0

Full Content is available to subscribers

Subscribe/Learn More  >

Validation of Diagnostic Data With Statistical Analysis and Embedded Knowledge

[+] Author Affiliations
Hans DePold, Jason Seigel, Allan Volponi

United Technologies – Pratt & Whitney, East Hartford, CT

Jonthan Hull

United technologies – Pratt & Whitney, East Hartford, CT

Paper No. GT2003-38764, pp. 573-579; 7 pages
doi:10.1115/GT2003-38764
From:
  • ASME Turbo Expo 2003, collocated with the 2003 International Joint Power Generation Conference
  • Volume 1: Turbo Expo 2003
  • Atlanta, Georgia, USA, June 16–19, 2003
  • Conference Sponsors: International Gas Turbine Institute
  • ISBN: 0-7918-3684-3 | eISBN: 0-7918-3671-1
  • Copyright © 2003 by ASME

abstract

The method described in this paper is intended to improve data stability and reduce dispersion of a given parameter signal by replacing any point deemed to be noise (e.g. an outlier) with an estimate of the parameter based on its recent history. This method determines a point is an outlier by utilizing learning and analytical elements that automatically adjust to the dispersion of the input data and automatically model the underlying process. The analytical elements of this method contain two types of embedded knowledge. The first is physics based engineering knowledge based on known interrelationships between the parameters to provide evidence when a parameter data point is physically unexplainable. It also contains the rule-based knowledge of real gas turbine trend characteristics such as persistency, polarity, and monotonic direction. Once a parameter data point is suspected to be noise, the system validates that assessment with a persistency check. If no parameter trend shift is occurring, as determined by a lack of persistency, the parameter is deemed noise and is replaced with the parameter’s last good state. Persistency checks enable removing noise without obscuring shifts in data that could be occurring due to real system changes. The process is designed to decide within two points if a data point should be considered noise. The goal of the process presented here for improving the quality of gas generator data is to automatically replace all probable noise without distorting the underlying parameter signals. The metrics for success are a 25% reduction in the dispersion (e.g. scatter) of the data with no bias in the parameter central tendency (e.g. mean value), with no reduction the granularity of the parameter signal (visibility of anomalies), and with no delay in the detection of a real trend change.

Copyright © 2003 by ASME

Figures

Tables

Interactive Graphics

Video

Country-Specific Mortality and Growth Failure in Infancy and Yound Children and Association With Material Stature

Use interactive graphics and maps to view and sort country-specific infant and early dhildhood mortality and growth failure data and their association with maternal

NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In