| Summary: | 博士 === 國立成功大學 === 航空太空工程學系碩博士班 === 92 === This thesis is concerned with the development of damage identification technique for a simple beam structure in order to monitor the structural health condition. The damaged structural stiffness and mass matrices were identified based on the specified loading and response signals. The finite element beam model was simplified to become a substructure model according to the Guyan dynamic reduction method and the dynamic reduction process. The orthogonal polynomial approach was subsequently used to extract the proposed distributed parameter model in question. The identified structural stiffness and mass matrices for the reduced model with or without damage were furthermore transformed to the corresponding global unreduced structural matrices by means of the back-propagation artificial neural network scheme and the binary coded genetic algorithm. Based on the identified damaged structural stiffness and mass matrices, the damage locations and the extent of damage of the beam structure were then determined.
In addition, the adaptive excitation method could also help to reduce the system parameters in a large structure. In this work, the adaptive excitation is also adopted to analyze the sensitivity when the structure has defect. This method belongs to the two stages identification process. By checking the sensitivities in different defect conditions, the defect location could be found first and the unknown system parameters will then be reduced to three (i.e. mass, damping coefficient, and stiffness ratios) or two (i.e. mass, and stiffness ratios) parameters. These unknown parameters can then be determined by using gene or neural networks identification methods.
In this thesis, rotation angle identification of a beam structure based on a laser displacement sensor tracked transverse displacement signal is also discussed. Finite element beam model was coded to compute the dynamic response of structure at the specified sensor locations. These numerically computed transverse displacements were then superimposed with a 20% noise-to-signal ratio random generated white noise signals and were used as the input for rotation angle identification. The orthogonal polynomial approach was used to expand the dynamic transverse displacement signal to the orthogonal functions and their associated coefficients. These coefficients can be used to identify the orthogonally expanded coefficients associated with the rotation angles of beam. The rotation angle response can finally be synthesized. In addition, a series of tests was conducted using the impact hammer to study the dynamic response of the cantilever metallic beam. Both the loading history and the transverse displacement history were recorded. After the displacement signal was measured, either a wavelet filter with the Daubechies scaling function or the cubic moving least square error method were used to reduce the noise from test. Similar to the numerical synthesis process mentioned above, the rotation angle of the beam can be determined.
Through numerical validation, current identification process is capable of monitoring the structural health condition by using specified time domain input and output signals. And the rotation angle identification discussed in this thesis is also useful through numerical and experiment validations.
|
|---|
