CN117251794A - Dry-type transformer fault diagnosis method and device - Google Patents

Dry-type transformer fault diagnosis method and device Download PDF

Info

Publication number
CN117251794A
CN117251794A CN202310946148.3A CN202310946148A CN117251794A CN 117251794 A CN117251794 A CN 117251794A CN 202310946148 A CN202310946148 A CN 202310946148A CN 117251794 A CN117251794 A CN 117251794A
Authority
CN
China
Prior art keywords
transformer
characteristic parameters
characteristic
extraction
neural network
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
Application number
CN202310946148.3A
Other languages
Chinese (zh)
Inventor
黄炜
张龙飞
韩占利
周昕
张磊
张晓星
黎大健
芦宇峰
陈梁远
喻敏
朱立平
王乐
刘鹏
饶夏锦
彭博雅
李锐
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Electric Power Research Institute of Guangxi Power Grid Co Ltd
Original Assignee
Electric Power Research Institute of Guangxi Power Grid Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Electric Power Research Institute of Guangxi Power Grid Co Ltd filed Critical Electric Power Research Institute of Guangxi Power Grid Co Ltd
Priority to CN202310946148.3A priority Critical patent/CN117251794A/en
Publication of CN117251794A publication Critical patent/CN117251794A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
    • G06F17/141—Discrete Fourier transforms
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
    • G06F17/147—Discrete orthonormal transforms, e.g. discrete cosine transform, discrete sine transform, and variations therefrom, e.g. modified discrete cosine transform, integer transforms approximating the discrete cosine transform
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2132—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on discrimination criteria, e.g. discriminant analysis
    • G06F18/21322—Rendering the within-class scatter matrix non-singular
    • G06F18/21324—Rendering the within-class scatter matrix non-singular involving projections, e.g. Fisherface techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/25—Fusion techniques
    • G06F18/253—Fusion techniques of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2123/00—Data types
    • G06F2123/02—Data types in the time domain, e.g. time-series data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/02—Preprocessing
    • G06F2218/04—Denoising
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Pure & Applied Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Discrete Mathematics (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Probability & Statistics with Applications (AREA)
  • Protection Of Transformers (AREA)

Abstract

The invention discloses a fault diagnosis method and device for a dry-type transformer, and belongs to the technical field of transformers. The method comprises the following steps: extracting the MFCC characteristic parameters and GFCC characteristic parameters of transformer noise, and obtaining fusion characteristic parameters MGCC by utilizing a characteristic fusion algorithm optimized by using Fisher criterion dimension reduction; inputting a characteristic parameter MGCC of a transformer noise signal into a convolutional neural network algorithm to perform first extraction of the characteristics, and then inputting a result of the first extraction into a long-short-time memory neural network algorithm to perform second time sequence signal extraction of the characteristics; and after the result of the extraction of the second time sequence signal is transformed by the flat layer and integrated by the full-connection layer, the result is sent to the Softmax layer to finish the identification and classification of faults of different working conditions of the transformer. The invention can identify and classify faults of different working conditions of the transformer, determine whether the dry-type transformer has abnormal operation, send out an alarm and inform operation and maintenance to carry out maintenance work, thereby reducing maintenance cost.

Description

Dry-type transformer fault diagnosis method and device
Technical Field
The present invention relates to transformer technology, and more particularly, to a method and apparatus for diagnosing faults of a dry-type transformer.
Background
The power transformer is responsible for important tasks such as voltage conversion, electric energy transmission, reactive compensation and the like, and the position of the transformer in a power system is more and more important due to the adjustment of a power grid structure in China and the continuous improvement of power consumption requirements, so that whether the whole power system can safely, stably and economically run is directly threatened. The dry type transformer refers to a transformer with an iron core and windings not immersed in insulating oil, and has the advantages of small maintenance workload, high operation efficiency, small volume, strong short circuit resistance and the like, thereby being widely applied to places such as high-rise buildings and the like. However, when a transformer fails, there is a great influence on social production and life. And effectively monitor the state of the transformer and diagnose faults, thus having very important significance.
In view of this, it is very necessary to provide a fault diagnosis method for a dry-type transformer, which prompts operation and maintenance to carry out maintenance work in the initial stage of fault occurrence, thereby reducing maintenance cost. The voice signal and the voice signal of the transformer have certain common characteristics, so with the rapid rise of the voiceprint technology, the voiceprint monitoring technology of the transformer becomes a current research hot spot. When the sound of a power transformer is measured on site, the sound is often interfered by various environmental noises, and the interference greatly influences the identification effect of the running state of the sound signal of the transformer. Therefore, further researches are needed in the aspects of on-line monitoring of the acoustic signals of the transformer, accurate extraction of the distribution characteristic parameters of the acoustic signals, monitoring and identifying of corresponding faults of the transformer and the like so as to improve the accuracy of monitoring mechanical faults of the transformer.
Disclosure of Invention
In view of this, in order to solve or improve the above-mentioned adverse phenomena in the prior art, the present invention proposes a method and an apparatus for diagnosing faults of a dry-type transformer, which can determine whether the dry-type transformer has an abnormal operation.
In order to achieve the above object, the present invention provides a fault diagnosis method for a dry-type transformer, comprising: extracting the MFCC characteristic parameters and GFCC characteristic parameters of transformer noise, and obtaining fusion characteristic parameters MGCC by utilizing a characteristic fusion algorithm optimized by using Fisher criterion dimension reduction; the convolutional neural network algorithm and the long-short-term memory neural network algorithm are connected in series to form a new characteristic extraction network CNN-LSTM model; inputting a characteristic parameter MGCC of a transformer noise signal into a convolutional neural network algorithm to perform first extraction of the characteristics, and then inputting a result of the first extraction into a long-short-time memory neural network algorithm to perform second time sequence signal extraction of the characteristics; and after the result of the extraction of the second time sequence signal is transformed by the flat layer and integrated by the full-connection layer, the result is sent to the Softmax layer to finish the identification and classification of faults of different working conditions of the transformer.
In one possible implementation manner, before the method, the method further includes: and extracting voiceprints of the transformer noise.
In one possible implementation manner, the method step of extracting the MFCC characteristic parameters of the transformer noise further includes: pre-processing the voiceprint signal of the transformer noise, such as pre-emphasis, framing, windowing and the like; and performing fast Fourier transform on the preprocessed voice print signal of the transformer noise, obtaining a power spectrum, sending the power spectrum into a Mel filter bank, and performing DCT (discrete cosine transform) to obtain the MFCC.
The method for extracting the GFCC characteristic parameters of the transformer noise further comprises the following steps: pre-processing the voiceprint signal of the transformer noise, such as pre-emphasis, framing, windowing and the like; and performing fast Fourier transform on the preprocessed voice print signal of the transformer noise, obtaining a power spectrum, sending the power spectrum into a gammatine filter bank, and performing DCT (discrete cosine transform) to obtain GFCC (frequency division multiplexing) characteristic parameters.
In one possible implementation, the method further includes: confirming whether the transformer has abnormal operation or not; if the transformer is abnormal, an alarm is sent out.
The invention also provides a fault diagnosis device of the dry-type transformer, which comprises the following steps: the first module is used for extracting the MFCC characteristic parameters and GFCC characteristic parameters of the transformer noise and obtaining fusion characteristic parameters MGCC by utilizing a characteristic fusion algorithm optimized by the Fisher criterion for dimension reduction; the second module is used for connecting the convolutional neural network algorithm and the long-short-term memory neural network algorithm in series to form a new characteristic extraction network CNN-LSTM model; the third module is used for inputting the characteristic parameter MGCC of the transformer noise signal into a convolutional neural network algorithm to extract the characteristic for the first time, and then inputting the result of the first extraction into a long-short-time memory neural network algorithm to extract the characteristic for the second time sequence signal; and after the result of the extraction of the second time sequence signal is transformed by the flat layer and integrated by the full-connection layer, the result is sent to the Softmax layer to finish the identification and classification of faults of different working conditions of the transformer.
In one possible implementation, the first module is further configured to extract a voiceprint of the transformer noise.
In one possible implementation manner, the method step of extracting the MFCC characteristic parameters of the transformer noise further includes: pre-processing the voiceprint signal of the transformer noise, such as pre-emphasis, framing, windowing and the like; and performing fast Fourier transform on the preprocessed voice print signal of the transformer noise, obtaining a power spectrum, sending the power spectrum into a Mel filter bank, and performing DCT (discrete cosine transform) to obtain the MFCC.
In one possible implementation manner, the method step of extracting GFCC characteristic parameters of the transformer noise further includes: pre-processing the voiceprint signal of the transformer noise, such as pre-emphasis, framing, windowing and the like; and performing fast Fourier transform on the preprocessed voice print signal of the transformer noise, obtaining a power spectrum, sending the power spectrum into a gammatine filter bank, and performing DCT (discrete cosine transform) to obtain GFCC (frequency division multiplexing) characteristic parameters.
In one possible implementation, the method further includes: a fourth module for confirming whether the transformer has abnormal operation; if the transformer is abnormal, an alarm is sent out.
Advantageous effects
Compared with the prior art, the technical scheme of the invention has the advantages that: the invention realizes the extraction of MFCC characteristic parameters and GFCC characteristic parameters according to the principles of a Mel filter and a Gamma filter, and provides a characteristic fusion algorithm optimized by Fisher criterion dimension reduction on the basis, thus obtaining fusion characteristic parameters MGCC, and extracting voiceprints of running sound as the input of a subsequent model; the two algorithms are connected in series by utilizing the advantages of a Convolutional Neural Network (CNN) algorithm and a long-short-term memory neural network (LSTM) algorithm to form a new feature extraction network CNN-LSTM model; the MGCC characteristic parameters of the noise signals of the transformer are input into a Convolutional Neural Network (CNN) algorithm to extract the characteristics for the first time, then the CNN output is used as the input of a long-short-time memory neural network (LSTM) algorithm to extract the characteristics for the second time, and finally the characteristics are transformed by a Flatten layer and integrated by the paving of a full-connection layer and sent into a Softmax layer to finish the identification and classification of faults of different working conditions of the transformer; the method can identify and classify faults of different working conditions of the transformer to determine whether the dry-type transformer has the problem of abnormal operation, so that when the transformer has abnormal conditions, an alarm is sent out to inform operation and maintenance to carry out maintenance work, and the maintenance cost is reduced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments of the present invention will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present invention, and other drawings may be obtained according to the drawings without inventive effort for a person having ordinary skill in the art. Like elements or portions are generally identified by like reference numerals throughout the several figures. In the drawings, elements or portions thereof are not necessarily drawn to scale.
Fig. 1 is a schematic diagram of a fault diagnosis method for a dry-type transformer according to an embodiment of the present invention.
Fig. 2 is a flow chart of fault diagnosis of a dry-type transformer according to an embodiment of the present invention.
Fig. 3 is a flowchart of an MFCC characteristic parameter extraction process provided by an embodiment of the present invention.
Fig. 4 is a flowchart of a GFCC feature parameter extraction process provided by an embodiment of the invention.
FIG. 5 is a flowchart of a feature fusion algorithm based on Fisher criteria, provided by an embodiment of the invention.
Fig. 6 is a schematic diagram of a fault diagnosis apparatus for a dry-type transformer according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It will be apparent that the described embodiments are only some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and thus should not be construed as limiting the present invention.
In the description of the present invention, a number means one or more, a number means two or more, and greater than, less than, exceeding, etc. are understood to not include the present number, and above, below, within, etc. are understood to include the present number. The terms "first," "second," "third," and the like, if any, are used for descriptive purposes only and for distinguishing between technical features and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated or implicitly indicating the precedence of the technical features indicated.
In the description of the present invention, it should be noted that, unless explicitly specified and limited otherwise, the terms "mounted," "connected," and "connected" are to be construed broadly, and may be either fixedly connected, detachably connected, or integrally connected, for example; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art. In addition, the technical features of the different embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
It should be understood that the terms "comprises" and "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.
It is also to be understood that the terminology used in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.
It should be further understood that the term "and/or" as used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes such combinations.
In order to send out alarm information at the initial stage of the abnormal state of the transformer and inform operation and maintenance to carry out maintenance work, the embodiment provides a fault diagnosis method of a dry-type transformer, which is used for monitoring operation sound according to the operation sound of the dry-type transformer, so that the operation state of equipment can be better mastered, the operation fault of the transformer is found, and the alarm information is sent out at the initial stage of the abnormal state of the transformer. The method is as shown in fig. 1, and includes steps S110 to S130:
step S110, extracting MFCC characteristic parameters and GFCC characteristic parameters of transformer noise, and obtaining fusion characteristic parameters MGCC by utilizing a characteristic fusion algorithm optimized by Fisher criterion dimension reduction;
step S120, a convolutional neural network algorithm and a long-short-term memory neural network algorithm are connected in series to form a new feature extraction network CNN-LSTM model;
step S130, inputting a characteristic parameter MGCC of a transformer noise signal into a convolutional neural network algorithm for first extraction of the characteristics, and then inputting a result of the first extraction into the convolutional neural network algorithm as a long-short-term memory neural network algorithm for second extraction of the characteristics; and after the result of the extraction of the second time sequence signal is transformed by the flat layer and integrated by the full-connection layer, the result is sent to the Softmax layer to finish the identification and classification of faults of different working conditions of the transformer.
Before the method, the method further comprises: and extracting voiceprints of the transformer noise.
After the method, the method further comprises: confirming whether the transformer has abnormal operation or not; if the transformer is abnormal, an alarm is sent out.
Specifically, the implementation flow of the method can be shown in fig. 2:
the extraction of the MFCC characteristic parameters and the GFCC characteristic parameters is realized according to the principles of the Mel filter and the Gamma filter, a characteristic fusion algorithm optimized by the Fisher criterion for reducing the dimension is provided on the basis, the fusion characteristic parameters MGCC are obtained, and the voiceprint of the running sound is extracted as the input of a subsequent model.
It should be noted that, the MFCC characteristic parameter extraction process is shown in fig. 3, and includes step 1, pre-emphasis: the operation sound signal of the dry type transformer passes through a high-pass filter, so that the frequency spectrum of the signal is flattened; 2. framing: n sampling points are integrated into one frame, the constant value of the sampling points is 256 or 512, and the corresponding time of each frame is about 20-30 ms; 3. windowing: adding a Hamming window function for improving the continuity of two ends of a frame signal, wherein the Hamming window function has the characteristics of widening and reducing main lobes and canceling side lobes; 4. fast fourier transform: after pre-emphasis and framing windowing are carried out on the sound signal, a time domain signal with the multiplicative windowing function is obtained, but the characteristics of the signal are generally difficult to distinguish by the transformation on the time domain, the fast Fourier transformation is needed, the time domain signal is converted into the frequency domain, and the frequency spectrum analysis is carried out; 5. mel filter bank: the sensitivity of the human ear to the sounds of different frequency ranges is different, the sound frequency heard by the human ear is almost linearly related to the actual sound frequency in a low frequency range, the sound frequency is basically in a logarithmic relationship in a high frequency range, and the triangular band-pass filter bank of the Mel scale is divided according to the sensitivity degree of the human ear; 6. logarithmic operation: carrying out logarithmic operation on the data output by each Mel filter group; 7. discrete cosine transform: discrete cosine transforming the logarithmic energy to obtain the characteristic parameter of the mel frequency cepstrum coefficient.
It should be noted that, as shown in fig. 4, the GFCC characteristic parameter extraction process is substantially the same as the MFCC characteristic parameter extraction process, and the only difference is that the filter bank is different, and the Mel filter bank is changed to a Gammatone filter bank, so that the Gammatone filter bank can better simulate the perceptual characteristic of the basement membrane, and compared with the Mel filter bank, the noise immunity of the voiceprint recognition is greatly improved.
It should be noted that, as shown in fig. 5, the flowchart of the feature fusion algorithm based on the Fisher criterion has a higher recognition rate in the noisy audio environment, and the classification accuracy is significantly reduced in the complex background noise environment; in contrast, GFCC has better interference immunity and robustness in a noisy-free audio environment, although the recognition rate is not as high as MFCC characteristic parameters. Therefore, the invention fuses the two characteristic quantities with different performances, and better expresses the voiceprint signal of the transformer. And introducing Fisher criterion to reduce the dimension of the mixed characteristic. The method comprises the following steps of 1, pretreatment: pre-emphasis, framing, windowing and the like are carried out on the voiceprint signal of the dry type transformer; 2. extracting MFCC and GFCC characteristic parameters: performing fast Fourier transform, power spectrum calculation, transmission into a Mel filter bank and a Gamma filter bank, and DCT to obtain MFCC and GFCC characteristic parameters; 3. fusing characteristic parameters: and the characteristic quantities with different performances are fused, so that the voiceprint signal of the transformer is better expressed. And introducing Fisher criterion to reduce the dimension of the mixed characteristic.
The two algorithms are connected in series by utilizing the advantages of a Convolutional Neural Network (CNN) algorithm and a long-short-term memory neural network (LSTM) algorithm to form a new feature extraction network CNN-LSTM model.
It should be noted that, the Convolutional Neural Network (CNN) algorithm has the advantage of extracting local features, so as to extract noise signal space features of the dry-type transformer; and the long-short-term memory neural network (LSTM) algorithm is more used for processing text sequences with different lengths, so that the extraction of the noise signal time characteristics of the dry-type transformer can be completed. Therefore, the invention connects the Convolutional Neural Network (CNN) algorithm and the long-short-time memory neural network (LSTM) algorithm in series to form a new characteristic extraction network CNN-LSTM model.
The MGCC characteristic parameters of the noise signals of the transformer are input into a Convolutional Neural Network (CNN) algorithm to extract the characteristics for the first time, then the CNN output is used as the input of a long-short-time memory neural network (LSTM) algorithm to extract the characteristics for the second time sequence signals, and finally the characteristics are transmitted into a Softmax layer to finish the identification and classification of faults of different working conditions of the transformer through the conversion of the Flatten layer and the paving integration of the full-connection layer. The method is used for determining whether the dry-type transformer has the problem of abnormal operation or not, so that when the transformer has abnormal conditions, an alarm is sent out to inform operation and maintenance to carry out overhaul work.
In summary, the fault diagnosis method for the dry-type transformer provided by the invention realizes the extraction of the MFCC characteristic parameter and the GFCC characteristic parameter according to the principles of the Mel filter and the Gamma filter, and provides a feature fusion algorithm optimized by Fisher criterion dimension reduction on the basis, so as to obtain a fusion characteristic parameter MGCC, and extract voiceprints of running sound as the input of a subsequent model; the two algorithms are connected in series by utilizing the advantages of a Convolutional Neural Network (CNN) algorithm and a long-short-term memory neural network (LSTM) algorithm to form a new feature extraction network CNN-LSTM model; the MGCC characteristic parameters of the noise signals of the transformer are input into a Convolutional Neural Network (CNN) algorithm to extract the characteristics for the first time, then the CNN output is used as the input of a long-short-time memory neural network (LSTM) algorithm to extract the characteristics for the second time sequence signals, and finally the characteristics are transmitted into a Softmax layer to finish the identification and classification of faults of different working conditions of the transformer through the conversion of the Flatten layer and the paving integration of the full-connection layer. The method is used for determining whether the dry-type transformer has the problem of abnormal operation or not, so that when the transformer has abnormal conditions, an alarm is sent out to inform operation and maintenance to carry out overhaul work.
As shown in fig. 6, the present embodiment further provides a dry-type transformer fault diagnosis apparatus, including: the first module is used for extracting the MFCC characteristic parameters and GFCC characteristic parameters of the transformer noise and obtaining fusion characteristic parameters MGCC by utilizing a characteristic fusion algorithm optimized by the Fisher criterion for dimension reduction; the second module is used for connecting the convolutional neural network algorithm and the long-short-term memory neural network algorithm in series to form a new characteristic extraction network CNN-LSTM model; the third module is used for inputting the characteristic parameter MGCC of the transformer noise signal into the convolutional neural network algorithm to extract the characteristic for the first time, and then inputting the result of the first extraction into the long-short-term memory neural network algorithm to extract the characteristic for the second time sequence signal; and (3) after the result of the extraction of the second time sequence signal is transformed by the flat layer and integrated by the full-connection layer, the result is sent to the Softmax layer to finish the identification and classification of faults of different working conditions of the transformer.
In this embodiment, the first module is further configured to extract a voiceprint of the transformer noise.
In this embodiment, the method step of extracting the MFCC characteristic parameter of the transformer noise further includes: pre-processing the voiceprint signal of the transformer noise, such as pre-emphasis, framing, windowing and the like; and performing fast Fourier transform on the preprocessed voice print signal of the transformer noise, obtaining a power spectrum, sending the power spectrum into a Mel filter bank, and performing DCT (discrete cosine transform) to obtain the MFCC.
In this embodiment, the method step of extracting GFCC characteristic parameters of transformer noise further includes: pre-processing the voiceprint signal of the transformer noise, such as pre-emphasis, framing, windowing and the like; and performing fast Fourier transform on the preprocessed voice print signal of the transformer noise, obtaining a power spectrum, sending the power spectrum into a gammatine filter bank, and performing DCT (discrete cosine transform) to obtain GFCC (frequency division multiplexing) characteristic parameters.
In this embodiment, further comprising: a fourth module for confirming whether the transformer has abnormal operation; if the transformer is abnormal, an alarm is sent out.
Those of ordinary skill in the art will appreciate that the elements of the examples described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the elements of the examples have been described generally in terms of functionality in the foregoing description to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the embodiments provided in the present invention, it should be understood that the division of the units is merely a logic function division, and there may be other division manners in actual implementation, for example, multiple units may be combined into one unit, one unit may be split into multiple units, or some features may be omitted.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a Read-0nlyMemory (ROM), a random access memory (RAM, randomAccessMemory), a removable hard disk, a magnetic disk, or an optical disk, or the like, which can store program codes.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention, and are intended to be included within the scope of the appended claims and description.

Claims (10)

1. A method for diagnosing faults in a dry-type transformer, comprising:
extracting the MFCC characteristic parameters and GFCC characteristic parameters of transformer noise, and obtaining fusion characteristic parameters MGCC by utilizing a characteristic fusion algorithm optimized by using Fisher criterion dimension reduction;
the convolutional neural network algorithm and the long-short-term memory neural network algorithm are connected in series to form a new characteristic extraction network CNN-LSTM model;
inputting a characteristic parameter MGCC of a transformer noise signal into a convolutional neural network algorithm to perform first extraction of the characteristics, and then inputting a result of the first extraction into a long-short-time memory neural network algorithm to perform second time sequence signal extraction of the characteristics; and after the result of the extraction of the second time sequence signal is transformed by the flat layer and integrated by the full-connection layer, the result is sent to the Softmax layer to finish the identification and classification of faults of different working conditions of the transformer.
2. The dry-type transformer fault diagnosis method according to claim 1, further comprising, before the method: and extracting voiceprints of the transformer noise.
3. The method of claim 1, wherein the step of extracting MFCC characteristic parameters of the transformer noise further comprises:
pre-processing the voiceprint signal of the transformer noise, such as pre-emphasis, framing, windowing and the like;
and performing fast Fourier transform on the preprocessed voice print signal of the transformer noise, obtaining a power spectrum, sending the power spectrum into a Mel filter bank, and performing DCT (discrete cosine transform) to obtain the MFCC.
4. The method of claim 1, wherein the step of extracting GFCC characteristic parameters of the transformer noise further comprises:
pre-processing the voiceprint signal of the transformer noise, such as pre-emphasis, framing, windowing and the like;
and performing fast Fourier transform on the preprocessed voice print signal of the transformer noise, obtaining a power spectrum, sending the power spectrum into a gammatine filter bank, and performing DCT (discrete cosine transform) to obtain GFCC (frequency division multiplexing) characteristic parameters.
5. The method of claim 1, further comprising, after the method: confirming whether the transformer has abnormal operation or not; if the transformer is abnormal, an alarm is sent out.
6. A dry-type transformer fault diagnosis apparatus, comprising:
the first module is used for extracting the MFCC characteristic parameters and GFCC characteristic parameters of the transformer noise and obtaining fusion characteristic parameters MGCC by utilizing a characteristic fusion algorithm optimized by the Fisher criterion for dimension reduction;
the second module is used for connecting the convolutional neural network algorithm and the long-short-term memory neural network algorithm in series to form a new characteristic extraction network CNN-LSTM model;
the third module is used for inputting the characteristic parameter MGCC of the transformer noise signal into a convolutional neural network algorithm to extract the characteristic for the first time, and then inputting the result of the first extraction into a long-short-time memory neural network algorithm to extract the characteristic for the second time sequence signal; and after the result of the extraction of the second time sequence signal is transformed by the flat layer and integrated by the full-connection layer, the result is sent to the Softmax layer to finish the identification and classification of faults of different working conditions of the transformer.
7. The dry-type transformer fault diagnosis apparatus according to claim 6, wherein the first module is further configured to extract voiceprints of transformer noise.
8. The dry transformer fault diagnosis apparatus according to claim 6, wherein the step of extracting MFCC characteristic parameters of the transformer noise further comprises:
pre-processing the voiceprint signal of the transformer noise, such as pre-emphasis, framing, windowing and the like;
and performing fast Fourier transform on the preprocessed voice print signal of the transformer noise, obtaining a power spectrum, sending the power spectrum into a Mel filter bank, and performing DCT (discrete cosine transform) to obtain the MFCC.
9. The dry-type transformer fault diagnosis apparatus according to claim 6, wherein the step of extracting GFCC characteristic parameters of the transformer noise further comprises:
pre-processing the voiceprint signal of the transformer noise, such as pre-emphasis, framing, windowing and the like;
and performing fast Fourier transform on the preprocessed voice print signal of the transformer noise, obtaining a power spectrum, sending the power spectrum into a gammatine filter bank, and performing DCT (discrete cosine transform) to obtain GFCC (frequency division multiplexing) characteristic parameters.
10. The dry-type transformer fault diagnosis apparatus according to claim 6, further comprising: a fourth module for confirming whether the transformer has abnormal operation; if the transformer is abnormal, an alarm is sent out.
CN202310946148.3A 2023-07-31 2023-07-31 Dry-type transformer fault diagnosis method and device Pending CN117251794A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310946148.3A CN117251794A (en) 2023-07-31 2023-07-31 Dry-type transformer fault diagnosis method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310946148.3A CN117251794A (en) 2023-07-31 2023-07-31 Dry-type transformer fault diagnosis method and device

Publications (1)

Publication Number Publication Date
CN117251794A true CN117251794A (en) 2023-12-19

Family

ID=89125464

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310946148.3A Pending CN117251794A (en) 2023-07-31 2023-07-31 Dry-type transformer fault diagnosis method and device

Country Status (1)

Country Link
CN (1) CN117251794A (en)

Similar Documents

Publication Publication Date Title
CN109357749B (en) DNN algorithm-based power equipment audio signal analysis method
CN109507510A (en) A kind of transformer fault diagnosis system
CN108490349A (en) Motor abnormal sound detection method based on Mel frequency cepstral coefficients
CN107610715A (en) A kind of similarity calculating method based on muli-sounds feature
CN111912519B (en) Transformer fault diagnosis method and device based on voiceprint frequency spectrum separation
CN105244038A (en) Ore dressing equipment fault abnormity audio analyzing and identifying method based on HMM
CN108332843A (en) A kind of noise diagnostics method of electrical equipment malfunction electric arc
CN112201260A (en) Transformer running state online detection method based on voiceprint recognition
CN111814872A (en) Power equipment environmental noise identification method based on time domain and frequency domain self-similarity
CN112880750A (en) Transformer multidimensional comprehensive online monitoring intelligent diagnosis system
CN112599134A (en) Transformer sound event detection method based on voiceprint recognition
CN114217149A (en) Transformer acoustic fingerprint uninterrupted power detection and state early warning method
Geng et al. Mechanical fault diagnosis of power transformer by GFCC time-frequency map of acoustic signal and convolutional neural network
CN114530166A (en) Transformer on-load tap-changer fault diagnosis method based on background sound texture
CN116705039A (en) AI-based power equipment voiceprint monitoring system and method
CN115376526A (en) Power equipment fault detection method and system based on voiceprint recognition
CN112016470A (en) On-load tap-changer fault identification method based on sound signal and vibration signal
CN114487952A (en) Quench detection system and method using acoustic optical fiber
CN112666430B (en) Intelligent fault detection method and system for voiceprint of transformer
Dang et al. Cochlear filter cepstral coefficients of acoustic signals for mechanical faults identification of power transformer
CN117292713A (en) Transformer voiceprint monitoring method and system based on end Bian Yun cooperation
CN117251794A (en) Dry-type transformer fault diagnosis method and device
CN114157023B (en) Distribution transformer early warning information acquisition method
CN111860241A (en) Power equipment discharge fault identification method based on wavelet packet analysis
CN116229991A (en) Motor fault diagnosis method based on MFCC voice feature extraction and machine learning

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