Methodology for temporal fault event isolation and identification

• US 7,233,884 B2
• Filed: 10/31/2002
• Issued: 06/19/2007
• Est. Priority Date: 10/31/2002
• Status: Active Grant

• Current Assignee: Rtx Corporation
• Original Assignee: United Technologies Corporation (Raytheon Technologies Corporation d/b/a RTX)
• Inventors: Volponi, Allan J.
• Primary Examiner(s): COSIMANO, EDWARD R
• Application Number:
  US10/286,504
• Publication Number:
  US 20040153815A1
• Time in Patent Office: 1,692 Days
• Field of Search: 702/33, 702/34, 702/189
• US Class Current: 702/189
• CPC Class Codes: G05B 23/0275 Fault isolation and identification, e.g. classify fault; estimate cause or root of failure

Abstract

A method of fault event identification which comprises the steps of receiving a plurality of parameter measurements, performing single fault isolation to establish one of the plurality of parameter measurements as an event start, one of the plurality of parameter measurements as an event detect, and one of the parameter measurements as an event end wherein a latency period extends from the event start to the event detect and a blackout period extends from the event start to the event end, performing multiple fault isolation to establish a first trend line for the plurality of parameter measurements prior to the blackout period and after the blackout period, reprocessing the parameter measurements in the latency period, processing the parameter measurements in the blackout period, calculating a model IC from the reprocessed parameter measurements, calculating a plurality of single fault vectors, calculating an estimate for each of the single fault vectors, calculating a normalized error for each of said plurality of single fault vectors, calculating a polarized error term for each of the plurality of single fault vectors, and selecting the single fault vector with smallest of the calculated normalized errors.

Claims

1. A method of fault event identification, comprising the steps of:

receiving a plurality of parameter measurements collected from a gas turbine engine;

performing single fault isolation upon said plurality of parameter measurements to establish at least one of said plurality of parameter measurements as an event start parameter measurement of at least one fault event based upon a first performance change of said gas turbine engine, one of said plurality of parameter measurements as an event detect parameter measurement of at least one fault event based upon a second performance change of said gas turbine engine, and one of said parameter measurements as an event end parameter measurement of at least one fault event based upon a third performance change of said gas turbine engine wherein a latency period extends from said event start parameter measurement to said event detect parameter measurement and a blackout period extends from said event start parameter measurement to said event end parameter measurement;

performing multiple fault isolation using said plurality of parameter measurements prior to said event start parameter measurement of said blackout period and after said event end parameter measurement of said blackout period;

reprocessing said parameter measurements from said event start parameter measurement to said event detect parameter measurement of said latency period to improve an estimation of each of a plurality of single fault vectors;

reprocessing said parameter measurements from said event start parameter measurement to said event end parameter measurement of said blackout period to improve said estimation of each of said plurality of single fault vectors;

calculating a said plurality of single fault vectors based upon reprocessing during said latency period and said blackout period;

calculating an estimate for each of said plurality of single fault vectors;

calculating a normalized error for each of said plurality of single fault vectors;

calculating a polarized error term for each of said plurality of single fault vectors;

selecting said single fault vector with the smallest of said calculated normalized errors; and

identifying at least one single fault event for said gas turbine engine based upon said single fault vector with the smallest of said calculated normalized errors.

2. The method of claim 1 wherein said performing single fault isolation comprises performing a statistical analysis.

3. The method of claim 1, wherein said performing single fault isolation comprises applying Kalman Filter parameter estimation.

4. The method of claim 1, wherein said plurality of parameter measurements are derived from an engine.

5. The method of claim 1, wherein said reprocessing said parameter measurements comprises the steps of:

computing a linear regression line from said plurality of parameter measurements comprising said latency period; and

recomputing each of said plurality of measurements comprising said latency period as a difference between said linear regression line and an average level prior to said event start.

6. The method of claim 1 wherein each of said plurality of parameter measurements comprises a differential of a current measurement and an immediately preceding measurement.

7. The method of claim 6 wherein each of said plurality of parameter measurement is a percentage differential.

8. A method of fault event identification, comprising the steps of:

receiving a plurality of parameter measurements collected from a gas turbine engine;

performing single fault isolation upon said plurality of parameter measurements to establish at least one of said plurality of parameter measurements as an event start parameter measurement of at least one fault event based upon a first performance change of said gas turbine engine, one of said plurality of parameter measurements as an event detect parameter measurement of at least one fault event based upon a second performance change of said gas turbine engine, and one of said parameter measurements as an event end parameter measurements of at least one fault event based upon a third performance change of said gas turbine engine wherein a latency period extends from said event start parameter event to said event detect parameter event and a blackout period extends from said event start parameter event to said event end parameter event;

performing multiple fault isolation using said plurality of parameter measurements prior to said event start parameter measurement of said blackout period and after said event end parameter measurement of said blackout period;

reprocessing said parameter measurements from said event start parameter measurement to said event detect parameter measurement of said latency period to improve an estimation of each of a plurality of single fault vectors;

reprocessing said parameter measurements from said event start parameter measurement to said event end parameter measurement of said blackout period to improve an estimation of each of a plurality of single fault vectors;

calculating said plurality of single fault vectors based upon reprocessing during said latency period and said blackout period;

calculating an estimate for each of said plurality of single fault vectors;

calculating a normalized error for each of said plurality of single fault vectors;

calculating a polarized error term for each of said plurality of single fault vectors; and

determining if a double fault calculation is required based upon said polarized error term for each of said plurality of single fault vectors;

calculating a plurality of double fault vectors based upon reprocessing during said latency period and said blackout period;

calculating an estimate for each of said plurality of double fault vectors;

calculating a normalized error for each of said plurality of double fault vectors;

calculating a polarized error term for each of said plurality of double fault vectors comprising a smallest normalized error for each of said plurality of single fault vectors and a smallest normalized error for each of said plurality of double fault vectors; and

identifying at least one single fault event or at least one multiple fault event for said gas turbine engine based upon said polarized error term for each of said plurality of double fault vectors.

9. The method of claim 8 comprising the additional steps of:

selecting the smaller of said smallest normalized error for each of said plurality of single fault vectors and said smallest normalized error for each of said plurality of double fault vectors.

10. The method of claim 8 wherein said performing single fault isolation comprises performing a statistical analysis.

11. The method of claim 8, wherein said performing single fault isolation comprises applying Kalman Filter parameter estimation.

12. The method of claim 8, wherein said plurality of parameter measurements are derived from an engine.

13. The method of claim 8, wherein said reprocessing said parameter measurements comprises the steps of:

computing a linear regression line from said plurality of parameter measurements comprising said latency period; and

recomputing each of said plurality of measurements comprising said latency period as a difference between said linear regression line and an average level prior to said event start.

14. The method of claim 8 wherein each of said plurality of parameter measurements comprises a differential of a current measurement and an immediately preceding measurement.

15. The method of claim 14 wherein each of said plurality of parameter measurement is a percentage differential.