from dc to holistic centrality: ai-driven estimation for enhanced power network analysis

This paper introduces a novel approach to power system analysis by integrating dc holistic centrality with advanced deep learning (dl) techniques to enhance the efficiency and accuracy of grid operation assessments. we propose dc holistic centrality, a computationally efficient measure derived from dc load flow data, which extends traditional operational centrality by incorporating generation and demand nodes as pendant buses. leveraging this new metric, we develop a suite of ai-driven estimation methods: a linear regression baseline for active holistic betweenness prediction, a deep neural network (dnn) for accurate cross-bus prediction of active holistic betweenness, a convolutional neural network (cnn) for voltage magnitude estimation from dc holistic dependency matrices, and a scalability assessment using the ieee 57-bus system to validate model robustness. the study utilizes a comprehensive dataset generated from varied operational scenarios, with feature selection guided by correlation analyses rather than additional extraction techniques. results demonstrate significant improvements in capturing inter-bus dependencies and system dynamics, offering a promising framework for real-time grid monitoring and management.