Short‐Term Load Forecasting of Multi‐Energy in Integrated Energy System Based on Efficient Information Extracting Informer

ABSTRACT With the advancement of the energy revolution and the proposal of carbon peaking and carbon neutrality goals, the integrated energy system (IES) has received increasing attention from researchers. The efficient planning and control of IES cannot be separated from accurate multi‐energy load forecasting, especially short‐term load forecasting (STLF). Based on the above requirements, the transformer‐based method is introduced, and an efficient information extracting informer (EI2) model is proposed to predict the electric, cooling, and heating loads in an IES. Firstly, the feature maps of electric, cold and heat loads are constructed from historical data, and then input to the parameter sharing encoder layer of the proposed STLF model. Secondly, to enable more efficient deep pattern information learning, we have added high‐dimensional MLP layers to the feed forward layers in both the encoder and decoder parts of the joint prediction of electric, cold, and heat loads. As a result, the training model has been optimized. Finally, the predicted values for electric, cold, and heat loads are output through three independent decoders. The proposed EI2 STLF model effectively increases the prediction accuracy of multi‐energy loads in an IES, as verified and compared with other models using actual examples.