ICA Mixture Modeling for the Classification of Materials in Impact-Echo Testing

摘要

This paper presents a novel ICA mixture model applied to the classification of different kinds of defective materials evaluated by impact-echo testing. The approach considers different geometries of defects build from point flaws inside the material. The defects change the wave propagation between the impact and the sensors producing particular spectrum elements which are considered as the sources of the underlying ICA model. These sources and their corresponding transfer functions to the sensors make a signature of the resonance modes for different conditions of the material. We demonstrate the model with several finite element simulations and real experiments.