Electricity Price Forecasting Based on Multi-Scale Feature Fusion and Hyperparameter-Optimized Bidirectional GRU

To address the strong volatility, nonlinearity, and time-varying nature of electricity prices in the electricity market, this paper proposes a Bidirectional Gated Recurrent Unit model integrating Multi-Scale Feature Fusion and Automated Hyperparameter Optimization (H-O-GRU). The proposed framework incorporates feature-importance gating, multi-scale temporal pooling, and a multi-head self-attention mechanism to model electricity price dynamics across multiple temporal scales.