Forecasting day-ahead electricity prices for the electricity market with dynamic time period
Accurate forecasting of day-ahead electricity prices is crucial for market participants in the electricity market due to their impact on trading strategies, cost management and grid stability. Unlike traditional daily hourly forecasts, sub-hourly (e.g., 15 min) forecasts capture intra-hour ramps and...
| Autores: | , , , , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Centre Tecnològic de Telecomunicacions de Catalunya (CTTC) |
| Repositorio: | r-CTTC. Repositorio Institucional Producción Científica del Centre Tecnològic de Telecomunicacions de Catalunya (CTTC) |
| OAI Identifier: | oai:cttc.fundanetsuite.com:p8778 |
| Acceso en línea: | https://cttc.fundanetsuite.com/Publicaciones/ProdCientif/PublicacionFrw.aspx?id=8778 |
| Access Level: | acceso abierto |
| Palabra clave: | Day ahead market Weighted ensemble learning Electricity price forecasting Electricity market |
| Sumario: | Accurate forecasting of day-ahead electricity prices is crucial for market participants in the electricity market due to their impact on trading strategies, cost management and grid stability. Unlike traditional daily hourly forecasts, sub-hourly (e.g., 15 min) forecasts capture intra-hour ramps and volatility caused by high wind and solar penetration. They enable faster imbalance detection and mitigation, more flexible operations such as battery energy storage and demand response, and improved bidding performance. There is a lack studies in the literature on forecasting day-ahead electricity prices over such shorter time period. In this paper, a novel AutoGluon-based framework for electricity price forecasting is presented. This framework solves optimization problem that minimizes the mean absolute error of a proposed weighted ensemble forecasting model for different electricity price periods. It dynamically adapts to different time periods during the day and improves the forecast accuracy by feature extraction and fine-tuning the AutoGluon parameters. An initial analysis of electricity price fluctuations over different time periods shows the importance of capturing temporal price patterns. The proposed method newly generates data set and it utilizes feature extraction to identify the key price determinants and employs the AutoGluon machine learning model for more granular price forecasting. Furthermore, the framework is extended to predict electricity prices for shorter time intervals, which improves the adaptability to market dynamics. The results show that the proposed approach improves the prediction accuracy compared to conventional models and provides valuable insights to market operators and participants. |
|---|