Numerical Simulation for Fuzzy Perturbed Integro-Differential Equations Using partially Neuro-Fuzzy System

Abstract

Recently, the study of fuzzy singular perturbed Fredholm integro- differential equations (FSPFIDEs) has been of increasing interest for a long time. Our paper has a design for a fast feed -forward partially neuro-fuzzy system (PFNN) to  adduce a new method for solving two- dimensions (FSPFIDEs). Employing a multi-layer that has one hidden layer consisting of seven a set of units and one linear output node. And the sigmoid activation for every unit is the hyperbolic tangent function and the Levenberg – Marquardt training algorithm(LM) . We compared our exact solution in illustrative examples with the results of numerical experiments, confirming the efficiency and accuracy of our presented scheme.