木材音射訊號模式之辨識

材料受力變形至破壞時，以數種彈性波之形式釋出訊號。假設此數種彈性波之波形分別代表材料不同變形階段之破壞來源，則可以利用此特性作為瞭解材料受力整體行為模式，並且有助於研發非破壞音射檢測系統。因此正確地分析波形就顯得格外重要。就辨識訊號模式方面，首先將音射訊號處理，擷取具代表訊號數個特徵參數作為進一步多變值分析，以有效及正確將訊號分類。茲將本研究之主要結果歸納如下：1.音射訊號特徵值彼此間之相關性，有的高，有的低。選取低相關性之特徵值，作進一步群落分析。2.成功地將音射訊號依據群落分析（cluster analysis）分為三類模式。3.此三類音射訊號模式在木材載重階段都有可能發生，因此無法以它代表任何載重階段。

關鍵字

模式辨識；音射；訊號處理；多變值分析；群落分析

並列摘要

When material is stressed from deformation to failure, it will release energy in several forms of elastic wave. If each forms stands for a distinct deformation or fracture source, that will help to understand failure mode of material under stress and develop non-destructive acoustic emission systems. Therefore, the analysis of waveforms is remarkably significant. As to the analysis of waveforms, the features of waveforms will be processed and extracted for multivariate statistical analysis to correctly recognize signals. The main results of this study were found as follows: 1. The relationships among the features of waveform were either high or low. Features with low relationships were selected for further cluster analysis. 2. The acoustic emission signals were sucessfully classified into three categories. 3. Each class might occur in different load stage, therefore each class could not indicate a load stage.

並列關鍵字

Pattern recognition ； Acoustic emission ； Signal processing ； Multivariate statistical analysis ； Cluster analysis

國際替代計量

木材音射訊號模式之辨識

未授權

主題瀏覽