Vibro-acoustic signature treatment process in high-voltage electromechanical switching system

• US 6,215,408 B1
• Filed: 04/30/1999
• Issued: 04/10/2001
• Est. Priority Date: 01/22/1999
• Status: Active Grant

• Current Assignee: Hydro-Quebec (Province of Québec)
• Original Assignee: Hydro-Quebec (Province of Québec)
• Inventors: Rajotte, Claude, Foata, Marc, Léonard, François
• Primary Examiner(s): Wu, Daniel J.
• Assistant Examiner(s): Huang, Sihong
• Application Number:
  US09/302,287
• Time in Patent Office: 711 Days
• Field of Search: 340/683, 340/680, 340/644, 340/635, 340/511, 340/540, 324/424, 324/415
• US Class Current: 340/644
• CPC Class Codes: G01R 31/3274 Details related to measuring, e.g. sensing, displaying or computing; Measuring of variables related to the contact pieces, e.g. wear, position or resistance measuring contact resistance G01R27/205
• CPC Class Codes: G01R 31/3275 Fault detection or status indication
• CPC Class Codes: H01H 2009/0061 Monitoring tap change switching devices

Abstract

In the present method and apparatus for processing a vibro-acoustic signal emitted by a high-voltage switching system, the vibro-acoustic signal is converted into a digital signal. The digital signal is rectified and the resulting rectified signal is applied to a convolutional filter with a spectral window to produce a smoothed signal. The smoothed signal is decimated to produce a decimated signal representing an envelope of the vibro-acoustic signal. A time realignment of the decimated signal is carried out with respect to a reference signature to produce a realigned signal. Time deviation values generate an alarm if they exceed a limit threshold. The realigned signal is added as a factor of means providing a reference signature and an actualized signature. Variances of the realigned signal are calculated with respect to the reference and actualized signatures. The realigned signal is compared with the actualized and reference signatures to detect a gradual behavior change or a sudden defect, taking the variances into consideration.

Claims

1. A method of processing a vibro-acoustic signal emitted by a high-voltage switching system, comprising the steps of:

(a) converting the vibro-acoustic signal into a digital signal;

(b) rectifying the digital signal to produce a rectified signal;

(c) applying a convolutional filter with a spectral window on the rectified signal to produce a smoothed signal;

(d) decimating the smoothed signal according to a predetermined decimation factor to produce a decimated signal representing an envelope of the vibro-acoustic signal;

(e) carrying a time realignment of the decimated signal with respect to a reference signature to produce a realigned signal;

(f) adding the realigned signal as a factor of a mean to produce the reference signature, the mean having a predetermined reference signal for initial factor;

(g) adding the realigned signal as a factor of a mean to produce an actualized signature, the mean having the first realigned signal produced by the step (e) for initial factor;

(h) calculating variances of the realigned signal with respect to the reference and actualized signatures; and

(i) comparing the realigned signal with the actualized and reference signatures to detect a gradual behavior change or a sudden defect, taking the variances into consideration.

2. The method according to claim 1, wherein the step (a) comprises the steps of:

sampling the vibro-acoustic signal to produce a sampled signal; and

converting the sampled signal in digital values forming the digital signal.

3. The method according to claim 2, wherein the step (a) comprises the additional step of:

passing the vibro-acoustic signal in an anti-aliasing filter prior to the sampling step.

4. The method according to claim 1, comprising the additional step of:

passing the digital signal produced at the step (a) in a phase filter.

5. The method according to claim 4, wherein the phase filter performs a Fourier transform of the digital signal, a phase unwrapping, an angular correction by addition of a phase ramp and a reversed Fourier transform to provide a compensated digital signal.

6. The method according to claim 5, wherein the phase ramp is of a second degree.

7. The method according to claim 4, wherein the phase filter carries out a convolution of the digital signal with a chirp.

8. The method according to claim 1, comprising the additional step of:

integrating the digital signal between the steps (a) and (b).

9. The method according to claim 1, comprising the additional step of:

separating the digital signal produced at the step (a) in distinct frequency bands, the steps (b) to (i) being applied on each of the distinct frequency bands.

10. The method according to claim 9, comprising the additional step of:

integrating the digital signal between the steps (a) and (b), on each of the distinct frequency bands.

11. The method according to claim 1, in which the step (b) is carried out by Hilbert transform.

12. The method according to claim 1, comprising the additional step of:

shaping the smoothed signal produced at the step (c) by a logarithm.

13. The method according to claim 1, comprising the additional step of:

passing the smoothed signal produced at the step (c) in an anti-aliasing filter.

14. The method according to claim 13, wherein the anti-aliasing filter carries out a filtration of a low amplitude portion at a band end of the smoothed signal.

15. The method according to claim 13, wherein the anti-aliasing filter is a low pass filter having a cut-off frequency between 50 Hz to 50 kHz with a sampling period between 10 ms to 10 μ

s.

16. The method according to claim 1, wherein the decimated signal produced at the step (d) has a final sampling rate comprised between 100 samples/s and 100 k samples/s.

17. The method according to claim 1, wherein the time realignment of the step (e) is carried out by an iterative process having an iterative cycle comprising the steps of:

selecting three values of time drift of zero order and three values of time drift of first order taking, for each order of drift, an initial value and values corresponding to the initial value ±

1/2 search range during a first iteration cycle, then a carry over value and values corresponding to the carry over value ±

1/2 search range during next iteration cycles;

sampling the decimated signal according to starting points set respectively by the values of drift of zero order and with a period equal to the initial value of drift of first order during the first iteration cycle then to the carry over value of drift of first order during the next iteration cycles, to produce three first sampled signals;

interpolating the first sampled signals by a reconstruction function sin(x)/x to which a Blackman-Harris type spectral window is applied, to produce three first interpolated signals, x representing samples'"'"' positions from the first sampled signals;

correlating the first interpolated signals with the reference signature, to produce three first correlated signals;

determining which one of the first correlated signals has a highest correlation value of drift of zero order, the value of drift corresponding to the highest correlation value of drift of zero order becoming then the carry over value of drift of zero order for the next iteration cycle;

sampling the decimated signal according to starting points set respectively by the values of drift of first order and with a period equal to the carry over value of drift of zero order, to produce three second sampled signals;

interpolating the second sampled signals by a reconstruction function sin(x)/x to which a Blackman-Harris type spectral window is applied, to produce three second interpolated signals, x representing samples'"'"' positions from the second sampled signals;

correlating the second interpolated signals with the reference signature, to produce three second correlated signals;

determining which one of the second correlated signals has a highest correlation value of drift of first order, the value of drift corresponding to the highest correlation value of drift of first order becoming then the carry over value of drift of first order for the next iteration cycle; and

dividing the search ranges of each order of drift by a predetermined factor;

if a predetermined stop criterion is reached, then resampling and interpolating the decimated signal using the carry over values of drift of zero and first order to produce the realigned signal, or else stepping back to the next iteration cycle.

18. The method according to claim 17, wherein the stop criterion is a maximum number of iterative cycles to be repeated.

19. The method according to claim 17, wherein the stop criterion is a dimension of search ranges corresponding to a desired resolution.

20. The method according to claim 17, wherein the correlating steps are carried out by simple square deviation calculation.

21. The method according to claim 17, comprising the additional step of:

separating the digital signal produced at the step (a) into distinct frequency bands, the steps (b), (c), (d), (f), (g), (h) and (i) being applied on each of the distinct frequency bands, and the iterative process of the step (e) being applied on the decimated signal. in one of the distinct frequency bands, the resampling and interpolating steps being carried out in each of the distinct frequency bands using the carry over values of drift of zero and first order obtained for said one of the distinct frequency bands.

22. The method according to claim 1, wherein the time realignment of the step (e) is carried out by multiscale correlation or by DTW.

23. The method according to claim 1, wherein the means of the steps (f) and (g) are running means.

24. The method according to claim 23, wherein the running means are updated by recurrence, the running means producing the reference and actualized signatures having respectively weight factors decreasing and increasing progressively with time or a number of realigned signals considered in the means.

25. The method according to claim 1, wherein the variances of the step (h) are updated by recurrence.

26. The method according to claim 1, wherein the step (i) comprises the steps of:

comparing the variances with each other; and

generating an alarm if the variances have a deviation that exceeds a predetermined tolerance threshold.

27. The method according to claim 1, wherein the realigned signal is compared in the step (i) with the actualized and reference signatures by correlation.

28. The method according to claim 1, wherein the realigned signal is compared in the step (i) with the actualized and reference signatures by mean square deviation.

29. The method according to claim 1, comprising the additional step of:

correlating the decimated signal with one of the reference and actualized signatures, to produce a validation indicia of the vibro-acoustic signal.

30. The method according to claim 17, comprising the additional steps of:

adding-the carry over values of drift of zero and first order to respective means;

calculating standard deviations of the carry over values of drift of zero and first order with respect to the respective means;

comparing the carry over values of drift of zero and first order with the respective means, and generating an alarm as soon as one of the carry over values or one of the standard deviations exceeds a predetermined tolerance threshold.

31. The method according to claim 1, comprising the additional steps of:

decompressing the realigned signal by interpolation to produce an interpolated signal; and

displaying the interpolated signal for visual analysis.

32. The method according to claim 1, comprising the additional step of:

comparing the decimated signal initially obtained with a typical signature particular to the switching system under surveillance, to detect setup and other defects when the switching system is put into service.

33. An apparatus for processing a vibro-acoustic signal emitted by a high-voltage switching system, comprising:

a converting means for converting the vibro-acoustic signal into a digital signal;

a rectifying means for rectifying the digital signal and producing a rectified signal;

a filtering means for applying a convolutional filter with a spectral window on the rectified signal and producing a smoothed signal;

a decimating means for decimating the smoothed signal according to a predetermined decimation factor and producing a decimated signal representing an envelope of the vibro-acoustic signal;

a realigning means for performing a time realignment of the decimated signal with respect to a reference signature and producing a realigned signal;

a first calculating means for adding the realigned signal as a factor of a mean to produce the reference signature, the mean having a predetermined reference signal for initial factor;

a second calculating means for adding the realigned signal as a factor of a mean to produce an actualized signature, the mean having the first realigned signal produced by the realigning means for initial factor;

a third calculating means for calculating variances of the realigned signal with respect to the reference and actualized signatures; and

a comparing means for comparing the realigned signal with the actualized and reference signatures to detect a gradual behavior change or a sudden defect, taking the variances into consideration.

34. The apparatus according to claim 33, wherein the converting means comprises:

a sampler sampling the vibro-acoustic signal and producing a sampled signal; and

a converter converting the sampled signal into digital values forming the digital signal.

35. The apparatus according to claim 34, comprising:

an anti-aliasing filter through which the vibro-acoustic signal is passed, the anti-aliasing filter being coupled upstream of the sampler.

36. The apparatus according to claim 33, comprising:

a phase filter through which the digital signal is passed, the phase filter being coupled between the converting and rectifying means.

37. The apparatus according to claim 36, wherein the phase filter carries out a Fourier transform of the digital signal, a phase unwrapping, an angular correction by adding a phase ramp and a reversed Fourier transform to provide a compensated digital signal.

38. The apparatus according to claim 37, wherein the phase ramp is of a second degree.

39. The apparatus according to claim 36, wherein the phase filter carries out a convolution of the digital signal with a chirp.

40. The apparatus according to claim 33, comprising:

an integrator integrating the digital signal, the integrator being coupled between the converting and rectifying means.

41. The apparatus according to claim 33, comprising:

a separator separating the digital signal into distinct frequency bands, the separator being coupled between the converting and rectifying means, each of the distinct frequency bands being processed separately by the rectifying, filtering, decimating, realigning, calculating and comparing means.

42. The apparatus according to claim 41, comprising:

an integrator integrating the digital signal on each of the distinct frequency bands, the integrator being coupled between the separator and the rectifying means.

43. The apparatus according to claim 33, wherein the rectifying means comprises a Hilbert convolutional filter.

44. The apparatus according to claim 33, comprising:

a means coupled between the filtering and decimating means, for shaping the smoothed signal produced by the filtering means by a logarithm.

45. The apparatus according to claim 33, comprising:

an anti-aliasing filter through which the smoothed signal is passed, the anti-aliasing filter being coupled between the filtering and decimating means.

46. The apparatus according to claim 45, wherein the anti-aliasing filter carries out a filtration of a low amplitude portion at a band end of the smoothed signal.

47. The apparatus according to claim 45, wherein the anti-aliasing filter comprises a low pass filter having a cut-off frequency between 50 Hz to 50 kHz with a sampling period between 10 ms to 10 μ

s.

48. The apparatus according to claim 33, wherein the decimated signal produced by the decimating means has a final sampling rate comprised between 100 samples/s and 100 k samples/s.

49. The apparatus according to claim 33, wherein the realigning means carries out an iterative process having an iterative cycle and comprises:

first and second selecting means for selecting respectively three values of time drift of zero order and three values of time drift of first order taking, for each order of drift, an initial value and values corresponding to the initial value ±

1/2 search range during a first iteration cycle, then a carry over value and values corresponding to the carry over value ±

1/2 search range during next iteration cycles;

a sampling means for sampling the decimated signal according to starting points set respectively by the values of drift of zero order and with a period equal to the initial value of drift of first order during the first iteration cycle then to the carry over value of drift of first order during the next iteration cycles, to produce three first sampled signals;

an interpolating means for interpolating the first sampled signals by a reconstruction function sin(x)/x to which a Blackman-Harris type spectral window is applied, to produce three first interpolated signals, x representing samples'"'"' positions from the first sampled signals;

a correlating means for correlating the first interpolated signals with the reference signature, to produce three first correlated signals;

a determining means for determining which one of the first correlated signals has a highest correlation value of drift of zero order, the value of drift corresponding to the highest correlation value of drift of zero order becoming then the carry over value of drift of zero order for the next iteration cycle;

a sampling means for sampling the decimated signal according to starting points set respectively by the values of drift of first order and with a period equal to the carry over value of drift of zero order, to produce three second sampled signals;

an interpolating means for interpolating the second sampled signals by a reconstruction function sin(x)/x to which a Blackman-Harris type spectral window is applied, to produce three second interpolated signals, x representing samples'"'"' positions from the second sampled signals;

a correlating means for correlating the second interpolated signals with the reference signature, to produce three second correlated signals;

a determining means for determining which one of the second correlated signals has a highest correlation value of drift of first order, the value of drift corresponding to the highest correlation value of drift of first order becoming then the carry over value of drift of first order for the next iteration cycle;

a dividing means for dividing the search ranges of each order of drift by a predetermined factor;

an iterating means for stopping the iterative process if a stop criterion is reached, or else stepping back to the next iteration from the selecting means; and

resampling and interpolating means for resampling and interpolating the decimated signal using the carry over values of drift of zero and first order to produce the realigned signal when the iterative process is stopped.

50. The apparatus according to claim 49, wherein the stop criterion is a maximum number of iterative cycles to be repeated.

51. The apparatus according to claim 49, wherein the stop criterion is a dimension of search ranges corresponding to a desired resolution.

52. The apparatus according to claim 49, wherein the correlating means performs simple square deviation calculations.

53. The apparatus according to claim 49, comprising:

a separator separating the digital signal into distinct frequency bands, the separator being coupled between the converting and rectifying means, each of the distinct frequency bands being processed separately by the rectifying, filtering, decimating, calculating and comparing means, the iterative process carried out by the realigning means being applied on the decimated signal in one of the distinct frequency bands, the resampling and interpolating means operating on each of the distinct frequency bands using the carry over values of drift of zero and first order obtained for said one of the distinct frequency bands.

54. The apparatus according to claim 33, wherein the realigning means realigns the decimated signal by a multiscale correlation or by DTW.

55. The apparatus according to claim 33, wherein the means are running means.

56. The apparatus according to claim 55, wherein the first and second calculating means update the running means by recurrence, the running means producing the reference and actualized signatures having respectively weight factors decreasing and increasing progressively with time or a number of realigned signals considered in the means.

57. The apparatus according to claim 33, wherein the third calculating means updates the variances by recurrence.

58. The apparatus according to claim 33, wherein the comparing means compares the variances with each other and generates an alarm if the variances have a deviation that exceeds a predetermined tolerance threshold.

59. The apparatus according to claim 33, wherein the comparing means comprises a correlator that compares the realigned signal with the actualized and reference signatures.

60. The apparatus according to claim 33, wherein the comparing means compares the realigned signal with the actualized and reference signatures by mean square deviation.

61. The apparatus according to claim 33, comprising a correlator correlating the decimated signal with one of the reference and actualized signatures, the correlator being coupled between the decimating means and the first calculating means, to produce a validation indicia of the vibro-acoustic signal.

62. The apparatus according to claim 49, comprising:

a fourth calculating means for adding the carry over values of drift of zero and first order to respective means;

a fifth calculating means for calculating standard deviations of the carry over values of drift of zero and first order with respect to the respective means; and

a comparing means for comparing the carry over values of drift of zero and first order with the respective means and generating an alarm as soon as one of the carry over values or one of the standard deviations exceeds a predetermined tolerance threshold.

63. The apparatus according to claim 33, comprising:

a means for decompressing the realigned signal by interpolation to produce an interpolated signal; and

a means for displaying the interpolated signal for visual analysis.

64. The apparatus according to claim 33, comprising:

a means for comparing the decimated signal initially obtained with a typical signature particular to the switching system under surveillance, for detecting setup and other defects when the switching system is put into service.