臺灣博碩士論文加值系統

模擬演化法(Simulated Evolution)是根據生物學中的達爾文進化論
(Darwinian Evolutionary Theory) 所設計出來的一套全域最佳化程序。
在本研究中，吾人將把模擬演化法應用於線性振動結構的模態參數識
別問題上。在本文中，模擬演化法將在函數最佳化問題上展示其能在
眾多的區間極小值(local minima)中搜尋全域最小值(global minimum)的
強健性。繼而，本文將探討在模態參數識別問題中，根據不同的收斂
準則對模擬演化法模態參數識別結果之影響。最後，本研究將對於在
量測數據含有不同程度大小的噪聲時，基於模擬演化法所得到的模態
參數識別結果之可靠性作探討。識別結果顯示，在中等程度的量測噪
聲下，模擬演化法仍然能有效並可靠地識別出主要的模態參數。

In this thesis, modal parameter identification of linear
vibrating structure based on optimization approach is
studied. A global optimization search algorithm
based on the Darwinian Evolutionary Theory is implemented
for identification of modal parameters of a linear
vibrating structure. The capability of the evolutionary method
in locating the global minimum among numerous local
minima will be demonstrated. Moreover, different
kinds of convergence criteria for the simulated evolution
and their effects on the identification results will be
studied. The reliability of the searching technique in modal
parameter identification under different measurement noise
levels will also be examined. Results show that the
technique is effective and reliable in identifying
dominant modes under moderate noisy conditions.