陈帅,黄念,李建乐,张佳奇,李腾腾,徐浩*,武湛君.基于分布式光纤传感信号的结构损伤识别主成分分析方法[J].实验力学,2022,37(6):838~846 |
基于分布式光纤传感信号的结构损伤识别主成分分析方法 |
Principal component analysis for structural damage identification based on distributed optical fiber sensing signals |
投稿时间:2021-07-07 修订日期:2021-09-08 |
DOI:10.7520/1001-4888-21-166 |
中文关键词: 结构健康监测 分布式光纤传感器 损伤识别 主成分分析 |
英文关键词:structural health monitoring distributed optical fiber damage identification principal component analysis |
基金项目:国家重点研发计划(2018YFA0702800);国家自然科学基金面上项目(12072056);国防基础科研项目(XXXX2018204B011) |
|
摘要点击次数: 547 |
全文下载次数: 109 |
中文摘要: |
航空航天复合材料结构在服役时需要对其健康状况进行监测。基于背向瑞利散射的分布式光纤传感器因其便于埋入、抗干扰能力强等优点被广泛应用于结构健康监测领域,但如何从光纤数据中识别其对应的损伤类型和位置一直是研究人员关注的问题。本文针对分布式光纤传感器应变测点密集、噪声影响显著等信号特征,提出了一种改进的主成分分析(Modified Principal Component Analysis, MPCA)方法,通过采用局部测点分量主成分分析(PCA)模型,并结合霍特林统计阈值进行损伤自动化识别。为了验证方法的可靠性,开展了预制裂纹损伤的铝合金板拉伸实验,通过表面粘贴的分布式光纤传感器精确测量了结构应变变化特征,光纤传感器的应变监测数据为PCA提供了有效模型数据。使用多级载荷下无损实验数据进行PCA建模,并采用预制裂纹后的含损应变数据进行了方法验证。结果表明,MPCA方法的训练模型可精确识别铝合金板内的损伤位置,并可为结构健康状态的在线评估提供有力支撑。 |
英文摘要: |
The health status of aerospace composite structures need to be monitored in service. Distributed optical fiber sensors based on back Rayleigh scattering are widely used in the field of structural health monitoring because of their advantages of easy embedding and strong anti-interference ability. However, how to identify their corresponding damage types and locations from optical fiber data has always been a concern of researchers. In this paper, a modified principal component analysis (MPCA) method is proposed according to the signal characteristics of distributed optical fiber sensor, such as dense strain measuring points and significant noise effect. By using the PCA model for local measurement regions and Hotelling statistical threshold, automatic damage identification is enabled. To verify the reliability of the method, the tensile loading test of an aluminum alloy plate with a crack damage was carried out. The strain variation characteristics of the structure were accurately measured by the distributed optical fiber sensor pasted on structural surface, and effective model data was provided for the principal component analysis. The PCA model is established by using the data of non-destructive test under multi-stage load, and the method was verified by using the data of measured strains subjected to the prefabricated crack. The results show that the training model can accurately identify the damage location in the aluminum alloy plate, providing technical support for on-line assessment of structural health. |
查看全文 下载PDF阅读器 |
关闭 |