Data-driven fault detection in nuclear power plants under sensor degradation

Xin Jin, Yin Guo, Robert M. Edwards, Asok Ray

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Tools of data-driven fault detection facilitate performance monitoring of complex dynamical systems if the physics-based models are either inadequate or not available. To this end, many data-driven methods have been developed. An inherent difficulty for a completely data-driven fault detection tool is that the detection performance can deduce drastically in the presence of sensor degradation. Symbolic dynamic filtering (SDF) is recently introduced in the literature as a real-time data-driven pattern identification tool, which is built upon the concepts of Symbolic Dynamics, Information Theory and Statistical Mechanics. This paper investigates a SDF-based fault detection algorithm for health monitoring in nuclear power plants under sensor degradation. The proposed fault detection methodology is validated on the International Reactor Innovative & Secure (IRIS) simulator of nuclear power plants, and its performance is comparatively evaluated with existing fault detection tools.

Original languageEnglish (US)
Title of host publication7th International Topical Meeting on Nuclear Plant Instrumentation, Control, and Human-Machine Interface Technologies 2010, NPIC and HMIT 2010
Pages1586-1599
Number of pages14
StatePublished - 2010
Event7th International Topical Meeting on Nuclear Plant Instrumentation, Control, and Human-Machine Interface Technologies 2010, NPIC and HMIT 2010 - Las Vegas, NV, United States
Duration: Nov 7 2010Nov 11 2010

Publication series

Name7th International Topical Meeting on Nuclear Plant Instrumentation, Control, and Human-Machine Interface Technologies 2010, NPIC and HMIT 2010
Volume3

Other

Other7th International Topical Meeting on Nuclear Plant Instrumentation, Control, and Human-Machine Interface Technologies 2010, NPIC and HMIT 2010
Country/TerritoryUnited States
CityLas Vegas, NV
Period11/7/1011/11/10

All Science Journal Classification (ASJC) codes

  • Human-Computer Interaction
  • Nuclear Energy and Engineering
  • Control and Systems Engineering

Fingerprint

Dive into the research topics of 'Data-driven fault detection in nuclear power plants under sensor degradation'. Together they form a unique fingerprint.

Cite this