Damage discrimination of composites based on wavelet decomposed acoustic emission signals

Composite materials are characterized by different types of failure mechanisms which are typically associated with matrix cracking, fiber-matrix debonding and fiber breakage and have a different AE signature. In this framework, neural networks are widely used for damage characterization of composites from the AE activity during loading. The neural networks are trained with typical AE signals originating from the structure in operation and then are used for clustering the incoming signals during the lifetime of a construction. The typical approach involves recording waveform features like the amplitude, duration, energy and average frequency and tries to associate them with the damage source. Nevertheless it has been proven that it is very hard to draw definite conclusions based on these features. In this work, we have used parameter-less self-organized mapping based on wavelet decomposed AE signals containing both time and frequency information.