A neural network approach for burn-up calculation and its application to the dynamic fuel cycle code CLASS

•Development of a neural network model to predict the requiered plutonium content.•The accuracy of this model is very good (0.37% of error).•Development of a neural network model to predict evolution of average cross sections.•Predictions allow calculating fuel depletion quickly and with a very good accuracy.•This approach has been applied to the PWR MOX case in a dynamic fuel cycle code.