Gear transmission condition monitoring method and apparatus

• US 6,507,789 B1
• Filed: 07/18/2000
• Issued: 01/14/2003
• Est. Priority Date: 07/18/2000
• Status: Expired due to Term

• Current Assignee: GE Global Sourcing LLC (Westinghouse Air Brake Technologies Corporation)
• Original Assignee: General Electric Company
• Inventors: Reddy, Suresh Baddam, Kliman, Gerald Burt, Yazici, Birsen
• Primary Examiner(s): Hilten, John S.
• Assistant Examiner(s): Sun, Xiuqin
• Application Number:
  US09/618,440
• Time in Patent Office: 910 Days
• Field of Search: 702/34, 731/62, 706/52, 356/319, 382/275
• US Class Current: 702/34
• CPC Class Codes: G01M 13/021 Gearings
• CPC Class Codes: G01M 13/028 Acoustic or vibration analysis
• CPC Class Codes: G05B 23/0256 injecting test signals and analyzing monitored process response, e.g. injecting the test signal while interrupting the normal operation of the monitored system; superimposing the test signal onto a control signal during normal operation of the monitored system

Abstract

A gear transmission monitoring method includes: forming a good operating condition baseline matrix by, for each of a plurality of different gear mesh frequencies, obtaining a good operating condition signal indicative of gear transmission conditions over a segment of time and transforming the obtained good operating condition signal into a good operating condition time-frequency spectrum; and then obtaining a gear mesh frequency and a test signal over a segment of time, transforming the obtained test signal into a test time-frequency spectrum, and using the gear mesh frequency and the good operating condition baseline matrix to examine the test time-frequency spectrum to monitor gear transmission conditions.

Claims

1. A gear transmission condition monitoring method comprising the steps of:

identifying gear mesh frequencies that are remote from interfering frequencies;

forming a good operating condition baseline matrix by, for each of the identified gear mesh frequencies, obtaining a good operating condition signal indicative of gear transmission conditions over a segment of time and transforming the obtained good operating condition signal into a good operating condition time-frequency spectrum;

forming a faulty operating condition baseline matrix by operating the gear transmission under a plurality of different faulty operating conditions while, for each of the identified gear mesh frequencies, obtaining a faulty operating condition signal over a segment of time and transforming the obtained faulty operating condition signal into a faulty operating condition time-frequency spectrum; and

then, obtaining a test gear mesh frequency and a test signal over a segment of time, transforming the obtained test signal into a test time-frequency spectrum, and using the test gear mesh frequency, the good operating condition baseline matrix, and the faulty operating condition baseline matrix to examine the test time-frequency spectrum to monitor gear transmission conditions.

2. The method of claim 1 wherein the good and faulty operating condition signals comprise signals representative of speed, current, voltage or torque.

3. The method of claim 1 wherein the plurality of different faulty operating conditions include worn gear and low lubrication conditions.

4. The method of claim 1 wherein identifying the gear mesh frequencies that are remote includes identifying machine speeds for which interfering frequencies are minimized and wherein obtaining the good operating condition, faulty operating condition, and test signals includes obtaining the good operating condition, faulty operating condition, and test signals at the identified machine speeds.

5. The method of claim 1 wherein forming good and faulty operating condition baseline matrices include, for each of the identified gear mesh frequencies, selecting a window of frequency components to form a feature vector.

6. The method of claim 5 wherein forming good and faulty operating condition baseline matrices further include, for each of the identified gear mesh frequencies, selecting a sub-window around the maximum frequency component within the window to form the feature vector.

7. The method of claim 6 further including forming the feature vectors into feature data sets to form distinct baselines for different operating modes.

8. The method of claim 7 further including compressing the feature data sets.

9. The method of claim 8 wherein compressing each respective feature data set includes:

calculating a covariance matrix of the feature data set;

performing an eigenvalue decomposition of the covariance matrix; and

using multiple columns of the eigenvalue decomposition matrix to transform the feature data set to a multi-dimensional compressed vector space.

13. The method of claim 1 wherein using the test gear mesh frequency, the good operating condition baseline matrix, and the faulty operating condition baseline matrix to examine the test time-frequency spectrum includes forming a test feature vector from the test time-frequency spectrum.

14. The method of claim 13 wherein using the test gear mesh frequency, the good operating condition baseline matrix, the faulty operating condition baseline matrix to examine the test time-frequency spectrum further includes obtaining a plurality of test gear mesh frequencies and a plurality of respective test signals over a plurality of respective segments of time, forming a plurality of respective test feature vectors, and tracking instances when the test feature vectors falls outside the good operating condition baseline matrix or within the faulty operating condition baseline matrix.

15. The method of claim 14 further including declaring a fault after a predetermined number of consecutive tracked instances.

10. A gear transmission condition monitoring method comprising the steps of:

forming a good operating condition baseline matrix by, for each of a plurality of different identified gear mesh frequencies, obtaining a good operating condition signal indicative of gear transmission conditions over a segment of time and transforming the obtained good operating condition signal into a good operating condition time-frequency spectrum;

forming a faulty operating condition baseline matrix by operating the gear transmission under a plurality of different faulty operating conditions while, for each of the identified gear mesh frequencies, obtaining a faulty operating condition signal over a segment of time and transforming the obtained faulty operating condition signal into a faulty operating condition time-frequency spectrum,

wherein forming the good and faulty operating condition baseline matrices include, for each of the identified gear mesh frequencies, selecting a window of frequency components and a sub-window around the maximum frequency component within the window to form the feature vector, forming the feature vectors into feature data sets to form distinct baselines for different operating modes, compressing the feature data sets, and characterizing a probability distribution function of each respective compressed feature data set; and

then, obtaining a test gear mesh frequency and a test signal over a segment of time, transforming the obtained test signal into a test time-frequency spectrum, and using the test gear mesh frequency, the good operating condition baseline matrix, and the faulty operating condition baseline matrix to examine the test time-frequency spectrum to monitor gear transmission conditions.

11. The method of claim 10 further including clustering each feature data set by a mean feature vector and a covariance matrix.

12. The method of claim 11 further including determining a statistical boundary for each clustered feature data set.

16. A method for forming a baseline matrix for use in gear transmission condition monitoring comprising the steps of;

identifying gear mesh frequencies that are remote from interfering frequencies; and

for each of a plurality of different ones of the identified gear mesh frequencies;

obtaining a signal indicative of gear transmission conditions over a segment of time and transforming the obtained signal into a time-frequency spectrum;

selecting a window of frequency components to form a feature vector; and

forming the feature vectors into a feature data sets to form a distinct baseline.

17. The method of claim 16 wherein identifying the gear mesh frequencies that are remote includes identifying machine speeds for which interfering frequencies are minimized and wherein obtaining the signal includes obtaining the signal at the identified machine speeds.

18. The method of claim 16 further including compressing the feature data set.

19. The method of claim 18 further including characterizing a probability distribution function of the compressed feature data set.

20. The method of claim 19 further including clustering the feature data set by a mean feature vector and a covariance matrix.

21. A gear transmission condition monitoring apparatus comprising a processor for:

identifying gear mesh frequencies that are remote from interfering frequencies;

forming a good operating condition baseline matrix by, for each of the identified gear mesh frequencies, obtaining a good operating condition signal indicative of gear transmission conditions over a segment of time and transforming the obtained good operating condition signal into a good operating condition time-frequency spectrum;

forming a faulty operating condition baseline matrix by operating the gear transmission under a plurality of different faulty operating conditions while, for each of the identified gear mesh frequencies, obtaining a faulty operating condition signal over a segment of time and transforming the obtained faulty operating condition signal into a faulty operating condition time-frequency spectrum; and

then, obtaining a test gear mesh frequency and a test signal over a segment of time, transforming the obtained test signal into a test time-frequency spectrum, and using the test gear mesh frequency, the good operating condition baseline matrix, and the faulty operating condition baseline matrix to examine the test time-frequency spectrum to monitor gear transmission conditions.

22. The apparatus of claim 21 wherein the good and faulty operating condition signals comprise signals representative of speed, current, voltage or torque.

23. The apparatus of claim 21 wherein the processor is adapted to identify the gear mesh frequencies that are remote by identifying machine speeds for which interfering frequencies are minimized, and to obtain the good operating condition, faulty operating condition, and test signals at the identified machine speeds.

24. The apparatus of claim 21 wherein the processor is adapted to form good and faulty operating condition baseline matrices by, for each of the identified gear mesh frequencies, selecting a window of frequency components to form a feature vector.

25. The apparatus of claim 24 wherein the processor is adapted to form good and faulty operating condition baseline matrices by further forming the feature vectors into feature data sets to form distinct baselines for different operating modes.

26. The apparatus of claim 25 wherein the processor is adapted to form good and faulty operating condition baseline matrices by further compressing the feature data sets.

27. A gear transmission condition monitoring apparatus comprising a processor for:

forming a good operating condition baseline matrix by, for each of a plurality of identified gear mesh frequencies, obtaining a good operating condition signal indicative of gear transmission conditions over a segment of time and transforming the obtained good operating condition signal into a good operating condition time-frequency spectrum;

forming a faulty operating condition baseline matrix by operating the gear transmission under a plurality of different faulty operating conditions while, for each of the identified gear mesh frequencies, obtaining a faulty operating condition signal over a segment of time and transforming the obtained faulty operating condition signal into a faulty operating condition time-frequency spectrum,

wherein the processor is adapted to form good and faulty operating condition baseline matrices by, for each of the identified gear mesh frequencies, selecting a window of frequency components to form a feature vector, forming the feature vectors into feature data sets to form distinct baselines for different operating modes, compressing the feature data sets, and characterizing a probability distribution function of each respective compressed feature data set; and

then, obtaining a test gear mesh frequency and a test signal over a segment of time, transforming the obtained test signal into a test time-frequency spectrum, and using the test gear mesh frequency, the good operating condition baseline matrix, and the faulty operating condition baseline matrix to examine the test time-frequency spectrum to monitor gear transmission conditions.

28. The apparatus of claim 27 wherein the processor is further adapted to cluster each feature data set by a mean feature vector and a covariance matrix.

29. The apparatus of claim 28 wherein the processor is further adapted to determine a statistical boundary for each clustered feature data set.