Pattern Recognition Analysis of Acoustic Emission from Composites


Proceedings of EWGAE - 23rd European Conference on AE Testing, Vienna, 6-8 May 1998, pp. 15-20

A. A. Anastassopoulos, T. P. Philippidis


Unsupervised Pattern Recognition (UPR) analysis is proposed as an alternative to conventional statistical techniques for the discrimination of AE signals from composites. The AE data were classified by means of partitional clustering algorithms and the resulting partitions were numerically validated by means of separability criteria and by direct comparison with conventional inspection techniques. In view of the above the AE signature from each damage mode was established for the different material configurations tested. The technique proved suitable for the discrimination of AE signals from composites and the results were successfully correlated with those obtained by the complementary inspection techniques. Damage accumulation from each mode is described by means of the cumulative AE hits of each class of the resulting partition as a function of the  applied load.