Cascading failure prediction and recovery in large-scale critical infrastructure networks: A survey

Context: Large-scale critical infrastructure (CI) networks are crucial to society but prone to cascading failures due to their dynamic and interconnected characteristics. Recent research focuses on their reliability, using network theories and real-world data to develop recovery functions and crash warning indicators. Objective: This review evaluates cascading failure prediction and recovery trends, examines verification methods, and addresses challenges in enhancing network reliability and topology recovery within CI systems. Methods: A comprehensive survey explores cascading failure prediction and recovery from two perspectives: inter-network and inter-module structures. It summarizes recent research trends, common verification platforms, and datasets for predicting and recovering from cascading failures. Results: The review focuses on low-dimensional static networks, revealing significant challenges in dynamic environments. It underscores the necessity for improved recovery techniques and enhanced network reliability. Conclusion: This article identifies future research directions and unresolved issues by analyzing existing work in cascading failure prediction and recovery. Understanding cascading failure mechanisms aims to inspire the design of more resilient and reliable network systems, contributing to developing cohesive and low-coupling CI systems.