Spatio-Temporal Cooperative Probability Forecasting Method for Distributed Photovoltaic Output Based on iDGA-LSTM
Due to the significant uncertainty of photovoltaic output,the high proportion of distributed photovoltaic integration will pose a huge challenge to the safe and stable operation of the new power system.Accurate and reliable distributed photovoltaic output forecasting is of great significance to improve the security of the new power system operation.Under this background,considering the spatio-temporal coupling characteristics and uncertainty of distributed photovoltaic output,this paper proposes a spatio-temporal cooperative probability forecasting method of distributed photovoltaic output based on the improved quantile regression based dynamic graph attention and the long short-term memory network(iDGA-LSTM).Firstly,considering the spatial correlation of distributed photovoltaic output across a wide area,a distributed photovoltaic spatial feature extraction and aggregation model based on the graph attention network is constructed.Secondly,regarding the extracted distributed photovoltaic spatial correlation features,a distributed photovoltaic spatio-temporal coupling feature extraction model based on the dynamic graph attention and the long short-term memory network is constructed.Then,taking into account the spatio-temporal coupling characteristics of distributed photovoltaic output and combining them with numerical meteorological forecasting features,a probability forecasting model for distributed photovoltaic output based on improved quantile regression is constructed.Finally,the proposed forecasting method is verified with some actual distributed photovoltaic data.The simulation results show that the proposed method improves the reliability and accuracy of distributed photovoltaic output forecasting and provides references for the power system operation strategies with different risk levels.
万方数据
维普
