目前系統識別已被廣泛應用在工程上，在許多識別方法中，頻率域分解法是一種較為簡單實用的技巧。因此本論文的研究乃利用此方法來求得橋梁斷面之顫振導數。該斷面模型試驗採用雙自由度之架構並結合FDD方法，再將其結果與傳統單自由度實驗結果加以比較 由本文之研究結果可知，透過雙自由度實驗求得之 與單自由度實驗之結果較為一致，但高風速下雙自由度實驗會有垂直振態不明顯之現象進而有些微差異，在 方面則顯示雙自由度實驗之結果有提早發散之現象，而透過雙自由度與單自由度實驗求得之 則相當一致。另外，以雙自由度為實驗架構並改變頻率比所求得之顫振導數結果顯示其差異性不大。在FDD方法中，最重要的是選取峰值其所對應的頻寬，而改變截取的頻寬將會造成不同的顫振導數結果。本文採用雙自由度進行試驗，由於其垂直與扭轉振態並未很接近，故FDD方法之優點則較不明顯。 The system identification methods have already been extensively used in engineering. Among these methods, Frequency Domain Decomposition (FDD) is easy to use and can be applied in civil engineering structures. Therefore, FDD method is adopted in this thesis to obtain the flutter derivatives of bridge decks. The sectional model test, with two degrees-of-freedom, in association with FDD can be more easily conducted compared to the conventional one. The flutter derivatives identified from this approach are compared with those in the conventional test, with single degree-of- freedom, associated with Scanlan’s identification method. The comparative results show that the flutter derivative identified from this approach is almost the same as those in the conventional test. However, at high reduced wind speeds, there are some differences. This is because the vertical mode is not obvious in the test with two degrees-of-freedom. For the torsional flutter derivative , the results identified from two methods have the same trend. Since the sectional model with two degrees-of-freedom tends to diverge at smaller reduced wind speed than the conventional test, the values identified from both the methods have some discrepancies. For the torsional flutter derivative , the results identified from both methods are in good agreement. The effect of the frequency ratio between the torsional and vertical modes in the test is not significant. The choices of frequency peak and the corresponding frequency bandwidth used in FDD are very important. Adopting different frequency bandwidths will result in different results. Since the frequencies of tosional and vertical modes are not close in this two degrees-of-freedom test, the advantages of the FDD method are not obvious.
