小波理論於管結構損壞識別之應用

Abstract

近年來小波轉換應用於各個研究領域之中，由於其對訊號具有多重解析的特性，以及提供以往傅力葉轉換只於頻率域上的解析外，訊號隨著時間變化關係的解析能力，表現出其對訊號處理方面上的優越性。 本研究提出一個應用於管結構的非破壞性損壞識別方法，先建立識別參數資料庫，將衝擊檢測獲得的訊號以小波分析後與資料庫進行比對，以診斷出損壞發生位置與損壞的程度。研究過程使用數值方法模擬管勁度衰減損壞下的加速度反應，利用小波分析的多重解析技術將加速度訊號分解成不同尺度頻率區間的子訊號，利用特定尺度的子訊號能量變化判斷損壞發生位置及損壞程度。使用四個案例對管損壞識別系統進行測試，顯示出除了靠近邊界時，對於損壞發生位置的識別以及損壞程度上的識別都有相當不錯的表現。

This thesis proposes an application of NDT method for damage identification of pipes. A numerical model is employed to process simulate accelerative responses of pipes which are subjected to the stiffness degradation damage in circular.First,a finite-element software is employed to establish the database of identification parameters.The wavelet transform is then adopted to process the signal from impact tests performed in FEM model.Finally, the damage location and the degree of damage is diagnosed by comparing analyzed results with parameters in the database.The accelerative responses are decomposed into several different frequency regions by the multi-resolution technique of wavelet analsys.It uses the energy variation of sub-signals in specific scale to diagnose damage.Testing the identify system by four cases,it reveal that the proposed model has excellent performances in damage identification except that damage is close to two supported ends.