Structural damage recognition by grouped data based on Principal Component Analysis theory

Damage always reduces structural stiffness, and changes the dynamic responses. Generally, the changes are too slight to reveal the damage information directly. Therefore, more explicit and efficient methods are needed to reveal the underlying damage information. This study recognizes the damage existence, quantification, and location by classifying structural responses data into groups. Firstly, four damaged scenarios are designed to be investigated. Secondly, structural responses and their statistical features are explored for damage recognition. Thirdly, the damage existence is recognized on grouped data based on the acceleration responses according to the general Principal Component Analysis (PCA) theory. Fourthly, the damage existence, quantification, and location are recognized by grouped data based on the model data according to the coordinate rotations in PCA. Finally, the damage information is recognized by modal shape data with two different noise levels to show the robustness of the method.