CN110310667A - A kind of equipment fault detection method based on acoustic feature - Google Patents
A kind of equipment fault detection method based on acoustic feature Download PDFInfo
- Publication number
- CN110310667A CN110310667A CN201910628092.0A CN201910628092A CN110310667A CN 110310667 A CN110310667 A CN 110310667A CN 201910628092 A CN201910628092 A CN 201910628092A CN 110310667 A CN110310667 A CN 110310667A
- Authority
- CN
- China
- Prior art keywords
- matrix
- signal
- following formula
- sample
- node
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 19
- 239000000284 extract Substances 0.000 claims abstract description 11
- 238000010586 diagram Methods 0.000 claims abstract description 10
- 238000004364 calculation method Methods 0.000 claims abstract description 9
- 238000001228 spectrum Methods 0.000 claims abstract description 9
- 238000005259 measurement Methods 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 44
- 230000007257 malfunction Effects 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 12
- 238000000034 method Methods 0.000 claims description 11
- 238000007373 indentation Methods 0.000 claims description 6
- 230000021615 conjugation Effects 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 230000000452 restraining effect Effects 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 238000003745 diagnosis Methods 0.000 abstract description 5
- 238000000513 principal component analysis Methods 0.000 abstract description 5
- 238000012880 independent component analysis Methods 0.000 abstract description 3
- 238000002955 isolation Methods 0.000 abstract description 3
- 238000000926 separation method Methods 0.000 description 4
- 230000004888 barrier function Effects 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000005299 abrasion Methods 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000010297 mechanical methods and process Methods 0.000 description 1
- 230000005226 mechanical processes and functions Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
Classifications
-
- 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
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- 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
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
- G10L25/18—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being spectral information of each sub-band
-
- 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
- G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
-
- 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
-
- 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
-
- 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
-
- 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)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (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)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The equipment fault detection method based on acoustic feature that the invention discloses a kind of, is related to fault detection technique field.The interference sound separates calculation method, using following steps: S11, using kurtosis as the measurement of signal non-Gaussian system.The equipment fault detection method based on acoustic feature, sound when transformer equipment work is acquired by microphone array, interference sound isolation technics based on negentropy independent component analysis extracts independent source voice signal, dimension-reduction treatment is carried out to voice data based on two-dimensional principal component analysis and extracts primary spectrum characteristic information, classified based on multivalued decision diagram to voice signal and realizes fault diagnosis, finally based on the electrical equipment fault positioning of weighting Multiple Signal Classification, realize the purpose of real-time online fault detection, it can judgement equipment fault type and information accurately and timely, the use for the person of being convenient to use while improving working efficiency.
Description
Technical field
The present invention relates to fault detection technique field, specially a kind of equipment fault detection method based on acoustic feature.
Background technique
Power equipment, which breaks down, refers to that the working condition of electric system is abnormal, and the partial function of power equipment loses
Effect or the performance indicator of power equipment exceed its rated range, are usually thus that power equipment enters malfunction.Electricity
Structure is complicated for power equipment, and at runtime, system easily breaks down.Failure cause is primarily referred to as in the device operating conditions,
Physics, chemistry, biology or the mechanical process for causing power equipment to fail, such as burn into creep, abrasion, heated, aging
Deng.
In the prior art, Fault Diagnosis for Electrical Equipment and the most common method of state-detection are exactly electrical measurements,
But electrical quantity generally can not significantly characterize electrical equipment fault information, and this parameter is relatively difficult to detect, it is difficult to quasi-
Really timely judge equipment fault type and information, and power equipment can work normally before breaking down and electrical quantity does not have
There is too big variation.
Summary of the invention
(1) the technical issues of solving
In view of the deficiencies of the prior art, the equipment fault detection method based on acoustic feature that the present invention provides a kind of, solution
It has determined in the prior art, Fault Diagnosis for Electrical Equipment and the most common method of state-detection are exactly electrical measurements, still
Electrical quantity generally can not significantly characterize electrical equipment fault information, and this parameter is relatively difficult to detect, it is difficult to it is accurate and
When judgement equipment fault type and information, and power equipment can work normally before breaking down and electrical quantity is without too
Big variation issue.
(2) technical solution
In order to achieve the above object, the present invention is achieved by the following technical programs: a kind of interference sound separation calculation method,
Using following steps:
S11, using kurtosis as the measurement of signal non-Gaussian system, specific algorithm is as follows:
Kurt (x)=E (x4)-3(E(x2))2；
S12, the algorithm based on negentropy maximum direction is realized and sequentially extracts Independent sources signal, and specific algorithm is as follows:
J (x)=H (xgauss)-H(x)；
S13, the negentropy independent element for finding isolated component, concrete operations are as follows:
S13.1, centralization is carried out to data, makes its mean value 0,；Whitened data provides z；
S13.2, the initialization vector w for choosing a unit norm placed in the middle, can randomly select；
S13.3, w is updated, with following formula:
w←w-E{zg(wTz)}-E{g′(wTz)}w
S13.4 standardizes w, with following formula:
w←w/||w||
The loop iteration if not restraining, until convergence；
S14, estimated with the average value of sample, then using Orthogonal Symmetric estimate multiple isolated components negentropy it is independent at
Point, concrete operations are as follows:
S14.1, centralization is carried out to data, makes its mean value 0；Whitened data provides z；
S14.2, selection m, that is, the isolated component number for needing to estimate；
S14.3, all wi of initialization, wherein each wi has unit norm.Then with the method for the 5th step to matrix
It is orthogonalized；
S14.4, each wi is updated, with following formula:
S14.5, Orthogonal Symmetric is carried out to matrix W, with following formula:
W←(WWT)-1/2W
W=(wI..., wm)T
If S14.6, do not restrained, return step S14.4.
A kind of equipment fault detection method based on acoustic feature, including above-mentioned interference sound separate calculation method, using such as
Lower step:
S21, lock out operation is carried out to interference sound according to the parameter that above-mentioned interference sound separation calculation method obtains, using two dimension
Principal Component Analysis Algorithm carries out dimension-reduction treatment to the spectrum signature of voice data, extracts main feature information；
S22, classified with algorithm of support vector machine to voice signal, to judge whether power equipment is in certain event
Barrier state；
S23, it averages to the eigenmatrix of power equipment voice training sample, all training sample eigenmatrix corresponding positions
The characteristic parameter numerical value set is summed and then is averaged, and is obtained Mean Matrix, is then subtracted the eigenmatrix of each training sample
Value matrix, and covariance matrix is sought according to following formula:
Sx=E (Y-EY) (Y-EY)T；
S24, characteristic value is sought covariance matrix, and each feature vector is found out, and according to following formula selected characteristic vector
Do optimal axis of projection:
{X1, X2... Xd}=argmaxJ (X)
J (X)=XTGtX
S25, by the two dimensional character matrix of each power equipment voice data, including training sample and test sample, Xiang Shangshu
Projection matrix projection, to obtain the data for projection after all sample sound two dimensional character matrix dimensionality reductions:
Y=A [X1, X2... Xd]=[Y1, Y2..., Yd]；
S26, H (X) and H (X | B) is calculated according to the following formula according to the distribution of normal condition type X, specific formula is as follows:
S27, according to H (X) and H (X | B), calculate the information gain index value IG (B) of fault type B according to the following formula:
IG (B)=H (X)-H (X | B)；
S28, according to information gain index calculated, from big to small fault type B is ranked up to obtain fault type
Sequence；
S29, the top node (Si, Bi) for constructing decision diagram D, i=1, wherein S1 is all sample sets, and B1 is information gain
Maximum malfunction；In (Si, Bi) indentation storehouse Stack；
The stack top element (Si, Bi) of S210, pop-up a stack stack；
S211, Hash table is searched for (Si, Bi), if " (Si, Bi), Di " exists record, then uses decision in hash table
Di in figure replaces (Si, Bi) node, if the ratio that the sample in Si includes signal characteristic Xi is more than Q, in decision diagram
Normal condition leaf node Xi replaces (Si, Bi) node；If Bi is the last one malfunction, the ratio of the signal characteristic Xi in Si
Example is no more than Q, then abandons (Si, Bi) node；If Bi is not the last one malfunction, the ratio of the signal characteristic Xi in Si is not
Then it is m class according to the possible value of Bi more than Q, stretches out m bifurcated from (Si, Bi), each bifurcated j represents a difference of Bi
Class, to form m decision node of graph (Sj, Bj }), Sj is that score value takes the sample set represented in class in bifurcated j in Si, and Bj is
Next malfunction of Bi in information gain sequence；And in m decision node of graph (Sj, Bj }) indentation storehouse Stack；Turn
2)；
S212, the failure of power equipment is positioned, concrete operations are as follows:
S212.1, conjugation permutatation is carried out to the data of linear array acquisition:
Y (t)=JX*(t)；
The orthogonality of S212.2, signal subspace and noise subspace are enhanced, and are constructed for new noise subspace
Weight matrix W adjusts noise subspace matrix to target signal direction vector in signal subspace matrix by weight matrix
Sensitivity:
Un=u (:, 1:M-K)；
S212.3, the position corresponding with noise subspace matrix Un weight matrix W is multiplied, obtains new noise subspace:
Unew=W.*Un；
S212.4, the Estimation of Spatial Spectrum function for weighting Multiple Signal Classification can calculate as follows:
(3) beneficial effect
The equipment fault detection method based on acoustic feature that the present invention provides a kind of.Have following the utility model has the advantages that the base
In the equipment fault detection method of acoustic feature, sound when transformer equipment work is acquired by microphone array, is based on negentropy
The interference sound isolation technics of independent component analysis extracts independent source voice signal, based on two-dimensional principal component analysis to voice data into
Primary spectrum characteristic information is extracted in row dimension-reduction treatment, is classified based on multivalued decision diagram to voice signal and realizes fault diagnosis, finally
Electrical equipment fault positioning based on weighting Multiple Signal Classification, realizes the purpose of real-time online fault detection, can it is accurate and
When judgement equipment fault type and information, improve working efficiency while the person of being convenient to use use.
Detailed description of the invention
Fig. 1 is electrical equipment fault overhaul flow chart.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Referring to Fig. 1, the present invention provides a kind of technical solution: a kind of interference sound separation calculation method, using following steps:
S11, using kurtosis as the measurement of signal non-Gaussian system, specific algorithm is as follows:
Kurt (x)=E (x4)-3(E(x2))2；
S12, the algorithm based on negentropy maximum direction is realized and sequentially extracts Independent sources signal, and specific algorithm is as follows:
J (x)=H (xgauss)-H(x)；
S13, the negentropy independent element for finding isolated component, concrete operations are as follows:
S13.1, centralization is carried out to data, makes its mean value 0,；Whitened data provides z；
S13.2, the initialization vector w for choosing a unit norm placed in the middle, can randomly select；
S13.3, w is updated, with following formula:
w←w-E{zg(wTz)}-E{g'(wTz)}w
S13.4 standardizes w, with following formula:
w←w/||w||
The loop iteration if not restraining, until convergence；
S14, estimated with the average value of sample, then using Orthogonal Symmetric estimate multiple isolated components negentropy it is independent at
Point, concrete operations are as follows:
S14.1, centralization is carried out to data, makes its mean value 0；Whitened data provides z；
S14.2, selection m, that is, the isolated component number for needing to estimate；
S14.3, all wi of initialization, wherein each wi has unit norm.Then with the method for the 5th step to matrix
It is orthogonalized；
S14.4, each wi is updated, with following formula:
S14.5, Orthogonal Symmetric is carried out to matrix W, with following formula:
W←(WWT)-1/2W
W=(w1..., wm)T
If S14.6, do not restrained, return step S14.4.
A kind of equipment fault detection method based on acoustic feature, including above-mentioned interference sound separate calculation method, using such as
Lower step:
S21, lock out operation is carried out to interference sound according to the parameter that above-mentioned interference sound separation calculation method obtains, using two dimension
Principal Component Analysis Algorithm carries out dimension-reduction treatment to the spectrum signature of voice data, extracts main feature information；
S22, classified with algorithm of support vector machine to voice signal, to judge whether power equipment is in certain event
Barrier state；
S23, it averages to the eigenmatrix of power equipment voice training sample, all training sample eigenmatrix corresponding positions
The characteristic parameter numerical value set is summed and then is averaged, and is obtained Mean Matrix, is then subtracted the eigenmatrix of each training sample
Value matrix, and covariance matrix is sought according to following formula:
Sx=E (Y-EY) (Y-EY)T；
S24, characteristic value is sought covariance matrix, and each feature vector is found out, and according to following formula selected characteristic vector
Do optimal axis of projection:
{X1, X2... Xd}=argmaxJ (X)
J (X)=XTGtX
S25, by the two dimensional character matrix of each power equipment voice data, including training sample and test sample, Xiang Shangshu
Projection matrix projection, to obtain the data for projection after all sample sound two dimensional character matrix dimensionality reductions:
Y=A [X1, X2... Xd]=[Y1, Y2..., Yd]；
S26, H (X) and H (X | B) is calculated according to the following formula according to the distribution of normal condition type X, specific formula is as follows:
S27, according to H (X) and H (X | B), calculate the information gain index value IG (B) of fault type B according to the following formula:
IG (B)=H (X)-H (X | B)；
S28, according to information gain index calculated, from big to small fault type B is ranked up to obtain fault type
Sequence；
S29, the top node (Si, Bi) for constructing decision diagram D, i=1, wherein S1 is all sample sets, and B1 is information gain
Maximum malfunction；In (Si, Bi) indentation storehouse Stack；
The stack top element (Si, Bi) of S210, pop-up a stack stack；
S211, Hash table is searched for (Si, Bi), if " (Si, Bi), Di " exists record, then uses decision in hash table
Di in figure replaces (Si, Bi) node, if the ratio that the sample in Si includes signal characteristic Xi is more than Q, in decision diagram
Normal condition leaf node Xi replaces (Si, Bi) node；If Bi is the last one malfunction, the ratio of the signal characteristic Xi in Si
Example is no more than Q, then abandons (Si, Bi) node；If Bi is not the last one malfunction, the ratio of the signal characteristic Xi in Si is not
Then it is m class according to the possible value of Bi more than Q, stretches out m bifurcated from (Si, Bi), each bifurcated j represents a difference of Bi
Class, to form m decision node of graph (Sj, Bj }), Sj is that score value takes the sample set represented in class in bifurcated j in Si, and Bj is
Next malfunction of Bi in information gain sequence；And in m decision node of graph (Sj, Bj }) indentation storehouse Stack；Turn
2)；
S212, the failure of power equipment is positioned, concrete operations are as follows:
S212.1, conjugation permutatation is carried out to the data of linear array acquisition:
Y (t)=JX*(t)；
The orthogonality of S212.2, signal subspace and noise subspace are enhanced, and are constructed for new noise subspace
Weight matrix W adjusts noise subspace matrix to target signal direction vector in signal subspace matrix by weight matrix
Sensitivity:
Un=u (:, 1:M-K)；
S212.3, the position corresponding with noise subspace matrix Un weight matrix W is multiplied, obtains new noise subspace:
Unew=W.*Un；
S212.4, the Estimation of Spatial Spectrum function for weighting Multiple Signal Classification can calculate as follows:
In conclusion being somebody's turn to do the equipment fault detection method based on acoustic feature, transformer equipment is acquired by microphone array
Sound when work, the interference sound isolation technics based on negentropy independent component analysis extract independent source voice signal, based on two dimension
Principal component analysis carries out dimension-reduction treatment to voice data and extracts primary spectrum characteristic information, based on multivalued decision diagram to voice signal
Fault diagnosis is realized in classification, finally based on the electrical equipment fault positioning of weighting Multiple Signal Classification, realizes real-time online failure
The purpose of detection, can judgement equipment fault type and information accurately and timely, be convenient to use while improving working efficiency
The use of person.
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with
A variety of variations, modification, replacement can be carried out to these embodiments without departing from the principles and spirit of the present invention by understanding
And modification, the scope of the present invention is defined by the appended.
Claims (2)
1. a kind of interference sound separates calculation method, it is characterised in that: use following steps:
S11, using kurtosis as the measurement of signal non-Gaussian system, specific algorithm is as follows:
Kurt (x)=E (x4)-3(E(x2))2；
S12, the algorithm based on negentropy maximum direction is realized and sequentially extracts Independent sources signal, and specific algorithm is as follows:
J (x)=H (xgauss)-H(x)；
S13, the negentropy independent element for finding isolated component, concrete operations are as follows:
S13.1, centralization is carried out to data, makes its mean value 0,；Whitened data provides z；
S13.2, the initialization vector w for choosing a unit norm placed in the middle, can randomly select；
S13.3, w is updated, with following formula:
w←w-E{zg(wTz)}-E{g’(wTz)}w
S13.4 standardizes w, with following formula:
w←w/||w||
The loop iteration if not restraining, until convergence；
S14, estimated with the average value of sample, then estimate the negentropy independent element of multiple isolated components, tool using Orthogonal Symmetric
Gymnastics is made as follows:
S14.1, centralization is carried out to data, makes its mean value 0；Whitened data provides z；
S14.2, selection m, that is, the isolated component number for needing to estimate；
S14.3, all wi of initialization, wherein each wi has unit norm.Then matrix is carried out with the method for the 5th step
Orthogonalization；
S14.4, each wi is updated, with following formula:
S14.5, Orthogonal Symmetric is carried out to matrix W, with following formula:
W←(WWT)-1/2W
W=(w1..., wm)T
If S14.6, do not restrained, return step S14.4.
2. a kind of equipment fault detection method based on acoustic feature, it is characterised in that: separate calculating side including above-mentioned interference sound
Method, using following steps:
S21, the parameter that obtains of calculation method is separated according to above-mentioned interference sound lock out operation is carried out to interference sound, using two dimension it is main at
Divide parser to carry out dimension-reduction treatment to the spectrum signature of voice data, extracts main feature information；
S22, classified with algorithm of support vector machine to voice signal, to judge whether power equipment is in certain failure shape
State；
S23, it averages to the eigenmatrix of power equipment voice training sample, all training sample eigenmatrixes corresponding position
Characteristic parameter numerical value is summed and then is averaged, and obtains Mean Matrix, the eigenmatrix of each training sample is then subtracted mean value square
Battle array, and covariance matrix is sought according to following formula:
Sx=E (Y-EY) (Y-EY)T；
S24, characteristic value is sought covariance matrix, and each feature vector is found out, and done most according to following formula selected characteristic vector
Excellent axis of projection:
{X1, X2... Xd}=arg maxJ (X)
J (X)=XTGt X
S25, by the two dimensional character matrix of each power equipment voice data, including training sample and test sample, to above-mentioned projection
Matrix projection, to obtain the data for projection after all sample sound two dimensional character matrix dimensionality reductions:
Y=A [X1, X2... Xd]=[Y1, Y2..., Yd]；
S26, H (X) and H (X | B) is calculated according to the following formula according to the distribution of normal condition type X, specific formula is as follows:
S27, according to H (X) and H (X | B), calculate the information gain index value IG (B) of fault type B according to the following formula:
IG (B)=H (X)-H (X | B)；
S28, according to information gain index calculated, be ranked up to obtain fault type sequence to fault type B from big to small；
S29, the top node (Si, Bi) for constructing decision diagram D, i=1, wherein S1 is all sample sets, and B1 is information gain maximum
Malfunction；In (Si, Bi) indentation storehouse Stack；
The stack top element (Si, Bi) of S210, pop-up a stack stack；
S211, Hash table is searched for (Si, Bi), if " (Si, Bi), Di " exists record, then in decision diagram in hash table
Di replace (Si, Bi) node, if the ratio that sample in Si includes signal characteristic Xi is more than Q, use normal in decision diagram
State leaf node Xi replaces (Si, Bi) node；If Bi is the last one malfunction, the ratio of the signal characteristic Xi in Si is not
More than Q, then (Si, Bi) node is abandoned；If Bi is not the last one malfunction, the ratio of the signal characteristic Xi in Si is no more than
Q is then m class according to the possible value of Bi, stretches out m bifurcated from (Si, Bi), and each bifurcated j represents an inhomogeneity of Bi, from
And m decision node of graph (Sj, Bj }) is formed, Sj is that score value takes the sample set represented in class in bifurcated j in Si, and Bj is information
Next malfunction of Bi in gain sequence；And in m decision node of graph (Sj, Bj }) indentation storehouse Stack；Turn 2)；
S212, the failure of power equipment is positioned, concrete operations are as follows:
S212.1, conjugation permutatation is carried out to the data of linear array acquisition:
Y (t)=JX*(t)；
The orthogonality of S212.2, signal subspace and noise subspace are enhanced, and construct weight for new noise subspace
Matrix W, by weight matrix adjust noise subspace matrix in signal subspace matrix target signal direction vector it is sensitive
Degree:
Un=u (:, 1:M-K)；
S212.3, the position corresponding with noise subspace matrix Un weight matrix W is multiplied, obtains new noise subspace:
Unew=W.*Un；
S212.4, the Estimation of Spatial Spectrum function for weighting Multiple Signal Classification can calculate as follows:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910628092.0A CN110310667A (en) | 2019-07-12 | 2019-07-12 | A kind of equipment fault detection method based on acoustic feature |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910628092.0A CN110310667A (en) | 2019-07-12 | 2019-07-12 | A kind of equipment fault detection method based on acoustic feature |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110310667A true CN110310667A (en) | 2019-10-08 |
Family
ID=68080091
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910628092.0A Pending CN110310667A (en) | 2019-07-12 | 2019-07-12 | A kind of equipment fault detection method based on acoustic feature |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110310667A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111461090A (en) * | 2020-06-17 | 2020-07-28 | 杭州云智声智能科技有限公司 | Sound vibration signal processing method and system based on environment sample basic cloud model |
| CN112098129A (en) * | 2020-09-11 | 2020-12-18 | 成都大学 | Method and system for detecting operation fault of machine in strong noise environment |
| CN112420074A (en) * | 2020-11-18 | 2021-02-26 | 麦格纳（太仓）汽车科技有限公司 | Method for diagnosing abnormal sound of motor of automobile rearview mirror |
| CN113362856A (en) * | 2021-06-21 | 2021-09-07 | 国网上海市电力公司 | Sound fault detection method and device applied to power Internet of things |
| CN114046968A (en) * | 2021-10-04 | 2022-02-15 | 北京化工大学 | Two-step fault positioning method for process equipment based on acoustic signals |
| CN115497501A (en) * | 2022-11-18 | 2022-12-20 | 国网山东省电力公司济南供电公司 | SW-MUSIC based transformer fault voiceprint positioning method and system |
| CN116359642A (en) * | 2023-03-10 | 2023-06-30 | 湖南金烽信息科技有限公司 | Transformer running state 5G intelligent monitoring system and method |
| CN117809696A (en) * | 2024-02-29 | 2024-04-02 | 南京迅集科技有限公司 | Industrial equipment health assessment and fault prediction method and system based on acoustic analysis |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104064186A (en) * | 2014-06-26 | 2014-09-24 | 山东大学 | Electrical equipment failure tone detection method based on independent component analysis |
| CN104155648A (en) * | 2014-08-26 | 2014-11-19 | 国家海洋局第一海洋研究所 | High-frequency ground-wave radar single-time snapshot MUSIC direction detecting method based on array data rearrangement |
| CN105260965A (en) * | 2015-11-18 | 2016-01-20 | 浙江师范大学 | Decision diagram-based intelligent course selection method |
| CN106934421A (en) * | 2017-03-16 | 2017-07-07 | 山东大学 | Converting station electric power transformer fault detecting system and detection method based on 2DPCA and SVM |
| CN109188244A (en) * | 2018-09-03 | 2019-01-11 | 长沙学院 | Based on the diagnostic method for failure of switch current circuit for improving FastICA |
| CN109543141A (en) * | 2018-11-05 | 2019-03-29 | 临沂大学 | A kind of independent component analysis innovatory algorithm |
-
2019
- 2019-07-12 CN CN201910628092.0A patent/CN110310667A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104064186A (en) * | 2014-06-26 | 2014-09-24 | 山东大学 | Electrical equipment failure tone detection method based on independent component analysis |
| CN104155648A (en) * | 2014-08-26 | 2014-11-19 | 国家海洋局第一海洋研究所 | High-frequency ground-wave radar single-time snapshot MUSIC direction detecting method based on array data rearrangement |
| CN105260965A (en) * | 2015-11-18 | 2016-01-20 | 浙江师范大学 | Decision diagram-based intelligent course selection method |
| CN106934421A (en) * | 2017-03-16 | 2017-07-07 | 山东大学 | Converting station electric power transformer fault detecting system and detection method based on 2DPCA and SVM |
| CN109188244A (en) * | 2018-09-03 | 2019-01-11 | 长沙学院 | Based on the diagnostic method for failure of switch current circuit for improving FastICA |
| CN109543141A (en) * | 2018-11-05 | 2019-03-29 | 临沂大学 | A kind of independent component analysis innovatory algorithm |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111461090A (en) * | 2020-06-17 | 2020-07-28 | 杭州云智声智能科技有限公司 | Sound vibration signal processing method and system based on environment sample basic cloud model |
| CN112098129A (en) * | 2020-09-11 | 2020-12-18 | 成都大学 | Method and system for detecting operation fault of machine in strong noise environment |
| CN112420074A (en) * | 2020-11-18 | 2021-02-26 | 麦格纳（太仓）汽车科技有限公司 | Method for diagnosing abnormal sound of motor of automobile rearview mirror |
| CN113362856A (en) * | 2021-06-21 | 2021-09-07 | 国网上海市电力公司 | Sound fault detection method and device applied to power Internet of things |
| CN114046968A (en) * | 2021-10-04 | 2022-02-15 | 北京化工大学 | Two-step fault positioning method for process equipment based on acoustic signals |
| CN115497501A (en) * | 2022-11-18 | 2022-12-20 | 国网山东省电力公司济南供电公司 | SW-MUSIC based transformer fault voiceprint positioning method and system |
| CN116359642A (en) * | 2023-03-10 | 2023-06-30 | 湖南金烽信息科技有限公司 | Transformer running state 5G intelligent monitoring system and method |
| CN116359642B (en) * | 2023-03-10 | 2024-01-23 | 湖南金烽信息科技有限公司 | Transformer running state 5G intelligent monitoring system and method |
| CN117809696A (en) * | 2024-02-29 | 2024-04-02 | 南京迅集科技有限公司 | Industrial equipment health assessment and fault prediction method and system based on acoustic analysis |
| CN117809696B (en) * | 2024-02-29 | 2024-05-10 | 南京迅集科技有限公司 | Industrial equipment health assessment and fault prediction method and system based on acoustic analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110310667A (en) | A kind of equipment fault detection method based on acoustic feature | |
| CN112699913A (en) | Transformer area household variable relation abnormity diagnosis method and device | |
| CN103488941B (en) | Hardware Trojan horse detection method and system | |
| CN106611106B (en) | Genetic mutation detection method and device | |
| Henry | Robust automatic bandwidth for long memory | |
| CN109828184B (en) | Voltage sag source identification method based on mutual approximate entropy | |
| CN109407649A (en) | A kind of fault type matching process based on fault signature variables choice | |
| Harvey et al. | Phylogenetic extinction rates and comparative methodology | |
| CN109002859B (en) | Sensor array feature selection and array optimization method based on principal component analysis | |
| Meese | Using the standard staircase to measure the point of subjective equality: A guide based on computer simulations | |
| CN110188090A (en) | A kind of distribution topological data method for evaluating quality and device based on data mining | |
| CN110133444A (en) | A kind of Fault Locating Method based on positive sequence voltage variable quantity, apparatus and system | |
| CN109767074A (en) | Effect comprehensive estimation method is planned in a kind of distribution of high reliability service area | |
| Guo et al. | Fault diagnosis for power system transmission line based on PCA and SVMs | |
| CN109753762A (en) | Based on the modified power distribution network two stages network topology identification method of classification and device | |
| CN112801135B (en) | Fault line selection method and device for power plant service power system based on characteristic quantity correlation | |
| Kangping et al. | Analysis on residential electricity consumption behavior using improved k-means based on simulated annealing algorithm | |
| CN108872742A (en) | Multi-stage characteristics towards home environment match non-intrusion type electrical equipment detection method | |
| CN113919430A (en) | Voltage sag monitoring and judging method and device | |
| CN116956745B (en) | Reliability analysis method for positioning and ensuring redundant objects of sealed electronic equipment | |
| CN112636328B (en) | Medium-voltage distribution network equipment load state identification method | |
| CN110008205A (en) | A kind of monitoring system redundant data cleaning method | |
| Yilmaz et al. | Exploratory study on clustering methods to identify electricity use patterns in building sector | |
| CN111199209A (en) | Bearing time-frequency spectrogram identification method based on IWO-KFCM algorithm | |
| CN114091593A (en) | Network-level arc fault diagnosis method based on multi-scale feature fusion |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191008 |
|
| RJ01 | Rejection of invention patent application after publication |