A Multi-model Fusion Method for Short-term Solar Irradiance Prediction

Abstract To improve the accuracy of prediction, this paper proposes a short-term solar irradiance prediction method based on multi-model fusion by classification. Firstly, the weather classification criteria under single-point conditions are used to achieve automated weather classification. Then, the correlated micro-meteorological information is selected through correlation analysis. Finally, multiple irradiance prediction models are constructed based on classification results and selected features, enhancing the role of weather type features in the prediction process. Experimental results show that the accuracy of this prediction method is superior to that of single-model algorithms.