Machine Learning-Based Nodal Loadability Estimation for Power System

This paper presents the results of experiments that investigated using a machine learning approach for nodal loadability estimation in power systems. It defines the concept of nodal loadability and the methodology of its estimation using a graph neural network, including dataset generation, model architecture, and training pipeline. The results obtained indicate a significant improvement in computation speed, as well as an average estimation error of up to 48% for 25,000-bus networks.