在發電廠中迴轉機之數量相當多，維持其正常之運轉乃是必需的，因此多數之迴轉機皆有線上之振動探測設備，以免因振動過大而跳機。但引起振動過大之因素相當多，由振動之頻譜及相角大多可診斷其引起振動過大之原因；其中油漩(oil whirl)係回轉機振動過大常發生之原因。其頻譜上會出現有1/2倍頻轉速，且其半倍頻之振幅會突然急遽產生，而造成跳機。 在油漩未發生前，很難由頻譜事先預知，因此本計畫以線上(on line)系統判別法，以實驗方式求得回轉機之非線性動態方程式，並以模擬該方程式之結果與實驗數據比較，而驗證該方程式之正確性。 The rotating machines in power plant play an important role, and keeping normal operation condition of the rotating machines is necessary. Thus on-line deception equipment are installed in most of the rotating machines to avoid shut-down due to large vibration. From frequency spectrum and phase angle one can judge what cause the trouble. However it is difficult to tell when the oil whirl will occur. The frequency of the oil whirl is one half of the operation speed. The amplitude of the oil whirl is higher than that of the synchronous vibration. In this project we propose to apply the system identification method experimentally to obtain a nonlinear equation for the rotating system. The comparison between the simulation result and experimented data shows the feasibility of the field.