Distributed Photovoltaic Power Prediction Based on Spatiotemporal Attention Convolutional Networks under Weather-free Conditions

With the massive integration of distributed photovoltaic (PV) systems into the power grid, accurate forecasting of distributed PV power generation is of great significance for the stable operation of power systems. Existing methods often neglect the spatiotemporal correlations among distributed PV systems, making it difficult to effectivelyutilize spatiotemporal feature information to improve power forecasting accuracy. Considering the strong spatiotemporal correlations among distributed PV outputs and the difficulty in obtaining meteorological data, this paper proposes a distributed PV power forecasting method based on Spatial- Time-Attention, Graph Convolutional Network (GCN), and Temporal Convolutional Network (TCN). This method achieves accurate PV power prediction under weather-free conditions based solely on historical power data. First, the spatial-time attention mechanism is used to extract the spatiotemporal dependencies of the distributed PV cluster output. Next, GCN and TCN are utilized to construct a spatiotemporal graph convolutional network for prediction training, obtaining the forecasting results. Finally, a simulation is conducted based on real data from a distributed PV power plant in a certain region of China. The case study results show that the proposed model has higher accuracy compared to traditional deep learning models.