Techniques for distributed machinery monitoring

• US 6,687,654 B2
• Filed: 09/06/2002
• Issued: 02/03/2004
• Est. Priority Date: 09/10/2001
• Status: Active Grant

• Current Assignee: Johns Hopkins University
• Original Assignee: Johns Hopkins University
• Inventors: Peacock, George S., Smith, George A. Jr., Vojtech, George L. Jr.
• Primary Examiner(s): Hoff, Marc S.
• Assistant Examiner(s): Raymond, Edward
• Application Number:
  US10/236,356
• Publication Number:
  US 20030061008A1
• Time in Patent Office: 515 Days
• Field of Search: 702/183, 702/33, 702/34, 702/48, 702/54, 702/56, 702/57, 702/58, 702/59, 702/75, 702/76, 702/77, 702/81, 702/84, 702/182, 702/184, 702/185, 702/188, 700/9, 700/21, 700/26, 700/108, 700/110, 700/174, 714/1, 714/25, 714/48, 324/500, 324/512, 324/520, 324/531, 324/772, 324/705, 324/707, 324/718, 324/76.41, 324/76.74
• US Class Current: 702/183
• CPC Class Codes: G05B 23/0235 based on a comparison with predetermined threshold or range, e.g. "classical methods", carried out during normal operation; threshold adaptation or choice; when or how to compare with the threshold
• CPC Class Codes: G05B 23/0283 Predictive maintenance, e.g. involving the monitoring of a system and, based on the monitoring results, taking decisions on the maintenance schedule of the monitored system; Estimating remaining useful life [RUL] preventive maintenance, i.e. planning maintenance according to the available resources without monitoring the system G06Q10/06

Abstract

Techniques for monitoring a machine for significant deviations from normal operations include collecting, at a first processing element, sensor data about the machine. The first processing element performs narrowband frequency domain processing to determine a segment of sensor data that indicates a deviation from normal operations that exceeds a threshold deviation. A message including the segment of sensor data is sent to a second processing element. In response to receiving the message, the second processing element performs different narrowband frequency domain processing to determine whether the deviation from normal operations is significant for maintaining the machine. If the deviation from normal operations is determined to be significant for maintaining the machine, then the deviation is reported to cause the machine to be maintained.

Claims

1. A method of monitoring a machine for significant deviations from normal operations, the method comprising the steps of:

collecting, at a first processing element, sensor data about the machine;

performing narrowband frequency domain processing by the first processing element to determine a segment of sensor data that indicates a deviation from normal operations that exceeds a threshold deviation;

sending a message including the segment of sensor data to a second processing element;

in response to receiving the message, performing different narrowband frequency domain processing at the second processing element to determine whether the deviation from normal operations is significant for maintaining the machine; and

if the deviation from normal operations is determined to be significant for maintaining the machine, then reporting the deviation to cause the machine to be maintained.

2. A method of monitoring a machine for significant deviations from normal operations, the method comprising the steps of:

receiving at a first processing element a message including a segment of sensor data about the machine from a second processing element, which segment is determined at the second processing element to indicate a deviation from normal operations that exceeds a threshold deviation;

in response to receiving the message, performing on the segment of sensor data, at the first processing element, first narrowband frequency domain processing to determine whether the deviation from normal operations is significant for maintaining the machine, the narrowband frequency domain processing different from second processing performed in the second processing element; and

if the deviation from normal operations is determined to be significant for maintaining the machine, then reporting the deviation to cause the machine to be maintained,

wherein narrowband frequency domain processing includes;

normalizing a frequency spectrum for a temporal portion of sensor data, which includes determining a broad background spectrum and removing the broad background spectrum from the frequency spectrum to produce a residual spectrum; and

determining in the residual spectrum a peak set of one or more spectral peaks that exceed a threshold amplitude.

3. A method as recited in claim 2, said step of performing first narrowband frequency domain processing further comprising the step of determining a first residual spectrum for a first temporal portion of the segment of sensor data with a finer frequency resolution than a second residual spectrum determined for the first temporal portion during the second processing performed in the second processing element.

4. A method as recited in claim 2, said step of performing first narrowband frequency domain processing further comprising the steps of:

dividing the segment of sensor data into a plurality of temporal portions;

determining a first plurality of peak sets in a corresponding plurality of residual spectra for the corresponding plurality of temporal portions; and

determining a track of related peaks having one peak from each peak set of a second plurality of peak sets of the first plurality of peak sets.

5. A method as recited in claim 4, said step of determining a track of related peaks further comprising the step of merging a plurality of adjacent peaks in a peak set into a single peak.

6. A method as recited in claim 2, said step of performing first narrowband frequency domain processing further comprising the step of performing modulated source processing to extract a low frequency signal that modulates a broadband signature of sensor data from the machine.

7. A method as recited in claim 2, the method further comprising the step of performing directional processing to determine a direction along which a signal propagates from a component of the machine to a sensor that generates the sensor data.

8. A method as recited in claim 4, said step of performing first narrowband frequency domain processing further comprising the steps of:

determining a set of one or more values for a set of one or more characteristics of the track; and

determining whether the deviation from normal operations is significant for maintaining the machine based at least in part on the set of one or more values.

9. A method as recited in claim 8, said step of determining whether the deviation from normal operations is significant for maintaining the machine further comprising the step of comparing the set of one or more values of the track to a reference set of one or more values for the set of one or more characteristics of a corresponding track found during normal operation of the machine.

10. A method as recited in claim 8, said step of determining whether the deviation from normal operations is significant for maintaining the machine further comprising the step of comparing the set of one or more values of the track to a reference set of one or more values for the set of one or more characteristics of a corresponding track found during operation of the machine when maintenance should be performed on a component of the machine.

11. A method as recited in claim 2, said step of performing first narrowband frequency domain processing further comprising determining a harmonic set of peaks in the peak set, each peak in the harmonic set separated in frequency from a lowest frequency peak in the harmonic set by an integer multiple of a constant frequency difference.

12. A method as recited in claim 2, said step of performing first narrowband frequency domain processing further comprising;

determining for the segment a set of one or more values for a set of one or more characteristics of the segment;

classifying the segment into a particular mode of a plurality of component maintenance modes based at least in part on the set of one or more values, each component maintenance mode associated with a component of the machine and an indication of maintenance to be performed; and

determining whether the deviation from normal operations is significant at least in part based on the particular mode into which the segment is classified.

13. A method as recited in claim 2, said step of performing first narrowband frequency domain processing further comprising:

determining a first numeric score based on a first set of one or more frequency domain characteristics of the segment;

determining a second numeric score based on a second set of one or more frequency domain characteristics of the segment; and

determining whether the deviation from normal operations is significant at least in part based on a sum of the first numeric score and the second numeric score.

14. A method as recited in claim 2, wherein the sensor data is vibration data.

15. A method of monitoring a machine for significant deviations from normal operations, the method comprising the steps of:

collecting, at a first processing element, sensor data about the machine;

performing first narrowband frequency domain processing on the first processing element to determine a segment of sensor data that indicates a deviation from normal operations that exceeds a threshold deviation; and

sending a message including the segment of sensor data to a second processing element for performing different second narrowband frequency domain processing to determine whether the deviation from normal operations is significant for maintaining the machine,

wherein narrowband frequency domain processing includes;

normalizing a frequency spectrum for a temporal portion of sensor data, which includes determining a broad background spectrum and removing the broad background spectrum from the frequency spectrum to produce a residual spectrum; and

determining in the residual spectrum a peak set of one or more spectral peaks that exceed a threshold amplitude.

16. A method as recited in claim 15, said step of performing the first narrowband frequency domain processing further comprising the step of determining a first residual spectrum for a first temporal portion of the segment of sensor data with a coarser frequency resolution than a second residual spectrum determined for the first temporal portion during the second processing performed in the second processing element.

17. A method as recited in claim 15, said step of performing the first narrowband frequency domain processing further comprising the steps of:

determining a set of one or more values for a set of one or more characteristics of the peak set for the segment; and

determining whether the segment indicates a deviation from normal operations based at least in part on the set of one or more values.

18. A method as recited in claim 17, said step of determining a set of one or more values for a set of one or more characteristics of the peak set for the segment including determining a set of values for at least one of a number of peaks in the peak set, a set of frequencies associated with the one or more peaks in the peak set, and a set of indications of amplitudes associated with the one or more peaks in the peak set.

19. A method as recited in claim 17, said step of determining a set of one or more values for a set of one or more characteristics of the peak set for the segment including determining a set of values for spacing between neighboring peaks in the peak set.

20. A method as recited in claim 17, said step of determining whether the segment indicates a deviation from normal operations further comprising the step of comparing the set of one or more values for the peak set to a reference set of one or more values for the set of one or more characteristics of a peak set found during normal operation of the machine.

21. A method as recited in claim 17, said step of determining whether the segment indicates a deviation from normal operations further comprising the step of comparing the set of one or more values for the peak set to a reference set of one or more values for the set of one or more characteristics of a peak set found during operation of the machine when maintenance should be performed on a component of the machine.

22. A method as recited in claim 19, said step of performing first narrowband frequency domain processing further comprising determining a harmonic set of peaks in the peak set, each peak in the harmonic set separated in frequency from a lowest frequency peak in the harmonic set by an integer multiple of a constant frequency difference.

23. A method as recited in claim 15, said step of performing first narrowband frequency domain processing further comprising;

determining for the segment a set of one or more values for a set of one or more characteristics of the segment;

classifying the segment into a particular mode of a plurality of deviation modes based at least in part on the set of one or more values, each deviation mode associated with a deviation from normal operation of the machine; and

determining whether the segment indicates a deviation from normal operations at least in part based on the particular mode into which the segment is classified.

24. A method as recited in claim 15, said step of performing first narrowband frequency domain processing further comprising:

determining a first numeric score based on a first set of one or more frequency domain characteristics of the segment;

determining a second numeric score based on a second set of one or more frequency domain characteristics of the segment; and

determining whether the segment indicates a deviation from normal operations at least in part based on a sum of the first numeric score and the second numeric score.

25. A method as recited in claim 15, wherein the sensor data is vibration data.

26. A computer-readable medium carrying one or more sequences of instructions for monitoring a machine for significant deviations from normal operations, wherein execution of the one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

receiving a message including a segment of sensor data about the machine from a second processor, which segment is determined at the second processor to indicate a deviation from normal operations that exceeds a threshold deviation;

in response to receiving the message, performing on the segment of sensor data first narrowband frequency domain processing to determine whether the deviation from normal operations is significant for maintaining the machine, the narrowband frequency domain processing different from second processing performed in the second processor; and

if the deviation from normal operations is determined to be significant for maintaining the machine, then reporting the deviation to cause the machine to be maintained,

wherein narrowband frequency domain processing includes;

normalizing a frequency spectrum for a temporal portion of sensor data, which includes determining a broad background spectrum and removing the broad background spectrum from the frequency spectrum to produce a residual spectrum; and

determining in the residual spectrum a peak set of one or more spectral peaks that exceed a threshold amplitude.

27. A computer-readable medium carrying one or more sequences of instructions for monitoring a machine for significant deviations from normal operations, wherein execution of the one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

collecting sensor data about the machine;

performing first narrowband frequency domain processing to determine a segment of sensor data that indicates a deviation from normal operations that exceeds a threshold deviation; and

sending a message including the segment of sensor data to a second processor for performing different second narrowband frequency domain processing to determine whether the deviation from normal operations is significant for maintaining the machine,

wherein narrowband frequency domain processing includes;

normalizing a frequency spectrum for a temporal portion of sensor data, which includes determining a broad background spectrum and removing the broad background spectrum from the frequency spectrum to produce a residual spectrum; and

determining in the residual spectrum a peak set of one or more spectral peaks that exceed a threshold amplitude.

28. An apparatus for monitoring a machine for significant deviations from normal operations, comprising:

a communications port for receiving messages;

a computer-readable medium storing one or more sequences of instructions; and

a processor connected to the communications port and computer readable medium,

wherein execution of the one or more sequences of instructions by the processor causes the processor to perform the steps of;

receiving a message including a segment of sensor data about the machine from a second processor, which segment is determined at the second processor to indicate a deviation from normal operations that exceeds a threshold deviation;

in response to receiving the message, performing on the segment of sensor data first narrowband frequency domain processing to determine whether the deviation from normal operations is significant for maintaining the machine, the narrowband frequency domain processing different from second processing performed in the second processor; and

if the deviation from normal operations is determined to be significant for maintaining the machine, then reporting the deviation to cause the machine to be maintained,

wherein narrowband frequency domain processing includes;

normalizing a frequency spectrum for a temporal portion of sensor data, which includes determining a broad background spectrum and removing the broad background spectrum from the frequency spectrum to produce a residual spectrum; and

determining in the residual spectrum a peak set of one or more spectral peaks that exceed a threshold amplitude.

29. An apparatus for monitoring a machine for significant deviations from normal operations, comprising:

a communications port for sending messages;

a sensor for generating sensor data about the machine;

a computer-readable medium storing one or more sequences of instructions; and

a processor connected to the communications port, the sensor and the computer readable medium, wherein execution of the one or more sequences of instructions by the processor causes the processor to perform the steps of;

collecting sensor data about the machine from the sensor;

performing first narrowband frequency domain processing to determine a segment of sensor data that indicates a deviation from normal operations that exceeds a threshold deviation; and

sending a message including the segment of sensor data to a second processor for performing different second narrowband frequency domain processing to determine whether the deviation from normal operations is significant for maintaining the machine,

wherein narrowband frequency domain processing includes;

normalizing a frequency spectrum for a temporal portion of sensor data, which includes determining a broad background spectrum and removing the broad background spectrum from the frequency spectrum to produce a residual spectrum; and

determining in the residual spectrum a peak set of one or more spectral peaks that exceed a threshold amplitude.

30. An apparatus for monitoring a machine for significant deviations from normal operations, comprising:

means for receiving a message including a segment of sensor data about the machine from a second processor, which segment is determined at the second processor to indicate a deviation from normal operations that exceeds a threshold deviation;

means for performing on the segment of sensor data, in response to receiving the message, first narrowband frequency domain processing to determine whether the deviation from normal operations is significant for maintaining the machine, the narrowband frequency domain processing different from second processing performed in the second processor; and

means for reporting the deviation to cause the machine to be maintained, if the deviation from normal operations is determined to be significant for maintaining the machine,

wherein narrowband frequency domain processing includes;

normalizing a frequency spectrum for a temporal portion of sensor data, which includes determining a broad background spectrum and removing the broad background spectrum from the frequency spectrum to produce a residual spectrum; and

determining in the residual spectrum a peak set of one or more spectral peaks that exceed a threshold amplitude.

31. An apparatus for monitoring a machine for significant deviations from normal operations, comprising:

means for collecting sensor data about the machine;

means for performing first narrowband frequency domain processing to determine a segment of sensor data that indicates a deviation from normal operations that exceeds a threshold deviation; and

means for sending a message including the segment of sensor data to a second processor for performing different second narrowband frequency domain processing to determine whether the deviation from normal operations is significant for maintaining the machine,

wherein narrowband frequency domain processing includes;

normalizing a frequency spectrum for a temporal portion of sensor data, which includes determining a broad background spectrum and removing the broad background spectrum from the frequency spectrum to produce a residual spectrum; and

determining in the residual spectrum a peak set of one or more spectral peaks that exceed a threshold amplitude.

32. A system for monitoring a machine for significant deviations from normal operations, comprising:

a communications channel;

a remote unit connected to the communications channel, the remote unit including;

a first communications port for sending messages over the communications channel;

a sensor for generating sensor data about the machine;

a first computer-readable medium storing a first set of one or more sequences of instructions and first data indicating sensor data during at least one of normal operations of the machine and operations when the machine should be maintained; and

a first processor connected to the communications port, the sensor and the computer readable medium, wherein execution of the first set of instructions by the first processor causes the first processor to perform the steps of

collecting sensor data about the machine from the sensor,

performing first narrowband frequency domain processing to determine a segment of sensor data that indicates a deviation from normal operations that exceeds a threshold deviation, and

sending a message including the segment of sensor data; and

a base unit connected to the communications channel, the base unit including;

a second communications port for receiving messages over the communications channel;

a second computer-readable medium storing a second set of one or more sequences of instructions and second data indicating sensor data during at least one of normal operations of the machine and during operations when the machine should be maintained; and

a second processor connected to the communications port and the computer readable medium, wherein execution of the second set of instructions by the second processor causes the second processor to perform the steps of;

receiving the message,

in response to receiving the message, performing different narrowband frequency domain processing to determine whether the deviation from normal operations is significant for maintaining the machine, and

reporting the deviation to cause the machine to be maintained, if the deviation from normal operations is determined to be significant for maintaining the machine.

33. A system for monitoring a machine for significant deviations from normal operations, comprising:

means for collecting, at a first processing element, sensor data about the machine;

means for performing narrowband frequency domain processing on the first processing element to determine a segment of sensor data that indicates a deviation from normal operations that exceeds a threshold deviation;

means for sending a message including the segment of sensor data to a more powerful second processing element;

means for performing different narrowband frequency domain processing, at the second processing element in response to receiving the message, to determine whether the deviation from normal operations is significant for maintaining the machine; and

means for reporting the deviation to cause the machine to be maintained, if the deviation from normal operations is determined to be significant for maintaining the machine.