- Conference Article
- 10.1109/appeec66370.2025.11382127
Improving Electricity Demand Forecasting in Regional Microgrids Using a Bayesian Hybrid Model with LightGBM
- Dec 02, 2025
- Hidetaka Baba + 5 more +5
Accurate electricity demand forecasting is essential for the efficient operation of energy management systems in regional microgrids. Various forecasting methods have been proposed, including statistical approaches, artificial intelligence (AI) techniques, and hybrid models that combine both. In our previous work, we developed a forecasting method based on Bayesian inference that achieves high prediction accuracy, ensures explainability, enables real-time processing, and allows for the quantification of uncertainty. By incorporating distinct demand patterns between weekdays and holidays, and applying Bayesian inference to historical data collected during the same time period on various past days, the method provides valuable insights for system operators. In this study, we further improve forecasting performance by introducing a hybrid approach that integrates the Bayesian method with other predictive models. A case study using actual demand data demonstrates that the proposed hybrid method statistically outperforms conventional models in key performance metrics, validating its effectiveness and practical applicability.
Read more