Photovoltaic power prediction based on multi-layer fusion model

Abstract Accurate forecasting of photovoltaic power output can provide technical guidance for smooth grid connection of photovoltaic power plants and avoid grid energy quality problems caused by large fluctuations in photovoltaic power. Therefore, it is very important to obtain more accurate photovoltaic energy forecast data. The traditional photovoltaic power forecasting model has the problems of low forecast accuracy and hysteresis.to further improve the prediction accuracy of pv power, this paper proposes a multi-layer fusion PV power prediction model based on dynamic modification of the weight parameters of each model. The model's benchmark algorithms include XGBoost, LigthGBM and ConvLSTM. In the case study, real photovoltaic station data were used to verify the simulation,The forecast results show that the model proposed in this paper has a better effect than any single model and has strong practicability.