Ultra-Short-Term Load Forecasting for Customer-Level Integrated Energy Systems Based on Composite VTDS Models

A method is proposed to address the challenging issue of load prediction in user-level integrated energy systems (IESs) using a composite VTDS model. Firstly, an IES multi-dimensional time series decomposed into multiple intrinsic mode functions (IMFs) variational decomposition (VMD). Then, each IMF, along with other influential features, subjected data dimensionality reduction and clustering denoising t-distributed stochastic neighbor embedding (t-SNE) fast density-based spatial applications noise (FDBSCAN) perform major feature selection. Subsequently, reduced denoised are reconstructed, time-aware long short-term memory (T-LSTM) artificial neural network employed fill missing by incorporating interval information. Finally, selected multi-factor used as input support vector regression (SVR) model optimized quantum particle swarm optimization (QPSO) algorithm for prediction. Using measured from specific at Tempe campus Arizona State University, USA, case study, comparative analysis between approaches conducted. The results demonstrate that this study achieved higher accuracy forecasting IES’s loads.