Energy Management of isolated DC microgrids with hybrid batteries-hydrogen storage system using Model Predictive Control and Wavelet Neural Networks based forecasts

The intermittency of renewable energy production is still a serious impediment for their development, and thus must be mitigated by efficient energy storages and adapted energy management systems. In this context, a novel control strategy for an islanded microgrid with a hybrid batteries-hydrogen energy storage is proposed. It is a model predictive control which originality lies simultaneously in the objective function formulation that does not focus on economic considerations and the use of wavelet neural network as a forecasts model. In order to validate our approach, three simulations have been done under realistic working conditions, comparing performances of the control with a classical ruled based control, and evaluating its potential for improvement. The results confirm that this method outclasses the ones usually used, but can still be enhanced by improving the forecast model.