AE波形の周波数領域での特徴の機械学習によるCFRPの破壊過程の推定

抄録

The aim of this study is to classify the damages in a Carbon Fiber Reinforced Plastics (CFRP) plate under tensile loading with a machine learning technique for acoustic emission (AE) waveforms. AE waveforms originate from micro cracks inside a material and their features are characterized by fracture mode. In this study, waveform energy in each frequency band obtained with the wavelet transform were used as features for classifying AE waveform using self-organization map(SOM) and k-means. Furthermore, the validity of the classification was evaluated by cross-sectional observation and by AE in unidirectional CFRP samples.