Research on Photovoltaic Power Generation Power Prediction Method Based on Spatio-Temporal Graph Convolutional Network Coupled with Tramsformer

With the development of renewable energy, photov oltaic power generation forecasting plays a crucial role in new ene rgy consumption. Traditional prediction methods still have obvio us shortcomings in modeling spatio-temporal dependence and im proving model generalization ability, and it is difficult to adapt to high-frequency fluctuations and multi-source heterogeneous data environments. Therefore, this paper proposes a photovoltaic pow er generation prediction model (STGCN-Transformer) that comb ines spatiotemporal graph convolutional network and transforme $\mathbf{r}$ encoder. Firstly, the model effectively captures the spatial correl ation between photovoltaic power plants through the spatio-temp oral graph convolutional network (STGCN). Secondly, a long-ter m time-dependent model is established based on the self-attention mechanism of Transformer. Finally, through the in-depth modeli ng of historical power data and meteorological data, the model ca n comprehensively consider the impact of environmental factors o n photovoltaic power. The experimental results show that the pro posed method significantly improves the accuracy of photovoltaic power generation power prediction.