Research On Structural Damage Identification Based On EEMD

Since the structural vibration responses under ambient excitation are always nonlinear and non-stationary signals,in this paper,Ensemble Empirical Mode Decomposition(EEMD),which is adaptive to deal with nonlinear and non-stationary signal,is employeed to decompose the vibration response signal,and combining with AR model,energy entropy and statistical control charts to identify structure damage,the main contents are as follows:(1)The methods of structural damage identification based on vibration response analysis are reviewed,and the applicability and existing problems of these methods are also discussed.(2)Some basic theories used in this paper are introduced,including principle and procedure of EEMD,EEMD energy entropy,AR model,statistical control chart.(3)A structural damage identification method based on EEMD and residual error of AR model is presented.Firstly,a series of intrinsic mode function(IMF)are abtained by decomposing the acceleration response signals with EEMD,the IMF which contains the low frequency component of undamaged structural response data is selected to establish AR prediction reference model,secondly,the IMF which contains the low frequency component of unknown condition structural response data is input into the prediction reference model,residual error of unknown condition is gained,at last,the varance ratio of the calculated residual error(unknown condition)and the residual error of the AR model(undamaged condition)is taken as damage index,which is used to identify the condition of the structure.The effectiveness of the presented method is verified by numerical simulation and model experiment of offshore platform structure.(4)The damage identification method based on EEMD energy entropy and statistical control chart is proposed in chapter 4,the EEMD energy entropy is taken as the damage characteristic parameter.First,each node EEMD energy entropy is used to construct the feature vector;Second,the Mahalanobis distances between each feature vector and the entire sample which is composed of multiple sets of feature vectors are calculated;Finally,the Mahalanobis distances are used as the sample statistic to establish the X-bar statistical control chart,and the principle of the control chart is emploeed to alarm the damage.Furthermore,the damage location can be preliminarily juded by the change of node energy entropy.The effectiveness of the presented method is verified by numerical simulation and experiment of offshore platform structure.