Engineering system fault detection using particle filters

Research output: Contribution to book or proceedingConference articlepeer-review

3 Scopus citations

Abstract

This paper presents an approach to machinery fault detection using particle filters (PF). The machine vibration signals are processed using morphological signal processing (MSP) to extract a novel entropy based health index (HI) characterizing the signal shape-size complexity. The evolution of HI is approximated as a nonlinear state space model using a computational intelligence (CI) technique. PF is used to estimate the progression of HI in presence of observation and process noise. The PF based approach is illustrated for estimation of state and parameters of a chaotic system. The feasibility of the approach is demonstrated through vibration dataset of a helicopter drive-train system gearbox. The results help understand the relationship of the system condition, the corresponding HI, the level of degradation and its progression in a stochastic environment using Bayesian learning.

Original languageEnglish
Title of host publicationASME 2009 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, IDETC/CIE2009
Pages95-101
Number of pages7
EditionPARTS A AND B
DOIs
StatePublished - 2009
EventASME 2009 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, IDETC/CIE2009 - San Diego, CA, United States
Duration: Aug 30 2009Sep 2 2009

Publication series

NameProceedings of the ASME Design Engineering Technical Conference
NumberPARTS A AND B
Volume1

Conference

ConferenceASME 2009 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference, IDETC/CIE2009
Country/TerritoryUnited States
CitySan Diego, CA
Period08/30/0909/2/09

Keywords

  • Failure prognostics
  • Nonlinear systems
  • Particle filtering
  • Uncertainty prediction

Fingerprint

Dive into the research topics of 'Engineering system fault detection using particle filters'. Together they form a unique fingerprint.

Cite this