基于l1正则化无迹卡尔曼滤波的结构损伤方法

发布时间：2018-06-16 00:40

  本文选题：无迹卡尔曼滤波 + l范数正则化　； 参考：《工程力学》2017年08期

【摘要】：采用传统卡尔曼滤波类算法对结构进行损伤识别时,损伤识别反问题的不适定性使得识别结果易受噪声干扰,甚至算法不收敛。为此,该文提出了一种结合l1范数正则化的无迹卡尔曼滤波损伤识别算法。根据结构出现局部损伤时其损伤参数分布具有稀疏性的特点,通过伪测量方法,将l1范数正则化引入到无迹卡尔曼滤波框架中,在改善反问题求解不适定性的同时,能有效地提高结构局部损伤识别能力。梁、桁架结构的数值分析与实验研究表明,该文方法可以对损伤的位置与程度进行准确识别,且具有良好的鲁棒性。
[Abstract]:When the traditional Kalman filtering algorithm is used to identify the damage of the structure, the ill-posed problem of the inverse problem of the damage identification makes the result of the identification easy to be disturbed by noise, and even the algorithm does not converge. In this paper, an unscented Kalman filter damage detection algorithm based on L 1 norm regularization is proposed. According to the sparse property of the damage parameter distribution when the local damage occurs in the structure, the L 1 norm regularization is introduced into the unscented Kalman filter framework by pseudo-measurement method, which can improve the ill-posed solution of the inverse problem at the same time. It can effectively improve the local damage identification ability of the structure. The numerical analysis and experimental study of beam and truss structures show that the proposed method can accurately identify the location and degree of damage and has good robustness.
【作者单位】： 南昌大学建筑工程学院;
【基金】：国家自然科学基金项目(51268045,51469016) 教育部高等学校博士点基金项目(20123601120011) 水沙科学与水利水电工程国家重点实验室开放研究基金项目(sklhse-2014-C-03)
【分类号】：TU317
，

本文编号：2024407