RNN-based Photovoltaic Power Identification Method for Distribution Networks

To ensure the safe operation of distribution networks with more and more photovoltaic (PV), it is crucial to study the PV output power identification technology. Thus, the ability of the system to withstand random fluctuations will be enhanced and the stability of distribution network will also be improved effectively. In this paper, a standard three-layer recurrent neural network (RNN) is proposed to build a nonlinear mapping model for efficiently identifying PV power. RNN can fit the nonlinear mapping relationship between the intensity of illumination and PV output power within the distribution station area effectively. The distributed PV output power from the power measurement data of the distribution station area is successfully separated. Finally, a case study of Yulara Solar power plant with 1.8 MVA in Australia is used to validate the feasibility and effectiveness of the proposed method.