Study On The Structural Damage Identification Based On IMF Component Hilbert Spectrum

In view of the fact that the structural vibration response signals under ambient excitation are generally nonlinear and non-stationary.In this paper,the acceleration response signals are analyzed by Hilbert-Huang transform(HHT)with good adaptive ability,the intrinsic mode functions(IMFs)are extracted.The effective IMFs which are sensitive to the structural damage are selected to be analyzed by Hilbert transform to obtain the Hilbert spectrum.The damage characteristic indexes are constructed with the Hilbert spectrum to identify the structural damage.Numerical simulation and model tests of offshore platform structures under stationary and non-stationary excitation verify the effectiveness of the method.The main contents are as follows:(1)In this paper,the state of the art of structural damage identification at home and abroad is reviewed,with an emphasis on the method of structural damage identification based on vibration response.(2)Some basic theories used in this paper are briefly described,including Hilbert-Huang transform,Modified Ensemble Empirical Mode Decomposition(MEEMD)and support vector machine.(3)Research on structural damage alarming based on the Hilbert spectrum of effective IMF component.Firstly,the acceleration response signal is decomposed into a series of IMF(intrinsic mode function)by MEEMD.The effective IMFs with the frequency consistent with the natural vibration frequency of the structure are selected by the analysis of amplitude-frequency characteristic of the IMFs.The effective IMFs are analyzed with Hilbert transform to get the Hilbert spectrum.The correlation coefficient,the relative variation and the confidence level of the Hilbert time spectrum before and after the damage are calculated and used as the damage alarming indexes to carry out the structural damage alarming.The influences of different noises are considered at the same time.(4)Research on structural damage identification based on the energy of Hilbert spectrum of IMF component.Firstly,a series of IMF components are obtained by MEEMD decomposition of the acceleration response signal,and the Hilbert spectrum isobtained by the analysis of IMF with Hilbert transformation.The feature vector is constructed with the energy of Hilbert spectrum with the cumulative energy contribution rate greater than 98%.The damage location and damage degree are evaluated with the classification algorithm and the regression algorithm of support vector machine respectively.Considering the cost-benefit,the optimal combination of different number and different location of measuring point are also studied.