A Renewable Energy Farm Forecasting Model Based on Ensemble Learning
Wanrong Bai, X. George Xu, Zhicheng Ma, Jinxiong Zhao, Lei Zhang, Xun Bao · 2024
With the continuous increase in the penetration rate of renewable energy sources such as wind and light in the power grid, their inherent randomness, volatility, and intermittency pose challenges to the safe and stable operation of the power system. Regional wind and solar power prediction is an effective measure to address the above issues. However, the prediction performance of a single model still has some limitations. As an effective method, ensemble learning can improve the prediction accuracy and enhance the generalization ability of the model. This paper proposes a wind and solar cluster power prediction model based on feature dimensionality reduction using the Stacking integration framework, which integrates Long Short-Term Memory (LSTM), Extreme Gradient Boosting (XGBoost), Least Absolute Shrinkage and Selection Operator (LASSO), and Support Vector Machine (SVM) models, and the model hyperparameters are optimized using grid search method. The results show that the proposed fusion model performs better in evaluation indicators such as Root Mean Squared Error (RMSE) and Mean Average Error (MAE) than other models. It has strong robustness and enhances the model's generalization ability, verifying the effectiveness of the proposed fusion model and effectively improving the wind and solar power prediction performance.