Ensemble Learning for Short-Term Load Forecasting: Integrated Methods of Multiple Individual Models

In this paper, ensemble learning-based short-term load forecasting (STLF) method that integrates multiple single models was proposed. Considering the importance of short-term load forecasting in power systems and its challenges, such as data complexity and nonlinear relationships, integrated learning strategies are adopted, which can overcome the limitations of individual models as well. The paper selects a variety of well-performing single models such as linear regression, neural networks etc. and combines them with the ensemble method under the Stacking framework in order to form a powerful integrated prediction model, which can effectively improve the prediction accuracy and stability. The experimental results indicated that the proposed approach performing higher prediction accuracy in STLF compared to individual model. The classification weighted ensemble model based on MLP as the best combined model with a MAPE of ${4. 4 2 \%}$ in prediction.