Cross-Border Electricity Price Forecasting with Deep Learning

Cross-Border Electricity Price Forecasting with Deep Learning (Proposals Track)

Hadeer El Ashhab (KIT); Benjamin Schäfer (Karlsruhe Institute of Technology)

NeurIPS 2024 Recorded Talk Cite

Abstract

Accurate electricity price forecasting (EPF) is critical for the efficient operation of energy markets, especially with the increasing integration of renewable energy sources. In this study, we explore the performance of advanced deep learning models, including Long Short-Term Memory (LSTM), vanilla Transformers, Adaptive Fourier Neural Operator (AFNO), and Mamba, in forecasting electricity prices across 16 bidding zones in the European Union. By utilizing a comprehensive dataset that includes cross-border electricity prices and generation data, we compare the effectiveness of these models under different learning strategies, including zero-shot, one-shot, and few-shot learning. We hope our results set a new benchmark for future EPF studies and offer valuable insights into the dynamics of electricity pricing in energy markets.