CN115219015A - Transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics - Google Patents
Transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics Download PDFInfo
- Publication number
- CN115219015A CN115219015A CN202211005429.0A CN202211005429A CN115219015A CN 115219015 A CN115219015 A CN 115219015A CN 202211005429 A CN202211005429 A CN 202211005429A CN 115219015 A CN115219015 A CN 115219015A
- Authority
- CN
- China
- Prior art keywords
- transformer
- voiceprint
- fault
- frequency
- vibration
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000001514 detection method Methods 0.000 claims abstract description 14
- 238000009432 framing Methods 0.000 claims abstract description 12
- 238000001914 filtration Methods 0.000 claims abstract description 6
- 238000012360 testing method Methods 0.000 claims description 27
- 239000013598 vector Substances 0.000 claims description 24
- 238000000605 extraction Methods 0.000 claims description 16
- 238000000926 separation method Methods 0.000 claims description 12
- 238000001228 spectrum Methods 0.000 claims description 12
- 230000002159 abnormal effect Effects 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 9
- 230000007547 defect Effects 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 230000000737 periodic effect Effects 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 238000013139 quantization Methods 0.000 claims description 5
- 238000012544 monitoring process Methods 0.000 claims description 4
- 230000005236 sound signal Effects 0.000 claims description 3
- 230000001755 vocal effect Effects 0.000 claims description 3
- 238000004804 winding Methods 0.000 abstract description 15
- 238000012423 maintenance Methods 0.000 abstract description 5
- 230000000694 effects Effects 0.000 abstract description 3
- 238000011156 evaluation Methods 0.000 abstract description 3
- 230000005856 abnormality Effects 0.000 abstract 1
- 230000006870 function Effects 0.000 description 8
- 238000009413 insulation Methods 0.000 description 7
- 230000015556 catabolic process Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000006866 deterioration Effects 0.000 description 2
- 239000006185 dispersion Substances 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 210000003462 vein Anatomy 0.000 description 2
- 230000032683 aging Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H17/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves, not provided for in the preceding groups
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Protection Of Transformers (AREA)
Abstract
The invention provides a transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics, and relates to the technical field of digital power. Firstly, extracting transformer noise by a Fourier transform and framing method, and filtering corona, fan and environment interference data; secondly, comparing the voiceprint library of the transformer in the whole life cycle with the current voiceprint of the transformer to judge whether the abnormality exists. On the basis, the multidimensional time-frequency characteristic evaluation weight is adjusted through an entropy weight method, so that the fault voiceprint of the transformer winding is identified. Compared with the prior art, the method has the advantages of high identification accuracy, no need of power failure maintenance, no need of additional detection equipment except a vibration sensor, capability of judging fault types and good identification effect on the winding faults.
Description
Technical Field
The invention relates to the technical field of digital power, in particular to a transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics.
Background
The main transformer is core equipment of a transformer substation, undertakes the tasks of voltage transformation and electric energy distribution, and has positive significance for ensuring safe and stable operation of a power grid in safe and reliable operation. The main transformer is complex in design, and particularly, the converter transformer is in a quality perfection stage, so that the failure rate is high, and if the transformer fails, the problems of power failure loss, high maintenance cost and the like are caused. The fault of the transformer is caused by a plurality of reasons, the state acquisition technology capability of the transformer is low, and the key hidden danger can not be pre-warned in real time. Therefore, identification of transformer faults is required to be intervened to ensure normal working operation of the transformer.
The identification of transformer faults mainly comprises two types of power failure maintenance and online detection. And the power failure maintenance mode is that after the transformer has power failure, the winding is observed whether to deform by adopting a core hanging method, or whether the transformer has a fault is judged by adopting a capacitance testing method, a short-circuit impedance method and a frequency response curve method. However, the method needs a power failure test, is long in inspection time, and cannot find potential transformer faults in time. The on-line detection method is to perform detection by partial discharge of the transformer, analysis of Dissolved Gas (DGA) of transformer oil, and the like. However, the method only has high fault identification rate on the sleeve, the oil tank and the like, and has poor effect on identifying the winding fault.
Therefore, it is necessary to provide a transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics to solve one of the above technical problems.
Disclosure of Invention
In order to solve one of the technical problems, the invention provides a transformer fault voiceprint recognition method based on multi-dimensional time-frequency characteristics, which comprises a transformer vibration signal acquisition step, a transformer noise data separation step, a transformer voiceprint characteristic extraction step, an equipment full life cycle standard voiceprint comparison step, a fault voiceprint time domain analysis step, a fault voiceprint information entropy analysis step and a fault voiceprint defect identification step.
Specifically, the transformer vibration signal acquisition step: the vibration parameters of the engineering machinery in the operation of the transformer are converted into electric signals through the vibration sensor, and the electric signals are measured, so that the voiceprint B of the transformer in the operation of the transformer is obtained a 。
Specifically, the transformer noise data separation step: will be provided withTransformer voiceprint B a Conversion to frequency domain signal F a And applying the frequency domain signal F by framing a Separating the noise data of the transformer to obtain a vibration signal F after the noise data of the transformer are separated b 。
Specifically, the transformer voiceprint feature extraction step comprises the following steps: vibration signal F by MFCC Mel frequency spectrum coefficient b Carrying out transformer voiceprint feature extraction and filtering to obtain transformer voiceprint feature M b 。
Specifically, the comparison step of the standard voiceprint of the whole life cycle of the equipment comprises the following steps: the method comprises the steps of collecting and comparing the standard voiceprint of the whole life cycle; the full-life-cycle standard voiceprint acquisition is to acquire standard voiceprint data of the transformer in the whole process from a factory test to a return operation, and establish a full-life-cycle standard voiceprint; full life cycle standard voiceprint comparison is about to transformer voiceprint feature M b Comparing with the standard voiceprint of the whole life cycle, and calculating the distortion degree; if the distortion degree exceeds a distortion threshold value, judging that the power converter fails, and obtaining a failure voiceprint; otherwise, the converter operates normally.
Specifically, the fault voiceprint time domain analysis step comprises: and carrying out time domain analysis on the fault voiceprint to obtain multi-dimensional time-frequency domain characteristics and corresponding weights.
Specifically, the fault voiceprint information entropy analysis step comprises: and carrying out characteristic analysis on the multi-dimensional time-frequency domain characteristics of the fault voiceprint by an entropy weight method to obtain the fault voiceprint information entropy of the transformer voiceprint multi-dimensional time-frequency joint analysis.
Specifically, a fault voiceprint defect identification step: and setting information entropy characteristics corresponding to various faults in a priori manner, and identifying and classifying the information entropy of the fault voiceprint and the information entropy characteristics of various faults to obtain the fault category of the power converter.
As a further solution, the frequency monitoring range of the vibration sensor is 5Hz to 30KHz, the contact type vibration sensor is arranged at a plurality of mechanical joints of the transformer for voiceprint extraction, and extracted transformer voiceprint B a Comprises the following steps:
wherein n is a The number of vibration sensors installed for the transformer; b si And the mechanical vibration signals of the transformer are collected by different vibration sensors.
As a further solution, the transformer noise data separation step converts the sound signal in the vibration sensor into a frequency domain signal F by Fourier transform a ：
Wherein x (τ) is a periodic function of τ; τ is the time of a periodic function; σ (τ -t) is a window function in the Fourier transform; e.g. of the type -2πjkt As a complex function in the fourier transform.
As a further solution, the separation of the transformer noise data is performed in a framing manner, and the noise signals between two frames after framing are overlapped, so that the number of transformer noise frames G is equal to the number of the noise frames a Comprises the following steps:
wherein n is c Total length of transformer noise data; o a Is the length of the sub-frame; c is the overlapping rate between two frame signals; number of through-transformer noise frames G a Extracting noise data of transformer to obtain vibration signal F b ：
Wherein n is b The upper limit of the integral multiple of 50Hz in the vibration data of the transformer; f. of 1 And f 2 Which are 1 and 2 times the 50Hz in the transformer vibration data, respectively.
As a further solution, MFCC Mel frequency spectrum coefficient to vibration signal F b Carrying out transformer voiceprint feature extraction and Mel conversion frequency M a Comprises the following steps:
wherein d is the vibration frequency of the transformer.
As a further solution, the filtered transformer voiceprint M b Comprises the following steps:
wherein n is d The number of points of Fourier transform; g bi The transformer vibration data after different Fourier transforms are obtained; delta i Filter parameters are extracted for different voiceprints.
As a further solution, the standard voiceprint data of the whole process from factory test to return operation comprises a transformer factory test voiceprint, a handover test voiceprint, a transformer normal operation voiceprint, an equipment abnormal live detection voiceprint, a fault power failure detection voiceprint and a fault test voiceprint.
As a further solution, a plurality of standard voiceprint data are constructed into vector data through vector quantization, and are quantized integrally in a vector space, so that distortion comparison between vectors is realized; average distortion ratio H of transformer standard voiceprint comparison b Comprises the following steps:
wherein n is g Training the number of vector sets for the transformer; d (x) i ,y i ) The distance between the vectors x and y for the different training sets.
As a further solution, the failure voiceprint time domain analysis step selects the sound intensity level, the high frequency energy proportion, the odd-even harmonic amplitude ratio and the frequency spectrum component as the multi-dimensional time-frequency domain characteristics.
As a further solution, an entropy weight method is adopted to carry out multidimensional time-frequency characteristic analysis on the transformer voiceprint, and fault voiceprint information entropy R (z) of the multidimensional time-frequency joint analysis on the transformer voiceprint is obtained a ,z b ,z c ,z d )：
R(z a ,z b ,z c ,z d )＝R(z a )+R(z b )+R(z c )+R(z d )+R(z a |z b |z c |z d )
Wherein R (z) a )、R(z b )、R(z c )、R(z d ) The information entropy of the transformer vocal print sound intensity level, the proportion of high-frequency energy, the amplitude ratio of odd-even sub-harmonics and the frequency spectrum components are respectively. R (z) a |z b |z c |z d ) Is R (z) a )、R(z b )、R(z c )、R(z d ) Four intersections of information entropy.
Compared with the related technology, the transformer fault voiceprint identification method based on the multi-dimensional time-frequency characteristics has the following beneficial effects:
firstly, extracting transformer noise by a Fourier transform and framing method, and filtering corona, fan and environment interference data; and secondly, comparing the voiceprint library of the whole life cycle of the transformer with the current voiceprint of the transformer to judge whether the abnormity exists or not. On the basis, the multidimensional time-frequency characteristic evaluation weight is adjusted through an entropy weight method, so that the fault voiceprint of the transformer winding is identified. Compared with the prior art, the method has the advantages of high identification accuracy, no need of power failure maintenance, no need of additional detection equipment except a vibration sensor, capability of judging fault types and good identification effect on the winding faults.
Drawings
FIG. 1 is a flowchart of a transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics according to the present invention;
fig. 2 is a diagram of a transmission path of the transformer vibration according to the present invention.
Detailed Description
Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. In the drawings, like reference numbers can indicate functionally identical or similar elements. While the various aspects of the embodiments are presented in drawings, the drawings are not necessarily drawn to scale unless specifically indicated.
The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration. Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.
Furthermore, in the following detailed description, numerous specific details are set forth in order to provide a better understanding of the present disclosure. It will be understood by those skilled in the art that the present disclosure may be practiced without some of these specific details. In some instances, methods, means, elements and circuits that are well known to those skilled in the art have not been described in detail so as not to obscure the subject matter of the present disclosure.
As shown in fig. 1 to 2, the transformer fault voiceprint recognition method based on the multi-dimensional time-frequency characteristics provided by the invention comprises a transformer vibration signal acquisition step, a transformer noise data separation step, a transformer voiceprint characteristic extraction step, an equipment full life cycle standard voiceprint comparison step, a fault voiceprint time domain analysis step, a fault voiceprint information entropy analysis step and a fault voiceprint defect identification step.
Specifically, the transformer vibration signal acquisition step: the vibration parameters of the engineering machinery in the operation of the transformer are converted into electric signals through the vibration sensor, and the electric signals are measured, so that the voiceprint B of the transformer in the operation of the transformer is obtained a 。
Specifically, the transformer noise data separation step: the sound pattern B of the transformer a Conversion to frequency domain signal F a And applying the frequency domain signal F by framing a Separating the noise data of the transformer to obtain a vibration signal F after the noise data of the transformer are separated b 。
Specifically, the transformer voiceprint feature extraction step comprises the following steps: vibration signal F by MFCC Mel frequency spectrum coefficient b Carrying out transformer voiceprint feature extraction and filtering to obtain transformer voiceprint featuresM b 。
Specifically, the comparison step of the standard voiceprint of the whole life cycle of the equipment comprises the following steps: the method comprises the steps of collecting and comparing the standard voiceprint of the whole life cycle; the full-life-cycle standard voiceprint acquisition is to acquire standard voiceprint data of the transformer in the whole process from a factory test to a return operation, and establish a full-life-cycle standard voiceprint; full life cycle standard voiceprint comparison is about to transformer voiceprint feature M b Comparing with the standard voiceprint of the whole life cycle, and calculating the distortion degree; if the distortion degree exceeds a distortion threshold value, judging that the power converter fails, and obtaining a failure voiceprint; otherwise, the converter operates normally.
Specifically, the fault voiceprint time domain analysis step comprises: and carrying out time domain analysis on the fault voiceprint to obtain multi-dimensional time-frequency domain characteristics and corresponding weights.
Specifically, the fault voiceprint information entropy analysis step comprises: and carrying out characteristic analysis on the multi-dimensional time-frequency domain characteristics of the fault voiceprint by an entropy weight method to obtain the fault voiceprint information entropy of the transformer voiceprint multi-dimensional time-frequency joint analysis.
Specifically, a fault voiceprint defect identification step: and setting information entropy characteristics corresponding to various faults in a priori manner, and identifying and classifying the information entropy of the fault voiceprint and the information entropy characteristics of various faults to obtain the fault category of the power converter.
It should be noted that: as shown in fig. 1, in the transformer voiceprint feature extraction step, sound data of a transformer are collected by a vibration sensor installed at a connection of a mechanical structure of the transformer, and after noise data are separated by a framing method, transformer voiceprint feature extraction is performed. In a standard voiceprint comparison link of the transformer, a typical transformer voiceprint library is formed by using a transformer handover test and normal operation data, and is compared with current transformer voiceprint data to judge whether the current voiceprint is abnormal or not, and if the current voiceprint is abnormal, the voiceprint is a transformer fault voiceprint. In the identification link of the transformer fault voiceprint defects, firstly, the multidimensional voiceprint time-frequency domain of the transformer is analyzed, and then the information entropy value of each dimension is calculated to form a transformer fault voiceprint identification result.
As a moreIn a further solution, the frequency monitoring range of the vibration sensor is 5Hz to 30KHz, the contact type vibration sensor is arranged at a plurality of mechanical joints of the transformer for voiceprint extraction, and extracted voiceprints B of the transformer a Comprises the following steps:
wherein n is a The number of vibration sensors installed for the transformer; b si And the mechanical vibration signals of the transformer are collected by different vibration sensors.
It should be noted that: the vibration sensor converts the engineering machinery vibration parameters of the transformer operation into electric signals and measures the electric signals so as to obtain the mechanical vibration characteristics of the transformer operation, and the frequency monitoring range of the vibration sensor is 5Hz to 30KHz. In order to ensure accurate acquisition of the vibration signals of the transformer, the contact type vibration sensors are arranged at a plurality of mechanical joints of the transformer to extract the voiceprints.
As a further solution, the transformer noise data separation step converts the sound signal in the vibration sensor into a frequency domain signal F by Fourier transform a ：
Wherein x (τ) is a periodic function of τ; τ is the time of the periodic function; σ (τ -t) is a window function in the Fourier transform; e.g. of the type -2πjkt Is a complex function in the fourier transform.
It should be noted that: the signals extracted by the transformer vibration sensor mainly comprise: transformer noise, fan noise, corona noise, and environmental noise. Wherein the noise of the transformer is a stable signal of integral multiple of 50Hz, and the frequency range is within 2 kHz; the fan noise is a full-frequency-band signal within 2 kHz; the corona noise is a short-time pulse signal of a wide frequency band; the environmental noise is a full-band signal of 20Hz to 20 kHz.
As a further solutionThe scheme is that the separation of the noise data of the transformer adopts a framing mode, and the noise signals between two frames after framing are overlapped, so the number G of the noise frames of the transformer is a Comprises the following steps:
wherein n is c Total length of transformer noise data; o. o a Is the length of the sub-frame; c is the overlapping rate between two frame signals; number of passing transformer noise frames G a Extracting the noise data of the transformer to obtain a vibration signal F b ：
Wherein n is b The upper limit is the integral multiple of 50Hz in the vibration data of the transformer; f. of 1 And f 2 Which are 1 and 2 times the 50Hz of the transformer vibration data, respectively.
It should be noted that: because the noise data of the transformer has fixed frequency and period, the noise of the transformer is a stable signal of integral multiple of 50Hz, and the frequency range is within 2 kHz; and other noises are obviously different from the noise, so that the noise frame number G of the transformer can be determined a To extract transformer noise data.
As a further solution, MFCC Mel spectral coefficients are applied to the vibration signal F b Carrying out transformer voiceprint feature extraction and Mel conversion frequency M a Comprises the following steps:
wherein d is the vibration frequency of the transformer.
As a further solution, the filtered transformer voiceprint M b Comprises the following steps:
wherein n is d The number of points of Fourier transform; g bi The transformer vibration data after different Fourier transforms are obtained; delta i Filter parameters are extracted for different voiceprints.
As a further solution, the standard voiceprint data of the whole process from factory test to return operation comprises a transformer factory test voiceprint, a handover test voiceprint, a transformer normal operation voiceprint, an equipment abnormal live detection voiceprint, a fault power failure detection voiceprint and a fault test voiceprint.
It should be noted that: the sound veins of the whole life cycle of the transformer refer to the sound vein data of the whole process from factory test to return operation of the transformer. The method comprises the following steps: the method comprises the following steps of transformer delivery test voiceprints, handover test voiceprints, transformer normal operation voiceprints, equipment abnormal live detection voiceprints, fault power failure detection voiceprints and fault test voiceprints.
Firstly, comparing the current voiceprint of the transformer with factory test voiceprints, handover test voiceprints and normal operation voiceprints, judging the distortion degree of the current voiceprint and the normal voiceprint of the transformer, and if the distortion degree of the current voiceprint and the normal voiceprint of the transformer is large, indicating that the transformer has a fault. Secondly, the voice print is detected with abnormal electrification of the equipment, the fault power failure detection voice print and the fault test voice print, and the distortion degree of the current voice print and the abnormal voice print is judged. And if the distortion degree of the current voiceprint and the abnormal voiceprint is small, the transformer is indicated to have a fault.
As a further solution, a plurality of standard voiceprint data are constructed into vector data through vector quantization, and are quantized integrally in a vector space, so that distortion comparison between vectors is realized; average distortion ratio H of standard voiceprint comparison of transformer b Comprises the following steps:
wherein n is g Training the number of vector sets for the transformer; d (x) i ,y i ) Vector x for different training setsAnd y.
It should be noted that: vector Quantization (VQ) is a signal correlation Quantization method based on shannon theory, and constructs a plurality of scalar data into Vector data, and quantizes the Vector data in a Vector space as a whole, so as to achieve distortion comparison between vectors.
In a specific embodiment, the distortion degrees of the current ratio transformer voiceprint and the normal and abnormal voiceprints of the transformer full life cycle are respectively counted.
In the formula: delta f 1 Factory test voiceprints; delta f 2 A handover test voiceprint; delta f 3 Normal running voiceprints; delta f 4 Detecting voiceprints for abnormal electrification of equipment; delta f 5 Detecting voiceprints for a fault power outage; delta f 6 Voiceprints were tested for failure.
As a further solution, the failure voiceprint time domain analysis step selects the sound intensity level, the high frequency energy proportion, the odd-even harmonic amplitude ratio and the frequency spectrum component as the multi-dimensional time-frequency domain characteristics.
It should be noted that: the transformer winding faults mainly comprise three types of permanent deformation, insulation degradation and insulation damage of the winding, wherein the permanent deformation of the winding is the conditions of inclination, distortion, displacement, collapse, bulge and the like of the winding; the insulation deterioration is insufficient short-circuit resistance of the winding and insulation aging between turns; insulation breakdown is insulation breakdown that occurs after deformation of the winding and deterioration of the insulation occur for a long time. The transformer winding faults described above all result in a change in the transformer voice print. The propagation path of the transformer vibration is shown in fig. 2. As can be seen from fig. 2, when the vibration propagation path of the transformer does not change, the vibration signal changes less, and when the transformer fails, the mechanical state of the transformer changes, and the vibration signal changes.
Because the difference of the transformer voiceprint characteristics of transformers of different voltage grades and different manufacturers under different operating conditions such as different loads, winding deformation and the like is large, the multidimensional time-frequency domain characteristics of the sound intensity level, the high-frequency energy proportion, the odd-even harmonic amplitude ratio and the frequency spectrum components are selected according to typical settings of transformer industry associations for analysis, as shown in table 1.
TABLE 1 Multi-dimensional time-frequency domain characteristic index Table
As a further solution, an entropy weight method is adopted to carry out multidimensional time-frequency characteristic analysis on the transformer voiceprint, and fault voiceprint information entropy R (z) of the multidimensional time-frequency joint analysis on the transformer voiceprint is obtained a ,z b ,z c ,z d )：
R(z a ,z b ,z c ,z d )＝R(z a )+R(z b )+R(z c )+R(z d )+R(z a |z b |z c |z d )
Wherein R (z) a )、R(z b )、R(z c )、R(z d ) The information entropy of the transformer vocal print sound intensity level, the proportion of high-frequency energy, the amplitude ratio of odd-even sub-harmonics and the frequency spectrum components are respectively. R (z) a |z b |z c |z d ) Is R (z) a )、R(z b )、R(z c )、R(z d ) Four intersections of information entropy.
It should be noted that: the entropy weight method is a system index weight evaluation method. In this method, the degree of dispersion of the index is determined by the size of the entropy value, and the larger the degree of dispersion, the smaller the entropy value, and the larger the weight of the index. Therefore, the entropy weight method is adopted in the text for carrying out the multi-dimensional time-frequency characteristic analysis of the transformer voiceprint.
After the fault voiceprint information entropy is obtained, the transformer winding fault type can be identified through manual classification or machine learning model classification, and the manual classification or the machine learning model classification is the existing classification technology and is not repeated herein.
The above description is only an embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. A transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics is characterized by comprising a transformer vibration signal acquisition step, a transformer noise data separation step, a transformer voiceprint characteristic extraction step, an equipment full-life cycle standard voiceprint comparison step, a fault voiceprint time domain analysis step, a fault voiceprint information entropy analysis step and a fault voiceprint defect identification step;
a transformer vibration signal acquisition step: the vibration parameters of the engineering machinery in the operation of the transformer are converted into electric signals through the vibration sensor, and the electric signals are measured, so that the voiceprint B of the transformer in the operation of the transformer is obtained a ；
A transformer noise data separation step: the sound pattern B of the transformer a Conversion to frequency domain signal F a And applying a framing operation to the frequency domain signal F a Separating the noise data of the transformer to obtain a vibration signal F after the noise data of the transformer are separated b ；
Transformer voiceprint feature extraction: vibration signal F by MFCC Mel frequency spectrum coefficient b Carrying out transformer voiceprint feature extraction and filtering to obtain transformer voiceprint feature M b ；
And comparing the standard voiceprint of the whole life cycle of the equipment: the method comprises the steps of collecting and comparing the standard voiceprint of the whole life cycle; the full-life-cycle standard voiceprint acquisition is to acquire standard voiceprint data of the transformer in the whole process from a factory test to a return operation, and establish a full-life-cycle standard voiceprint; full life cycle standard voiceprint comparison is about to transformer voiceprint feature M b Comparing with the standard voiceprint of the whole life cycle, and calculating the distortion degree; if the distortion degree exceeds the distortion threshold value, judging that the power converter fails, and obtaining a failure voiceprint; otherwise, the converter normally operates;
a fault voiceprint time domain analysis step: carrying out time domain analysis on the fault voiceprint, and obtaining multi-dimensional time-frequency domain characteristics and corresponding weights;
analyzing the entropy of the fault voiceprint information: carrying out characteristic analysis on the multi-dimensional time-frequency domain characteristics of the fault voiceprint by an entropy weight method to obtain fault voiceprint information entropy of the transformer voiceprint multi-dimensional time-frequency joint analysis;
identifying the fault voiceprint defect: and setting information entropy characteristics corresponding to various faults in a priori manner, and identifying and classifying the information entropy of the fault voiceprint and the information entropy characteristics of the various faults to obtain the fault category of the power converter.
2. The method for recognizing the voiceprint of the transformer fault based on the multi-dimensional time-frequency characteristics according to claim 1, wherein the frequency monitoring range of the vibration sensor is 5Hz to 30KHz, the contact type vibration sensors are arranged at a plurality of mechanical joints of the transformer to extract the voiceprint, and the extracted voiceprint B of the transformer is extracted a Comprises the following steps:
wherein n is a The number of vibration sensors installed for the transformer; b si And the mechanical vibration signals of the transformer are collected by different vibration sensors.
3. The method for recognizing the transformer fault voiceprint based on the multi-dimensional time-frequency characteristics according to claim 1, wherein the transformer noise data separation step converts a sound signal in the vibration sensor into a frequency-domain signal F through Fourier transform a ：
Wherein x (τ) is a periodic function of τ; τ is the time of the periodic function; σ (τ -t) is a window function in the Fourier transform; e.g. of the type -2πjkt As a complex function in the fourier transform.
4. The method according to claim 1, wherein the separation of the transformer noise data is performed in a framing manner, and the noise signals between two frames are overlapped after framing, so the number of transformer noise frames G is G a Comprises the following steps:
wherein n is c Total length of transformer noise data; o a Is the length of the sub-frame; c is the overlapping rate between two frame signals;
number of passing transformer noise frames G a Extracting the noise data of the transformer to obtain a vibration signal F b ：
Wherein n is b The upper limit of the integral multiple of 50Hz in the vibration data of the transformer; f. of 1 And f 2 Which are 1 and 2 times the 50Hz in the transformer vibration data, respectively.
5. The method for recognizing the voiceprint of the transformer fault based on the multi-dimensional time-frequency characteristics as claimed in claim 1, wherein the MFCC Mel frequency spectrum coefficient is used for the vibration signal F b Carrying out transformer voiceprint feature extraction and Mel conversion frequency M a Comprises the following steps:
wherein d is the vibration frequency of the transformer.
6. A process according to claim 5The transformer fault voiceprint identification method based on the multi-dimensional time-frequency characteristics is characterized in that the transformer voiceprint M obtained through filtering is b Comprises the following steps:
wherein n is d The number of points of Fourier transform; g bi The transformer vibration data after different Fourier transforms are obtained; delta i Filter parameters are extracted for different voiceprints.
7. The method for identifying the transformer fault voiceprint based on the multidimensional time-frequency characteristics according to claim 1, wherein the standard voiceprint data of the whole process from factory test to return operation comprises a transformer factory test voiceprint, a handover test voiceprint, a transformer normal operation voiceprint, an equipment abnormal live detection voiceprint, a fault power failure detection voiceprint and a fault test voiceprint.
8. The transformer fault voiceprint recognition method based on the multi-dimensional time-frequency characteristics according to claim 7, wherein a plurality of standard voiceprint data are constructed into vector data through vector quantization, and the vector data are integrally quantized in a vector space, so that distortion comparison among vectors is realized; average distortion ratio H of standard voiceprint comparison of transformer b Comprises the following steps:
wherein n is g Training the number of vector sets for the transformer; d (x) i ,y i ) The distance between the vectors x and y for the different training sets.
9. The method for identifying the transformer fault voiceprint based on the multi-dimensional time-frequency characteristics according to claim 1, wherein in the fault voiceprint time domain analysis step, a sound intensity level, a high-frequency energy proportion, an odd-even sub-harmonic amplitude ratio and a spectrum component are selected as the multi-dimensional time-frequency domain characteristics.
10. The method for identifying the transformer fault voiceprint based on the multi-dimensional time-frequency characteristics according to claim 9, wherein the entropy weight method is adopted to analyze the multi-dimensional time-frequency characteristics of the transformer voiceprint to obtain fault voiceprint information entropy R (z) of the multi-dimensional time-frequency joint analysis of the transformer voiceprint a ,z b ,z c ,z d )：
R(z a ,z b ,z c ,z d )＝R(z a )+R(z b )+R(z c )+R(z d )+R(z a |z b |z c |z d )
Wherein R (z) a )、R(z b )、R(z c )、R(z d ) The information entropy of the transformer vocal print sound intensity level, the proportion of high-frequency energy, the amplitude ratio of odd-even sub-harmonics and the frequency spectrum components are respectively. R (z) a |z b |z c |z d ) Is R (z) a )、R(z b )、R(z c )、R(z d ) Four intersections of information entropy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211005429.0A CN115219015A (en) | 2022-08-22 | 2022-08-22 | Transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211005429.0A CN115219015A (en) | 2022-08-22 | 2022-08-22 | Transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115219015A true CN115219015A (en) | 2022-10-21 |
Family
ID=83615557
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211005429.0A Pending CN115219015A (en) | 2022-08-22 | 2022-08-22 | Transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115219015A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116189711A (en) * | 2023-04-26 | 2023-05-30 | 四川省机场集团有限公司 | Transformer fault identification method and device based on acoustic wave signal monitoring |
| CN116884417A (en) * | 2023-07-14 | 2023-10-13 | 国网江苏省电力有限公司南京供电分公司 | Weight distribution-based transformer voiceprint spectrum feature enhancement method and system |
| CN117116290A (en) * | 2023-08-03 | 2023-11-24 | 中科航迈数控软件(深圳)有限公司 | Method and related equipment for positioning defects of numerical control machine tool parts based on multidimensional characteristics |
| CN117607598A (en) * | 2023-12-12 | 2024-02-27 | 国网山东省电力公司莒县供电公司 | Transformer fault detection method and system based on voiceprint characteristics |
-
2022
- 2022-08-22 CN CN202211005429.0A patent/CN115219015A/en active Pending
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116189711A (en) * | 2023-04-26 | 2023-05-30 | 四川省机场集团有限公司 | Transformer fault identification method and device based on acoustic wave signal monitoring |
| CN116189711B (en) * | 2023-04-26 | 2023-06-30 | 四川省机场集团有限公司 | Transformer fault identification method and device based on acoustic wave signal monitoring |
| CN116884417A (en) * | 2023-07-14 | 2023-10-13 | 国网江苏省电力有限公司南京供电分公司 | Weight distribution-based transformer voiceprint spectrum feature enhancement method and system |
| CN116884417B (en) * | 2023-07-14 | 2024-09-03 | 国网江苏省电力有限公司南京供电分公司 | Weight distribution-based transformer voiceprint spectrum feature enhancement method and system |
| CN117116290A (en) * | 2023-08-03 | 2023-11-24 | 中科航迈数控软件(深圳)有限公司 | Method and related equipment for positioning defects of numerical control machine tool parts based on multidimensional characteristics |
| CN117116290B (en) * | 2023-08-03 | 2024-05-24 | 中科航迈数控软件(深圳)有限公司 | Method and related equipment for positioning defects of numerical control machine tool parts based on multidimensional characteristics |
| CN117607598A (en) * | 2023-12-12 | 2024-02-27 | 国网山东省电力公司莒县供电公司 | Transformer fault detection method and system based on voiceprint characteristics |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115219015A (en) | Transformer fault voiceprint identification method based on multi-dimensional time-frequency characteristics | |
| CN109116193B (en) | Electrical equipment dangerous discharge distinguishing method based on partial discharge signal comprehensive entropy | |
| Zhao et al. | Detection of power transformer winding deformation using improved FRA based on binary morphology and extreme point variation | |
| Jayasinghe et al. | Winding movement in power transformers: a comparison of FRA measurement connection methods | |
| Zhao et al. | Diagnosing transformer winding deformation faults based on the analysis of binary image obtained from FRA signature | |
| Contin et al. | Classification and separation of partial discharge signals by means of their auto-correlation function evaluation | |
| SE515387C2 (en) | Monitoring of internal partial discharges in a power transformer | |
| CN110728257A (en) | Transformer winding fault monitoring method based on vibration gray level image | |
| CN103792462B (en) | Power transformer winding turn-to-turn short circuit failure detecting method based on resistance frequency curve | |
| US20120330871A1 (en) | Using values of prpd envelope to classify single and multiple partial discharge (pd) defects in hv equipment | |
| CN109991508B (en) | Transformer winding state diagnosis method based on dynamic nonlinear characteristic sequence | |
| Al-Ameri et al. | Understanding the influence of power transformer faults on the frequency response signature using simulation analysis and statistical indicators | |
| CN108169583B (en) | Autotransformer direct-current magnetic bias discrimination method and system with neutral point grounded through capacitor | |
| Refaat et al. | A review of partial discharge detection techniques in power transformers | |
| Hong et al. | State classification of transformers using nonlinear dynamic analysis and Hidden Markov models | |
| CN110703149B (en) | Method and system for detecting vibration and sound of running state of transformer by utilizing character spacing | |
| Abidullah et al. | Real-time power quality disturbances detection and classification system | |
| Vosoughi et al. | Transformer fault type discrimination based on window calculation method and frequency response analysis | |
| RU2495445C2 (en) | Determination of deteriorated insulating property in insulation between two members of inductive operating component | |
| CN110161363B (en) | Transformer running state vibration and sound detection method and system based on main frequency characteristic quantity | |
| CN112257227A (en) | Dielectric modulus fingerprint database based assessment method for insulation state of sleeve | |
| CN106526433A (en) | Transformer current waveform feature testing method and device | |
| CN110702215B (en) | Transformer running state vibration and sound detection method and system using regression tree | |
| Zheng et al. | An insulation evaluation method of transformer OIP bushings based on WGAN under unbalanced samples | |
| Chang et al. | DC Bias and Type Identification Methods of Various Transformers Based on Vibration Signals |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |