This article presents a alternative synchronization method for three phase systems based on Artificial Neuronal Networks. The training signals are obtained from a synchronism method designed with a conventional control whose objective is to achieve a sampling frequency at a N times higher frecuency with regard to the input signals. As a consequence of the complexity of the system, it is modelled with two different neural networks. The objective of the first one is to estimate the input signal phase and the objective of the second one is to generate a variable sampling frequency. The system is evaluated with typical disturbances, obtaining a similar behaviour in comparison with the conventional system. MATLAB simulations results are presented.