In this paper a new control system on basis of group method for data handling neural networks (GMDH-NNs) is designed for voltage and power regulation in the photovoltaic (PV)/Fuel/Battery systems. The dynamics of all subsystems are considered to be fully uncertain. The suggested GMDH-NN is learned using online tuning rules that are concluded through the robustness investigation. The challenging operation conditions such as variable unknown dynamics, unknown temperature and irradiation and suddenly changes in output load are taken into account and are handled by suggested control system. The superiority of the suggested method is shown by simulation in several scenarios and comparison with other techniques.
