Adaptive Online Learning with LSTM Networks for Energy Price Prediction

TL;DR

The paper tackles day-ahead electricity price forecasting in California using an LSTM-based model that integrates historical prices, weather data, and generation mix. It introduces a composite loss $Loss = MAE + \alpha JSD + \beta SM$ to align predicted distributions with actual ones and enforce smoothness, and an adaptive online learning scheme for incremental updates to maintain relevance. Empirical results show that online learning outperforms static and dynamic baselines, especially when fuel mix and same-day features are included, with notable reductions in $MSE$, $MAE$, and $RMSE$ compared to traditional benchmarks like SVR and ARIMA. The framework provides a robust, real-time forecasting tool for grid operators and market participants facing volatile, non-linear electricity prices in dynamic markets.

Abstract

Accurate prediction of electricity prices is crucial for stakeholders in the energy market, particularly for grid operators, energy producers, and consumers. This study focuses on developing a predictive model leveraging Long Short-Term Memory (LSTM) networks to forecast day-ahead electricity prices in the California energy market. The model incorporates a variety of features, including historical price data, weather conditions, and the energy generation mix. A novel custom loss function that integrates Mean Absolute Error (MAE), Jensen-Shannon Divergence (JSD), and a smoothness penalty is introduced to enhance the prediction accuracy and interpretability. Additionally, an online learning approach is implemented to allow the model to adapt to new data incrementally, ensuring continuous relevance and accuracy. The results demonstrate that the custom loss function can improve the model's performance, aligning predicted prices more closely with actual values, particularly during peak intervals. Also, the online learning model outperforms other models by effectively incorporating real-time data, resulting in lower prediction error and variability. The inclusion of the energy generation mix further enhances the model's predictive capabilities, highlighting the importance of comprehensive feature integration. This research provides a robust framework for electricity price forecasting, offering valuable insights and tools for better decision-making in dynamic electricity markets.

Figures (5)

• Figure 1: An Overview of the Model Architecture
• Figure 2: Adaptive online learning framework
• Figure 3: Prediction results for the first 3 months of the test set - The Static Model
• Figure 4: Comparison of Actual vs. Predicted Electricity Prices Before and After Applying Custom Loss Function. The top row shows overall time series predictions, while the bottom row highlights hourly predictions for two specific days.
• Figure 5: Results and Residuals Comparison Between Dynamic and Online Learning Models