Skip to main navigation Skip to search Skip to main content

A symbolic reasoning based anomaly detection for gas turbine subsystems

  • School of Computer Science and Technology, Harbin Institute of Technology
  • School of Energy Science and Engineering, Harbin Institute of Technology

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The detection of gas turbine engine anomalies is of great significance to its reliable economic operation. Considering the collective anomaly data to be detected sensitively, this paper presents a symbolic approach and applies it to anomaly detection of gas turbine subsystem. The trained finite state machine evaluates the posterior probabilities of observed symbol sequence. Thus, an anomaly detection strategy based on FSM estimation is used to detect the defects. Experimental results indicate that, despite the high performance of the model, the robustness of the model is strong, especially within a certain sequence length. Therefore, the proposed method can be a good way to promote the existing anomaly detection performance in gas turbine subsystem.

Original languageEnglish
Title of host publication2017 Prognostics and System Health Management Conference, PHM-Harbin 2017 - Proceedings
EditorsBin Zhang, Yu Peng, Haitao Liao, Datong Liu, Shaojun Wang, Qiang Miao
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781538603703
DOIs
StatePublished - 20 Oct 2017
Externally publishedYes
Event8th IEEE Prognostics and System Health Management Conference, PHM-Harbin 2017 - Harbin, China
Duration: 9 Jul 201712 Jul 2017

Publication series

Name2017 Prognostics and System Health Management Conference, PHM-Harbin 2017 - Proceedings

Conference

Conference8th IEEE Prognostics and System Health Management Conference, PHM-Harbin 2017
Country/TerritoryChina
CityHarbin
Period9/07/1712/07/17

Keywords

  • anomaly detection
  • finite state machine
  • gas turbine
  • symbolic dynamic analysis

Fingerprint

Dive into the research topics of 'A symbolic reasoning based anomaly detection for gas turbine subsystems'. Together they form a unique fingerprint.

Cite this