Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

A Transformer approach for Electricity Price Forecasting

Created by
  • Haebom

Author

Oscar Llorente, Jose Portela

Outline

This paper presents a novel Electric Power Price Forecasting (EPF) method using a pure Transformer model. Unlike other methods, it demonstrates that the attention layer alone can sufficiently capture temporal patterns, without using recurrent neural networks combined with attention mechanisms. Furthermore, we utilize the open-source EPF toolbox to provide a fair comparison of models, and we make the code publicly available to enhance the reproducibility and transparency of EPF research. The results demonstrate that the Transformer model outperforms existing methods and is a promising solution for reliable and sustainable power system operation.

Takeaways, Limitations

•
Takeaways:
◦
Demonstrating the effectiveness of electricity price prediction using a pure transformer model.
◦
We demonstrate that attention mechanisms alone can achieve sufficient performance for power price prediction.
◦
Improving reproducibility and transparency of EPF research through open-source toolboxes and code disclosure.
◦
Presenting the potential to contribute to the operation of a reliable and sustainable power system.
•
Limitations:
◦
Further validation is needed on the generalization performance and robustness of the Transformer model presented in the paper across various datasets.
◦
A more comprehensive comparative analysis with other advanced electricity price prediction models is needed.
◦
Further research and verification are needed for application to actual power system operation.
👍
Made with Slashpage