基于LTSA和RVM的复合材料损伤预测分析

发布时间：2018-11-28 16:26

【摘要】：提出一种基于局部切空间排序(local tangent space alignment,简称LTSA)和相关向量机(relevance uector machine,简称RVM)相结合的复合材料结构损伤演化与预测模型。针对复合材料结构损伤特性,采用疲劳振动试验进行结构损伤预测研究。首先,采用总体平均经验模态分解(ensemble empirical mode decomposition,简称EEMD)方法对多传感器采集的复合材料结构健康信息进行自适应分解,得到不同传感器下的多个本征模态分量(intrinsic mode function,简称IMF),并对IMF进行希尔伯特(Hilbert)变换,得到相应的Hilbert边际谱能量作为各传感器的特征信息;然后,采用LTSA进行多特征降维融合得到特征能量,对降维融合后得到特征能量采用距离形态相似度方法定义结构健康指数;最后,将结构健康指数作为建模数据,创建RVM预测模型,并通过预测结构健康指数完成复合材料结构损伤预测研究。验证结果表明,该模型可有效地对复合材料结构损伤进行预测。
[Abstract]:A damage evolution and prediction model for composite structures based on local tangent space sorting (LTSA) and correlation vector machine (RVM) is proposed. According to the damage characteristics of composite structure, fatigue vibration test was used to predict the damage of composite structure. Firstly, the average empirical mode decomposition (ensemble empirical mode decomposition,) method is used to decompose the health information of composite structures collected by multiple sensors, and the multiple intrinsic modal components (intrinsic mode function,) of different sensors are obtained. IMF), and Hilbert (Hilbert) transform for IMF are used to obtain the corresponding marginal spectrum energy of Hilbert as the characteristic information of each sensor. Then, the feature energy is obtained by multi-feature dimensionality reduction fusion with LTSA, and the structural health index is defined by the distance morphological similarity method. Finally, the structural health index is taken as the modeling data, and the RVM prediction model is established, and the damage prediction of composite structures is completed by predicting the structural health index. The results show that the model can effectively predict the damage of composite structures.
【作者单位】：沈阳航空航天大学自动化学院;沈阳飞机设计研究所;
【基金】：航空科学基金资助项目(20153354005) 国防预研资助项目(A0520110023) 国防基础科研资助项目(Z052012B002) 辽宁省自然科学基金资助项目(2014024003)
【分类号】：TB33

【相似文献】