Identification and peak load forecasting of sensitive electricity-consuming industries under energy-climate challenges
WEI Yusi
ZHOU Ying
SHANG Yongpeng
WU Yajie
JI Ling
Abstract：Under the dual pressures of carbon neutrality and increasing climate extremes,enhancing the adaptability of regional power systems and the resilience of energy-intensive industries has become essential for energy security.This study proposes an integrated forecasting framework that combines feature extraction and model optimization.The Uniform Information Coefficient(UIC)is first used to identify electricity-consuming industries highly sensitive to meteorological factors.Then,Variational Mode Decomposition(VMD)and Sample Entropy(SE)are applied to extract multi-scale load features,and Grey Relational Analysis(GRA)is used to select key meteorological inputs.A Long Short-Term Memory(LSTM)neural network is developed to model each component and integrate the outputs for daily peak load forecasting.Using the non-metallic mineral products industry in Shandong Province as a case,the model shows superior accuracy,trend tracking,and robustness compared to benchmark models.The results provide technical support for power dispatch,load risk warning,and operational management of high-energy-consuming industries such as coal.
Keywords：daily peak load forecastingvariational mode decompositionlong short-term memory neural networkhybrid modelenergy security
Publication Date：2025-11-28
Online Publishing Date：2025-12-18(First online date of this platform, not the publication date of the document)
Pages：9( 15-23 )