Mathematical modeling and simulation of natural gas conversion process by water vapor provided in this thesis. The mathematical model is pseudo homogeneous one-dimensional based on mass and energy balances. Ergun equation is used to determine pressure drop and non-linear reaction kinetics for reactions in a fixed bed catalytic reactor filled with catalyst cylindrical particles. The model equations was solved with MATLAB software and in order to validate the model, the results were compared with data from the refinery, The maximum error is 17.31%. Then an artificial neural network for this catalytic unit was designed and developed in order to predict the operating conditions parameters of temperature, pressure and mole fraction of hydrogen according to 4 input parameters of temperature, pressure, furnaces heat load and the ratio of water vapor to feed. For training, validation and testing of the artificial neural networks presented in this thesis set of 700 data were used. Finally, based on the highest correlation coefficient and the lowest mean error, an artificial neural network with 3 layers using 4 neurons in input layer, 9 neurons in the middle layer and 3 neurons in output layer was selected. Function of the neurons was tangent sigmoid and for training the neural network Levenberg-Marquart algorithm was used. The correlation coefficient of real data and output of selected neural network model is 0.9791. Keywords: Modeling, Simulation, Reforming methane, Artificial neural networks