模糊聚类法在试验模态参数识别分析中的应用  

Application of Fuzzy Clustering Theory in Experimental Modal Parameter Identification Analysis

作　　者：姜金辉[1] 陈国平[1] 张方[1] 张茂争[1] 

机构地区：[1]南京航空航天大学航空宇航学院,南京210016

出　　处：《南京航空航天大学学报》2009年第3期344-347,共4页Journal of Nanjing University of Aeronautics & Astronautics

摘　　要：为了搜索系统的真实极点,将模糊聚类理论引入试验模态参数分析。根据极点之间内在的亲疏关系,将其合理地聚为若干类,搜索每一类的最佳极点,即系统的真实极点,在稳态图中自动完成真实极点的选择。该方法能够避免人为因素造成的识别误差,提高识别结果的可信度。在正交多项式曲线拟合和多参考点最小二乘复频域识别系统中引入该方法,实现了稳态图中极点的自动选择,提供了真实极点选择的一种新思路。通过仿真和实验证明了模糊聚类法对极点选择的可行性和稳定性。Fuzzy clustering theory is introduced into the experimental modal parameter identification anal- ysis to search the real systematical poles. Based on the correlation of poles, poles are categorized into certain categories to search the optical pole in each class, called the real pole of the system, so that actu- al poles can be automatically selected in the built stabilization diagram. The method can avoid the identi- fication error caused by man carelessness and enhance the identification reliability. In the orthogonal polynomial curve-fitting and poly-reference implementation of the least-squares complex frequency-do- main, the method can automatiically find actual poles in the built stabilization diagram and provide a new way to select actual poles. The feasibility and the reliability of the fuzzy clustering technology for the pole selection are proved by the experiment and the simulation.

关 键 词：模态参数识别 模糊聚类 稳态图 正交多项式曲线拟合 多参考点最小二乘复频域法 

分 类 号：TB52[理学—物理] TH113[理学—声学]