基于自回归模型和核主成分分析的结构损伤识别方法

Structural Damage Detection Based on Autoregressive Model and Kernel Principal Component Analysis

针对实际工程中采集的结构响应信号存在环境因素干扰的问题,提出一种基于自回归模型和核主成分分析的结构损伤识别方法。主要思想是建立结构在环境激励下的响应信号自回归模型,通过核主成分分析提取隐藏在自回归系数中的环境因素影响,从而提高自回归模型在诊断损伤时的准确度。最后以佛山平胜大桥的加速度数据验证该方法的有效性。

A new damage identification method based on autoregressive model and kernel principal component analysis is proposed for elimination of environmental disturbance. The main idea is to establish a regression model from structure response signal under environmental excitation, then adopt principal component analysis to extract environmental factors hidden in the regression coefficients. Thereby, the accuracy of autoregressive model in damage diagnosis is improved. Finally, the acceleration data collected from Pingsheng Bridge was used to verify the viability of this method.