CN109061354A - A kind of electrical energy power quality disturbance recognition methods based on improvement PSO and SVM - Google Patents
A kind of electrical energy power quality disturbance recognition methods based on improvement PSO and SVM Download PDFInfo
- Publication number
- CN109061354A CN109061354A CN201811048304.XA CN201811048304A CN109061354A CN 109061354 A CN109061354 A CN 109061354A CN 201811048304 A CN201811048304 A CN 201811048304A CN 109061354 A CN109061354 A CN 109061354A
- Authority
- CN
- China
- Prior art keywords
- signal
- particle
- power quality
- svm
- algorithm
- 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
- 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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of based on the electrical energy power quality disturbance recognition methods for improving PSO and SVM, comprising steps of 1) building weights morphological filter, collected voltage and current signal is filtered, reduces interference of the noise to signal, and extract corresponding characteristic value；2) traditional PSO algorithm is improved, SVM parameter is optimized using modified particle swarm optiziation, construct sorter model；3) using the Power Quality Disturbance characteristic signal of extraction as the input of classifier, after the identification of classifier, corresponding disturbing signal classification is exported.The present invention quickly filters out the noise in signal first, by the historical data of training electric energy quality signal, fast and accurately realizes electrical energy power quality disturbance identification classification.
Description
Technical field
The present invention relates to the technical fields of fault diagnosis and Classification and Identification, refer in particular to a kind of based on improvement PSO's and SVM
Electrical energy power quality disturbance recognition methods.
Background technique
With increasing, the disturbance of power quality of nonlinear load in the extensive use and electric system of electronic equipment
Have become a particularly significant problem in electric system.Any event for causing voltage or current deviation can be regarded as electricity
The disturbance of energy quality, these disturbances significantly reduce the stability of electric system.Therefore, to power quality carry out detection be must
It wants, is capable of detecting when whether electric system occurs to disturb and then various power failures are classified.How feature is extracted, so
After identify they become electric energy quality monitoring and analysis in sixty-four dollar question.
According to the relevant criterion of IEEE 1159-2009, electrical energy power quality disturbance is divided into 7 classes substantially.Since power quality is disturbed
Dynamic includes different types, and its classification standard is also sufficiently complex, and traditional classification method is caused mostly to there is certain lack
It falls into, most of method is combined with artificial intelligence by mathematic(al) manipulation to be identified and be classified at present.Common feature
Extracting method includes Fourier transform, short time discrete Fourier transform etc..It is compared to the above, wavelet transformation have temporal frequency,
Multiple dimensioned and multiresolution advantage.But the calculation amount of these methods is too big, and operation time is longer.Common Intelligence Classifier packet
Include artificial neural network, Bayes classifier etc..
The present invention provides a kind of electrical energy power quality disturbance recognition methods based on improvement PSO and SVM, will weight morphological filter
Combine with the SVM classifier after improving PSO optimization, is utilized in morphology principle that operation is simple and processing speed
The characteristics of rapid and good Nonlinear Processing ability, quickly original signal is handled, it is excellent by improved PSO algorithm
After the classifier identification for changing relevant parameter, Power Quality Disturbance is classified in realization.This method effectively increases electric energy
The speed and classification accuracy of quality disturbance signal identification.
Summary of the invention
It is an object of the invention to overcome the shortcomings of to optimize existing method, propose a kind of based on the electricity for improving PSO and SVM
Can quality disturbance recognition methods, break through that recognition accuracy in traditional electrical energy power quality disturbance recognition methods is low, and recognition speed is slow asks
Topic is trained the precision that can effectively improve identification to disturbance failure using historical data, realizes the fast of electrical energy power quality disturbance
Speed accurately identifies.
To achieve the above object, technical solution provided by the present invention are as follows: a kind of based on the electric energy matter for improving PSO and SVM
Disturbance identification method is measured, this method is to filter out signal first by collected voltage and current signal by weighting morphological filter
In interference noise, and then again to voltage and current signal extract individual features value；Then to particle swarm optimization algorithm, that is, PSO algorithm
It improves, the parameter of support vector machines is optimized using improved PSO algorithm, construct SVM classifier model,
By extracting input of the characteristic value of 8 class Power Quality Disturbances as classifier, the class of Power Quality Disturbance is exported
Not, the requirement of identification classification can be met；Itself the following steps are included:
1) building weighting morphological filter, is filtered collected voltage and current signal, reduces noise to letter
Number interference, and extract corresponding characteristic value；
2) traditional PSO algorithm is improved, SVM parameter is optimized using modified particle swarm optiziation, constructed
Sorter model；
3) using the Power Quality Disturbance characteristic signal of extraction as the input of classifier, by the identification of classifier
Afterwards, corresponding disturbing signal classification is exported.
In step 1), building weighting morphological filter filters out the interference noise in voltage and current signal, and extract correspondence
Input feature vector amount, comprising the following steps:
1.1) weighting morphological filter is made of the different combinations of Mathematical Morphology operator, two kinds of bases of mathematical morphology
This morphological operator is expansion and corrosion:
WhereinIndicate expansion,Indicating corrosion, f is initial signal, and g is morphological structuring elements,With
Signal respectively after expansion and erosion operation, DfAnd DgIndicate the domain of f and g, max and min indicate maximum value and
Minimum value；
Opening operation and closed operation are made of the various combination for expanding and corroding respectively:
It wherein zero indicates opening operation, indicates closed operation,Indicate expansion,Indicate corrosion, f is initial signal, and g is shape
State structural element；
The composition for weighting morphological filter is as follows:
Wherein y is the signal after filter process,It indicates opening operation, indicates closed operation, g1And g2It is not respectively
Same structural element, λ1And λ2It is weighting coefficient and meets λ1+λ2=1；
1.2) after by initial signal by weighting morphological filter, using following formula, to treated, signal carries out mathematics
Transformation:
Wherein s (k) is signal to be transformed, and L is the hits in a cycle, f1And f2It is the signal after mathematic(al) manipulation,For Hilbert transform, expression formula isτ is integration variable；
1.3) transformed signal is passed through into following processing, obtains feature vector as subsequent input variable:
Wherein v1,v2,v3,v4It is the feature vector extracted, p is the hits in a cycle, | f (k) | it is the width of f (k)
It is worth absolute value, k is the starting point of sampled value；
Input variable is constructed after above-mentioned transformation
In step 2), traditional PSO algorithm is improved, SVM parameter is carried out using improved PSO algorithm excellent
Change, construct sorter model, concrete condition is as follows:
In particle swarm algorithm, referred to as particle, the speed of t j-th of particle of generation and position each possible solution
It can be updated by following formula:
vj(t)=wvj(t-1)+c1·r1j·(pbest·j-xj(t-1))+c2·r2j(gbest-xj(t-1))
xj(t)=xj(t-1)+vj(t)
Wherein xjIt is the position of j-th of particle, vjIt is the speed of j-th of particle, w is weight coefficient, c1And c2It is to accelerate system
Number, r1jAnd r2jIt is two random numbers between [0,1], pbest·jIndicate the current optimum position particle j, gbestIndicate whole
Current optimum position in a population；
2.1) particle in genetic algorithm is mutated by the shortcomings that being easy precocity for traditional PS O algorithm, falling into local optimum
It is introduced into PSO, TSP question can at random initialize particle, widened the search space of particle, particle is made to jump out part
Optimal location prevents precocity；
2.2) for the ability of searching optimum of balanced algorithm and local search ability, inertia weight w is set and is gradually reduced,
Global search stage reduction speed is fast, and it is slow to reduce speed in the local search stage:
Wherein w (k) is inertia weight, wstartIt (k) is initial inertia coefficient, wendIt (k) is to terminate inertia coeffeicent, TmaxIt is most
Big the number of iterations；
2.3) parameter for initializing above-mentioned improvement PSO algorithm, calculates the fitness value of each particle in population: updating respectively
The maximum adaptation angle value of each particle in entire population and population, the position and speed of each particle, then sentences in Population Regeneration
Whether disconnected termination condition meets, and continues to update if being unsatisfactory for condition；If meeting condition, C is substituted intobestAnd gbestAfter optimization
SVM parameter；
In step 3), according to previous step 1.2) -1.3) the corresponding characteristic quantity of Power Quality Disturbance is obtained, and make
For input variableThen by having trained the Classification and Identification of the classifier of relevant parameter in step 2), final output is corresponding
Power Quality Disturbance classification.
Compared with prior art, the present invention have the following advantages that with the utility model has the advantages that
1, invention introduces weighting morphological filter, processing nonlinear properties energy simple using morphology operations principle
The strong feature of power, realizes accurate quick filter.
2, the present invention realizes the detection identification of electrical energy power quality disturbance fault point, is able to detect be out of order generation and end
Moment.
3, the present invention improves aiming at the problem that conventional particle colony optimization algorithm is easy to precocious, falls into local optimum, will
TSP question introduces wherein, solves the disadvantage that fall into precocious and local optimum.
4, the present invention, will for local search and the unbalanced problem of ability of searching optimum in conventional particle colony optimization algorithm
Weight coefficient improves by invariable to gradually reduce, overall balance overall situation and partial situation's search capability of particle.
5, the method for the present invention has extensive use space in electrical energy power quality disturbance identification classification, and recognition speed is fast, divides
Class accuracy rate is high, there is bright prospects in power system failure diagnostic.
Detailed description of the invention
Fig. 1 is logical flow diagram of the present invention.
Fig. 2 is the normal voltage signal for being weighted morphological filter processing.
Fig. 3 is the characteristic quantity that disturbing signal is extracted.
Fig. 4 is to improve PSO algorithm optimization SVM classifier parameter logistics flow chart.
Fig. 5 is to improve particle fitness value variation diagram in PSO algorithm optimization SVM classifier parametric procedure.
Specific embodiment
Below with reference to specific Power Quality Disturbance, the invention will be further described.
As shown in Figure 1, obtaining Power Quality Disturbance according to IEEE 1159-2009 relevant criterion, the present embodiment is mentioned
The electrical energy power quality disturbance recognition methods based on improvement PSO and SVM supplied, comprising the following steps:
1) building weighting morphological filter, filters out the interference noise in voltage and current signal, and extract corresponding input feature vector
Amount, comprising the following steps:
1.1) weighting morphological filter is made of the different combinations of Mathematical Morphology operator, two kinds of bases of mathematical morphology
This morphological operator is expansion and corrosion:
WhereinIndicate expansion,Indicating corrosion, f is initial signal, and g is morphological structuring elements,With
Signal respectively after expansion and erosion operation, DfAnd DgIndicate the domain of f and g, max and min indicate maximum value and
Minimum value；
Opening operation and closed operation are made of the various combination for expanding and corroding respectively:
WhereinIt indicates opening operation, indicates closed operation,Indicate expansion,Indicate corrosion, f is initial signal, and g is shape
State structural element.
The composition for weighting morphological filter is as follows:
Wherein y is the signal after filter process,It indicates opening operation, indicates closed operation, g1And g2It is not respectively
Same structural element, λ1And λ2It is weighting coefficient and meets λ1+λ2=1.
1.2) after by initial signal by weighting morphological filter, first with following formula, to treated, signal is carried out
Mathematic(al) manipulation:
Wherein s (k) is signal to be transformed, and L is the hits in a cycle, f1And f2It is the signal after mathematic(al) manipulation,For Hilbert transform, expression formula isτ is integration variable.
1.3) by transformed signal pass through following processing, obtain feature vector as subsequent input variable:
Wherein v1,v2,v3,v4It is the feature vector extracted, p is the hits in a cycle, | f (k) | it is the width of f (k)
It is worth absolute value, k is the starting point of sampled value.
Input variable is constructed after above-mentioned transformation
The weighting morphological filter of building is as follows:
WhereinIt indicates opening operation, indicates closed operation, f is signal to be processed, g1For the structural element of sinusoidal shape, g2
For the structural element of smooth-shaped.
It is weighted morphological filter treated that signal is as shown in Figure 2.
Every kind of Power Quality Disturbance type chooses 20 respectively, and characteristic quantity is as shown in Figure 3.
2) traditional PSO algorithm is improved, SVM parameter is optimized using modified particle swarm optiziation, constructed
Sorter model, comprising the following steps:
In particle swarm algorithm, referred to as particle, the speed of t j-th of particle of generation and position each possible solution
It can be updated by following formula:
vj(t)=wvj(t-1)+c1·r1j·(pbest·j-xj(t-1))+c2·r2j(gbest-xj(t-1))
xj(t)=xj(t-1)+vj(t)
Wherein xjIt is the position of j-th of particle, vjIt is the speed of j-th of particle, w is weight coefficient, c1And c2It is to accelerate system
Number, r1jAnd r2jIt is two random numbers between [0,1], pbest·jIndicate the current optimum position particle j, gbestIndicate entire
Current optimum position in population.
2.1) the shortcomings that being easy precocity for traditional PS O algorithm, falling into local optimum, first by the particle in genetic algorithm
Mutation is introduced into PSO.TSP question can at random initialize particle, widened the search space of particle, jumped out particle
Local optimum position, prevents precocity.
2.2) for the ability of searching optimum of balanced algorithm and local search ability, inertia weight w is set and is gradually reduced,
Global search stage reduction speed is fast, and it is slow to reduce speed in the local search stage:
Wherein w (k) is inertia weight, wstartIt (k) is initial inertia coefficient, wendIt (k) is to terminate inertia coeffeicent, TmaxIt is most
Big the number of iterations.
2.3) parameter for initializing above-mentioned improvement PSO algorithm, calculates the fitness value of each particle in population: updating respectively
The maximum adaptation angle value of each particle in entire population and population, the position and speed of each particle in Population Regeneration.Then sentence
Whether disconnected termination condition meets.Continue to update if being unsatisfactory for condition；If meeting condition, C is substituted intobestAnd gbestAfter optimization
SVM parameter.
Using above-mentioned steps, the optimized parameter for obtaining SVM classifier is respectively Cbest=760.175 and gbest=
297.143。
Fig. 4 is to improve PSO algorithm optimization SVM classifier parameter logistics flow chart.
Fig. 5 is to improve particle fitness value variation diagram in PSO algorithm optimization SVM classifier parametric procedure.
3) the corresponding characteristic quantity of Power Quality Disturbance 1) is obtained according to above-mentioned, and as input variableThen it passes through
Cross the Classification and Identification that the classifier of relevant parameter has been trained in 2), the class of the corresponding Power Quality Disturbance of final output
Not.
Final classification results are as shown in the table:
| D0 | D1 | D2 | D3 | D4 | D5 | D6 | D7 | Accuracy rate | |
| D0 | 100 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 100% |
| D1 | 0 | 100 | 0 | 0 | 0 | 0 | 0 | 0 | 100% |
| D2 | 0 | 0 | 97 | 3 | 0 | 0 | 0 | 0 | 97% |
| D3 | 0 | 0 | 0 | 100 | 0 | 0 | 0 | 0 | 100% |
| D4 | 0 | 0 | 0 | 0 | 100 | 0 | 0 | 0 | 100% |
| D5 | 0 | 0 | 0 | 0 | 0 | 100 | 0 | 0 | 100% |
| D6 | 0 | 0 | 1 | 0 | 0 | 2 | 97 | 0 | 97% |
| D7 | 4 | 0 | 0 | 0 | 0 | 0 | 0 | 96 | 96% |
(D0 be it is normal, D1 be temporarily rise, D2 be temporarily drop, D3 be interrupt, D4 be fluctuation, D5 be concussion, D6 is harmonic wave, and D7 is
Cut mark)
In conclusion the present invention provides new method for electrical energy power quality disturbance identification classification after using above scheme,
The speed and accuracy of Power Quality Disturbance Classification and Identification can be effectively improved, there is actual promotional value, be worthy to be popularized.
Embodiment described above is only the preferred embodiments of the invention, and but not intended to limit the scope of the present invention, therefore
Change made by all principles according to the present invention, should all be included within the scope of protection of the present invention.
Claims (2)
1. a kind of based on the electrical energy power quality disturbance recognition methods for improving PSO and SVM, it is characterised in that: this method is will to adopt first
The voltage and current signal collected filters out the interference noise in signal by weighting morphological filter, and then believes again voltage and current
Number extract individual features value；Then particle swarm optimization algorithm, that is, PSO algorithm is improved, utilizes improved PSO algorithm pair
The parameter of support vector machines optimizes, and constructs SVM classifier model, by the spy for extracting 8 class Power Quality Disturbances
Input of the value indicative as classifier, exports the classification of Power Quality Disturbance, can meet the requirement of identification classification；It includes
Following steps:
1) building weighting morphological filter, is filtered collected voltage and current signal, reduces noise to signal
Interference, and extract corresponding characteristic value；
2) traditional PSO algorithm is improved, SVM parameter is optimized using modified particle swarm optiziation, building classification
Device model；
3) defeated after the identification of classifier using the Power Quality Disturbance characteristic signal of extraction as the input of classifier
Corresponding disturbing signal classification out.
2. according to claim 1 a kind of based on the electrical energy power quality disturbance recognition methods for improving PSO and SVM, feature exists
In:
In step 1), building weighting morphological filter filters out the interference noise in voltage and current signal, and extract corresponding input
Characteristic quantity, comprising the following steps:
1.1) weighting morphological filter is made of the different combinations of Mathematical Morphology operator, two kinds of fundamental forms of mathematical morphology
State operator is expansion and corrosion:
WhereinIndicate expansion,Indicating corrosion, f is initial signal, and g is morphological structuring elements,WithRespectively
For the signal after expansion and erosion operation, DfAnd DgIndicate the domain of f and g, max and min indicate maximum value and minimum
Value；
Opening operation and closed operation are made of the various combination for expanding and corroding respectively:
WhereinIt indicates opening operation, indicates closed operation,Indicate expansion,Indicate corrosion, f is initial signal, and g is morphology
Structural element；
The composition for weighting morphological filter is as follows:
Wherein y is the signal after filter process,It indicates opening operation, indicates closed operation, g1And g2It is different respectively
Structural element, λ1And λ2It is weighting coefficient and meets λ1+λ2=1；
1.2) after by initial signal by weighting morphological filter, using following formula, to treated, signal carries out mathematics change
It changes:
Wherein s (k) is signal to be transformed, and L is the hits in a cycle, f1And f2It is the signal after mathematic(al) manipulation,For
Hilbert transform, expression formula areτ is integration variable；
1.3) transformed signal is passed through into following processing, obtains feature vector as subsequent input variable:
Wherein v1,v2,v3,v4It is the feature vector extracted, p is the hits in a cycle, | f (k) | it is exhausted for the amplitude of f (k)
To value, k is the starting point of sampled value；
Input variable is constructed after above-mentioned transformation
In step 2), traditional PSO algorithm is improved, SVM parameter is optimized using improved PSO algorithm, structure
Sorter model is built, concrete condition is as follows:
In particle swarm algorithm, referred to as particle each possible solution, the speed of t j-th of particle of generation and position can
It is updated by following formula:
vj(t)=wvj(t-1)+c1·r1j·(pbest·j-xj(t-1))+c2·r2j(gbest-xj(t-1))
xj(t)=xj(t-1)+vj(t)
Wherein xjIt is the position of j-th of particle, vjIt is the speed of j-th of particle, w is weight coefficient, c1And c2It is accelerator coefficient,
r1jAnd r2jIt is two random numbers between [0,1], pbest·jIndicate the current optimum position particle j, gbestIndicate entire kind
Current optimum position in group；
2.1) particle in genetic algorithm is mutated and introduces by the shortcomings that being easy precocity for traditional PS O algorithm, falling into local optimum
In PSO, TSP question can at random be initialized particle, widened the search space of particle, particle is made to jump out local optimum
Position prevents precocity；
2.2) for the ability of searching optimum of balanced algorithm and local search ability, inertia weight w is set and is gradually reduced, in the overall situation
Search phase reduction speed is fast, and it is slow to reduce speed in the local search stage:
Wherein w (k) is inertia weight, wstartIt (k) is initial inertia coefficient, wendIt (k) is to terminate inertia coeffeicent, TmaxIt is that maximum changes
Generation number；
2.3) parameter for initializing above-mentioned improvement PSO algorithm, calculates the fitness value of each particle in population: updating respectively entire
The maximum adaptation angle value of each particle in population and population, the position and speed of each particle in Population Regeneration, then judgement is eventually
Only whether condition meets, and continues to update if being unsatisfactory for condition；If meeting condition, C is substituted intobestAnd gbestAs the SVM after optimization
Parameter；
In step 3), according to previous step 1.2) -1.3) the corresponding characteristic quantity of Power Quality Disturbance is obtained, and as defeated
Enter variableThen by having trained the Classification and Identification of the classifier of relevant parameter, the corresponding electricity of final output in step 2)
The classification of energy quality disturbance signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811048304.XA CN109061354A (en) | 2018-09-10 | 2018-09-10 | A kind of electrical energy power quality disturbance recognition methods based on improvement PSO and SVM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811048304.XA CN109061354A (en) | 2018-09-10 | 2018-09-10 | A kind of electrical energy power quality disturbance recognition methods based on improvement PSO and SVM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109061354A true CN109061354A (en) | 2018-12-21 |
Family
ID=64760929
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811048304.XA Pending CN109061354A (en) | 2018-09-10 | 2018-09-10 | A kind of electrical energy power quality disturbance recognition methods based on improvement PSO and SVM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109061354A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109765420A (en) * | 2019-01-25 | 2019-05-17 | 东北电力大学 | Rapid detection method temporarily drops in a kind of single-phase voltage converted based on Morphologic filters and method of derivation α β-dq |
| CN110648088A (en) * | 2019-11-26 | 2020-01-03 | 国网江西省电力有限公司电力科学研究院 | Electric energy quality disturbance source judgment method based on bird swarm algorithm and SVM |
| CN110780655A (en) * | 2019-07-01 | 2020-02-11 | 烟台宏远氧业股份有限公司 | Remote fault diagnosis and operation and maintenance method and system for hyperbaric oxygen chamber based on Internet of things |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102944773A (en) * | 2012-10-29 | 2013-02-27 | 华南理工大学 | Method for detecting and classifying power disturbances based on space conversion |
| CN103901255A (en) * | 2014-04-22 | 2014-07-02 | 湖南工业大学 | Hybrid power filter harmonic current detection method based on generalized weighted morphological filtering theory |
| CN107123987A (en) * | 2017-05-11 | 2017-09-01 | 华东交通大学 | Electrical energy power quality disturbance recognition methods based on on-line training weighed SVM |
-
2018
- 2018-09-10 CN CN201811048304.XA patent/CN109061354A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102944773A (en) * | 2012-10-29 | 2013-02-27 | 华南理工大学 | Method for detecting and classifying power disturbances based on space conversion |
| CN103901255A (en) * | 2014-04-22 | 2014-07-02 | 湖南工业大学 | Hybrid power filter harmonic current detection method based on generalized weighted morphological filtering theory |
| CN107123987A (en) * | 2017-05-11 | 2017-09-01 | 华东交通大学 | Electrical energy power quality disturbance recognition methods based on on-line training weighed SVM |
Non-Patent Citations (2)
| Title |
|---|
| ZHEN LEI: "Power Quality Disturbance Identification Using Improved Particle Swarm Optimizerand Support Vector Machine", 《2018 IEEE INNOVATIVE SMART GRID TECHNOLOGIES - ASIA (ISGT ASIA)》 * |
| 李梦诗: "基于形态小波范数熵和支持向量机的电能质量分类研究", 《电机与控制学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109765420A (en) * | 2019-01-25 | 2019-05-17 | 东北电力大学 | Rapid detection method temporarily drops in a kind of single-phase voltage converted based on Morphologic filters and method of derivation α β-dq |
| CN110780655A (en) * | 2019-07-01 | 2020-02-11 | 烟台宏远氧业股份有限公司 | Remote fault diagnosis and operation and maintenance method and system for hyperbaric oxygen chamber based on Internet of things |
| CN110648088A (en) * | 2019-11-26 | 2020-01-03 | 国网江西省电力有限公司电力科学研究院 | Electric energy quality disturbance source judgment method based on bird swarm algorithm and SVM |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Detection of power grid disturbances and cyber-attacks based on machine learning | |
| CN110082640B (en) | Distribution network single-phase earth fault identification method based on long-time memory network | |
| WO2022067562A1 (en) | Method and device for diagnosing fault arc, and computer-readable storage medium | |
| CN103177188B (en) | The power system load dynamic characteristic sorting technique that a kind of feature based maps | |
| CN109061354A (en) | A kind of electrical energy power quality disturbance recognition methods based on improvement PSO and SVM | |
| CN112200694B (en) | Dominant instability mode identification model construction and application method based on graph neural network | |
| CN111191740A (en) | Fault diagnosis method for rolling bearing | |
| CN106203531A (en) | A kind of transmission line fault sorting technique based on the sparse autocoder of convolution | |
| CN116361718A (en) | Single-phase earth fault line selection method based on improved variational modal decomposition and dispersion entropy | |
| CN107579846A (en) | A kind of cloud computing fault data detection method and system | |
| CN109753762A (en) | Based on the modified power distribution network two stages network topology identification method of classification and device | |
| CN110866366A (en) | XGboost algorithm-based island detection method for photovoltaic microgrid containing PHEV | |
| CN112821559A (en) | Non-invasive household appliance load depth re-identification method | |
| CN105406461A (en) | Adaptive dynamic load monitoring method for power distribution network power failure events | |
| CN113836817A (en) | Axial flow compressor rotating stall prediction method based on stacked long-short term memory network | |
| CN113447759A (en) | Multi-classification RVM power grid fault discrimination method and system | |
| CN115879048A (en) | Series arc fault identification method and system based on WRFMDA model | |
| CN115345255A (en) | Fault diagnosis method, control device, terminal and storage medium | |
| CN115859702A (en) | Permanent magnet synchronous wind driven generator demagnetization fault diagnosis method and system based on convolutional neural network | |
| CN114386024A (en) | Power intranet terminal equipment abnormal attack detection method based on ensemble learning | |
| CN116400168A (en) | Power grid fault diagnosis method and system based on depth feature clustering | |
| CN114548555B (en) | Axial flow compressor stall surge prediction method based on deep autoregressive network | |
| CN113610148B (en) | Fault diagnosis method based on bias weighted AdaBoost | |
| Samantaray et al. | Decision tree based fault detection and classification in distance relaying | |
| CN116227320A (en) | Double-fed fan control parameter identification method based on LSTM-IPSO |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20181221 |
|
| WD01 | Invention patent application deemed withdrawn after publication |