Short-term power load forecasting based on VMD and LSTM-Transformer

In order to accurately capture the nonlinear fluctuation characteristics and long-term dependence of short-term power load, a fusion prediction model combining variational modal decomposition (VMD), long-short-term memory network (LSTM) and Transformer is proposed. Firstly, the original power load sequence is decomposed into multiple modal components by VMD, and reorganized into feature sequences after passing the smoothness test; then LSTM is used to extract the local time-series features, and the global dependencies are captured by Transformer's self-attention mechanism; finally, the prediction results are outputted by the fully-connected layer. Experiments show that the model is better than the comparison model in short-term power load forecasting, which verifies its effectiveness.