Study on the influencing factors and influencing paths of photovoltaic power generation

Photovoltaic power generation is greatly affected by the weather, and with the growth of photovoltaic installed capacity, large-scale grid-connection poses challenges to the security and stability of the grid. In order to solve this problem, 10 regression models were trained by using 10-fold cross-validation combined with the operating parameters of photovoltaic panels and weather parameters. The optimal four models were selected and the random grid search method was used to adjust their hyperparameters, and then the feature importance of the four models was obtained respectively. Then, three optimal models that are adjusted with Stacking are integrated. Through the study of the importance of features, the influencing factors of photovoltaic power generation are studied. The results showed that the MAPE of the integrated model was 0.0243%. Considering that the integration model may ignore the dependence between factors, this study uses Bayesian networks to construct the influence path of photovoltaic power generation and form a visual network. In the model established above, the paper introduces the operating parameters of photovoltaic power generation panels and presents the important influencing factors and influencing paths of photovoltaic power generation through the model of Stacking and Bayesian network, which provides an effective reference for photovoltaic power generation prediction.