Parameters Identification Of Time Varying Structures Based On Hilbert-Huang Transform

Methods of parameters and damage identification of time varying structures and parameters identification of original structures by means of structures under control based on Hilbert-Huang Transform (HHT) are proposed in the thesis.The emphasis and innovation of this thesis are in the use of HHT and mathematical programming to identify the parameters of time varying multi-degrees of freedom (MDOF) system. In the analysis, the time varying instantaneous frequency of each mode of the structure is first obtained by the methods of HHT, then the stiffness or mass parameters at given time are selected to be design variables and are identified by minimizing the summation of the square of errors between the instantaneous frequencies and computed frequencies. This minimization procedure is repeated for all given instants under consideration in order to obtain time-varying stiffness or mass values during the duration. Numerical results show that the proposed method is capable of identifying time varying parameters of multi-degrees of freedom systems by means of response time histories of the systems subjected to harmonic excitation.In the thesis, HHT method is also employed to identify the onset of damage of time varying MDOF system. The signal is processed by HHT to obtain the instantaneous frequency of each mode, which is further used as the damage indicator to identify the onset of damage. Numerical results show that the HHT method is capable of distinguishing variation of response time histories of time varying multi-degrees of freedom system caused by suddenly change in excitation or loss of structural stiffness. Thereby the use of HHT can identify whether damage occurs and the instant based on the response time histories of the system.If the frequency of excitation is close to one of the natural frequencies of the structure, the window function can not filter the effect of steady response effectively. In this situation, the signal processed by the window function cannot be decomposed into the structure modes by Empirical Mode Decomposition (EMD) effectively. The sub-structure is then appended to original structure to create a. new structure named controlled structure or structure under control, by which the frequencies of the new structure will depart from the excitation frequency. Therefore, parameters of controlled structure can be identified and the parameters of the original structure consequently. Numerical results show that parameters of original structures can be identified effectively by means of response time history of the structure under control.