Global control of electrical supply: A variational mode decomposition-aided deep learning model for energy consumption prediction

Global energy consumption has increased significantly in recent decades due to changes in the industrial and economic sectors. Accurate demand estimates are critical for decision-makers to save operation and maintenance costs, improve energy reliability, and make informed decisions for future development. This study evaluates a newly proposed soft technique called Variational Mode Decomposition (VMD) to improve the accuracy of power consumption forecasts. To validate the experimental results, we compared the predicted energy consumption values with measured values from five geographically diverse countries, including developed and developing countries. The study examined different time horizons and performed seasonal evaluations. The VMD-BiGRU and VMD-LSTM models show consistent and superior prediction accuracy, outperforming other models by 20% to 50% on all evaluation measures. In addition, we evaluated the efficiency of VMD-based models over different forecast horizons and find that they are most effective for short-to medium-term forecasts (1 to 12 months). For longer-term forecasts, we recommend combining VMD with specialized techniques. Overall, this study recommends using VMD to forecast electricity consumption in different regions, emphasizing carefully considering forecast horizons for optimal results. (c) 2023 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Automated summary

This study evaluates the application of a new soft technique called Variational Mode Decomposition (VMD) in improving the accuracy of power consumption forecasts. The results show that the VMD-BiGRU and VMD-LSTM models outperform other models by 20% to 50% on all evaluation measures. Additionally, the study finds that VMD is most effective for short-to medium-term forecasts.