Multi-Energy Load Forecasting via Two-Stage Decomposition and Informer-LSTM Hybrid Model

Le résumé fourni par la source

Accurate forecasting of multi-energy loads is vital for the optimal scheduling and economic operation of integrated energy systems. However, existing methods often struggle to fully capture trend and seasonal patterns in load data and are vulnerable to high-frequency noise and measurement errors, limiting their predictive performance. To address these challenges, this paper proposes a novel two-stage decomposition and Informer-Long Short-Term Memory (LSTM) hybrid model for multi-energy load forecasting. The proposed method first applies seasonal-trend decomposition using Loess to extract the trend and seasonal components. The residual component is then further decomposed using variational mode decomposition to isolate intrinsic mode functions with multi-scale features. The hybrid forecasting model consists of two branches: the Informer network, which captures long-term dependencies through attention mechanisms, and the LSTM network, which focuses on short-term temporal dynamics. The outputs of both branches are concatenated and fused through a fully connected layer to generate the final predictions. Experiments show that the proposed model achieves excellent performance in multi-energy load forecasting.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.