本文提出一直接由量測輸出入資料來鑑定結構動態系統矩陣的程序,以ARX模式來描述輸出入資料間的互動行為。ARX模式之參數矩陣以最小平方技術來估計。並於其後轉換至離散狀態空間模式。為了除去無意義之幽靈模態與增加參數估計之精確度。我們亦發展將高階狀態空間模式降階至適當因次的凝縮技術。結構系統之動態矩陣可直接由已凝縮鑑定之連續狀態空間模式來萃取出。多起數值模擬結果說明本文所提程序可自訊號污染之量測中識別出結構動態系統矩陣。
A method of identification for dynamical system matrices of structures directly from the measured I/O data is presented. in which ARX model is developed to describe the interactive behaviors among the measured input and output data. The parameter matrices of ARX model arc determined by the least squares technique, and then the ARX model is transformed to a time-discrete state equation. In order to remove the ghost mode and improve the accuracy of parameter estimation, a reduced technique is developed to reduce the high dof state space model to adequate dimension. The system matrices of structures will be identified directly from the reduced continuous state space model. Numerical simulation show that the proposed procedure has a good results to identify the system matrices of structrues from noisy I/O measurements.