A diagnostic system using neural networks and natural frequencies

The identification and determination of damages in a structure is a very interesting problem in engineering, being present in the production phase of a component (non-destructive testing) as well as during its operation (evaluation of structural integrity). In this the damages of the structure are accomplished using an artificial neural network (ANN) with supervised training, making use of natural frequencies as physical parameter used in the identification. The ANN training data is obtained through numeric simulation, finite element method. The proposed approach is exemplified through the localisation of a damage in a free-free beam, based on the experimental measurement of its first five natural frequencies