Short Term Power Load Forecasting Based on PSVMD-CGA Model

Short-term power load forecasting is critical for ensuring system stability. A new algorithm that combines CNN, GRU, and an attention mechanism with the Sparrow to optimize variational mode decomposition (PSVMD–CGA) proposed address problem of effect random fluctuations on accuracy short-term forecasting. To avoid manual selection VMD parameters, adopted by decomposing data into multiple subsequences, thus significantly reducing volatility data. Subsequently, CNN (Convolution Neural Network) introduced fact GRU (Gated Recurrent Unit) difficult use extract high-dimensional features. Finally, selected when sequence too long, important information cannot be weighted highly. On basis original model, PSVMD–CGA model suggested in this paper has been considerably enhanced. MAE dropped 288.8%, MAPE 3.46%, RMSE 326.1 MW, R2 risen 0.99. At same time, various evaluation indicators show outperforms SSA–VMD–CGA GA–VMD–CGA models.