A local specific stiffness identification method based on a multi-scale "weak" formulation

A local specific stiffness identification method based on a multi-scale "weak" formulation
复制标题

基于多尺度“弱”公式的局部比刚度识别方法

DOI:
10.1016/j.ymssp.2020.106650
复制
发表时间:
2020-06-01
影响因子:
8.4
通讯作者:
Wu, Yipeng
Wu, Yipeng
中科院分区:
工程技术1区
文献类型:
--
作者:
Zhang, Chao;Ji, Hongli;Wu, Yipeng

文献摘要

被引文献

相似文献

提出了一种基于多尺度“弱”列式的局部比刚度识别方法。基于局部运动方程,结构的比刚度可以从其测量的振动位移中提取,它可以进一步被用作结构内部发生损伤的指示。然而,通过有限差分格式估计测量位移的高阶导数容易受到测量噪声的影响。为了解决这个问题,权重函数被用作扫描窗口,它转换为一个“逐点”的识别策略,以“区域逐区域”的范例。通过适当的权函数参数设置,最终得到的局部比刚度的数学表达式避免了直接计算高阶导数,从而提高了噪声测量条件下的辨识精度。作为一个概念验证的例子,铝悬臂梁进行了研究,以验证所提出的方法。研究了测量间隔、标度因子和被测振动位移导数阶数等关键参数对测量结果的影响。所提出的方法的有效性进行了数值模拟和实验验证,使用一个阶梯形梁。(C)2020由Elsevier Ltd.出版
This paper presents a novel local specific stiffness identification method based on a multi-scale "weak" formulation. Based on the local equation of motion, the specific stiffness of a structure can be extracted from its measured vibration displacement, which can further be used as an indicator of damage occurrence inside the structure. However, the estimation of the high order derivative of the measured displacement via a finite difference scheme is prone to the measurement noise. To tackle this problem, a weight function is utilized as a scanning window, which transforms a "point-by-point" identification strategy to a "region-by-region" paradigm. Through a proper parameter setting of the weight function, the final mathematical expression of the local specific stiffness allows avoiding the direct calculation of the high order derivative, thus improving the identification accuracy under noisy measurement conditions. As a proof-of-concept example, an aluminum cantilever beam is investigated for validating the proposed method. The influences of key parameters, such as measurement interval, scale factor and derivative order of the measured vibration displacement, are investigated. The effectiveness of the proposed method is demonstrated numerically and validated experimentally using a step-shaped beam. (C) 2020 Published by Elsevier Ltd.