Distributed energy power prediction of the variational modal decomposition and Gated Recurrent Unit optimization model based on the whale algorithm

Abstract

Based on the load characteristics of industrial parks, this paper optimizes the load prediction model of industrial parks, in order to provide data support for the research of scheduling algorithms. Aiming at the influence of Variational Modal Decomposition (VMD) modal parameters K and penalty factor α on the prediction accuracy of short-term power load forecasting method based on variational modal decomposition (VMD) and Gated Recurrent Unit (GRU), Whale Optimization Algorithm (WOA) was proposed. In this paper, WOA is used to optimize VMD decomposition parameters. Then, the optimized decomposition parameters decompose the original load data, and a set of more regular modal components are obtained. Finally, each mode decomposed by the WOA-VMD algorithm was sent to GRU for power prediction. And the prediction results were superimposed and reconstructed to obtain the final result. WOA optimized the search process and found more appropriate parameters for better prediction results. After whale algorithm optimization, Root Mean Square Error (RMSE) decreased from 108.8 MW to 38.29 MW, Mean Absolute Error (MAE) decreased from 83.09 MW to 24.26 MW.

Keywords

• Short-term load prediction
• Gated recurrent unit
• Variational mode decomposition
• Model combination
• Deep learning