Research On Damage Identification Of Beam Structure Based On Pattern Recognition

Under the adverse effects of various loads and sudden disasters,the beam structures will have a certain degree of damage during the operation processes.With the continuous accumulation of damage,the safety performance of structures is constantly decreasing and their carrying capacities are continuously reduced.This will not only bring hidden dangers for the safe service of structures,but also even cause major engineering accidents.In order to grasp the state of structural damage timely and provide scientific basis for reasonable and effective protective strengthening measures,accurate and reliable damage identification methods need to be proposed.This paper regards beam structures as research objects.From the viewpoint of energy,the influences of structural local stiffness and quality changes on their dynamic characteristics are deduced.Based on the theory,combining pattern recognition with structural damage identification,this paper identifies the damage location of structures by constructing the proper eigenvectors and pattern vectors and calculating their distances with the distance classifier.Thus,the corresponding damage pattern recognition methods are established.The major contents and innovative points of this paper are as follow:For beam structures,the relations between local stiffness and quality changes with the relative changes of natural frequencies are derived from the viewpoint of energy.It is proved that when the structure produces single stiffness damage,the relative changes of its natural frequencies are positively correlated with the squared values of the corresponding curvature modes which are normalized by stiffness;when the structure produces single quality damage,the relative changes of its natural frequencies are positively correlated with the squared values of the corresponding displacement modes which are normalized by quality.Based on the relationship between the structural natural frequency and its stiffness damage,the ratio of the relative changes of adjacent two natural frequencies is selected as the characteristic parameters which are used to construct structural damage pattern vectors and the actual damage eigenvectors.By measuring the similarities between the eigenvectors and the pattern vectors with a distance classifier,the structural damage is identified and a structural stiffness damage pattern recognition method is formed.Through two examples and the cantilever stiffness weakening test,this method is verified and the results show a certain precision.According to the functional relation between the displacement modes and the relative changes of the natural frequencies when structure produces single quality damage,the structural damage pattern vectors are built by displacement modes and the actual damage eigenvectors are constructed by the relative changes of natural frequencies.By defining the criterion of maximum normalized distance,the location of structural damage is realized and a structural quality damage pattern recognition method is proposed.Combining finite element numerical simulation with experimental verification,the reliability of this method is verified and the factors that affect the accuracy of the recognition results are analyzed.