无人机结构小损伤程度识别研究  

Research of Small Damage Severity Identification for SUAV

作　　者：李春锦[1] 闫云聚[1] 

机构地区：[1]西北工业大学力学与土木建筑学院,陕西西安710129

出　　处：《机械设计与制造》2014年第9期191-194,共4页Machinery Design & Manufacture

基　　金：国家自然科学基金资助项目(50875212);教育部博士点基金资助项目(200806990019);2012年博士点基金资助课题(20126102130004)

摘　　要：小型无人机结构的局部小损伤识别对于其及时检修有重要意义。针对当前仅能获得有限试验样本而造成识别精度不高的问题,提出了一种将小波能量谱和支持向量机(SVM)相结合的新方法,对无人机结构进行局部小损伤程度识别研究。利用ANSYS软件构建无损和有损仿真模型,于不同损伤工况下对其进行瞬态分析并将响应用小波包分解,在SVM良好小样本泛化能力基础上建立了准确的损伤识别学习机,并通过网格寻优对参数进行优化处理,最后得出识别结果。为了验证识别结果的可信性,同时取用另一组核函数和BP神经网络方法进行对比试验。结果表明,小样本条件下,用小波包能量谱和SVM结合来识别无人机的小损伤程度的方法是可靠有效的。The local small damage identification of SUAV（Small Unmanned Aerial Vehicle） has important significance for the timely maintenance. For the problem of limited samples and low accuracy, a new approach combining wavelet energy spectrum and support vector machine （SVM） was proposed to identify the local small damage severity of SUAV. The intact and damaged simulation model was constructed by the ANSYS. After transient analysis and wavelet packet decomposition under different damage conditions, a damage identification model was established based on the perfect generalization ability for small sample of SVM, and the parameters were optimized through the grid optimization. At last the identification results were given. The comparisons with the other group of kernel function and BP neural network method were made in order to validate the credibility of the results. The simulation results show that the approach combining wavelet energy spectrum and SVM to identify the local small damage severity of SUAV is reliable and effective under the condition of limited samples.

关 键 词：小型无人机 损伤识别 小波能量谱 支持向量机 

分 类 号：TH16[机械工程—机械制造及自动化] V215.7[航空宇航科学与技术—航空宇航推进理论与工程]