A FA-GWO-GRNN Method for Short-Term Photovoltaic Output Prediction

摘要

High-precision prediction of photovoltaic (PV) output is essential in PV system access to the power grid. To realize the security, stability and economic operation of power system, this paper proposes a hybrid factor analysis, gray wolf optimization, and generalized regression neural network (FA-GWO-GRNN) framework for short-term PV output forecast. In order to reduce the dimension of input features to PV output forecasting, the paper first develops a factor analysis (FA) to extract effective information from meteorological inputs. A generalized regression neural network (GRNN) algorithm is then employed to make the forecast, whose parameters are optimized by the gray wolf optimization (GWO) for its global searching capacity and fast convergence. The proposed GWO-GRNN framework effectively achieves high precision in short-term PV output forecasting, demonstrated in a case study on the measured power of a real world PV plant, which validated the accuracy and applicability of the proposed method in real-world scenarios.