Structural Damage Identification Of Nonlinear System Based On HHT

Due to nonlinear factors of structural material, connection and deformation,structural systems often exhibit a certain degree of nonlinearity. When structure issubjected to damage, damages may even increase its nonlinearity, and lead tonon-stationary dynamic responses of structures. Structural damages and nonlinearitywill cause the failure of the structural damage identification theory based on linearstructural behavior assumption. In this thesis, structural damage identification ofnonlinear system is studied based on empirical modal decomposition andHilbert-Huang Transform. Researches and results in this thesis are summarized asfollows.First, methods for nonlinear system identification are systematicallysummarized and analyzed. These methods include time-domain methods,frequency-domain methods, time and frequency domain method and methods basedon chaos theory. The method employed in this study is the Hilbert-Huang-Transformmethod in time and frequency domain.Ensemble Empirical Mode Decomposition is empployed in this research. Toremedy the shortage of shifting stopping criterion of the decomposition, which maylead to the inaccurate docomposing result and the non-self-adaptive decomposingprocedure, we combine using of the sum of the difference (SD parameter) and theestablishment condition of the Intristic Mode Function (IMF) for stopping criterion.Numerical examples demonstrate that the EEMD incorporated with the newstopping criterion increases signal decomposition accuracy and is adaptive.The influence of the end effect on the different characteristic time scales isstudied. High order IMFs have a short characteristic time scale, and low oder IMFshave a long characteristic time scale. It proved that the end effect is smaller for highorder IMFs than those for low oder IMFs. Therefore, higher-order IMFs are a betterchoice for damage detection when not all signal scales are to be considered.The improved EEMD method is applied to the2-D MDOF model undersimulated earthquake excitation. The instantaneous frequency (IF) of the first IMFof the output signal is utilized as the damage characteristic factor. It is proved thatthe time-domain information of the IFs can be used to identify the present and themoment of damage.A nonlinear element is incorporated into the model of athree-story frame structure, and section loss is introduced to simulate damage.Improved HHT is employed to identify danages in this structural model.Acceleration on the top story is analyzed, and the instantaneous frequency and instantaneous energy of the first IMF are used as the damage characteristic factors.The results show that choosing appropriate signals according to the symmetry of thestructures, the instantaneous frequency and instantaneous energy of the first IMFcan be used to identify not only the presence of damage but also the severity ofdamage.