摘要:In this paper, we present a new method based onmodification on single layer Legendre Neural Network(LeNN)method to be Legendre Neural Network DDY2 (LeNNDDY2)method to solve ordinary and partial differential equation. Theactivation function on the hidden layer is changed by Legendrepolynomial expansion. The optimization method used in weightand bias updates is the DDY2 conjugate gradient method. Forexample problems, the numerical results have been comparedwith the other methods and gotten better results.