Evaluation of fatigue damage process in a CFRP plate by a neural network for AE waveform

DOI Web Site

Bibliographic Information

Other Title
  • ニューラルネットワークを用いたAE波形分類によるCFRP板の疲労損傷過程の評価

Abstract

<p>The final goal of this study is to realize real-time evaluation for fatigue damage process in a carbon fiber reinforced plastics (CFRP) with an acoustic emission (AE) technique. AE signals from CFRP under static and cyclic three point bending tests were observed. The waveform classification with self-organizing maps (SOM) was attempted with AE features extracted from these AE waveforms. In the SOM method, each AE waveform was assigned to either node in the map and similar AE waves clustered on the map. We used 10 x 10 map in this study. The separation boundaries among the damage types on the map can be determined with an x-means method for representative characteristics in each node. AE waveforms detected during the static test were classified into seven clusters. According to the observation of sample cross-section after testing, cracks were observed only in the compression side of bending. Based on both AE waveform classification and observation results, the clusters showed the feature with low frequency components and weak amplitude can be assigned to matrix crack and the clusters with high frequency components and weak amplitude can be assigned to fiber/matrix interface fracture. Other clusters in which AE generated immediately before the fracture are assigned to a large fracture including a fiber breakage. With the learned map obtained from the AE waveforms on the static test, AE waveforms due to fatigue damage on the cyclic loading could be classified into corresponding fracture modes. The SOM method would be a powerful tool for understanding the fracture process of CFRP under various loading condition.</p>

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top