聚类分析技术在航天复合材料粘接质量敲击检测中的应用  

The Applied of Clustering Analysis Technology on Coin-Tap Test System of Aerospace Composite Material

作　　者：刘钊 李艳军[2] 张祥林 许振腾 樊喜刚 

机构地区：[1]北京星航机电装备有限公司,北京100074 [2]南京航空航天大学民航学院,江苏南京210016

出　　处：《飞机设计》2017年第2期40-43,共4页Aircraft Design

摘　　要：粘接类复合材料在航天领域得到应用广泛,但是粘接质量的判断一直是航天复合材料发展的一个重要课题。敲击检测在应用于复合材料粘接质量的检测时,受复合材料自身特性的影响较小,因此受到越来越多的关注;但是,敲击检测方法的数据处理问题一直没有得到解决,因此这也制约了其进一步发展。本文以敲击检测的数据处理为出发点,提出了聚类分析技术用以解决这一难题。经过选取数据样本进行验证,发现基于自组织竞争神经网络的聚类分析技术,可以很好地解决敲击检测的数据处理问题。Adhesive composite materials are widely used in the aerospace, but to judge the quality of bonding has been an important topic aerospace composite materials development. When the coin-tap test used in bonded composite material quality testing, it is much less effected by the individual char- acteristics of the composite material, and therefore more and more attentions were attracted. Howev- er ,data processing problem of coin-tap test has not been resolved, so its further development also been restricted. Based on the test data processing as a start point, a clustering analysis technique was proposed in this paper to solve the problem. By selecting the data sample for verification, it is found that clustering analysis based on selforganizing competitive neural network technology is a good way to solve the problem of data processing of coin-tap test.

关 键 词：聚类分析 自组织神经网络 复合材料 敲击检测 数据分析 

分 类 号：V414.1[航空宇航科学与技术—航空宇航推进理论与工程]