Research onfine-grained decomposition of regional electricity demand and factor matching technology based on GAM-DLM
ZHAO Qian
ZHENG Kangzhen
QIU Jia
CAO Youxia
WANG Pin
Abstract：Electricity demand is subject to the coupled impacts of multiple factors,including economic conditions,meteorological variables,and business expansion.However,existing studies struggle to effectively decompose the independent effects of each factor and lack quantitative methods for measuring their influence weights.This results in difficulties in accurately attributing electricity anomalies and limited reliability of demand forecasting.To identify and quantify the impacts of different factors on changes in electricity consumption,an interpretable driving component decomposition system is constructed.The GAM(Generalized Additive Model)combined with the DLM(Dynamic Linear Model)is adopted to decompose the total electricity demand,and a matching mechanism between each component and its influencing factors is established based on the IStOMP(Improved Orthogonal Matching Pursuit)algorithm,thereby quantifying the weights of key factors.Taking the full-year daily electricity consumption data of the new energy vehicle whole-vehicle manufacturing industry in a certain province as a verification case,the results demonstrate that this approach can effectively isolate the independent effects of multiple factors and achieve weight quantification,providing a methodological basis for the precise management and predictive analysis of electricity demand.
Keywords：electricity demand forecastingGAMDLMdriving-component decompositionquantification of influencing factors
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：7( 89-95 )