaei Mohammadreza;

  • 作者单位
  • 年度 2013
  • 总页数
  • 原文格式 PDF
  • 正文语种 en
  • 中图分类

相似文献

  • 外文文献
  • 中文文献
  • 专利

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号