結構物的損傷偵測在工程上一直是個很受重視的課題，若能及時的檢測缺陷出來，則可以預防許多災難的發生。近年來，小波轉換與類神經網路的強大功能，逐漸受到重視並被應用於非破壞檢測方面。本研究首先將利用二維離散小波轉換來偵測平板的裂縫位置，其次以類神經網路配合二維離散小波轉換來判別裂縫的損傷程度，包含偵測裂縫長短與裂縫深度。在數值模擬上，先探討雙裂縫平板的損傷指標趨勢與裂縫位置及長度變化的關係，之後嘗試改變損傷指標的選取範圍，並改以單裂縫平板之損傷指標做為類神經網路的訓練樣本，來判別多裂縫平板上各單一裂縫的破損程度，建立一個可以適用於多裂縫平板的偵測方法。實驗部分，會將實驗量測無裂縫平板與模擬無裂縫平板間的模態差異，來修正含裂縫平板之模態振型，讓實驗更貼近模擬的情況，藉此探討所提出之方法對於損傷程度判別誤差的影響。

Structural damage detection has been a very important issue in engineering. If damages can be detected in a timely manner, many disasters can be prevented. In recent years, wavelet transform methods and artificial neural network have been widely used in Non-Destructive Testing . The present investigation, at first the two-dimensional discrete wavelet transform method to detect the position of cracks in plate. Secondly, the artificial neural network along with the two-dimensional discrete wavelet transform method is used to identify the degree of damage in this study. During the numerical simulation, different from the previous study, only the area around a single crack is selected and used to calculated the damage indexes for a single crack on a plate. The obtained damage indexes are used as training sample for artificial neural network analysis, which then are utilized to evaluate the degree of damage of a single crack on a multiple cracks plate. In the experiment analysis, in order to eliminated the mode shapes, difference between simulations and experiments, the mode shapes of a perfect simulated plate and a plate without cracks used in experiment are studied. It is our intention to obtain more precise method to estimate the degree of damage of cracks on plate.