CN104237713A - Transformer winding deformation diagnostic method based on discrete wavelet transform - Google Patents

Abstract

The invention provides a transformer winding deformation diagnostic method based on discrete wavelet transform. The transformer winding deformation diagnostic method is characterized in that the discrete wavelet transform method is applied to the frequency response analytic diagnosis of transformer winding deformation. The transformer winding deformation diagnostic method comprises performing decomposition process on a measured waveform obtained by virtue of frequency response analysis and a reference waveform at a plurality of layers by virtue of discrete wavelet transform, and then performing mathematical statistical analysis by use of the processed smooth data and diagnosing the deformation degree of a transformer winding according to a corresponding criterion. Such a diagnostic mode is capable of effectively reducing the errors of a traditional diagnostic mode due to white noise interference during frequency response measurement and some local relatively strong interference points and accurately diagnosing the deformation degree of the transformer winding, and therefore, a new judgment method is provided for estimating transformer fault types and the development trend of faults.

Description

Based on the deformation of transformer winding diagnostic method of wavelet transform

Technical field

The present invention relates to deformation of transformer winding diagnostic field, specifically a kind of deformation of transformer winding diagnostic method based on wavelet transform.

Background technology

According to the incomplete statistics of State Grid Corporation of China, national grid system was at 2002 to 2006 these five-year periods, what electric pressure had an accident at 110kV and above transformer has 162 times, and this has become electric system and has realized the Tough questions that safe and stable operation faces.Especially when main-transformer is directly connected with generator, its the impaired generator that will force stops generating, have an accident when large-scale power transformer runs in systems in which, large-area power-cuts may be caused, maintenance after coil is impaired is very difficult, and its turn(a)round generally wants half a year more than, costs a lot of money, influence surface is very wide, seriously affects the safe and reliable operation of electric system.

At present, the method for existing detection deformation of transformer winding comprises: short circuit impedance method, analysis of vibration signal method, frequency response analysis.Short circuit impedance method is the classic method judging winding failure, have the advantage such as short consuming time, but sensitivity is not high; Analysis of vibration signal method is the state judging transformer according to the vibration signal on fuel tank, can not reflect short circuit in winding fault, by external influence many factors comprehensively; Frequency response analysis is a kind of frequency response characteristic of Measurement and analysis winding on wider frequency band, judges the method for winding state, has the advantages such as highly sensitive, wiring is simple, bandwidth, has been widely applied to during deformation of transformer winding detects.

But further research shows, frequency response analysis still exists following defect or deficiency: there is the easily problem such as, driving source energy localizes lower by the interference of ground unrest, low frequency resolution.In the deformation of transformer winding diagnosis of frequency response analysis, the difference degree of practical frequency response curve and reference waveform is compared in main dependence, and thus the process of signal and the extraction of characteristic parameter are vital.The interference of measured signal mainly comprises two classes: a class is white noise, and signal amplitude is lower, but runs through whole measurement always; Two classes are the relatively strong noise spot of some local.To more adequately reflect winding deformation, need the impact of effectively rejecting undesired signal.Therefore, in correlative technology field, the deformation of transformer winding diagnostic method that searching is more perfect is needed badly, to address the deficiencies of the prior art.

Summary of the invention

The invention provides a kind of deformation of transformer winding diagnostic method based on wavelet transform, can effectively reduce to be disturbed by white noise during frequency response measurement and some local relatively strong noise spot and cause the error of conventional diagnostic pattern, can Accurate Diagnosis deformation of transformer winding degree, for the estimation of transformer fault type and fault progression trend provides new determination methods.

Based on a deformation of transformer winding diagnostic method for wavelet transform, it is characterized in that comprising the steps:

Step one: the sine wave exciting signal low pressure winding terminal of transformer being applied to different frequency encourages, gather the frequency response signal of transformer, frequency response signal is obtained transformer frequencies response measured waveform divided by pumping signal, and the frequency range of measured waveform and reference waveform is 1Hz-1MHz;

Step 2: respectively multilayer wavelet transform is carried out to measured waveform and reference waveform, realize the smoothing of denoising and waveform;

Step 3: the data difference of measured waveform and reference waveform after successively comparative analysis smoothing processing from high to low: detect the measured waveform of the most high layer smoothing processing and the data difference of reference waveform inspection to high-frequency range from 1Hz, when described data difference exceedes a certain threshold value, record to should the frequency of discrepancy as abnormal frequency point, simultaneously using frequency starting point that this abnormal frequency point compares as measured waveform and the reference waveform of lower one deck smoothing processing, if the no abnormal Frequency point of last layer smoothing processing, one deck smoothing processing is then descended to compare from 1Hz, until ground floor smoothing processing Difference test completes,

Step 4: the frequency range two abnormal frequency point that the measured waveform after two-layer smoothing processing contiguous in step 3 and reference waveform contrast obtain comprised is as abnormal frequency band, carry out comparing based on the ratio MM of min-max value and the mathematical criterion of related coefficient CC to the Wave data in abnormal frequency band, the data difference of the abnormal frequency band obtained is contrasted in order to measured waveform after quantitative test smoothing processing and reference waveform, carry out the supplemental diagnostics of fault, the ratio MM of min-max value and the expression formula of related coefficient CC are

$\mathrm{MM}(y_1,y_2)=\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\mathrm{Min}\left(\right|y_{1i}|,|y_{2i}\left|\right)}{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\mathrm{Max}\left(\right|y_{1i}|,|y_{2i}\left|\right)}$

$\mathrm{CC}(y_1,y_2)=\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{y}_{1i}{y}_{2i}}{\sqrt{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{y_{1i}}^2\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{y_{2i}}^2}}$

Wherein, y1, y2 are respectively the measured waveform after smoothing processing and reference waveform, and n is the number of data point;

Step 5: according to the supplemental diagnostics criterion based on mathematical criterion, when the ratio MM of min-max value and related coefficient CC meets certain condition, determines fault type and the order of severity.

Further, described step 4 is specially: as MM<0.9 and CC<0.96 time, think to there is obvious winding deformation; As 0.9<MM<0.96 and 0.96<CC<0.98 time, may slight deformation be there is; As MM>0.96 and CC>0.98 time, without winding deformation.

The present invention compared with prior art, mainly possesses following technological merit:

1, by the change of multilayer discrete wavelet to the smoothing process of the experiment curv of frequency response analysis, can the strong jamming of simultaneously filtering white noise and local, retain the feature of original signal, improve the precision, the reliability that contrast with reference signal and diagnose;

2, by being analyzed the measured waveform of multilayer wavelet transform and reference waveform, can draw the scope occurring abnormal frequency band, more traditional fixed frequency band divides has higher diagnostic accuracy;

3, by the supplemental diagnostics comprising abnormal frequency band and carry out based on mathematical criterion, on the basis detecting frequency response curve difference, the abnormal relevance with actual winding deformation of further determination frequency response comparative analysis, improves the reliability of judgement.

Accompanying drawing explanation

Fig. 1 is the deformation of transformer winding diagnostic system block diagram constructed by the present invention;

Fig. 2 is the schematic flow sheet of the deformation of transformer winding diagnostic method that the present invention is based on wavelet transform.

In figure: 1-broadband excitation signal source, 2-signal Input matching resistance, 3-signal input cable, 4-transformer, 5-signal output cable, 6-signal output matching resistance, 7-high precision oscillograph, 8-host computer.

Embodiment

In order to make object of the present invention, technical scheme and advantage clearly understand, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.In addition, if below in described each embodiment of the present invention involved technical characteristic do not form conflict each other and just can mutually combine.

Fig. 1 is according to the deformation of transformer winding diagnostic system block diagram constructed by the present invention.As shown in fig. 1, this deformation of transformer winding diagnostic system mainly comprises broadband excitation signal source 1, signal Input matching resistance 2, signal input cable 3, transformer 4, signal output cable 5, signal output matching resistance 6, high precision oscillograph 7 and host computer 8.In the present embodiment, broadband excitation signal source adopts FRAX-101, and its output signal amplitude is 5V, and frequency range is 1Hz-10MHz.With signal input cable 3, broadband excitation signal source 1 is connected with the low pressure winding terminal X of transformer 4, and the sine wave exciting signal that this end imposes different frequency is encouraged, utilize signal output cable 5 that pumping signal is sent into high precision oscillograph 7, by coaxial shielding line, the responder x of transformer 4 is connected with output signal build-out resistor 6, high precision oscillograph 7, the response signal of high pressure winding is sent into high precision oscillograph 7.In the present embodiment, shown transformer 4 is the A phase winding test wiring of DYn type transformer, and high precision oscillograph 7 adopts Tektronix TDS3052C, and its bandwidth is 500MHz, and sample frequency is 5GS/s.Diagnosing system software part in host computer 8 of the present invention adopts Matlab platform development, has data and extracts, analyzes and management function, automatically can show the functions such as abnormal frequency band, process progress, processing graphics output, result output.This system meets the requirements such as deformation of transformer winding diagnosis accurate measurement, state estimation.

See Fig. 1, Fig. 2, utilize network service, the pumping signal gather high precision oscillograph 7 and the data file of frequency response signal upload to host computer 8.The data file of reference waveform is kept in host computer 8 in advance.First obtain transformer frequencies response measured waveform by frequency response signal divided by pumping signal, the frequency range of measured waveform and reference waveform is 1Hz-1MHz.Measured waveform and reference waveform are carried out multilayer wavelet transform, in the implementation case, the highest wavelet transform carrying out 7 layers.First compare measured waveform and the reference waveform L7 of the 7th layer of smoothing processing, any one represents a winding deformation extremely.To high-frequency range checkout discrepancy from 1Hz, when the measured waveform of the 7th layer of smoothing processing detected and the data difference of reference waveform inspection are more than 1.5dB, record to should the frequency of discrepancy as abnormal frequency point, simultaneously as the frequency starting point of the 6th layer of smoothing processing waveform comparison.If indifference, then the 6th layer of smoothing processing compares from 1Hz, until the 1st layer of smoothing processing Difference test completes.

The frequency range that two abnormal frequency point obtained for the measured waveform after contiguous two-layer smoothing processing and reference waveform contrast comprise is as abnormal frequency band.Utilize the ratio MM of min-max value and related coefficient CC to carry out the supplemental diagnostics based on mathematical criterion to the Wave data in abnormal frequency band, the ratio MM of min-max value and the expression formula of related coefficient CC are

$\mathrm{MM}(y_1,y_2)=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\mathrm{Min}\left(\right|y_{1i}|,|y_{2i}\left|\right)}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\mathrm{Max}\left(\right|y_{1i}|,|y_{2i}\left|\right)}$

$\mathrm{CC}(y_1,y_2)=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{y}_{1i}{y}_{2i}}{\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{y_{1i}}^2\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{y_{2i}}^2}}$

Wherein, y ₁, y ₂be respectively the trial curve after smoothing processing and reference waveform, n is the number of data point.In the implementation case, as MM<0.9 and CC<0.96 time, think to there is obvious winding deformation; As 0.9<MM<0.96 and 0.96<CC<0.98 time, may slight deformation be there is; As MM>0.96 and CC>0.98 time, without winding deformation.

The above; be only the specific embodiment of the present invention, but protection scope of the present invention is not limited thereto, anyly belongs to those skilled in the art in the technical scope that the present invention discloses; the change that can expect easily or replacement, all should be encompassed within protection scope of the present invention.

Claims (2)

1., based on a deformation of transformer winding diagnostic method for wavelet transform, it is characterized in that comprising the steps:

Step one: the sine wave exciting signal low pressure winding terminal of transformer being applied to different frequency encourages, gather the frequency response signal of transformer, frequency response signal is obtained transformer frequencies response measured waveform divided by pumping signal, and the frequency range of measured waveform and reference waveform is 1Hz-1MHz;

Step 2: respectively multilayer wavelet transform is carried out to measured waveform and reference waveform, realize the smoothing of denoising and waveform;

Step 3: the data difference of measured waveform and reference waveform after successively comparative analysis smoothing processing from high to low: detect the measured waveform of the most high layer smoothing processing and the data difference of reference waveform inspection to high-frequency range from 1Hz, when described data difference exceedes a certain threshold value, record to should the frequency of discrepancy as abnormal frequency point, simultaneously using frequency starting point that this abnormal frequency point compares as measured waveform and the reference waveform of lower one deck smoothing processing, if the no abnormal Frequency point of last layer smoothing processing, one deck smoothing processing is then descended to compare from 1Hz, until ground floor smoothing processing Difference test completes,

Step 4: the frequency range two abnormal frequency point that the measured waveform after two-layer smoothing processing contiguous in step 3 and reference waveform contrast obtain comprised is as abnormal frequency band, carry out comparing based on the ratio MM of min-max value and the mathematical criterion of related coefficient CC to the Wave data in abnormal frequency band, the data difference of the abnormal frequency band obtained is contrasted in order to measured waveform after quantitative test smoothing processing and reference waveform, carry out the supplemental diagnostics of fault, the ratio MM of min-max value and the expression formula of related coefficient CC are

$\mathrm{MM}(y_1,y_2)=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\mathrm{Min}\left(\right|y_{1i}|,|y_{2i}\left|\right)}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\mathrm{Max}\left(\right|y_{1i}|,|y_{2i}\left|\right)}$

$\mathrm{CC}(y_1,y_2)=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{y}_{1i}{y}_{2i}}{\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{y_{1i}}^2\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{y_{2i}}^2}}$

Wherein, y1, y2 are respectively the measured waveform after smoothing processing and reference waveform, and n is the number of data point;

Step 5: according to the supplemental diagnostics criterion based on mathematical criterion, when the ratio MM of min-max value and related coefficient CC meets certain condition, determines fault type and the order of severity.

2., as claimed in claim 1 based on the deformation of transformer winding diagnostic method of wavelet transform, it is characterized in that described step 5 is specially: as MM<0.9 and CC<0.96 time, think to there is obvious winding deformation; As 0.9<MM<0.96 and 0.96<CC<0.98 time, may slight deformation be there is; As MM>0.96 and CC>0.98 time, without winding deformation.