在風工程領域中，以傳統自由振動的實驗方法識別顫振導數已漸趨成熟，但是實驗結果易受週遭環境影響而產生不可靠的結果。為改善缺點，本研究使用間接強制振動的實驗方式，研擬出一套新的顫振導數識別方法。由伺服馬達給予振動平台白噪音強制振動，透過彈簧振動橋面板斷面結構模型，然後量測在平滑流場下之氣彈互制效應。 實驗流程分為非耦合顫振導數識別與耦合顫振導數識別，均藉由氣彈互制反應之轉換函數實驗值與理論值比較，在頻率域以曲線擬合最佳化識別出理論式中最佳參數，最後得到橋梁之顫振導數。其中於理論部分引用狀態空間方程式之觀念進行推導，而最佳化過程則引用基因演算法(Genetic Algorithm) 求解，以確保得到最佳解。 本文以機翼斷面模型及不同寬深比之橋面板為例，使用淡江大學土木系風洞實驗室進行識別試驗，並分別和Theodorsen函數(理論解)及M.Matsumoto所做實驗之結果做比較。識別結果顯示除A4*、H4*外其值趨勢接近，顯示間接式強制振動新識別法能可靠識別顫振導數。A4*及H4*存再之誤差，仍待後續研究進一步釐清。 本文所使用之間接強制振動，和直接強制振動的差別，在於識別過程已引入了結構動力學的方程式，具有如下優點(1)使用過程中所獲得之狀態方程式，配合給定之抖振外力，即可模擬抖振反應之時間歷時。(2)藉由檢驗狀態方程式中系統矩陣Ac之穩定與否，即可求得顫振臨界風速。 In wind engineering application, although the conventional technique using free oscillation method to obtain flutter derivatives of bridge decks has been mature, the results thus obtained might be sensitive to the test and/or environmental conditions. To improve the reliability of test results, this thesis presents a new identification approach by utilizing indirect forced actuation in the wind tunnel tests. In the experiment setup, the bridge section model is connected to a two-axis actuating device through serial connection of springs. Under the excitation of the actuation device and smooth wind flow, the aero-elastic response of the section model is thus measured for identifying the flutter derivatives. The identification scheme proposed is composed of two parts, one is for uncoupled term flutter derivatives and the other is for couple ones. By comparing the frequency response function of aero-elastic responses with the theoretical values that are derived based on state space equation theory, the optimal parameters involved in the theoretical formula can be determined by using curve-fitting optimization which employs the Genetic Algorithm (GA) in the searching process to ensure achieving the global optimum. For the demonstration of this approach, the section model of an air foil and the models with width/depth ratios of 5 ~12.5 were used for identification. The Theodorsen functions (theoretical solution for flat plate) and the experimental results of M. Matsumoto were used for comparison. The identified results showed that, except for A4* and H4*, the comparisons of the flutter derivatives are quite matched. Hence, the indirect forced vibration approach is a reliable method for identifying bridge flutter derivatives. The existence of the discrepancy on A4* and H4* should be clarified in the future study. Compared with the direct forced oscillation method, the indirect forced oscillation method introduces the dynamics of the structure in the identification process, and therefore provides advantages in two folds: (1) with the buffeting disturbance given, the state space equation introduced can be slightly modified and used to simulate the response in time domain; (2) the critical flutter speed can be determined by examining the stability of the system matrix Ac in the state space equation.
