Low-cost delamination monitoring of CFRP beams using electrical resistance changes with neural networks

Delamination is a significant defect of laminated composites. The present study employs an electrical resistance change method in an attempt to identify internal delaminations experimentally. The method adopts reinforcing carbon fibers as sensors. In our previous paper, an actual delamination crack in a carbon fiber reinforced plastic (CFRP) laminate was experimentally identified with artificial neural networks (ANNs) or response surfaces created from a large number of experiments. The experimental results were used for the learning of the ANN or for regressions of the response surfaces. For the actual application of the method, it is necessary to minimize the number of experiments in order to keep the cost of the experiments to a minimum. In the present study, therefore, finite-element method (FEM) analyses are employed to make sets of data for the learning of the ANN. First, the electrical conductivity of the CFRP laminate is identified by means of the least estimation error method. After that, the results of the FEM analyses are used for the learning of the ANN. The method is applied to the actual delamination monitoring of CFRP beams. As a result, the method successfully monitored the delamination location and size using only ten experiments.