Load Forecasting in Power Grids Using Spatio-Temporal Graph Neural Networks

The increasing penetration of renewable energy and distributed generation has significantly increased the complexity of modern power grids. Consequently, accurate power load forecasting requires consideration of not only temporal patterns but also spatial dependencies among geographically adjacent nodes. In this study, we propose a spatio-temporal forecasting model based on an extended Graph Neural Network (GNN) architecture. The model is applied to a combined environment of the IEEE 14-bus test system and real-world load data. Experimental results demonstrate that the proposed model outperforms traditional time-series-based approaches in forecasting accuracy, highlighting the substantial contribution of spatial information to forecasting performance.