CN107392090A - Optimize Classification of Power Quality Disturbances device ELM method - Google Patents
Optimize Classification of Power Quality Disturbances device ELM method Download PDFInfo
- Publication number
- CN107392090A CN107392090A CN201710415018.1A CN201710415018A CN107392090A CN 107392090 A CN107392090 A CN 107392090A CN 201710415018 A CN201710415018 A CN 201710415018A CN 107392090 A CN107392090 A CN 107392090A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- mtr
- mtd
- power quality
- 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
Classifications
-
- 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
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
-
- 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
-
- 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
Abstract
The invention discloses optimization Classification of Power Quality Disturbances device ELM method, electrical energy power quality disturbance S-transformation signal analysis is carried out first, establish Power Quality Disturbance model, feature extraction is carried out according to the extraction principle of feature on basis herein, then learning training link is carried out, GA is recycled to be optimized for ELM design parameter, carry out a large amount of statistical tests and select preferably experimental program, finally determine that parameter setting carries out electrical energy power quality disturbance identification classification checking, proposed to be applied to classifier design scheme in practice according to the experimental result drawn；So as to realize that ELM graders have higher classification accuracy under various noise circumstances, classificating requirement advantage in practice disclosure satisfy that.
Description
Technical field
The present invention relates to electrical energy power quality disturbance identification technology field, in particular it relates to optimize Classification of Power Quality Disturbances device
ELM method.
Background technology
As the extensive use of power electronic equipment, power quality problem are more obvious.Power quality analysis basis and
On condition that Power Quality Transient Disturbance Signal is classified.The design of existing more single grader is by noise etc.
Influence it is more serious, the shortcomings that design is excessively complicated and cost is higher be present, it is difficult to meet magnanimity disturbance letter in power system
Number analysis classification effectiveness requirement.
It is transient voltage problem that Power Quality Transient, which disturbs the problem of its is substantive,.The disturbance type of general transient state mainly has
Voltage swell, voltage dip, shock oscillation, short interruptions etc..In practice, voltage swell, voltage dip, transient state can typically be shaken
Swing, a variety of phenomenons such as short time voltage is interrupted, harmonic wave and flickering while make a distinction and identify.The analysis of transient disturbance is typically divided to two
Individual step, signal transacting and pattern-recognition.
Current most of graders are affected by noise larger, and time-consuming longer in the training process, can not meet reality
The electrical energy power quality disturbance identification classificating requirement of environment.Domestic related product focuses primarily upon the work(such as the collection of power quality data
Can, lack the research to power quality event classification of type.
The content of the invention
It is an object of the present invention in view of the above-mentioned problems, propose a kind of side for optimizing Classification of Power Quality Disturbances device ELM
Method, think the advantages of analysis of electrical energy power quality disturbance identification is with administering provides effective foundation.
To achieve the above object, the technical solution adopted by the present invention is：It is a kind of to optimize Classification of Power Quality Disturbances device ELM's
Method, mainly include：
Step 1：S-transformation is carried out to Power Quality Disturbance, establishes Power Quality Disturbance model, and extract electric energy
Quality disturbance characteristic；
Step 2：Data after being trained and test according to the electrical energy power quality disturbance characteristic in step 1, ELM are carried out
Parameter selection is set, then training and test setting parameter；
Step 3：The parameter that step 2 obtains is optimized based on genetic iteration algorithm；
Step 4：ELM after optimizing to step 3 carries out fitness evaluation, judges whether to meet end condition, if it is,
Optimal feature subset setting is obtained, if it is not, then continuing executing with step 3.
Further, step 1 is S (τ, f) after S-transformation specifically, input signal is h (t)
Wherein, g (τ, f) is Gaussian window, and parameter τ is used for adjusting the diverse location on Gaussian window time shaft, σ=1/ | f |；
The discrete representation of S-transformation is
Wherein, N represents total sampling number, j and n=0,1 ..., N-1；
A two-dimensional complex number matrix is obtained after discrete S-transformation, i.e., S-transformation time-frequency modular matrix STMM is obtained to its modulus, should
Matrix row vector reflects time domain distribution of the signal under specific frequency, and column vector describes certain time-ofday signals amplitude versus frequency characte.Further
Ground, the ELM are specifically included, input layer, hidden layer and output layer, and the input layer has n neuron, and corresponding n input becomes
Amount；The hidden layer has l neuron；The output layer has m neuron, corresponds to m output variable, wherein, it is corresponding implicit
Layer h-th of interneuronal connection weight a of k-th of neuron and input layer be,
Correspond to i-th of neuron of hidden layer is with k-th of neuron connection weight b of output layer：
The threshold value c of hidden layer neuron is
C=[c1 c2 … cl]T l×1 (6)
Assuming that there is Q sample training collection, if input matrix X and output matrix Y are respectively：
Assuming that the activation primitive of hidden layer neuron is k (x), then network output F is
F=[t1 t2 … tQ]n×Q (9)
Wherein ai=[ai1 ai2 … ain],xj=[x1j x2j … xnj]T
Formula (10) can be deformed into
Ha=F'(11)
Wherein, F' is the transposition of matrix F；H represents the hidden layer output matrix of neutral net.
Further, in step 2, the ELM carries out parameter selection and set, then training and test setting parameter, specifically
It is trained and tests to input the parameter selection of weights and bias vector to ELM hidden layers.
Further, the step 3 specifically includes, according to ELM electrical energy power quality disturbance combinations of features data, construction heredity
Population at individual in computing, is encoded to chromosome, is then carried out parameter setting to the combinations of features of genetic group, is obtained most
Excellent character subset.
Further, the combinations of features to genetic group carries out parameter setting specifically, feature to population at individual
Combine into row variation and crossing operation.
Further, the Power Quality Disturbance specifically includes voltage swell signal, voltage dip signal, flickering letter
Number, interrupt signal, harmonic signal and oscillator signal composition compound disturbing signal.
A kind of storage device, wherein being stored with a plurality of instruction, the instruction is loaded and performed by processor：
Step 1：S-transformation is carried out to Power Quality Disturbance, establishes Power Quality Disturbance model, and extract electric energy
Quality disturbance characteristic；
Step 2：Data after being trained and test according to the electrical energy power quality disturbance characteristic in step 1, ELM are carried out
Parameter selection is set, then training and test setting parameter；
Step 3：The parameter that step 2 obtains is optimized based on genetic iteration algorithm；
Step 4：ELM after optimizing to step 3 carries out fitness evaluation, judges whether to meet end condition, if it is,
Optimal feature subset setting is obtained, if it is not, then continuing executing with step 3.
The optimization Classification of Power Quality Disturbances device ELM of various embodiments of the present invention method, carries out electrical energy power quality disturbance S first
Signal analysis is converted, establishes Power Quality Disturbance model, feature is carried out according to the extraction principle of feature on basis herein
Extraction, learning training link is then carried out, recycle GA to be optimized for ELM design parameter, carry out a large amount of statistical tests
Preferably experimental program is selected, finally determines that parameter setting carries out electrical energy power quality disturbance identification classification checking, according to the reality drawn
Result is tested to propose to be applied to classifier design scheme in practice；So as to realize that ELM graders have under various noise circumstances
Higher classification accuracy, it disclosure satisfy that classificating requirement advantage in practice.
Other features and advantages of the present invention will be illustrated in the following description, also, partly becomes from specification
Obtain it is clear that or being understood by implementing the present invention.
Below by drawings and examples, technical scheme is described in further detail.
Brief description of the drawings
Accompanying drawing is used for providing a further understanding of the present invention, and a part for constitution instruction, the reality with the present invention
Apply example to be used to explain the present invention together, be not construed as limiting the invention.In the accompanying drawings：
Fig. 1 is the method flow diagram of the optimization Classification of Power Quality Disturbances device ELM described in the embodiment of the present invention；
Fig. 2 is the ELM network structures described in the embodiment of the present invention；
Fig. 3 is the signal to noise ratio 20db experimental result pictures described in the embodiment of the present invention；
Fig. 4 is the signal to noise ratio 50db experimental result pictures described in the embodiment of the present invention；
Fig. 5 is the random noise experimental result picture described in the embodiment of the present invention；
Fig. 6 is the method overview flow chart of the optimization Classification of Power Quality Disturbances device ELM described in the embodiment of the present invention.
Embodiment
The preferred embodiments of the present invention are illustrated below in conjunction with accompanying drawing, it will be appreciated that described herein preferred real
Apply example to be merely to illustrate and explain the present invention, be not intended to limit the present invention.
In the case of taking into account actual environment and strong noise, there is provided a kind of optimization ELM grader
Design method, for electrical energy power quality disturbance in practice, voltage swell, voltage dip, flickering, interruption, harmonic wave, vibration etc. are multiple
Disturbance is closed to be classified.Can be the analysis of electrical energy power quality disturbance identification and administer provides effective foundation.
By being tested respectively under 20,50db and not 3 kinds of conditions environmentals of Noise, verify that the grader can reach
The classificating requirement of practical application.The characteristics of ELM has adaptive ability and autonomous learning to most non-precision rules, only need to be
Suitable hidden layer neuron number is chosen before training, implementation procedure is once completed, and optimal solution is achieved with without iteration, is had good
Good classification accuracy and classification effectiveness.Parameter optimization is carried out to ELM using GA, can so obtain more preferable effect.With showing
Some neural network classifiers are compared, it is possible to reduce the time of training and the internal memory taken.Enter respectively in the environment of 3 kinds of noises
Perturbation features are classified by row experiment by the grader, under the premise of guarantee is efficient, and it is accurate with higher classification
True rate, the grader can be applied in actual Classification of Power Quality Disturbances, had to power quality analysis field certain
Positive effect.
Idea of the invention is that using neutral net carry out electrical energy power quality disturbance identification basis of classification on extension, be
ELM has carries out parameter optimization proposition on good identification classification capacity basis.Flow is as shown in Figure 1：Electricity is carried out first
Energy quality disturbance S-transformation signal analysis, establishes Power Quality Disturbance model, former according to the extraction of feature on basis herein
Feature extraction is then carried out, then carries out learning training link, recycles GA to be optimized for ELM design parameter, is carried out big
Amount statistical test selects preferably experimental program, finally determines that parameter setting carries out electrical energy power quality disturbance identification classification checking, root
It is proposed to be applied to classifier design scheme in practice according to the experimental result drawn.
Iing is proposed for this method can be by the classifier design in pairs in the software of disturbed depth classification, can also be in electronic die
The classifier modules dedicated for disturbed depth classification are designed in block design, the new method identifies applied to electrical energy power quality disturbance
Improvement for power quality problem and prediction are provided to the foundation of an analysis in classification.
Below by concrete analysis, to describe the technical scheme of invention in detail.
1. the Power Quality Disturbance feature extraction based on S-transformation
Input signal is h (t), is S (τ, f) after S-transformation
Wherein, g (τ, f) is Gaussian window, and parameter τ is used for adjusting the diverse location on Gaussian window time shaft, σ=1/ | f |.S
The discrete representation of conversion is：
Wherein, N represents total sampling number, j and n=0,1 ..., N-1.
A two-dimensional complex number matrix can be obtained after discrete S-transformation, i.e., S-transformation time-frequency modular matrix (STMM) is obtained to its modulus.
The matrix row vector reflects time domain distribution of the signal under specific frequency, and column vector describes certain time-ofday signals amplitude versus frequency characte.Follow
The basic principle of feature extraction, 25 kinds of perturbation features in STMM in extraction such as table 1 below are used to describe disturbing signal.
The electrical energy power quality disturbance characteristic set of table 1
ELM is a kind of single improved new algorithm of hidden layer feedforward neural network.Input layer has n neuron in Fig. 2, corresponding
N input variable；Hidden layer has l neuron；Output layer has m neuron, corresponding m output variable.
Wherein, corresponding k-th of neuron of hidden layer and h-th of interneuronal connection weight of input layer.Corresponding hidden layer
I-th of neuron and k-th of neuron connection weight of output layer.
The connection weight a of input layer and implicit interlayer is
Hidden layer and export interlayer connection weight b be
The threshold value c of hidden layer neuron is
C=[c1 c2 … cl]T l×1 (6)
Assuming that there is Q sample training collection, if input matrix X and output matrix Y are respectively
Assuming that the activation primitive of hidden layer neuron is k (x), then network output F is
F=[t1 t2 … tQ]n×Q (9)
Wherein ai=[ai1 ai2 … ain],xj=[x1j x2j … xnj]T
Formula (10) can be deformed into
Ha=F'(11)
Wherein, F' is the transposition of matrix F；H represents the hidden layer output matrix of neutral net.
The GA of prioritization scheme based on to(for) ELM graders
Because the hidden layer input weights and bias vector of ELM sorter models generate at random in the training process,
Tested under numerous conditions, selection improperly situation occurs in random setting value, can reduce and electrical energy power quality disturbance is identified
The accuracy rate of classification.Analyzed, proposed using GA optimization ELM models hidden layer input weights and bias vector according to problem above
Parameter selection optimizes processing, to ensure to obtain preferably sorter model.The small ELM populations of error are selected by GA,
And the diversity that can ensure that combination by GA variation and crossing operation in combinations of features (population size in figure, is exactly
Characteristic species group closes the diversity considered, and the more operands of choosing are big, and it is not comprehensive enough that consideration has been lacked in choosing, utilizes variation and crossing operation
Exactly in order to increase the diversity of colony, it is believed that consider in an experiment it is enough comprehensively), so as to can be with by a large amount of statistical experiments
Obtain higher classification accuracy.The classifier design that the disaggregated model for finally meeting classificating requirement is defined as applying in practice
Method.
Analysis of experimental results：
According to experiment porch Matlab7.10 IEEE1159 normal datas, the Classification and Identification that disturb of the present invention is tested
Card.The data of 25 features every group 500 of 8 kinds of disturbance types are tested in experiment.By all data in experiment
Random sequence processing is carried out, is trained link using preceding 2000 groups of data, 2000 groups of data carry out test link after.Through
Statistical experiment arrange parameter, GA parameters：Population size：30；Crossover probability：0.7；Mutation probability：0.02；ELM parameters：Hidden layer
Neuron number：30；Activation primitive：sig.
(1) experimental result is as shown in Figure 3 in the case of signal to noise ratio 20db.
After genetic algorithm iteration 100 times, feature 4,5,16,18,19 may finally reach classification accuracy 98.6%.
(2) experimental result is as shown in Figure 4 in the case of signal to noise ratio 50db.
After genetic algorithm iteration 100 times, feature 9,18,19,20 may finally reach classification accuracy 99.8%.
(3) experimental result is as shown in Figure 5 in the case of random noise.
After genetic algorithm iteration 150 times, feature 7,9,19,20,22 may finally reach classification accuracy 99.4%.
Can by the comparison under three kinds of noise circumstances of signal to noise ratio 20db, 50db in Fig. 3, Fig. 4 and Fig. 5 and random noise
Know, by the feature selected after 100 iteration can be that classification accuracy reaches more than 99% under 50db and random noise environment；
Eventually pass through to throw away after iteration 100 times under 20db high-noise environment and can reach classification accuracy more than 98%.Illustrate that the present invention carries
The new method gone out has certain noise immunity, therefore, has more certain adaptability in the application of reality.
The present invention proposes a kind of classifier design method for electrical energy power quality disturbance identification classification, and the grader is used
In experimental verification, higher classification accuracy can be obtained under three kinds of noise circumstances, for power quality problem in practice
Analysis and improvement provide the foundation of a research, have important practical significance for power quality analysis field.
Following beneficial effect can at least be reached：Compared with existing electrical energy power quality disturbance recognition classifier designs, for ginseng
Number is set, and is trained and is generated by GA interative computations；ELM has faster pace of learning；ELM graders are under various noise circumstances
There is higher classification accuracy, disclosure satisfy that classificating requirement in practice.
Finally it should be noted that：The preferred embodiments of the present invention are the foregoing is only, are not intended to limit the invention,
Although the present invention is described in detail with reference to the foregoing embodiments, for those skilled in the art, it still may be used
To be modified to the technical scheme described in foregoing embodiments, or equivalent substitution is carried out to which part technical characteristic.
Within the spirit and principles of the invention, any modification, equivalent substitution and improvements made etc., it should be included in the present invention's
Within protection domain.
Claims (8)
1. optimize Classification of Power Quality Disturbances device ELM method, it is characterised in that comprise the following steps：
Step 1：S-transformation is carried out to Power Quality Disturbance, establishes Power Quality Disturbance model, and extract the quality of power supply
Perturbation features data；
Step 2：Data after being trained and test according to the electrical energy power quality disturbance characteristic in step 1, ELM carry out parameter
Selection is set, then training and test setting parameter；
Step 3：The parameter that step 2 obtains is optimized based on genetic iteration algorithm；
Step 4：ELM after optimizing to step 3 carries out fitness evaluation, judges whether to meet end condition, if it is, obtaining
Optimal feature subset is set, if it is not, then continuing executing with step 3.
2. optimization Classification of Power Quality Disturbances device ELM according to claim 1 method, it is characterised in that step 1 is specific
For input signal is h (t), is S (τ, f) after S-transformation
<mrow>
<mi>S</mi>
<mrow>
<mo>(</mo>
<mi>&tau;</mi>
<mo>,</mo>
<mi>f</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msubsup>
<mo>&Integral;</mo>
<mrow>
<mo>-</mo>
<mi>&infin;</mi>
</mrow>
<mi>&infin;</mi>
</msubsup>
<mi>h</mi>
<mrow>
<mo>(</mo>
<mi>t</mi>
<mo>)</mo>
</mrow>
<mi>g</mi>
<mrow>
<mo>(</mo>
<mi>&tau;</mi>
<mo>-</mo>
<mi>t</mi>
<mo>,</mo>
<mi>f</mi>
<mo>)</mo>
</mrow>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<mi>i</mi>
<mn>2</mn>
<mi>&pi;</mi>
<mi>f</mi>
<mi>t</mi>
</mrow>
</msup>
<mi>d</mi>
<mi>t</mi>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>g</mi>
<mrow>
<mo>(</mo>
<mi>&tau;</mi>
<mo>,</mo>
<mi>f</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<mo>|</mo>
<mi>f</mi>
<mo>|</mo>
</mrow>
<msqrt>
<mrow>
<mn>2</mn>
<mi>&pi;</mi>
</mrow>
</msqrt>
</mfrac>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<mrow>
<mo>(</mo>
<msup>
<mi>t</mi>
<mn>2</mn>
</msup>
<mo>/</mo>
<mn>2</mn>
<msup>
<mi>&sigma;</mi>
<mn>2</mn>
</msup>
<mo>)</mo>
</mrow>
</mrow>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, g (τ, f) is Gaussian window, and parameter τ is used for adjusting the diverse location on Gaussian window time shaft, σ=1/ | f |；
The discrete representation of S-transformation is
<mrow>
<mi>h</mi>
<mo>&lsqb;</mo>
<mi>k</mi>
<mi>t</mi>
<mo>&rsqb;</mo>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>N</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>n</mi>
<mo>=</mo>
<mn>0</mn>
</mrow>
<mrow>
<mi>N</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</munderover>
<mo>{</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>0</mn>
</mrow>
<mrow>
<mi>N</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</munderover>
<mi>S</mi>
<mo>&lsqb;</mo>
<mfrac>
<mi>n</mi>
<mrow>
<mi>N</mi>
<mi>T</mi>
</mrow>
</mfrac>
<mo>,</mo>
<mi>j</mi>
<mi>T</mi>
<mo>&rsqb;</mo>
<mo>}</mo>
<msup>
<mi>e</mi>
<mrow>
<mi>i</mi>
<mn>2</mn>
<mi>&pi;</mi>
<mi>n</mi>
<mi>k</mi>
</mrow>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, N represents total sampling number, j and n=0,1 ..., N-1；
A two-dimensional complex number matrix is obtained after discrete S-transformation, i.e., S-transformation time-frequency modular matrix STMM, the matrix are obtained to its modulus
Row vector reflects time domain distribution of the signal under specific frequency, and column vector describes certain time-ofday signals amplitude versus frequency characte.
3. optimization Classification of Power Quality Disturbances device ELM according to claim 2 method, it is characterised in that the ELM tools
Body includes, input layer, hidden layer and output layer, and the input layer has n neuron, corresponding n input variable；The hidden layer
There is l neuron；The output layer has m neuron, corresponds to m output variable, wherein, k-th of neuron of corresponding hidden layer
It is with h-th of interneuronal connection weight a of input layer,
Correspond to i-th of neuron of hidden layer is with k-th of neuron connection weight b of output layer：
The threshold value c of hidden layer neuron is
C=[c1 c2 … cl]T l×1 (6)
Assuming that there is Q sample training collection, if input matrix X and output matrix Y are respectively：
Assuming that the activation primitive of hidden layer neuron is k (x), then network output F is
F=[t1 t2 … tQ]n×Q (9)
<mrow>
<msub>
<mi>t</mi>
<mi>j</mi>
</msub>
<mo>=</mo>
<msub>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mi>t</mi>
<mrow>
<mn>1</mn>
<mi>j</mi>
</mrow>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>t</mi>
<mrow>
<mn>2</mn>
<mi>j</mi>
</mrow>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>t</mi>
<mrow>
<mi>m</mi>
<mi>j</mi>
</mrow>
</msub>
</mtd>
</mtr>
</mtable>
</mfenced>
<mrow>
<mi>m</mi>
<mo>&times;</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>=</mo>
<msub>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>l</mi>
</munderover>
<msub>
<mi>b</mi>
<mrow>
<mi>i</mi>
<mn>1</mn>
</mrow>
</msub>
<mi>k</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>a</mi>
<mi>i</mi>
</msub>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>+</mo>
<msub>
<mi>b</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>l</mi>
</munderover>
<msub>
<mi>b</mi>
<mrow>
<mi>i</mi>
<mn>2</mn>
</mrow>
</msub>
<mi>k</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>a</mi>
<mi>i</mi>
</msub>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>+</mo>
<msub>
<mi>b</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>l</mi>
</munderover>
<msub>
<mi>b</mi>
<mrow>
<mi>i</mi>
<mi>m</mi>
</mrow>
</msub>
<mi>k</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>a</mi>
<mi>i</mi>
</msub>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>+</mo>
<msub>
<mi>b</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mrow>
<mi>m</mi>
<mo>&times;</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>,</mo>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
<mo>,</mo>
<mn>...</mn>
<mo>,</mo>
<mi>Q</mi>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>10</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein ai=[ai1 ai2 … ain],xj=[x1j x2j … xnj]T
Formula (10) can be deformed into
Ha=F'(11)
Wherein, F' is the transposition of matrix F；H represents the hidden layer output matrix of neutral net.
4. optimization Classification of Power Quality Disturbances device ELM according to claim 3 method, it is characterised in that in step 2,
The ELM progress parameter selection is set, then training and test setting parameter, specially to ELM hidden layers input weights and partially
The parameter selection for putting vector is trained and tested.
5. optimization Classification of Power Quality Disturbances device ELM according to claim 4 method, it is characterised in that the step 3
Specifically include, according to ELM electrical energy power quality disturbance combinations of features data, the population at individual in genetic operation is constructed, to chromosome
Encoded, parameter setting then is carried out to the combinations of features of genetic group, obtains optimal feature subset.
6. optimization Classification of Power Quality Disturbances device ELM according to claim 5 method, it is characterised in that in step 3,
The combinations of features to genetic group carries out parameter setting specifically, entering row variation and intersection to the combinations of features of population at individual
Computing.
7. optimization Classification of Power Quality Disturbances device ELM according to claim 1 method, it is characterised in that the electric energy
Quality disturbance signal specifically includes voltage swell signal, voltage dip signal, flickering signal, interrupt signal, harmonic signal and shaken
Swing the compound disturbing signal of signal composition.
8. a kind of storage device, wherein being stored with a plurality of instruction, the instruction is loaded and performed by processor：
Step 1：S-transformation is carried out to Power Quality Disturbance, establishes Power Quality Disturbance model, and extract the quality of power supply
Perturbation features data；
Step 2：Data after being trained and test according to the electrical energy power quality disturbance characteristic in step 1, ELM carry out parameter
Selection is set, then training and test setting parameter；
Step 3：The parameter that step 2 obtains is optimized based on genetic iteration algorithm；
Step 4：ELM after optimizing to step 3 carries out fitness evaluation, judges whether to meet end condition, if it is, obtaining
Optimal feature subset is set, if it is not, then continuing executing with step 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710415018.1A CN107392090A (en) | 2017-06-05 | 2017-06-05 | Optimize Classification of Power Quality Disturbances device ELM method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710415018.1A CN107392090A (en) | 2017-06-05 | 2017-06-05 | Optimize Classification of Power Quality Disturbances device ELM method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107392090A true CN107392090A (en) | 2017-11-24 |
Family
ID=60332942
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710415018.1A Pending CN107392090A (en) | 2017-06-05 | 2017-06-05 | Optimize Classification of Power Quality Disturbances device ELM method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107392090A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109271975A (en) * | 2018-11-19 | 2019-01-25 | 燕山大学 | A kind of electrical energy power quality disturbance recognition methods based on big data multi-feature extraction synergetic classification |
| CN110378286A (en) * | 2019-07-19 | 2019-10-25 | 东北大学 | A kind of Power Quality Disturbance classification method based on DBN-ELM |
| CN110610203A (en) * | 2019-08-30 | 2019-12-24 | 天津大学 | Electric energy quality disturbance classification method based on DWT and extreme learning machine |
| CN113486731A (en) * | 2021-06-17 | 2021-10-08 | 国网山东省电力公司汶上县供电公司 | Abnormal state monitoring method for power transmission equipment |
| CN114358042A (en) * | 2021-11-29 | 2022-04-15 | 国网安徽省电力有限公司马鞍山供电公司 | Power quality signal disturbance classification method based on T-S fuzzy model |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102831433A (en) * | 2012-06-06 | 2012-12-19 | 西南交通大学 | Method for classifying electric energy quality mixing disturbances based on multi-feature quantity of time-frequency domain |
| KR101352204B1 (en) * | 2012-07-05 | 2014-01-16 | 성균관대학교산학협력단 | Apparatus and method for classification of power quality disturbances at power grids |
| CN105447502A (en) * | 2015-11-05 | 2016-03-30 | 国家电网公司 | Transient power disturbance identification method based on S conversion and improved SVM algorithm |
| CN106778923A (en) * | 2017-03-23 | 2017-05-31 | 广东电网有限责任公司珠海供电局 | A kind of Power Quality Disturbance sorting technique and device |
-
2017
- 2017-06-05 CN CN201710415018.1A patent/CN107392090A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102831433A (en) * | 2012-06-06 | 2012-12-19 | 西南交通大学 | Method for classifying electric energy quality mixing disturbances based on multi-feature quantity of time-frequency domain |
| KR101352204B1 (en) * | 2012-07-05 | 2014-01-16 | 성균관대학교산학협력단 | Apparatus and method for classification of power quality disturbances at power grids |
| CN105447502A (en) * | 2015-11-05 | 2016-03-30 | 国家电网公司 | Transient power disturbance identification method based on S conversion and improved SVM algorithm |
| CN106778923A (en) * | 2017-03-23 | 2017-05-31 | 广东电网有限责任公司珠海供电局 | A kind of Power Quality Disturbance sorting technique and device |
Non-Patent Citations (2)
| Title |
|---|
| 于志勇,张卫辉,王新库,黄南天,黄喜旺: "基于GA和ELM的电能质量扰动识别特征选择方法", 《电测与仪表》 * |
| 吕忠，周强，周琨等: "基于遗传算法改进极限学习机的变压器故障诊断", 《高压电器》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109271975A (en) * | 2018-11-19 | 2019-01-25 | 燕山大学 | A kind of electrical energy power quality disturbance recognition methods based on big data multi-feature extraction synergetic classification |
| CN109271975B (en) * | 2018-11-19 | 2020-08-21 | 燕山大学 | Power quality disturbance identification method based on big data multi-feature extraction collaborative classification |
| CN110378286A (en) * | 2019-07-19 | 2019-10-25 | 东北大学 | A kind of Power Quality Disturbance classification method based on DBN-ELM |
| CN110378286B (en) * | 2019-07-19 | 2023-03-28 | 东北大学 | DBN-ELM-based electric energy quality disturbance signal classification method |
| CN110610203A (en) * | 2019-08-30 | 2019-12-24 | 天津大学 | Electric energy quality disturbance classification method based on DWT and extreme learning machine |
| CN113486731A (en) * | 2021-06-17 | 2021-10-08 | 国网山东省电力公司汶上县供电公司 | Abnormal state monitoring method for power transmission equipment |
| CN114358042A (en) * | 2021-11-29 | 2022-04-15 | 国网安徽省电力有限公司马鞍山供电公司 | Power quality signal disturbance classification method based on T-S fuzzy model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107392090A (en) | Optimize Classification of Power Quality Disturbances device ELM method | |
| CN109800875A (en) | Chemical industry fault detection method based on particle group optimizing and noise reduction sparse coding machine | |
| CN102074955B (en) | Method based on knowledge discovery technology for stability assessment and control of electric system | |
| CN101404071B (en) | Electronic circuit fault diagnosis neural network method based on grouping particle swarm algorithm | |
| CN109271975A (en) | A kind of electrical energy power quality disturbance recognition methods based on big data multi-feature extraction synergetic classification | |
| CN109214460B (en) | Power transformer fault diagnosis method based on relative transformation and nuclear entropy component analysis | |
| CN103927550B (en) | A kind of Handwritten Numeral Recognition Method and system | |
| CN103995237A (en) | Satellite power supply system online fault diagnosis method | |
| CN108848068A (en) | Based on deepness belief network-Support Vector data description APT attack detection method | |
| CN104052612B (en) | A kind of Fault Identification of telecommunication service and the method and system of positioning | |
| CN106326984A (en) | User intention identification method and device and automatic answering system | |
| CN106650022A (en) | Method for predicting fault of complex electronic device | |
| CN104155574A (en) | Power distribution network fault classification method based on adaptive neuro-fuzzy inference system | |
| CN103729678A (en) | Navy detection method and system based on improved DBN model | |
| CN106327357A (en) | Load identification method based on improved probabilistic neural network | |
| CN103617147A (en) | Method for identifying mine water-inrush source | |
| CN103489009A (en) | Pattern recognition method based on self-adaptation correction neural network | |
| CN111353153A (en) | GEP-CNN-based power grid malicious data injection detection method | |
| CN105388402A (en) | Cross-linked cable partial discharge mode recognition method based on parameter optimization SVM (Support Vector Machine) algorithm | |
| CN110135167A (en) | A kind of edge calculations terminal security grade appraisal procedure of random forest | |
| CN110689069A (en) | Transformer fault type diagnosis method based on semi-supervised BP network | |
| CN110082082B (en) | GIS state identification method based on vibration signal principal component analysis method | |
| CN113591728A (en) | Electric energy quality disturbance classification method based on integrated deep learning | |
| CN103473556A (en) | Hierarchical support vector machine classifying method based on rejection subspace | |
| CN111062036A (en) | Malicious software identification model construction method, malicious software identification medium and malicious software identification equipment |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171124 |