Electricity Price Forecasting Based on BiGRU-Attention Model

Abstract

With the continuous advancement of electricity market reform, the electricity spot market has been gradually improved, and electricity price forecasting has become increasingly important in power system operation and market decision-making. Due to the strong nonlinearity, volatility, and temporal dependence of electricity price series, it remains difficult to achieve high forecasting accuracy using conventional methods. To address this issue, this paper proposes an electricity price forecasting model based on a Bidirectional Gated Recurrent Unit (BiGRU) and an Attention mechanism. The electricity price data from a certain province in China are selected as the research object. First, time-dimensional features and temporal features are extracted from the original electricity price series, and logarithmic transformation and normalization are adopted to improve data stability and model training effectiveness. Then, the Attention mechanism is introduced to enhance the model’s ability to focus on key temporal information. Finally, the proposed model is compared with XGBoost and GRU models through forecasting experiments. The results show that the BiGRU-Attention model achieves the best performance among the compared models, with a coefficient of determination (R²) of 0.94, a root mean square error (RMSE) of 9.54, and a mean absolute error (MAE) of 7.61. The experimental results demonstrate that the proposed model can effectively improve electricity price forecasting accuracy in the spot market and has good practical application value.