主催: 一般社団法人 日本機械学会
会議名: 第26回機械材料・材料加工技術講演会
開催日: 2018/11/16
The authors proposed a new classification method of AE signals using a machine learning to identify damage types occurring inside CFRP laminates. In this method, the temporal waveforms of AE signals were used as training data to develop the autoencoder. An AE signal generated by a transverse crack and that by delamination can be differently characterized by A0 and S0 modes in AE depending on the different source orientations. The results of the data analysis indicated that the autoencoder has a good performance to identify AE signals with different source orientations. Therefore, the proposed method has a reliability to identify damage types in CFRP laminates.