| Summary: | 碩士 === 國立成功大學 === 航空太空工程學系碩博士班 === 95 === This thesis studies Non-Stationary Time Series for the application of modal-parameter identification from non-stationary ambient vibration data. The original Time Series uses ARMA (Autoregressive Moving-Average) model, which contains autoregressive part and moving average part, to reconstruct the stationary ambient vibration data, and obtains modal parameter with autoregressive part of ARMA model. However, the original time series method is not applicable to non-stationary signal which is closer to natural environment. So we propose two ways to build a non-stationary time series model—by curve-fitting of amplitude and by introducing the basis function. We use this model to describe the non-stationary amplitude of data and we also apply it to modal-parameter identification from non-stationary ambient vibration data. Through numerical simulation, applicability and effectiveness of the proposed method of modal parameter identification from non-stationary ambient vibration data is demonstrated.
|
|---|
