A general approach for extracting sensitivity analysis from a neuro-fuzzy model

The problem of obtaining sensitivity analysis information from an even M-input N-membership function neuro-fuzzy model is addressed in this paper. Subsequently, this permits the application of classic deterministic optimization methods in order to find the global optimum of any objective function approximated using neuro-fuzzy modeling. For nondifferentiable functions this approach is of great benefit. Results from a practical electromagnetic optimization problem are presented.