Automated Vulnerability and Resilience Analysis of Power Grids under Attacks (And Digital Twins)
Description
The increasing complexity of smart grids has made them susceptible to failures caused by both natural events and targeted attacks. This study presents an automated vulnerability and resilience analysis framework for power grids using digital twins. Our research focuses on identifying critical nodes whose failure could significantly impact grid performance. By leveraging complex network analysis and centrality measures, we assess the effects of various attack strategies on power grids. We enhance GridLab-D simulations with real-time monitoring using currdump and voltdump, develop Python libraries for Monte Carlo simulations, and quantify power losses using FOE, SOC, and SOLC metrics. Our approach integrates graph-theoretic techniques to construct adjacency matrices, enabling the computation of betweenness and eigenvector centrality. These metrics help identify the most vulnerable components in the network. Using Glimpse, we enhance network visualization, allowing for a more intuitive understanding of grid topology. Preliminary results from the IEEE-123 test network highlight the cascading failures caused by node and load-based attacks. By systematically analyzing power flow disruptions, we propose a Vulnerability Predictive Measure (VPM) to enhance predictive capabilities in power grid resilience assessments. Future work includes running 10,000 Monte Carlo simulations to refine our centrality-based vulnerability model and develop optimized grid protection strategies.