Identifying critical nodes in power networks: A group-driven framework

• Published in: Expert systems with applications Vol. 196; p. 116557
• Main Authors: Liu, Yangyang, Song, Aibo, Shan, Xin, Xue, Yingying, Jin, Jiahui
• Format: Journal Article
• Language: English
• Published: New York Elsevier Ltd 15-06-2022; Elsevier BV
• Subjects: Blackout; Clustering; Critical nodes; Group structure; Group-driven framework; Networks; Nodes; Optimization; Power network; Tightness; Critical nodes; Group-driven framework; Power network; Group structure
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2022.116557

Cascading failures can easily occur and cause a major blackout in power systems when a critical devices breaks down. It is an essential problem to evaluate the importance of devices/nodes for power networks. Even though approaches for identifying a power network’s critical nodes have been investigated in the past, accurately achieving critical nodes’ identification has proven to remain a challenging task. Here, we propose a group-driven framework, called GDF-ICN, in which the potential group information is introduced for the first time, to enhance the performance of critical nodes identification. It is a novel framework performed in an iterative manner, the introduction of nodes clustering effect improves the reliability of identifying critical nodes, while the identification of critical nodes promotes the characterization of a denser group structure. Specifically, we adopt the electrical and group information of a power network simultaneously to define each node’s electrical coupling that might affect the importance of nodes. To produce denser groups, we propose a fuzzy tightness metric that can be regarded as group optimization’s objective function. We also provide a preliminary but systematic research on how to transform any metric of evaluating hard group structure into a metric of evaluating fuzzy group structure. Comprehensive experiments on several benchmark power networks show the necessity of considering group information to evaluate the importance of nodes and the efficiency of GDF-ICN.
•We proposed a group-driven critical nodes identification framework.•A fuzzy tightness metric was designed to measure the agglomeration phenomenon.•We provided a preliminary tool to soften the metrics of evaluating hard groups.•We demonstrated GDF-ICN’s performance on several benchmark power networks.