Research On Damage Identification Of Beam Structure Based On Pattern Recognition Technology

The structure is damaged due to environmental actions,material aging,load effects and other factors in the course of service.Evaluating the structural damage condition reasonably,therefore,is necessary in order to avoid bearing capacity decline or its failure owing to structural damage accumulation.Generally,natural frequency is global to the whole structure,based on which the structural damage identification method has a broad application prospect in practical engineering for its convenient testing and higher accuracy.The pattern identification,based on the perturbation theory,was applied to study single and multiple damage of beam structure in this paper,and a damage pattern recognition technology which can effectively identify the damage location and degree is proposed.Firstly,the effect of structural local mass and stiffness changes on the natural frequency is investigated using the perturbation theory.The study shows that the deviation squared natural frequency is proportional to squared displacement mode value at the location of mass change when the local mass of structure changes,and the deviation in squared natural frequency is proportional to squared curvature mode value at the location of stiffness change when the local stiffness of structure changes.The mass is considered to be constant,while the stiffness is variable when the structure is damaged,and the relationship among the deviation in squared natural frequency and that in stiffness,and the curvature mode is the theoretical basis of the damage identification.Then the pattern recognition technology was employed to the damage identification to study the damage location and degree,and the damage identification process can be roughly divided into four steps.The first step is to establish an accurate finite element model,in which the displacement modes are extracted and the curvature modes are calculated using the central difference method.The second step is to establish damage patterns according to the functional relationship among the deviation in squared natural frequency and that in stiffness,and the curvature modes.The third step is to obtain the natural frequency of damaged structure and establish the feature vector.The fourth step is to measure the Euclidean distance between the feature vector and the pattern vector by Euclidean distance function,the structural damage location and degree are accurately identified by minimizing the Euclidean distance function.At least r+1 modes are required to identify the damage location and degree when there are r damages(e.g.,at least 2 modes can assess the single damage).Finally,the identification error of the damage pattern recognition technology was expounded.It is applicable to the stage of weak nonlinear function relationship(regarded as a linear relationship)between the deviation in squared natural frequency,the structural damage location and degree are effectively identified at the weak nonlinear stage.In addition,the identification result of damage degree is more sensitive to the size of element in structural finite model than identification of damage location,the damage degree identification result is roughly inversely related to the size of the element,and the damage location identification result is accurate with different element size.