Method for determining rotational speed from machine vibration data

• US 5,744,723 A
• Filed: 05/10/1996
• Issued: 04/28/1998
• Est. Priority Date: 05/10/1996
• Status: Expired due to Term

• Current Assignee: Computational Systems Incorporated (Emerson Electric Company)
• Original Assignee: CSI Technology Incorporated (Emerson Electric Company)
• Inventors: Piety, Kenneth R.
• Primary Examiner(s): Oda, Christine K.
• Application Number:
  US08/644,176
• Time in Patent Office: 718 Days
• Field of Search: 73/660, 73/650, 73/658, 73/659, 73/602, 364/485, 364/508, 364/551.02
• US Class Current: 73/660
• CPC Class Codes: G01P 3/44 for measuring angular speed G01P3/56 takes precedence
• CPC Class Codes: G01P 3/48 by measuring frequency of generated current or voltage in general G01R23/00
• CPC Class Codes: G01P 3/4802 by using electronic circuits in general

Abstract

A method of determining the rotational speed of a rotating shaft from machine vibration data. The vibration produced by the shaft is sensed to produce a test vibration signal at an unknown rotational speed of the shaft, which test vibrational signal is converted to a test frequency spectrum. A reference frequency spectrum, corresponding to a known rotational speed of the shaft, is provided, and a stretch factor is determined. The stretch factor provides optimum correlation between the test frequency spectrum and the reference frequency spectrum. The unknown speed of the rotating shaft is calculated using the known speed of the reference frequency spectrum and the stretch factor.

Claims

1. A method of determining the rotational speed of a rotating shaft, comprising:

sensing vibration produced by the shaft to produce a test vibration signal at an unknown rotational speed of the shaft,

converting the test vibration signal to a test frequency spectrum,

compiling a table of peaks found in the test frequency spectrum,

providing a reference frequency spectrum corresponding to a known rotational speed of the shaft,

compiling a table of peaks found in the reference frequency spectrum, and

determining a stretch factor that provides optimum correlation between the table of test frequency spectrum peaks and the table of reference frequency spectrum peaks, and

calculating the unknown rotational speed of the rotating shaft using the known rotational speed of the reference frequency spectrum and the stretch factor.

2. The method of claim 1 further comprising removing from the table of peaks of the test frequency spectrum, prior to determining the stretch factor, peaks corresponding to line frequency and multiples of the line frequency.

3. The method of claim 1 further comprising:

the table of test frequency spectrum peaks having peak frequency and peak amplitude information from the test frequency spectrum, and

the table of reference frequency spectrum peaks having peak frequency and peak amplitude information from the reference frequency spectrum.

4. The method of claim 1 wherein the step of determining the stretch factor further comprises:

partitioning the test frequency spectrum into a predetermined number of partitions;

independently identifying peaks in each partition of the test frequency spectrum,

partitioning the reference frequency spectrum into a predetermined number of partitions;

independently identifying peaks in each partition of the reference frequency spectrum,

determining the stretch factor that provides optimum correlation between the identified test frequency spectrum peaks and the identified reference frequency spectrum peaks.

5. The method of claim 4 wherein the predetermined number of test frequency spectrum partitions is equal to the predetermined number of reference frequency spectrum partitions.

6. The method of claim 4 wherein the test frequency spectrum partitions are all the same size.

7. The method of claim 6 wherein the reference frequency spectrum partitions are all the same size.

8. The method of claim 1 wherein the step of determining the stretch factor further comprises:

determining peak amplitudes for the test frequency spectrum peaks,

normalizing the peak amplitudes of the test frequency spectrum peaks to produce normalized peaks,

determining peak amplitudes for the reference frequency spectrum peaks,

normalizing the peak amplitudes of the test frequency spectrum peaks to produce normalized peaks, and

determining the stretch factor that provides optimum correlation between the normalized test frequency spectrum peaks and the normalized reference frequency spectrum peaks.

9. The method of claim 8 wherein the step of normalizing the peak amplitudes of the test frequency spectrum peaks further comprises:

locating a largest peak in the test frequency spectrum table of peaks, designated as PKHI,

locating a smallest peak in the test frequency spectrum table of peaks, designated as PKLO,

calculating normalized values of test frequency spectrum peaks, designated as PKX, using the peak amplitudes of the test frequency spectrum peaks, designated as PKV, according to the equation;

##EQU6##

10. The method of claim 1 wherein the step of determining a stretch factor comprises:

providing a trial stretch factor,

iteratively repeating the following steps;

multiplying the test frequency spectrum peaks by the trial stretch factor,

determining a correlation index between the reference frequency spectrum peaks and the test frequency spectrum peaks as multiplied by the trial stretch factor,

if it is not the final iteration, then adjusting the trial stretch factor,

stopping the iteration after a predetermined number of iterations,

identifying a largest correlation index, and

designating the trial stretch factor that produced the largest correlation index as the optimum stretch factor.

11. The method of claim 1 wherein the step of determining a stretch factor comprises:

selecting a trial stretch factor,

iteratively repeating the following steps;

producing a set of trial stretch factors based on the selected trial stretch factor,

independently multiplying the test frequency spectrum peaks by each of the trial stretch factors in the set,

determining correlation indexes between the reference frequency spectrum peaks and the test frequency spectrum peaks as multiplied by the trial stretch factors in the set,

identifying a largest correlation index,

selecting a trial stretch factor in the set which produces the largest correlation index between the reference frequency spectrum peaks and the test frequency spectrum peaks,

stopping the iteration after a predetermined number of iterations, and

designating the selected trial stretch factor as an optimum stretch factor.

12. The method of claim 1 further comprising:

providing a flag which indicates whether the rotational speed of the shaft has already been determined, and if the flag indicates that the rotational speed of the shaft has already been determined, then the flag also indicates how the rotational speed of the shaft was determined, and

selectively by-passing all other steps of the method based on the flag.

13. The method of claim 12 wherein the flag further indicates whether the rotational speed of the shaft was determined by a specific one of a plurality of methods selected from the group consisting of calculated from a test spectrum table of peaks, measured, and manually set.

14. The method of claim 12 wherein the step of selectively by-passing further comprises selectively performing all other steps of the method regardless of the flag.

15. The method of claim 1 further comprising:

providing a flag which indicates whether the rotational speed of the shaft has already been determined, and if the flag indicates that the rotational speed of the shaft has already been determined, then the flag also indicates how the rotational speed of the shaft was determined,

selectively by-passing all other steps of the method based on the flag,

the step of determining the stretch factor further comprising;

partitioning the test frequency spectrum into a predetermined number of partitions;

independently identifying peaks having frequency and amplitude in each partition of the test frequency spectrum,

normalizing the amplitude of the peaks of the test frequency spectrum,

compiling a table of frequencies and amplitudes for a predetermined number of largest peaks identified in each partition of the test frequency spectrum,

partitioning the reference frequency spectrum into a predetermined number of partitions;

independently identifying peaks having frequency and amplitude in each partition of the reference frequency spectrum,

normalizing the amplitude of the peaks of the test frequency spectrum,

compiling a table of frequencies and amplitudes for a predetermined number of largest peaks identified in each partition of the test frequency spectrum,

providing the stretch factor,

iteratively repeating the following steps;

multiplying the test frequency spectrum peaks by the stretch factor,

determining a correlation index between the reference frequency spectrum peaks and the test frequency spectrum peaks as multiplied by the stretch factor,

adjusting the stretch factor unless it is the final iteration, and

stopping the iteration after a predetermined number of iterations.

16. The method of claim 15 wherein the predetermined number of test frequency spectrum partitions is equal to the predetermined number of reference frequency spectrum partitions.

17. The method of claim 15 wherein the test frequency spectrum partitions are all the same size.

18. The method of claim 15 wherein the reference frequency spectrum partitions are all the same size.

19. The method of claim 15 wherein the step of normalizing the amplitude of the peaks of the test frequency spectrum further comprises;

locating a largest peak in the test frequency spectrum table of peaks, designated as PKHI,

locating a smallest peak in the test frequency spectrum table of peaks, designated as PKLO,

calculating a normalized value of each peak in the test frequency spectrum, designated as PKX, using the amplitude of the peak, designated as PKV, according to the equation;

##EQU7##

20. The method of claim 15 wherein the flag further indicates whether the rotational speed of the shaft was determined by a specific one of a plurality of methods selected from the group consisting of calculated, measured, and manually set.

21. The method of claim 15 wherein the step of selectively by-passing further comprises selectively performing all steps of the method regardless of the flag.

22. The method of claim 15 wherein the step of determining a correlation index further comprises:

determining a smaller of a frequency resolution of the test frequency spectrum and a frequency resolution of the reference frequency spectrum, multiplied by a factor and designated as DFRQ,

determining a positive difference in frequency, designated as FDEL, between each peak of the test frequency spectrum which can be matched to one of the peaks of the reference frequency spectrum which is at most a frequency distance of DFRQ away,

designating the normalized amplitudes of the test frequency spectrum peaks as PKX_T,

designating the normalized amplitudes of the reference frequency spectrum peaks as PKX_R, and

calculating the correlation index according to the equation;

##EQU8##

23. A method of determining the rotational speed of a rotating shaft, comprising:

a. sensing vibration produced by the shaft to produce a test vibration signal at an unknown rotational speed of the shaft,

b. converting the test vibration signal to a test frequency spectrum,

c. providing a reference frequency spectrum corresponding to a known rotational speed of the shaft,

d. partitioning the reference frequency spectrum into a predetermined number of partitions,

e. independently identifying reference frequency spectrum peaks having frequency and amplitude in each partition of the reference frequency spectrum,

f. removing reference frequency spectrum peaks corresponding to line frequency in the reference frequency spectrum,

g. independently identifying a predetermined number of largest reference frequency spectrum peaks in each partition of the reference frequency spectrum,

h. compiling a table of frequencies and amplitudes for the largest reference frequency spectrum peaks identified from each partition of the reference frequency spectrum,

I. normalizing the amplitude of the reference frequency spectrum peaks,

j. partitioning the test frequency spectrum into a predetermined number of partitions,

k. independently identifying test frequency spectrum peaks having frequency and amplitude in each partition of the test frequency spectrum,

l. removing test frequency spectrum peaks corresponding to line frequency in the test frequency spectrum,

m. independently identifying a predetermined number of largest test frequency spectrum peaks from each partition of the test frequency spectrum,

n. compiling a table of frequencies and amplitudes for the largest test frequency spectrum peaks identified in each partition of the test frequency spectrum,

o. normalizing the amplitude of the test frequency spectrum peaks,

p. providing an estimated low unknown rotational speed,

q. providing an estimated high unknown rotational speed,

r. dividing the estimated low unknown rotational speed by the known rotational speed to produce a low stretch factor,

s. dividing the estimated high unknown rotational speed by the known rotational speed to produce a high stretch factor,

t. subtracting the low stretch factor from the high stretch factor to produce a range,

u. dividing the range by a predetermined number of steps to produce an increment,

v. compiling a set of stretch factors including the low stretch factor and the high stretch factor, and additional stretch factors between the low stretch factor and the high stretch factor, where each of the additional stretch factors is determined by adding to the low stretch factor a product of the increment and a serialized number, where the serialized number is an integer ranging from 1 to the predetermined number of steps minus 1,

w. selecting a selected stretch factor from the set of stretch factors until all stretch factors in the set of stretch factors have been selected, and for each selected stretch factor in the set of stretch factors;

25. The method of claim 23 further comprising analyzing the test frequency spectrum using, in part, the calculated unknown rotational speed.

26. The method of claim 23 wherein the predetermined number of test frequency spectrum partitions is equal to the predetermined number of reference frequency spectrum partitions.

27. The method of claim 23 wherein the rotating shaft is driven by a motor characterized by a motor type and the estimated high and low unknown rotational speeds are provided based at least on the motor type of the motor driving the rotating shaft.

28. The method of claim 23 wherein the test frequency spectrum partitions are all the same size.

29. The method of claim 23 wherein the reference frequency spectrum partitions are all the same size.

30. The method of claim 23 wherein the step of normalizing the amplitude of the test frequency spectrum peaks further comprises;

locating a largest peak in the test frequency spectrum table of peaks, designated as PKHI,

locating a smallest peak in the test frequency spectrum table of peaks, designated as PKLO,

calculating a normalized value of each of the test frequency spectrum peaks, designated as PKX, using the amplitude of the peak, designated as PKV, according to the equation;

##EQU9##

31. The method of claim 23 further comprising:

providing a flag which indicates whether the rotational speed of the shaft has already been determined, and if the flag indicates that the rotational speed of the shaft has already been determined, then the flag also indicates how the rotational speed of the shaft was determined, and

selectively by-passing all other steps of the method based on the flag.

32. The method of claim 30 wherein the flag further indicates whether the rotational speed of the shaft was determined by a specific one of a plurality of methods selected from the group consisting of calculated, measured, and manually set.

33. The method of claim 30 wherein the step of selectively by-passing further comprises selectively performing all other steps of the method regardless of the flag.

34. The method of claim 23 wherein the step of comparing to produce a correlation index further comprises:

determining a smaller of a frequency resolution of the test frequency spectrum and a frequency resolution of the reference frequency spectrum, multiplied by a factor and designated as DFRQ,

determining a positive difference in frequency, designated as FDEL, between each one of the test frequency spectrum peaks of the test frequency spectrum table of peaks which can be matched to one of the reference frequency spectrum peaks of the reference frequency spectrum table of peaks which is at most a frequency distance of DFRQ away,

designating the normalized amplitudes of the test frequency spectrum peaks as PKX_T,

designating the normalized amplitudes of the reference frequency spectrum peaks as PKX_R, and

calculating the correlation index according to the equation;

##EQU10##

35. The method of claim 23 wherein the estimated low unknown rotational speed and the estimated high unknown rotational speed are manually provided.

36. The method of claim 23 wherein the estimated low unknown rotational speed and the estimated high unknown rotational speed are calculated based on at least a rated rotational speed of the shaft.

37. The method of claim 23 wherein the predetermined number of steps is different for the first iteration than for all subsequent iterations.

38. The method of claim 23 wherein the predetermined number of steps is the same for all iterations.

24. multiplying the test frequency spectrum peaks by the selected stretch factor in the set of stretch factors to produce adjusted peaks,comparing the adjusted peaks to the reference frequency spectrum peaks to produce a correlation index, where the correlation index has a greater value when the adjusted peaks are closer to the reference frequency spectrum peaks,

x. selecting a first stretch factor from the set of stretch factors which yields a largest correlation index as a best stretch factor,

y. selecting a second stretch factor from the set of stretch factors which is just less than the best stretch factor to be the low stretch factor,

z. selecting a stretch factor which is just greater than the best stretch factor to be the high stretch factor,

aa. iteratively repeating steps t through z for a predetermined number of iterations, and

bb. calculating the unknown speed of the rotating shaft by dividing the known rotational speed of the shaft corresponding to the reference frequency spectrum by the best stretch factor produced by the last repetition of steps t through z, which is designated as an optimum stretch factor.