Study On The Structural Damage Alarming Based On Singular Spectrum Analysis

The structural damage identification method based on vibration will be affected by measurement noise and variation of environmental factors in practical application,which reduces the accuracy of structural damage identification method.Singular Spectrum Analysis(SSA)is a kind of non-parametric frequency-spectrum evaluation technique,which is suitable for analyzing time series collected by dynamic systems without prior knowledge.In this paper,the measurement signal is analyzed with SSA to obtain the components which are less affected by the measurement noise and contain more damage information,the damage index(DI)constructed with the AR model residuals of the components is used to identify the damage under the influence of noise.The mode frequency series is analyzed by SSA,the trend component is identified and removed,and the rest of the components are selected to reconstruct the frequency series,and the DI constructed with Euclidean distance is used to alarm the structural damage under the influence of environmental factors.The main contents are as follows:(1)In this paper,the research background and significance of structural damage identification is introduced,the classification of structural damage identification methods is reviewed,with an emphasis on the method of structural damage identification based on vibration.And the two main problems in structural damage identification are pointed out: noise and environmental factors.(2)Some basic theories used in this paper are introduced,such as SSA,AR model,peak spectral method and single-valued control chart.(3)Research on structural damage alarming based on SSA and AR model residuals under the influence of noise.Firstly,the ratio of the singular value of each order is obtained by analyzing the measured acceleration response signal with SSA,the components corresponding to the singular value with relatively large change before and after the damage of the structure,are selected to reconstruct the signal,of which the noise is removed.The AR model is used to fit the noise free acceleration response signal,and the parameters of AR model are obtained.The damage characteristic index constructed by using the AR model residuals before and after structural damage is used to identify the structural damage.Numerical simulation and shaking table model tests of five-story steel frame structure under white noise excitation and Qian'an seismic wave excitation verify the effectiveness of the damage alarming method,and the effect ofdifferent degrees of noise are also considered.(4)Research on structural damage alarming based on SSA and the modal frequency under the influence of environmental factors.Firstly,the measured acceleration response signals are segmented,and the first order modal frequency of each segment is obtained by using the spectral peak method,and arranged into the first order modal frequency sequences in the order of the segments.The singular value and the left and right singular vector of the first order modal frequency sequence are obtained by analyzing the first order modal frequency sequences with SSA.The time history curve of the right singular vector is used to determine whether it is a trend component or not.The components except the trend component are selected to construct the new frequency sequences removing the influence of environmental factors.It is considered that the trend components are the influence of environmental factors on the first order modal frequency series.The damage characteristic indexes constructed with the Euclidean distance between the reconstructed frequency series before and after the damage are taken as the samples of the single value control charts to realize the structural damage alarming.Numerical simulation and shaking table model tests of five-story steel frame structure under white noise excitation verify the effectiveness of the damage alarming method,and the effect of different degrees of noise are also considered.