Short-Term Photovoltaic Power Forecasting Based on Patch-Transformer Model

Photovoltaic (PV) power generation occupies an increasing proportion in the new energy power generation system. Accurate prediction of PV power is crucial for ensuring the stable operation of the power grid and enhancing the reliability of dispatch operations. However, traditional neural network models perform poorly in predicting the fluctuations in PV power out, leading to an inability to effectively anticipate and manage instability in the power grid. This paper introduces a PV power prediction model based on the innovative Patch-Transformer architecture. By integrating Patch and Transformer algorithms, the model significantly enhances scalability, perception, and generalization capabilities. The performance of model has been rigorously evaluated in spring, summer, autumn, and winter, with the prediction accuracies for PV power generation over the next seven days being 0.9995, 0.9994, 0.9996, and 0.9986, respectively. These results underscore the model's robustness and its potential to significantly improve power prediction in the renewable energy sector, promising enhanced grid reliability.