This paper presents a method to determine excitation angles for a Switched Reluctance Generator (SRG) by using Artificial Neural Network (ANN). The ANN model consists of the feed-forward neural network and the back-propagation learning with a linear activation function (the linear function) and a nonlinear function (the hyperbolic tangent). The ANN model with two layers; the hidden layer and the output layer, is derived from the current and flux linkage of the SRG. The SRG model is built from the magnetization curve which the flux linkage versus current at different rotor positions is analyzed from the finite element method (FEM). An 8/6 SRG was set up to validate the proposed ANN method.