目前土木工程上對於訊號處理仍以傅利葉轉換為主。然而一般自然界 中的訊號乃至橋梁結構中受震反應,往往具有時變性與非線性。由於 傅利葉頻譜以全域諧和波的疊加逼近原始的訊號,對於訊號處理的可 適性造成相當大的限制,尤其是對於會隨時間改變特性的訊號。 本研究主要是突破這些限制,以原本用傅利葉分析的方式進行希伯特 頻譜的分析,進行橋梁損傷的非破壞檢測。藉由所擷取訊息做希伯特 -黃頻譜(Hilbert-Huang Transform)[1],以求得橋梁損壞的位置與 程度﹔此分析方式採用了經驗模態分解法(The Empirical Mode Decomposition)之拆解模態的功能,而具有濾波性質,故能成功地將 振動記錄拆解成各模態,呈現於時間域上同時結合希伯特頻譜分析 (Hilbert-Spectral Analysis),處理具有非線性(nonlinear)及非穩 態(nonstationary)特性之時間序列,此法用於檢測橋梁結構損傷評 估頗具潛力,突破以往採用傅利葉基礎不夠敏感之缺陷。 希伯特轉換最強大的功能為對於具有時變性的資料能表現出結構特 性。本研究除了理論架構的建立之外,也將於實驗室內進行控制損壞 之實驗,對於損壞與未損壞的RC 梁與鋁梁進行試驗。主要分析方法 Today, civil engineers still mainly use Fast Fourier transform to analyze the behavior of structures. However, most signals, for example, earthquakes, have the property of non-stationary and time-dependent. As Fourier transform use global harmonic waves to approach original signals, it is not a proper tool for Fourier transform to analyze these signals. We purpose a new analytic method for non-destructive instrumentation for damage assessment of bridges. The new analytic method is based on the Hilbert Huang transform, which is a powerful tool to analyze non-stationary and non-linear time series. It consisted of the empirical mode decomposition method and Hilbert spectra analysis. The diagnosis on the health of the bridge structure is based on the change of dynamic property of the bridge response. We here improve the skill to detect the damage of bridge. The primary analytic methods are Hilbert spectrum, Hilbert energy spectrum, and Hilbert modal analysis.