Multivariate Load Forecasting for Regional Integrated Energy Systems Based on Error Compensation

Abstract

For the characteristics of multiple loads coupled and non-linear in the regional integrated energy system, an error compensated LSTM-GRU based multivariate load forecasting method for regional integrated energy systems is proposed in the paper. First, the processing of the residual data and abnormal data in the historical data, the factors that have a greater impact on the multi-load load of this paper are screened out through grey correlation analysis, the load is predicted using a traditional Long and Short-Term Memory (LSTM) network. Secondly, the prediction error is trained by the Gated Recurrent Unit (GRU) to obtain the error compensation value. Finally, a more accurate load forecast is obtained by reconstructing the load forecast value and the error forecast value. The feasibility of the model proposed in this paper is verified through examples and comparisons with other models.