Forecasting Electricity Price During Extreme Events Using a Hybrid Model of LSTM and ARIMA Architecture

Forecasting electricity prices during extreme events is crucial for grid stability, cost management, and risk mitigation in the energy sector. This study presents a novel hybrid forecasting framework that integrates Long Short-Term Memory (LSTM) and Autoregressive Integrated Moving Average (ARIMA) models, enhanced with the Holt-Winters method. The issues caused by extreme events in energy markets, where abrupt and unpredictably changing price variations are typical, are addressed by this hybrid architecture. The ARIMA model adds statistical rigour to account for subtle underlying patterns, while the LSTM component captures intricate temporal connections and quick price fluctuations linked to intense occurrences. The Holt-Winters approach further enhances the model’s ability to handle seasonality and trend variations, which become especially pronounced during extreme events. This paper intends to present the current project development stage and guide research on precisely investigating optimal predictive modelling methods in extreme event scenarios.