橋梁之振動訊號可以反應出橋梁之訊息,本文期望藉由模擬橋面板之實驗,來建立出類神經網路應用於二維橋樑之非破壞檢測的識別系統,識別系統包含破壞程度及檢測破壞點的識別。 首先在模擬橋面板上鑽取各種不同的裂縫大小和敲擊距裂縫不同距離的面板區塊,藉由量測儀器以震波量測法擷取各種裂縫大小及損壞位置的訊號,經由小波理論作訊號處理,擷取訊號的特徵,建立資料庫(database),再將訊號的特徵作為類神經網路輸入的特徵向量。經由類神經網路的訓練,可識別橋面板不同裂縫大小及損壞位置。 經學習向量量化網路及機率類神經網路識別分析後,經由結果可知,在學習向量量化網路方面,裂縫大小識別的準確率達89%,損壞位置識別的準確率則達97.6%;在機率類神經網路方面,裂縫大小識別的準確率達92%,損壞位置識別的準確率則達99.2%。由此可知,小波理論非常適用於訊號處理及特徵擷取,再經由類神經網路之訓練建立識別系統後可診斷橋樑之破壞程度和破壞點位置。
Vibration of the bridge always carries information about its function.The purpose of this paper hopes to establish the identifiable system of NonDestructive Testing (NDT) of two-dimensional bridge using in Artificial Neural Network (ANN) theory by modeling bridge floor testing.The identifiable system includes identification of different destruction degree and breakpoint position of the bridge. Firstly, a crack different size is initiated and knocking different position of the blocks on the bridge floor model. Then, signals of different crack size and knocking positions are generated by seismic wave measure method. Using the wavelet theory,we can extract features of signal and establish database, which forms the input eigenvector of ANN. After ANN training, we can judge different crack size and breakpoint position of the bridge floor model. By experiment results of Learning Vector Quantization(LVQ) network and Probabilistic Neural Network(PNN), we found that,in LVQ netwok,the identification accuracy of crack size achieved 89% and the identification accuracy of breakpoint position achieved 97.6%,in PNN,the identification accuracy of crack size achieved 92% and the identification accuracy of breakpoint position achieved 99.2%. It is obvious that wavelet theory is a good methodology on signal processing and feature extraction and after ANN training to establish identifiable system which can diagnoses destruction degree and breakpoint position of bridge floor.