A Spectral Approach for the Efficient Identification of Power Transmission Network Uncontrolled Separation

Youwei Jia, Chun Sing Lai, Zhao Xu

Research output: Journal article publicationJournal articleAcademic researchpeer-review

2 Citations (Scopus)


Uncontrolled network separation (system island formation) is one of the most critical contingencies in power systems. The integrity of the whole power transmission network is a prerequisite for reliable system operation. Therefore, in order to safeguard the system operation and control, it is imperative to quickly identify the topological changes, especially the formation of system islands. In this paper, we developed a spectral clustering-based approach to efficiently detect the existing or the potential network islands. The core of this approach is a graph-algebraic model, which combines the real-power deliverability and topological information of a power transmission network. Based on an improved spectral clustering algorithm, the proposed approach can efficiently identify the critical situations of the network splitting under multiple line outages. This approach has been successfully tested by using the New England 39-bus and 118-bus systems under different scenarios.
Original languageEnglish
Pages (from-to)2050-2061
Number of pages12
JournalElectric Power Components and Systems
Issue number18
Publication statusPublished - 8 Nov 2017


  • cascading failure
  • operation security
  • spectral clustering
  • transmission network
  • uncontrolled separation

ASJC Scopus subject areas

  • Energy Engineering and Power Technology
  • Mechanical Engineering
  • Electrical and Electronic Engineering

Cite this