Strain based method for monitoring the health state of composite structures

A method able to detect and localize a change of the local compliance in composite structures is presented. The method is based on a numerical algorithm requiring as input only strains measured in different locations of the structure and does not require the knowledge of loads acting on the structure and the details in terms of materials and geometry. The strains measured in each location are correlated with the strains measured in the neighbourhood by means of neural networks and the series of correlations between strains are representative of the reference structure's finger print. Any eventual change of the structure from its reference state is recognized and localized thanks to the change of the rules of the correlation between the strains at a given point. From a sensitivity analysis it is shown that the method is able to detect the damages from their early development, e.g. micro-cracks, up to the delamination mechanisms, that makes it particularly useful for structural health monitoring. The approach has been experimentally validated on composite structures subjected to complex loadings and deliberately damaged in different locations with different categories of damages of known entity.