Detection of a direct current component in an inductive device

Abstract

A method for detecting a direct current component in an inductive device, for example in a transformer or choke, includes using a computer for recording an oscillation signal, either of sound emitted from the device or of mechanical oscillation of the device, determining the frequency range of the oscillation signal, determining the value of at least one odd frequency in the frequency range, comparing the value of the odd frequency with the value of at least one even frequency in the frequency range, and determining a direct current component when the value of the odd frequency differs from the even frequency by a predefined amount. The method can be carried out without measuring equipment in the interior of an inductive device and without the involvement of an expert. A computer program product for carrying out the method is also provided.

Claims

1. A method for detecting a direct current component in an inductive device, transformer or inductor, the method comprising using a computer for: recording an oscillation signal of a sound emitted by the device or of a mechanical oscillation of the device; determining a frequency spectrum of the oscillation signal; determining a value of at least one odd harmonic in the frequency spectrum; comparing the value of the at least one odd harmonic with a value of at least one even harmonic in the frequency spectrum; determining a direct current component when the value of the at least one odd harmonic differs by a predefined extent from the value of the at least one even harmonic; and specifying a probability of a presence of a direct current component for determining the direct current component, and increasing the probability as an amount by which the value of the at least one odd harmonic differs from the value of the at least one even harmonic becomes greater.

2. The method according to claim 1, which further comprises predefining different differences for different probabilities of the presence of a direct current component.

3. The method according to claim 1, which further comprises comparing the value of one or more even harmonics with a value of a noise of the oscillation signal, and providing the presence of a direct current component with an uncertainty factor in an event of a difference being smaller than a predefined difference between the values.

4. The method according to claim 1, which further comprises relating two largest values of the even harmonics to one another, and providing the presence of a direct current component with an uncertainty factor in an event of a ratio being greater than a predefined ratio.

5. The method according to claim 1, which further comprises relating two largest values of the odd harmonics, not being equal to a fundamental frequency, to one another, and providing the presence of a direct current component with an uncertainty factor in an event of a ratio being greater than a predefined ratio.

6. The method according to claim 1, which further comprises relating a value of a fundamental oscillation to a value of remaining measured frequencies, and providing the presence of a direct current component with an uncertainty factor in an event of a ratio being greater than a predefined ratio.

7. The method according to claim 1, which further comprises assessing a presence of a direct current component with a lower probability when the value of the odd harmonic differs by a predefined extent from the value of the even harmonic and the difference increases with an increasing load current.

8. The method according to claim 1, which further comprises repeating the method with at least one of a different number of even harmonics or a different number of odd harmonics, and using a deviation as a measure of a robustness of a result when determining a direct current component.

9. A non-transitory computer program product, comprising a program to be directly loaded into a processor of a computer for carrying out the steps of claim 1 when executed by the processor.

Description

DETAILED DESCRIPTION OF THE INVENTION

(1) The invention is now explained in more detail on the basis of exemplary embodiments. These embodiments are exemplary and are intended to explain the concept of the invention, but not to restrict it in any way or to conclusively describe it at all. The starting point for the method according to the invention is an existing so-called narrowband analysis of an oscillation or noise signal from a transformer or an inductor. The following variables are used in this case:

(2) TABLE-US-00001 Variable Designation Explanation S.sub.xN signal value Magnitude of the individual component or of the signal, linearly represented (no dB value); x represents e (even), o (odd) or n (noise) eN even Type of spectral signal oN odd component nN noise N ordinal number Ordinal number of the signal: N = 1, 2, 3, . . . N.sub.max ? 6 (max. 10) N = 1: fundamental oscillation, above it: harmonic f.sub.eN even frequency f.sub.eN = 50.sup.1 .Math. 2 .Math. N in the case of a network voltage frequency of 50 Hz f.sub.oN odd frequency f.sub.oN = 50.sup.1 .Math. 2 .Math. (N + 1/2) in the case of a network voltage frequency of 50 Hz f.sub.neN noise frequency f.sub.nN? = 50.sup.1 .Math. 2 .Math. (N ? 1/4) and f.sub.nN+ = 50.sup.1 .Math. 2 .Math. (N + 1/4) S.sub.nN = (S.sub.fnN? + S.sub.fN+)/2 in the case of a network voltage frequency of 50 Hz A.sub.lin linear factor f-dependent A-assessment of the A- from acoustics, assessment delogarithmized S.sub.xNA S, A-assessed S.sub.xNA = S.sub.xN .Math. A(f.sub.xN) S.sub.xTotA S-total, A- assessed $?\begin{array}{c}\mathrm{Summed}\mathrm{signal}\mathrm{of}\mathrm{the}\mathrm{individual}\\ \begin{array}{c}\mathrm{coponents}{S}_{\mathrm{xN}}\\ S_{x-\mathrm{TotA}}=\underset{N=1}{\overset{N_{\max }}{.Math.}}{S}_{\mathrm{xN}}.Math.A\left(f_{\mathrm{xN}}\right)\end{array}\end{array}$ Se.sub.Tot?1, A S-total without fundamental oscillation, A- assessed $?\begin{array}{c}\mathrm{Summed}\mathrm{signal}\mathrm{of}\mathrm{the}\mathrm{individual}\\ \begin{array}{c}\mathrm{components}{S}_{\mathrm{xN}}\\ S_{\mathrm{eTot}-1,A}=\underset{N=2}{\overset{N_{\max }}{.Math.}}{S}_{\mathrm{xN}}.Math.A\left(f_{\mathrm{xN}}\right)\end{array}\end{array}$ lg ? log.sub.10 R ? ratio $S_{\mathrm{xN}}?S_{\mathrm{fxN}}$

(3) The index 1 indicates that the value 50 should be used for a network voltage frequency of 50 Hz. In the case of a different network voltage frequency, for example in case of a network voltage frequency of 60 Hz, the corresponding frequency should then be used, for example 60 instead of 50. The exact value of the network voltage frequency should be ensured from the frequency of the individual S.sub.eN values up to N.sub.max.

(4) The even frequency components S.sub.eN of an oscillation or of a noise which are cited below are typical of a transformer (or an inductor.) They are intended to clearly stand out from the noise which is described below using the signal components with the noise frequency S.sub.nN. Part of the method in this exemplary embodiment involves detecting and/or filtering out measurement results which cannot be used.

(5) More than 30 seconds are recommended as the duration for measuring an oscillation or noise signal. If there is a suspicion of interfering noises which are not based on a direct current component, a longer measurement duration would be advisable, for instance more than one minute.

(6) The criteria for the usability of the measurement signals are used first.

(7) In this case, the usability of the even measurement signals can be determined as the first criterion by determining the signal-to-noise ratio in dB, specifically first of all for a plurality of individual signals at an even frequency:

(8) $R_{\mathrm{Noise},\mathrm{eN}}\left[\mathrm{dB}\right]=10.Math.\lg \left(\frac{S_{eN}}{S_{nN}}\right)$

(9) The total signal-to-noise ratio of the even frequencies considered is then determined therefrom:

(10) $R_{\mathrm{Noise},e}=\frac{1}{N_{\max }}.Math.\underset{N=1}{\overset{N_{\max }}{.Math.}}{R}_{\mathrm{NoiseN}}$

(11) If this total signal-to-noise ratio is less than a predefined value Limit_R.sub.Noise,e in dB (for example: 6 dB), the noise component of the measurement signals makes the evaluation uncertain and the presence of a direct current component is then provided with an uncertainty factor or with a lower probability.

(12) Prominent individual values of the even and/or odd frequencies can be determined as the second criterion for the usability of the measurement signals.

(13) For this purpose, the two largest S.sub.eN values: S.sub.emax1,A and S.sub.emax2,A are first of all selected and a check is carried out in order to determine whether their ratio

(14) $R_{e\max }\left[\mathrm{dB}\left(A\right)\right]=10.Math.\lg \left(\frac{s_{e\max 1,A}}{s_{e\max 2,A}}\right)$

(15) is greater than a predefined value Limit_R.sub.Se (for example 10 dB (A)). In this case, there is a suspicion of interference or resonance of the even frequencies. The presence of a direct current component is then provided with an uncertainty factor or a lower probability.

(16) Secondly, the same method is carried out for the two largest values of the odd frequencies. The two largest S.sub.oN values: S.sub.omax1,A and S.sub.omax2,A are selected and a check is carried out in order to determine whether their ratio

(17) $R_{o\max }\left[\mathrm{dB}\left(A\right)\right]=10.Math.\lg \left(\frac{s_{o\max 1,A}}{s_{o\max 2,A}}\right)$

(18) is greater than a predefined value Limit_R.sub.So (for example 10 dB (A)). In this case, there is the suspicion of interference or resonance of the odd frequencies. The presence of a direct current component is then provided with an uncertainty factor or a lower probability.

(19) As the third criterion, it is investigated whether the fundamental oscillation excessively dominates, because this is an indication that the current noise does not originate from a direct current component, but rather is dominated by the fundamental oscillation. For this purpose, the following ratio is calculated:

(20) $R_{\mathrm{fund}}\left[\mathrm{dB}\left(A\right)\right]=10.Math.\lg \left(\frac{s_{e1,A}}{s_{eTot-1,A}}\right)$

(21) If this ratio is greater than a predefined value Limit_R.sub.fund (for example 0 dB (A)), the presence of a direct current component is provided with an uncertainty factor or a lower probability.

(22) The odd proportion of the frequencies is then determined as the fourth criterion which constitutes the main criterion for determining the direct current component. The following ratio is formed for this purpose:

(23) $R_{\mathrm{ons}}\left[\mathrm{dB}\left(A\right)\right]=10.Math.\lg \left(S_{o\mathrm{Tot},A}/S_{e\mathrm{Tot},A}\right)$

(24) If this ratio is below a first predefined value Limit_R1.sub.ons for example ?9 dB (A), there is certainly no odd proportion and it is therefore determined that there is no direct current component.

(25) If this ratio is above the first predefined value Limit_R1.sub.ons for example ?9 dB (A), but is below a second predefined value Limit_R2.sub.ons for example ?4 dB (A), there is a small odd proportion of frequencies and there is a suspicion of the presence of a direct current component with a particular probability.

(26) If this ratio R.sub.ons is at or above the second predefined Limit_R2.sub.ons there is a high odd proportion of frequencies and therefore there is a significant suspicion of DC. The probability of the presence of a direct current component is then greater than when the ratio R.sub.ons is only between the first predefined value Limit_R1.sub.ons and the second predefined value Limit_R2.sub.ons.

(27) The probability of the presence of a direct current component, which is determined using the fourth criterion, can be accordingly reduced by the uncertainties or probabilities determined in the first to third criteria.

(28) Finally, the evaluation in order to determine whether there is a direct current component can also be assessed by determining and stating the spectral quality. For this purpose, the evaluation is repeated with another spectral range. For this purpose, the number N of frequencies considered is varied, for example one frequency less, N.sub.max?1, or three frequencies more, N.sub.max+3. If 7 frequencies were therefore used during the first assessment of whether there is a direct current component, the same assessment is repeated again using 6 and/or 10 frequencies.

(29) If the result, that is to say the probability of the presence of a direct current component, then still remains substantially the same, there is a spectrally robust result. Otherwise, there may be an assessment or output to the user indicating that the result is spectrally unstable or spectrally sensitive.