Search

Advancing Electricity Price Forecasting: A Comprehensive Framework Integrating Ensemble Learning, Deep Learning, and Large Language Models

Author(s)

Guanlin Tang*

Affiliation(s)

Xi’an Jiaotong-Liverpool University, Suzhou, China *Corresponding Author

Abstract

Electricity price forecasting (EPF) is essential for daily energy market operations, helping participants optimize bidding strategies and make informed decisions by predicting price fluctuations accurately. Electricity price forecasting (EPF) plays a pivotal role in the operational dynamics of energy markets, facilitating optimized bidding strategies and informed decision-making for market participants through accurate prediction of price fluctuations. This study proposes a comprehensive methodological framework for EPF, integrating four principal approaches: ensemble learning, deep learning, large language models (LLMs), and the fusion of LLMs with retrieval-augmented generation (RAG). Ensemble learning leverages the complementary strengths of base models such as ARIMA, Support Vector Regression (SVR), and Random Forest Regression, while deep learning architectures, including LSTM, GRU, and Transformer, are employed to model complex temporal dependencies and long-term patterns inherent in electricity price data. Large language models, fine-tuned with domain-specific datasets, enhance predictive accuracy by harnessing advanced natural language processing capabilities. The RAG-enhanced LLM approach further refines forecasting precision by incorporating external knowledge retrieved from a structured knowledge graph. Collectively, these methodologies form a robust and versatile framework designed to address the inherent complexities and uncertainties of electricity markets, thereby supporting efficient power dispatch and market operations. This research underscores the evolving landscape of EPF methodologies and their potential to enhance grid reliability, operational efficiency, and the integration of renewable energy sources, ultimately contributing to the development of a more resilient and sustainable energy ecosystem.

Keywords

Electricity Price Forecasting (EPF); Large Language Models (LLMs); Renewable Energy Integration; Ensemble Learning

References

[1]A. Kitsatoglou, G. Georgopoulos, P. Papadopoulos, and H. Antonopoulos, “An ensemble approach for enhanced Day-Ahead price forecasting in electricity markets,” Expert Systems with Applications, vol. 256, p. 124971, Dec. 2024, doi: https://doi.org/10.1016/j.eswa.2024.124971. [2] G. Wang, E. Sbai, M. S. Sheng, and M. Tao, “News sentiment, climate conditions, and New Zealand electricity market: A real-time bidding policy perspective,” Energy, vol. 318, p. 134784, Mar. 2025, doi: https://doi.org/10.1016/j.energy.2025.134784. [3] X. Serrano-Guerrero, M. Briceño-León, J.-M. Clairand, and G. Escrivá-Escrivá, “A new interval prediction methodology for short-term electric load forecasting based on pattern recognition,” Applied Energy, vol. 297, p. 117173, Sep. 2021 [4] K. Li, W. Huang, G. Hu, and J. Li, “Ultra-short term power load forecasting based on CEEMDAN-SE and LSTM neural network,” Energy and Buildings, p. 112666, Nov. 2022, doi: https://doi.org/10.1016/j.enbuild.2022.112666.