小波轉換應用於板架結構的損傷偵測

Damage Detection in Stiffened Plates by Wavelet Transform

近年來將小波分析方法應用於結構的損傷偵測是一個相當熱門的研究領域，但是其研究範圍多半侷限在樑、平板、軸承等簡單的結構上。然而應用在實際的工程問題上時，這些簡單的結構極少單獨存在。為了使破損偵測的研究能使用於實際的船舶結構上，本研究針對板架結構，探討使用小波分析方法偵測其損傷位置的可能性。本研究首先以有限元素模擬含有缺陷的板架結構，並以二維離散小波分析板架結構的模態振型以找出缺陷的位置，然後對四個不同的板架結構進行實驗。實驗時先以激振器給予板架結構一個衝擊，然後以加速規量測板架結構的振動訊號，接下來以小波包轉換從振動訊號中找出板架結構的模態振型，最後以二維離散小波分析找出裂縫的位置。有限元素模擬及實驗結果顯示本研究所提出之損傷偵測方法有良好的損傷識別能力，並且較以往的方法能大幅地減少所需的量測點。 Wavelet analysis has been a useful tool to detect damage in structures during recent decades. However, most researches are focused on the damage identification in basic structures, such as beams, plates, and bearings. In practical application, the basic structures do not exist alone. Instead, they are usually accompanied with one another. Stiffened structures are commonly used in ship structures and are investigated in this study. The present investigation focuses on the estimation of damage location of damaged stiffened structures by using the wavelet analysis. In numerical analysis, the first mode shape of the damaged stiffened panels is simulated by finite element method. In experimental analysis proposed in this study, four different damaged stiffened structures are considered. The damage locations are detected by two-dimensional discrete wavelet analysis, A step by step procedure for damage detection is listed in present investigation. Firstly, a shaker was used to make an impact on the damaged structure. Secondly, accelerometers were used to measure the vibration responses of the structure. In the third step, the first mode shape of the structure is obtained by using the wavelet packet. Finally, two-dimensional discrete wavelet analysis was applied to determine the location of cracks. The results of numerical analysis and experimental investigation reveal that the proposed method is applicable to detect single crack or multi-cracks of a stiffened structure. The experimental results also show that fewer measurement points are required by utilizing the proposed method in comparison to those presented in previous studies.