Real-Time Energy Price Forecasting using BiLSTM-Autoencoder Deep Learning Model

Electricity price forecasting is crucial in the energy markets as it affects various stakeholders, including electricity consumers, retailers, suppliers, and grid operators. Still, it is enormously challenging due to high volatility, rapid spikes, seasonality, and others. Traditional forecasting methods often fail to capture the complex dynamics of electricity pricing, especially in real-time markets. Therefore, this paper introduces a new approach to forecasting real-time electricity prices by synergizing a bi-directional Long Short-Term Memory (BiLSTM) network with an autoencoder. Integrating BiLSTM into autoencoders allows a nuanced understanding of temporal data, significantly improving forecast accuracy. The result demonstrates lower Mean Absolute Error, Root Mean Square Error, and Mean Absolute Percentage Error compared to standard models, highlighting the potential of this hybrid approach in complex data-driven forecasting scenarios.