Electricity Price Forecasting on Day Ahead Market via a Multivariate CNN-LSTM Model

The day-ahead market (DAM) is vital for maintaining the demand-supply balance necessary for the proper operation of AC power systems. Various stakeholders within the energy sector, such as consumers, utility companies and investors heavily rely on a precise electricity price forecasting. Nonetheless, factors like load behavior, renewable energy generation, holidays, and weather conditions contribute to the electricity price uncertainty and volatility accession. Thus, modern techniques must be approached for obtaining accurate prediction results. A multivariate deep learning model using a hybrid convolutional-long short-term memory network is proposed in this paper to forecast future electricity prices on the United Kingdom electricity market. For increased accuracy, influence factors such as historical data for power demand, generation, generation mix (wind, solar, storage units, gas, fossil), weather (temperature, solar radiation, wind speed), day ahead prices and seasonality features extracted from the initial data are implemented, resulting in a percentage error of ${10\%}$.