本論文是利用ARX模型(Autoregressive with exogenous input model)與特徵正交分解(Proper Orthogonal Decomposition)的方法在時間域上進行系統參數識別,以結構系統的輸入(施加之外力)歷時與輸出(位移或加速度)歷時求得系統的自然頻率、阻尼比與模態形狀。首先,以數值模擬的方式產生輸入、輸出歷時資料探討系統在不同型式外力作用下,識別所需的取樣時間間隔、取樣區間與量測資料筆數,以獲得精確的模態參數。接下來,以實際的結構試驗歷時資料來進行系統參數識別,並將試驗與數值結果作比較,以供後續討論。 此外,更進一步使用層間偏移模態(Inter - story Drift Mode Shape)損壞指標來尋找結構系統的損壞位置,其僅需結構系統的第一個模態向量即可進行結構健康檢測。因此,吾人以數值模擬與試驗的方式,根據各種不同損壞組合之結構模型,進行結構模態參數識別與損壞指標計算,證明損壞指標確實能偵測結構損壞位置。
In this paper, the autoregressive with exogenous input model and the proper orthogonal decomposition (POD) method are used to identify system parameters such as natural frequencies, damping ratio and modal shapes in time domain. First, we numerical simulate a structure subject to different types of external force (inputs) and obtain the time response of the displacement or acceleration of the structure (outputs). Then, we tune the sampling interval、sampling range and length of measurement data in the process of system identification to obtain accurate system parameters. Next, we conduct a real structure test and collect the time - history data of inputs and outputs. The system parameters are identified based on the experimental data, and compared with results the ones identified from numerical results. Moreover, the identified system parameters are exploited for structure health monitoring. Inter - Drift Mode Shape (IDMS) that only needs the first mode shape of the structure is used to detect the damage that indicates inter – story stiffness reduction. We consider different kinds of damage combinations. The numerical and experimental results show that IDMS can accurately detect the damage location.