...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Smart Materials & Structures >Finite element model updating of a small steel frame using neural networks
【24h】

Finite element model updating of a small steel frame using neural networks

机译：基于神经网络的小型钢框架有限元模型更新

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

This paper presents an experimental and analytical dynamic study of a small-scale steel frame. The experimental model was physically built and dynamically tested on a shaking table in a series of different configurations obtained from the original one by changing the mass and by causing structural damage. Finite element modelling and parameterization with physical meaning is iteratively tried for the original undamaged configuration. The finite element model is updated through a neural network, the natural frequencies of the model being the net input. The updating process is made more accurate and robust by using a regressive procedure, which constitutes an original contribution of this work. A novel simplified analytical model has been developed to evaluate the reduction of bending stiffness of the elements due to damage. The experimental results of the rest of the configurations have been used to validate both the updated finite element model and the analytical one. The statistical properties of the identified modal data are evaluated. From these, the statistical properties and a confidence interval for the estimated model parameters are obtained by using the Latin Hypercube sampling technique. The results obtained are successful: the updated model accurately reproduces the low modes identified experimentally for all configurations, and the statistical study of the transmission of errors yields a narrow confidence interval for all the identified parameters.
机译：本文介绍了小型钢框架的实验和分析动力学研究。通过改变质量并引起结构损坏，实验模型是在振动台上进行物理构建的，并在振动台上进行了动态测试，并具有一系列原始配置，这些配置从原始配置中获得。对于原始的未损坏配置，反复尝试了具有物理意义的有限元建模和参数化。有限元模型通过神经网络进行更新，模型的固有频率为净输入。通过使用回归过程使更新过程更加准确和健壮，这是这项工作的原始贡献。已开发出一种新颖的简化分析模型来评估由于损坏而导致的元件弯曲刚度的降低。其余配置的实验结果已用于验证更新的有限元模型和解析模型。评估已识别模态数据的统计属性。从这些数据中，可以使用Latin Hypercube采样技术获得估计模型参数的统计属性和置信区间。获得的结果是成功的：更新后的模型准确地再现了针对所有配置通过实验确定的低模态，并且对误差传递的统计研究得出了所有确定的参数的窄置信区间。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

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