Ahmad Shirzadi

The subject of this thesis is the study of a single layer Legendre Neural Network (LeNN) model to solve Ordinary Differential Equations with initial and boundary conditions. An extension of the equation solution in terms of Legendre polynomials is simulated by an ANN. To update Network parameters (weights), a back propagation algorithm based on the minimization of an error term is used. The method is used to solve several kind of initial and boundary value problems, as well as a coupled system of ordinary differential equations. The results reveal the effciency of the method.