| Summary: | 碩士 === 國立臺灣大學 === 造船及海洋工程學研究所 === 90 === The system matrices of equation of motion contain the dynamical characteristics of a structure, which play an important role for vibration control (active and passive control), damage detecting and monitoring, and model updating of structures.
In this thesis, an efficient method has been developed to identify the equivalent system matrices of structures from the measured input force (or ground motion) and output response data on the multiple measured points. A vector type of backward auto-regressive with exogenous variable model, termed as VBARX model for short, is adopted to set up an assembled state equation, which describes the relationship between the measure data of different time steps. The equivalent system matrices of structure can be constructed from the assembled state equation and its eigen matrix.
In general, a higher order auto-regressive model can improve the accuracy of identified dynamical characteristics, however, it increases the number of numerical modes, and only some of them are the system modes. The proposed method provide a clear board to separate the system from the spurious modes.
For structures with higher damping system, the response of free vibration decays very fast. The available measured data for modal identification may be relative short in some cases. Compared with conventional identification approaches, the proposed method needs relative short data points. And hence it has advantage for modal identification of structures with higher damping system.
The system identifications of a numerical example for a 3-DOF lumped mass dynamic problem and two experimental examples for the offshore platform model and NTU 5 floors model have been carried out to validate the applicability and effectiveness of the proposed method.
|
|---|
