A Short-term Power Load Forecasting Method Based on An Improved VMD-LSTM Algorithm

Abstract Accurate power load prediction is a crucial basis to ensure the balance of supply and demand. This paper provides modified variational modal decomposition (VMD) - long short-term memory (LSTM) algorithm, using genetic algorithm (GA) iteratively optimize number layers penalty factor parameters, VMD obtain components, fuse them with filtered features. Then reconstruction data obtained through normalization according parameters original characteristic set. Finally, LSTM constructed excavate temporal characteristics results. Simulation fully demonstrates superior performance this model in terms effect.