基于移动主成分分析特征的智能损伤诊断方法

Intelligent damage identification method with feature based on moving principal component analysis

文章在移动主成分分析(moving principal component analysis,MPCA)基础上,提出一种优化的MPCA特征——特征向量差方向角(directional angle of eigenvector variation,DAEV),并将其作为机器学习的输入建立损伤识别模型。利用双跨连续梁的仿真应变监测数据验证了以DAEV建立机器学习模型诊断结构损伤的有效性。结果表明,与MPCA特征向量相比,DAEV能更好地表征桥梁状态的变化,以DAEV为输入的机器学习模型损伤识别能力更强;对于早期损伤,以DAEV特征为输入的模型识别准确率比以MPCA特征向量为输入的模型高38%~79%。

Based on the moving principal component analysis(MPCA),this paper proposes an enhanced feature of directional angle of eigenvector variation(DAEV),which is subsequently used as the input of machine learning algorithms to establish damage identification models.The methods are verified by using the simulated monitoring strain data of a continuous double-span beam.The results show that compared with the feature of eigenvector derived from MPCA,DAEV is more sensitive to the change of bridge states,and the machine learning models with DAEV as the input perform better in damage identification.With regard to early damage identification,the accuracy of the models with DAEV as the input is 38%79%higher than that of the models with the MPCA eigenvector as the input.