A fuzzy modeling approach for the solution of an inverse electrostatic problem

A numerical technique based on a suitable combination of artificial neural networks (ANNs) and fuzzy logic (FL) is presented. It is shown how the ANN solution of typical inverse problems can take advantage of the introduction of fuzzy information. The study case is an inverse electrostatic problem of some relevance for nondestructive testing (NDT) applications. The performance of both standard ANNs and the novel hybrid neuro-fuzzy model are compared, and it is shown that the structured approach is superior to the unstructured one, particularly in terms of speed of the learning phase.