Abstract

Critical component outages can lead to widespread cascading propagation, which is however typically ignored in existing investment planning approaches. To address this gap, this paper seamlessly integrates advanced cascading failure analysis into resilient investment planning. It first deploys a stochastic simulator to generate spatiotemporal high-impact low-probability (HILP) events, which are then assessed using a cascading failure model, generating various cascading quantification metrics (CQMs). The framework explicitly quantifies tail risks (i.e., HILP events) using Conditional Value-at-Risk (CVaR) with a confidence level determined by unsupervised clustering, instead of using a predetermined confidence level. This enables the more tailored identification of a set of worst-case scenarios for the system under investigation, improving its practicality. An optimization model then utilizes the outputs of the cascading analysis and the defined CVaR confidence level to identify investment portfolios that provide a hedge against cascading failures. The proposed work is demonstrated on the IEEE 39-bus system, revealing reduced cascading propagation.