Enhancing Electrical Network Vulnerability Assessment with Machine Learning and Deep Learning Techniques

Abstract

This research utilizes advanced machine learning techniques to evaluate node vul- nerability in power grid networks. Utilizing the SciGRID and GridKit datasets, con- sisting of 479, 16,167 nodes and 765, 20,539 edges respectively, the study employs K-nearest neighbor and median imputation methods to address missing data. Cen- trality metrics are integrated into a single comprehensive score for assessing node criticality, categorizing nodes into four centrality levels informative of vulnerability. This categorization informs the use of traditional machine learning (including XG- Boost, SVM, Multilayer Perceptron) and Graph Neural Networks in the analysis. The study not only benchmarks the capabilities of these models in network analy- sis but also explores their potential in identifying critical nodes using features be- yond centrality metrics alone, enhancing their applicability in real-world scenarios. The research addresses a significant gap in effectively assessing the vulnerability of electrical networks, marked by isolated use of traditional centrality metrics and a lack of integration between these combined metrics with both tradiational and ad- vanced machine learning models. The study integrates various centrality measures into a comprehensive metric and advocates for the application of advanced ma- chine learning models, particularly GNNs. It underscores the need for larger and more complex datasets to unlock the full potential of GNNs in network vulnerabil- ity assessments. By comparing the performance of GNN models with traditional machine learning approaches across datasets of different sizes and complexities, the study provides insights into optimizing model selection for network analysis, thereby contributing significantly to the field of network vulnerability assessment.