Comprehensive optimization strategy research on power load control and electricity carbon metering based on graph neural networks

Predicting and controlling power loads play a vital role in the energy market, supporting efficient administration and maintenance of power systems. Accurate power load control ensures balanced energy production and demand management. Effective power forecasting is essential for scheduling and sustainable energy operations. Developing an optimal power load control model helps decarbonize the building sector by promoting cost-effective carbon emission reduction. While conventional design optimization has focused on building energy devices, deep learning-based optimization methods remain underexplored. This study proposes a deep learning-assisted optimization strategy for power load control and carbon metering in electricity systems. A Graph Neural Network (GNN)-based operational control and forecasting technique is introduced to approximate complex and costly control optimization. The GNN model utilizes power load design variables to minimize carbon emissions. Furthermore, a Pelican Optimization Algorithm (POA) is employed to enhance the optimization process due to its high performance and efficiency. When applied to residential buildings, the proposed GNN-POA model demonstrates superior performance in optimizing power load design and significantly reducing carbon emissions, offering a promising approach for intelligent, low-carbon energy management.