Research on Short-term Load Forecasting Based on VMD-TCN
WANG Shudong
LI Runqing
CAO Wanshui
Abstract：In order to improve the prediction accuracy of the model,this paper adopts a method based on the maximal informa-tion coefficient(MIC).Sparrow algorithm(SSA)optimizes variational mode decomposition(VMD).The prediction model of Tempo-ral convolutional network and temporal pattern attention is combined.Firstly,considering the fluctuation and non-stationary of the original load signal,the VMD optimized by sparrow algorithm is used to decompose the original load sequence into different modal components,and the prediction difficulty of the neural network is reduced by sample entropy reconstruction.Considering the weath-er,electricity price and other influencing factors,MIC is adopted to screen the external features strongly associated with the current moment load signal,so as to achieve feature optimization and dimension reduction.Secondly,the decomposed modal components and the external features are screened by MIC respectively constitute the training set.Finally,the TPA-TCN model of time convolu-tion network based on time mode attention mechanism is constructed to predict.The practical example shows that the proposed pre-diction model can effectively improve the accuracy of prediction.
Keywords：short-term load forecastingsequential convolutional networkvariational modal decompositionmaximum mu-tual information coefficientsample entropytime model attention mechanismsparrow algorithm
Publication Date：2025-01-19
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：7( 96-102 )