@article{Zhou2025, 
author = {Xutao Zhou and Haixu Zhao and Yufeng Jiang and Shuqing Wang and Yu Liu},
title = {Iterative Damage Identification Method for Offshore Structures Based on K-Means Clustering Particle Swarm Algorithm},
year = {2025},
journal = {Periodical of Ocean University of China},
volume = {55},
number = {4},
pages = {134-147},
keywords = {K-means clustering particle swarm algorithm, damage identification, offshore wind turbine structure, iterative method},
url = {https://www.sciopen.com/article/10.16441/j.cnki.hdxb.20220500},
doi = {10.16441/j.cnki.hdxb.20220500},
abstract = {In order to solve the problem that the traditional intelligent optimization algorithm is easy to fall into the local optimal solution in structural damage identification, which leads to many misjudged units and poor identification accuracy in damage identification, an iterative structural damage identification method is proposed.To accelerate algorithm convergence and avoid failing into local optimum, a clustering particle swarm algorithm based on K-means is proposed. The results of traditional damage identification methods are improved by iteration, and the results of damage identification are updated iteratively to obtain accurate damage elements and damage severity. The three-legged offshore wind turbine structure is used as an example. Firstly, the effect of non-iterative methods on identifying structural damage is investigated with or without noise pollution. Secondly, the structural damage identification effect of the iterative damage identification methods under the influence of noise-free and noise pollution is studied. Thirdly, the convergence and stability of the proposed method are discussed. Finally, the proposed methods are verified by physical model experiment. The results show that the proposed iterative K-means clustering particle swarm algorithm could obtain accurate damage location and damage severity compared with the traditional structural damage identification methods, has good noise robustness, and has fewer iterations and a stable identification effect.}
}