Day-Ahead Zonal Electricity Price Forecasting using 1D-LSTM with Neighbouring Zones Data

In 2022 the electricity market saw a dramatic increase in electricity prices and increased price volatility. In this new situation, the need for instruments that can provide good forecasts in the day-ahead market has become acute. This work documents research, which attempts to use neighbouring zonal prices as a basis for forecasting. Different cross-border connections allow the transmission of electricity between price zones. Depending on the capacities of these transmission lines price covariances can be observed. The work reported here tries to exploit this for forecasting purposes. In this regard, this paper proposes a method to forecast the electricity prices of a particular zone (NO2 alias Kristiansand) in Norway by taking the electricity prices of other zones into account. Other zones are the zones of Norway and also from the neighbouring countries (UK, Sweden, Belgium, Austria, Netherlands and Denmark). A total of 10 time-series are considered as input for the day-ahead price forecasting of NO2 which makes it a multi-variate input. In this paper, we are using the 1D- LSTM for forecasting the day-ahead prices of NO2. We also compared the performance of this model with complex models like CNN and hybrid CNN-LSTM. The process of data pre-processing for fitting the LSTM model is also explained in detail. The results show the significant performance of a simple and computationally efficient LSTM model over other complex models for such multi-variate time-series datasets. Thus, these models can be utilized in future when there is complex data available from several different zones and other financial markets.