Artificial neural networks to interpret acoustic emission signals to detect early delamination during carbonization of pre-fabricated components of carbon-carbon composite material

This study applies Artificial Neural Network Systems (ANN) to identify the features in the acoustic emission signals that may be used to predict delamination defects. Processing time and temperature are known to be strong influences in the pyrolysis process. Gas evolution and flow rates, and mass change data were available from testing providing additional process characterization. In this paper we propose the use of Acoustic Emission (AE) to detect early delamination, and Artificial Neural Network (ANN) techniques to interpret in-situ AE signals and to control fabrication parameters. Results of six carbonization runs are presented including those for components with and without delamination. The overall results of the preliminary research are very encouraging and demonstrate the benefit of combined Acoustic Emission and Artificial Neural Network techniques