Structural Health Monitoring ( IF 5.7 ) Pub Date : 2021-06-04 , DOI: 10.1177/14759217211021938 Rongrong Hou 1 , Xiaoyou Wang 2 , Yong Xia 2, 3
The l1 regularization technique has been developed for damage detection by utilizing the sparsity feature of structural damage. However, the sensitivity matrix in the damage identification exhibits a strong correlation structure, which does not suffice the independency criteria of the l1 regularization technique. This study employs the elastic net method to solve the problem by combining the l1 and l2 regularization techniques. Moreover, the proposed method enables the grouped structural damage being identified simultaneously, whereas the l1 regularization cannot. A numerical cantilever beam and an experimental three-story frame are utilized to demonstrate the effectiveness of the proposed method. The results showed that the proposed method is able to accurately locate and quantify the single and multiple damages, even when the number of measurement data is much less than the number of elements. In particular, the present elastic net technique can detect the grouped damaged elements accurately, whilst the l1 regularization method cannot.
中文翻译:
使用模态数据通过弹性网法进行稀疏损伤检测
的升1正则化技术已被利用的结构损伤的稀疏特征开发了用于损伤检测。然而,损伤识别中的灵敏度矩阵表现出很强的相关性结构,不足以满足l 1正则化技术的独立性标准。本研究采用弹性网法,结合l 1和l 2正则化技术来解决该问题。此外,所提出的方法能够同时识别分组的结构损伤,而l 1正则化不能。数值悬臂梁和实验三层框架被用来证明所提出方法的有效性。结果表明,即使在测量数据的数量远小于单元数量的情况下,所提出的方法也能够准确定位和量化单个和多个损伤。特别是目前的弹性网技术可以准确地检测出分组的损坏单元,而l 1正则化方法则不能。