Load forecasting based on the BWO-RVMD method and the integrated Informer-BiGRU deep learning architecture

Power load forecasting is a key support for ensuring the safe and stable operation of power systems and achieving dynamic balance between supply and demand. However, the volatility and complexity of load fluctuations pose significant challenges to load forecasting. Therefore, an integrated forecasting model based on the White Whale Optimization Algorithm (BWO) is proposed to optimize Rolling Variational Mode Decomposition (RVMD) and fuse the Informer-BiGRU deep learning architecture. The model constructs the power load dataset by performing feature correlation analysis using the Maximum Information Coefficient (MIC). The BWO algorithm is used to adjust the key parameters in RVMD, significantly improving the smoothness and predictability of the signal. The Informer model is employed to effectively capture global temporal patterns and long-term trends, while the BiGRU network accurately captures local temporal changes and short-term fluctuations. Experimental results show that the proposed model outperforms traditional methods across various evaluation metrics, demonstrating superior forecasting performance.