Structural Damage Identification Based On The Vibration Transmissibility Function And Support Vector Machine

The structural damage identification methods based on the vibration had been developed rapidly in recent years.However,the effect of structural damage identification methods based on the vibration in the process of actual application was poor due to the difficulty of accurately measuring the excitation information,and measurement noise.Therefore,the structure vibration response were directly used to calculate the vibration transmissibility function,the structure damage characteristic vectors are constructed by wavelet packet analysis,principal component analysis,and the support vector machines was employed to identify the structure damage.The main research contents were as follows:(1)Reviewing the research developments of structural damage identification methods based on vibration at home and abroad;(2)Researching on the structural damage identification based on the wavelet packet energy of vibration transmissibility function and support vector machine.Firstly,the acceleration response signals of the structure were extracted,and the vibration transmissibility function between two adjacent nodes were calculated;Then,the amplitudes of vibration transmissibility function were analyzed with wavelet packet decomposition and reconstruction algorithm to obtain the energy of each band,which were used as the structure damage characteristic vectors;Finally,the classification and regression algorithms were employed to identify the damage location and degree of structure separately.The results of numerical simulation and shaking table model test of five layer three dimensional frame structure showed that the proposed method did not need the excitation information,and which could give relative good damage identification results in the conditions of less samples,and had a certain anti noise ability.(3)Researching on the structural damage identification based on the principal components of vibration transmissibility function and support vector machine.Firstly,the vibration transmissibility function were grouped to construct a matrix of vibration transmissibility function;Then,the principal component analysis was employed toprocess the matrix,and extract the first N order principal components,which were combined as a one-dimensional damage characteristic vector;Finally,the classification and regression algorithms were employed to identify the damage location and degree of structure separately.The results of numerical simulation and shaking table model test of five layer three dimensional frame structure showed that the proposed method did not need the excitation information,and had a certain anti noise ability.(4)By comparing the damage identification results of two proposed methods,it was found that the structural damage identification method based on principal component of vibration transmissibility function and support vector machine was superior than the structural damage identification method based on wavelet packet energy of vibration transmissibility function and support vector machine.