敲擊回音法於傾斜裂縫影像修正之探討

Abstract

敲擊回音法能有效檢測混凝土內部缺陷，該方法在結構表面敲擊以激發應力波，接著量取表面位移進行傅立葉轉換，再由頻譜之頻率尖峰值判讀缺陷位置。當以敲擊回音試驗偵測傾斜介面時，因試體表面與反射介面不平行，應力波在兩介面之間的來回反射並不像平行面一樣規則，因此頻譜的回音尖峰並不一定會對應到測點正下方之介面，且回音頻率尖峰不太明顯。若以影像法繪製頻譜影像時，會發現介面的傾角及位置都略有偏差。 本研究之目的在開發一有效之頻譜影像傾斜介面修正法，以修正來自傾斜介面上各點應力波回傳所造成的互相干擾。本研究所提出之頻譜移位影像法，與震測移位影像法類似，但處理之訊號並非時間域訊號，而是頻率域訊號。移位法利用橢圓疊加的原理，將所有測點的頻譜中各頻率的能量散布到整個半橢圓中，各像素得到來自所有測點的能量疊加。由於頻譜移位影像法中，頻率振幅對頻譜影像各像素均有相同比重貢獻，會導致非傾斜介面位置產生鬼影干擾。因此本文再利用權重分配法以增強傾斜介面反射能量並去除影像修正後產生的鬼影。此外，本文也對頻譜影像進行色階轉換以增強影像對比。藉由本文提出方法可使垂直斷面影像清楚顯示傾斜裂縫的形狀以及位置。 本研究先以有限元素法模擬六個不同斜率的傾斜裂縫，再以數值結果驗證本研究提出方法之可行性，由數值模擬結果可看出經由頻譜移位影像法後，能加強裂縫影像，修正傾斜裂縫位置，且能消除影像中的鬼影。本研究同時進行三組模型試驗，實驗結果顯示，頻譜移位影像法能有效應用於實驗訊號，且所顯現的影像是更為直觀的試體剖面圖，能準確地定出傾斜裂縫的位置與大小，在實際應用上應可大幅提升混凝土非破壞檢測之效能。

Impact echo method can effectively detect internal defects of concrete, this method is to impact the structure surface to stimulate the stress wave, and then to measure the surface displacement of the Fourier transform, and then interpreted by the peak value of the spectrum of frequencies of defect sites. When using Impact echo method to detect tilt interface, the specimen surface and the reflection interface is not parallel, so force wave between the two interfaces coming back reflection is not like the rule between two parallel surfaces. Therefore, spectrum echo peak is not necessarily correspond to the measuring point just below the interface, and the echo frequency peak is not too obvious.If you use the image method to draw a spectral tomography, you will find out the inclination and position of the interface have a slight deviation. The purpose of this study is to develop an effective Spectral Tomography Tilt Interface Amendment Method to amend mutual interference cause by the stress wave returns from tilt interface . This paper presents a Spectral Migration Image Method, similar to the Earth’s Interior Migration Method, but the signal processing is not a time domain signal, but the frequency domain signal. Migration uses the oval superposition principle, the energy of each frequency in the spectrum of all measured points spread to the entire semi-elliptical, each pixel from all points of the energy overlay. As the Spectral Migration Image Method, frequency amplitude spectrum of each pixel of the image have the same proportion of the contribution will result in non-tilt interface location of the Interference signal. This article re-use of the method of weight distribution to enhance the tilt interface reflection energy and remove the image correction resulting from the Interference signal. In addition, the gradation conversion of spectrum images to enhance image contrast. By this paper, the method can make the vertical cross-section image clearly shows the shape and position of the inclined cracks. This study use the Finite Element Method to simulate six different slope of the inclined cracks, then use numerical results verify the feasibility of the proposed method. From numerical simulation results, we can see that through the Spectrum Migration Amendment Method can fix tilt crack location, enhance the signal strength of the crack area and focus on the correct thickness of the defect, and eliminate Interference signal, solve the problem of error in judgment. This study sets three model tests at the same, the experimental results can also be found that the amendment method proposed in this article can be effectively applied to the experimental signal, accurately identified the location and size of the inclined cracks.