Abstract:This paper presents a novel approach to electricity price forecasting (EPF) using a pure Transformer model. As opposed to other alternatives, no other recurrent network is used in combination to the attention mechanism. Hence, showing that the attention layer is enough for capturing the temporal patterns. The paper also provides fair comparison of the models using the open-source EPF toolbox and provide the code to enhance reproducibility and transparency in EPF research. The results show that the Transformer model outperforms traditional methods, offering a promising solution for reliable and sustainable power system operation.

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve This paper aims to address the issue of Electricity Price Forecasting (EPF). Specifically, the authors propose a new method based on a pure Transformer model to predict electricity prices for the next 24 hours. Compared to traditional statistical methods and other deep learning models, this method can better capture the complex dynamics in time series, thereby providing more accurate prediction results. ### Background and Motivation 1. **Development of the Electricity Market**: - With the competition and deregulation in the electricity market, accurate and efficient electricity price forecasting has become increasingly important. - These forecasts are a fundamental decision-making tool for market participants such as electricity producers, consumers, traders, and grid operators. 2. **Impact of Renewable Energy**: - The intermittency and unpredictability of renewable energy increase the volatility and complexity of electricity prices. - Therefore, more sophisticated forecasting methods are needed to capture these dynamics and provide accurate predictions. 3. **Limitations of Existing Methods**: - Traditional statistical methods find it difficult to capture the complex nonlinear dynamics of electricity prices in some cases. - Although Recurrent Neural Networks (RNN) and Long Short-Term Memory networks (LSTM) perform well in certain tasks, they are often combined with other attention mechanisms, increasing the complexity of the model. 4. **Advantages of the Transformer Model**: - The Transformer model can effectively capture long-term dependencies in time series through the attention mechanism. - The authors propose a pure Transformer model, not combined with other recurrent networks, to verify whether the attention layers alone are sufficient to capture time patterns. ### Main Contributions 1. **Proposing a Pure Transformer Model**: - The authors designed a pure Transformer model specifically for electricity price forecasting. - This model uses only the attention mechanism and does not rely on other recurrent networks, demonstrating the effectiveness of attention layers in time series forecasting. 2. **Fair Benchmarking**: - The model comparison was conducted using an open-source EPF toolbox, ensuring the reproducibility and transparency of the experiments. - The code is provided to enhance the reproducibility and transparency of the research. 3. **Superior Performance**: - Experimental results show that the proposed Transformer model outperforms traditional methods on multiple datasets, providing a reliable solution. ### Conclusion By proposing and validating the application of a pure Transformer model in electricity price forecasting, the authors demonstrate the powerful capability of the attention mechanism in time series forecasting. This method not only improves the accuracy of predictions but also provides strong support for the reliable and sustainable operation of power systems.

A Transformer approach for Electricity Price Forecasting

An Intra-Day Electricity Price Forecasting Based on a Probabilistic Transformer Neural Network Architecture

Interpretable Transformer Model for Capturing Regime Switching Effects of Real-Time Electricity Prices

Transformers for Energy Forecast

Probabilistic electricity price forecasting based on penalized temporal fusion transformer

Data-Driven Techniques for Short-Term Electricity Price Forecasting through Novel Deep Learning Approaches with Attention Mechanisms

Transformer-Based Model for Electrical Load Forecasting

A Hybrid Approach for Probabilistic Forecasting of Electricity Price.

Transformer-based power system energy prediction model

A New State-of-the-Art Transformers-Based Load Forecaster on the Smart Grid Domain

A Novel Multiscale Transformer Network Framework for Natural Gas Consumption Forecasting

Forecasting electricity prices from the state-of-the-art modeling technology and the price determinant perspectives

Artificial intelligence-based power market price prediction in smart renewable energy systems: Combining prophet and transformer models

ETSformer: Exponential Smoothing Transformers for Time-series Forecasting

Using Generative Pre-Trained Transformers (GPT) for Electricity Price Trend Forecasting in the Spanish Market

Short-Term Electricity Load Forecasting Using the Temporal Fusion Transformer: Effect of Grid Hierarchies and Data Sources

Power substation load forecasting using interpretable transformer-based temporal fusion neural networks

Deep Learning-Based Electricity Price Forecast for Virtual Bidding in Wholesale Electricity Market

A feature extraction- and ranking-based framework for electricity spot price forecasting using a hybrid deep neural network

TransPVP: A Transformer-Based Method for Ultra-Short-Term Photovoltaic Power Forecasting

Itransformer: Inverted Transformers Are Effective for Time Series Forecasting