A Holistic Approach to Power Systems Using Innovative Machine Learning and System Dynamics

The digital revolution requires greater reliability from electric power systems. However, predicting the growth of electricity demand is challenging as there still much uncertainty in terms demographics, industry changes, and irregular consumption patterns. Machine learning has emerged a powerful tool, particularly with latest developments deep learning. Such tools can predict and, thus, contribute to better decision-making by energy managers. it important recognize that are no efficient methods for forecasting peak growth. In addition, features add complexity, such climate change economic growth, take time model. Therefore, these new be integrated other proven used model specific system structures, dynamics. This research proposes unique framework support decision-makers dealing daily activities while attentively tracking monthly demand. approach integrates advances machine integration potential more precise forecasts, which help develop strategies deal supply variations. A real-world case study was comprehend needs environment effects COVID-19 on systems; also helps demonstrate use leading-edge tools, convolutional neural networks (CNNs), Three well-known CNN variants were studied: multichannel CNN, CNN-LSTM, multi-head CNN. found outperformed all models, an R2 0.92 MAPE value 1.62% month-ahead consists one main processes four input separate channel, resulting feature map. Furthermore, dynamics introduced sector’s dynamic behavior (i.e., residential, commercial, government demands, etc.). calibrated reproduced historical data curve fairly well between 2005 2017, 0.94 4.8%.