CN107482683A - A kind of wind farm group transient voltage clustering recognition method based on principal component analysis - Google Patents
A kind of wind farm group transient voltage clustering recognition method based on principal component analysis Download PDFInfo
- Publication number
- CN107482683A CN107482683A CN201710834264.0A CN201710834264A CN107482683A CN 107482683 A CN107482683 A CN 107482683A CN 201710834264 A CN201710834264 A CN 201710834264A CN 107482683 A CN107482683 A CN 107482683A
- Authority
- CN
- China
- Prior art keywords
- voltage
- wind
- principal component
- wind farm
- power plant
- 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.)
- Granted
Links
- 230000001052 transient effect Effects 0.000 title claims abstract description 64
- 238000000513 principal component analysis Methods 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000000605 extraction Methods 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 29
- 230000005684 electric field Effects 0.000 claims description 10
- 239000013598 vector Substances 0.000 claims description 8
- 230000009467 reduction Effects 0.000 claims description 6
- 230000017105 transposition Effects 0.000 claims description 6
- 238000003064 k means clustering Methods 0.000 claims description 5
- NRZWYNLTFLDQQX-UHFFFAOYSA-N p-tert-Amylphenol Chemical compound CCC(C)(C)C1=CC=C(O)C=C1 NRZWYNLTFLDQQX-UHFFFAOYSA-N 0.000 claims description 4
- 230000014509 gene expression Effects 0.000 claims 3
- 230000005611 electricity Effects 0.000 abstract description 11
- 238000004458 analytical method Methods 0.000 abstract description 9
- 230000000694 effects Effects 0.000 abstract description 2
- 230000010354 integration Effects 0.000 abstract description 2
- 238000005259 measurement Methods 0.000 abstract description 2
- 230000004888 barrier function Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 7
- 230000008859 change Effects 0.000 description 6
- 238000004088 simulation Methods 0.000 description 4
- 241000233855 Orchidaceae Species 0.000 description 2
- 235000008331 Pinus X rigitaeda Nutrition 0.000 description 2
- 235000011613 Pinus brutia Nutrition 0.000 description 2
- 241000018646 Pinus brutia Species 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 206010003591 Ataxia Diseases 0.000 description 1
- 206010010947 Coordination abnormal Diseases 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- 241000283086 Equidae Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 208000016290 incoordination Diseases 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 238000012847 principal component analysis method Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- H02J3/386—
-
- 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
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/70—Wind energy
- Y02E10/76—Power conversion electric or electronic aspects
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
A kind of wind farm group transient voltage clustering recognition method based on principal component analysis, the invention belongs to the transient safe and stable after wind power integration system and control technology field, more particularly to a kind of wind farm group transient voltage recognition methods based on principal component analysis.During electric network fault, only need each wind farm grid-connected measurement voltage data in system, the garbage of redundancy in voltage time sequence is rejected by principal component analysis, the only maximum the first two principal component of extraction contribution rate, projected to two dimensional surface, pass through the close and distant relation of each wind power plant of k means focusing solutions analysis again, on-line dynamic analysis wind power plant divides the influence factor of group, the quick identification of the wind farm group of transient behavior strong correlation can be achieved, analysis wind-powered electricity generation is contributed to access regional systems failure collection on a large scale, plan the configuration capacity of each Reactive Compensation in Wind Farm device, and the unified management between wind farm group controls with coordinating.With the advantages of scientific and reasonable, strong applicability, reliability is high, and effect is good.
Description
Technical field
The invention belongs to the transient safe and stable after wind power integration system and control technology field, more particularly to one kind to be based on
The wind farm group transient voltage recognition methods of principal component analysis.
Background technology
Influenceed by China's natural wind resource with power load distributing feature, adjoint voltage peace after large-scale wind power cluster is grid-connected
Full problem shows that it is substantially the strong Weak link sending of voltage sensitivity.With the large-scale wind electricity base of ten million kilowatt of level capacity
Ground is progressively built up, and the wind-powered electricity generation off-grid accident under high wind-powered electricity generation permeability environment happens occasionally, and causes wind power resources utilization rate to be limited
System, therefore wind power plant dynamic stability has become researcher and pays close attention to problem.Unbalanced grid faults generally trigger wind
Voltage of electric field falls, and blower fan is prevents overcurrent damage internal electronic device and off-grid., will when off-grid blower fan reaches certain scale
Have a strong impact on the stability of a system.Therefore, blower fan low voltage crossing (low voltage ride through, LVRT) energy is improved
Power, reformation of blower performance can effectively alleviate blower fan off-grid situation, reduce the generation of extensive chain off-grid accident.
The high voltage pulse that transient voltage refers to quickly jump, which is added on supply voltage, causes voltage to raise, and it is a kind of
Decay slower overvoltage in duration ranges.Transient overvoltage is due to breaker operator or short trouble occurs, and makes
The overvoltage occurred in the case of reaching certain temporary stabilization again after power system experience transient process, also known as power frequency electric
Pressure rise.
Have research and propose using frequency converter transient voltage control and award setting improve double-fed wind field transient voltage it is steady
It is qualitative to ensure the realization of Wind turbines LVRT functions.Above method is required to carry out raising to blower fan low voltage ride-through capability to change
Make, in terms of the current research about wind power plant LVRT lays particular emphasis on its principle and control technology, carried out in units of single game more
Independent control slowly shows the features such as its incoordination and complexity, it is impossible to takes into account the pressure regulation demand of global system.Therefore and
Still needed further research on the problem of influencing each other between the wind power plant cluster of access system.
PCA (principal component analysis, PCA) is also referred to as principal component analysis, it is intended to utilizes
The thought of dimensionality reduction, multi objective is converted into a few overall target (i.e. principal component), wherein each principal component can reflect
The most information of original variable, and information contained does not repeat mutually.This method is while many-sided variable is introduced by complexity
Factor is attributed to several principal components, simplifys a problem, while the more scientific and effective data message of obtained result.It is this to incite somebody to action
The wind field group that the ability of problem reduction is just being suitably for complexity carries out point group so as to identify off-grid wind-powered electricity generation by whether transient behavior is similar
.But the research of this respect is still in blank stage.Therefore being needed badly among prior art wants a kind of new technical scheme
Solves this problem.
The content of the invention
The technical problems to be solved by the invention are：A kind of wind farm group transient voltage based on principal component analysis is provided to gather
Class recognition methods, the transient voltage degree of susceptibility difference to be happened suddenly according to wind farm group grid entry point is come by the similar wind of transient behavior
Electric field group carries out a point group, so as to accurate judgement off-grid wind farm group.
The object of the present invention is achieved like this：A kind of wind farm group transient voltage clustering recognition based on principal component analysis
Method, it is characterised in that comprise the following steps：
Step 1: control room, which gathers each wind-powered electricity generation field output voltage Data Concurrent, gives electrical power services device, electrical power services device
The magnitude of voltage received and default standard voltage value are compared；
Step 2: the magnitude of voltage that electrical power services device recognizes reception forces down 10% than standard electric, whole wind field grid entry points are recorded
Voltage value signal S (n)=[s1(n),s2(n),...,sm(n)]T, n=1,2 ..., N,
The voltage signal quantity and N >=2, T representing matrixs transposition, S that wherein m is the quantity of wind power plant, n is each wind field
(n) matrix, the s being made up of magnitude of voltagem(n) n-th of voltage value signal of m-th of wind park is represented；
Step 3: it is U=[u to obtain m wind power plant voltage signal average1,u2,...,um]T, obtain each wind field voltage
Value signal removes result S (n)-U=X that equalization is handled,
Wherein X is the voltage value matrix after equalizationI.e., U is that wind power plant voltage signal is equal
Value, T representing matrix transposition；
Step 4: obtain X covariance matrix XXT=A, passes through PTAP=diag [λ1,λ2,...,λm], obtain orthogonal moment
Battle array P, wherein λ1≥λ2≥...≥λm>=0 be matrix A m characteristic value, PTFor P transposed matrix, wherein diag represents one
Diagonal matrix, λ are referred to as A characteristic value；
Step 5: pass through PTX, the data Y after dimensionality reduction is obtained, and have λi=E (yi 2), X=PY=p1y1+p2y2+...+
pmym, wherein p1,p2,...,pmFor A characteristic vector, i.e. principal component load, represent related to corresponding wind farm grid-connected voltage
Coefficient, coefficient correlation corresponding to first, second principal component is down to 2 dimensions as horizontal, ordinate, so far N-dimensional voltage signal before extraction,
Wherein E represents to seek formula of variance；
Step 6: each magnitude of voltage under the two-dimensional coordinate in the step 5 is believed by k-means (k averages) clustering algorithms
Number coefficient correlation be divided into k classes, i.e., m wind farm group is divided into k group；
Step 7: calculate and obtain m point in two-dimensional space to the central point of k group in the step 6 Euclidean away from
From m point is incorporated into closest classification respectively；
Step 8: according to cluster result, the arithmetic average of all elements Euclidean distance in each group is recalculated simultaneously
Obtain k center；
Step 9：Repeating said steps seven are constant up to cluster result, draw the similar wind farm group of transient behavior, individually
Electric field in groups is off-grid electric field.
It is described Step 6: by k-means clustering algorithms by each voltage signal under the two-dimensional coordinate in the step 5
Coefficient correlation is divided into 3 classes, i.e., m wind power plant is divided into 3 groups.
It is described Step 6: each voltage signal under the two-dimensional coordinate in the step 5 is pressed by k-means clustering algorithms
Coefficient correlation is divided into 4 classes, i.e., m wind power plant is divided into 4 groups.
The wind farm group transient voltage clustering recognition method based on principal component analysis is realized using above step.
The present invention can bring following beneficial effect：System is by after failure, it is only necessary to each wind farm grid-connected in system
Point measurement voltage data, the garbage of redundancy in voltage time sequence is rejected by principal component analysis, only extracts contribution rate
Maximum the first two principal component, is projected to two dimensional surface, then pass through each wind power plant of k-means focusing solutions analysis
Close and distant relation, on-line dynamic analysis wind power plant divide the influence factor of group, can be achieved transient behavior strong correlation wind farm group it is fast
Speed identification, contribute to analysis wind-powered electricity generation to access regional systems failure collection on a large scale, plan each Reactive Compensation in Wind Farm device
Unified management between configuration capacity, and wind farm group controls with coordinating.With scientific and reasonable, strong applicability, reliability is high,
The advantages of effect is good.
Brief description of the drawings
Below in conjunction with the drawings and specific embodiments, the present invention is further illustrated：
Fig. 1 is a kind of flow chart of the wind farm group transient voltage clustering recognition method based on principal component analysis of the present invention.
Fig. 2 is that the two dimension in a kind of wind farm group transient voltage clustering recognition method based on principal component analysis of the present invention is main
Constituent analysis explanation figure.
Fig. 3 is the Liao Dynasty in a kind of wind farm group transient voltage clustering recognition embodiment of the method based on principal component analysis of the present invention
The geographical wiring diagram of peaceful western Fuxin power network.
Fig. 4 is a kind of wind farm group transient voltage clustering recognition method first time failure based on principal component analysis of the present invention
The curve map one of lower wind power plant transient voltage.
Fig. 5 is a kind of wind farm group transient voltage clustering recognition method first time failure based on principal component analysis of the present invention
The curve map two of lower wind power plant transient voltage.
Fig. 6 is a kind of wind farm group transient voltage clustering recognition method first time failure based on principal component analysis of the present invention
The curve map three of lower wind power plant transient voltage.
Fig. 7 is a kind of wind farm group transient voltage clustering recognition method first time failure based on principal component analysis of the present invention
The curve map four of lower wind power plant transient voltage.
Fig. 8 is a kind of wind farm group transient voltage clustering recognition method first time failure based on principal component analysis of the present invention
When PCA analysis results schematic diagram.
Fig. 9 is a kind of wind farm group transient voltage clustering recognition second of failure of method based on principal component analysis of the present invention
When the secondary curve map one fallen of wind power plant transient voltage.
Figure 10 is a kind of second of the event of wind farm group transient voltage clustering recognition method based on principal component analysis of the present invention
The wind power plant transient voltage secondary curve map two fallen during barrier.
Figure 11 is a kind of second of the event of wind farm group transient voltage clustering recognition method based on principal component analysis of the present invention
The wind power plant transient voltage secondary curve map three fallen during barrier.
Figure 12 is a kind of second of the event of wind farm group transient voltage clustering recognition method based on principal component analysis of the present invention
The wind power plant transient voltage secondary curve map four fallen during barrier.
Figure 13 is a kind of second of the event of wind farm group transient voltage clustering recognition method based on principal component analysis of the present invention
The secondary schematic diagram for falling PCA analysis results during barrier.
Figure 14 is a kind of wind farm group transient voltage clustering recognition method third time event based on principal component analysis of the present invention
The curve map one that wind power plant transient voltage falls three times during barrier.
Figure 15 is a kind of wind farm group transient voltage clustering recognition method third time event based on principal component analysis of the present invention
The curve map two that wind power plant transient voltage falls three times during barrier.
Figure 16 is a kind of wind farm group transient voltage clustering recognition method third time event based on principal component analysis of the present invention
The curve map three that wind power plant transient voltage falls three times during barrier.
Figure 17 is a kind of wind farm group transient voltage clustering recognition method third time event based on principal component analysis of the present invention
The curve map four that wind power plant transient voltage falls three times during barrier.
Figure 18 is a kind of wind farm group transient voltage clustering recognition method third time event based on principal component analysis of the present invention
The curve map five that wind power plant transient voltage falls three times during barrier.
Figure 19 is a kind of wind farm group transient voltage clustering recognition method third time event based on principal component analysis of the present invention
The curve map six that wind power plant transient voltage falls three times during barrier.
Figure 20 is a kind of wind farm group transient voltage clustering recognition method third time event based on principal component analysis of the present invention
Fall the schematic diagram of PCA analysis results during barrier three times.
In figure, 1- Long Wan, 2- Fu Xi, 3- Wang Ying, 4- peaces ditch, 5- prick blue mountain, 6- pools mill, 7- post horses pond, 8- Fu Bei,
9- Fuxins.
Embodiment
As shown in figure 1, a kind of wind farm group transient voltage clustering recognition method based on principal component analysis, its feature exist
In it comprises the following steps：Step 1: control room, which gathers each wind-powered electricity generation field output voltage Data Concurrent, gives electrical power services device,
The magnitude of voltage received and default standard voltage value are compared by electrical power services device；
Step 2: the magnitude of voltage that electrical power services device recognizes reception forces down 10% than standard electric, start recording whole wind field
The voltage value signal S (n) of grid entry point=[s1(n),s2(n),...,sm(n)]T, n=1,2 ..., N,
M is wind power plant number, and n is number of signals, s1(n),s2(n),...,sm(n) it is each wind farm grid-connected voltage
Be worth column vector, be m row n column matrix S (n) after transposition, i.e., a corresponding wind power plant per data line.
Step 3: it is U=[u to obtain m wind power plant voltage signal average1,u2,...,um]T, obtain each wind field voltage
Value signal removes result S (n)-U=X that equalization is handled,
Wherein X is the voltage value matrix after equalizationThat is U is that wind power plant voltage signal is equal
Value, T representing matrix transposition；
u1,u2,...,umFor corresponding s1(n),s2(n),...,sm(n) average, [u after transposition1,u2,...,um]TIt is every
The corresponding wind power plant of data line, therefore X (n)=S (n)-U goes equalization to handle for S (n).
Step 4: obtain X covariance matrix XXT=A, passes through PTAP=diag [λ1,λ2,...,λm], obtain orthogonal moment
Battle array P, wherein λ1≥λ2≥...≥λm>=0 be matrix A m characteristic value, PTFor P transposed matrix, wherein diag represents one
Diagonal matrix, λ are referred to as A characteristic value；
The theoretical foundation explanation of this case：1. covariance property theorem：Covariance is that both timing explanations are positive correlations,
For covariance to be negative correlativing relation when bearing, covariance is that both 0 interval scales can mathematically use two vectorial associations independently of each other
Variance represents its correlation, due to going average to allow m vectorial averages to be 0, when covariance is 0, represents that two vectors are complete
Full independence, i.e. irrelevance.When sample is multidimensional data, their covariance is actually covariance matrix；2. characteristic value
Basic definition theorem：If AX=λ X, the A characteristic values that λ is referred to as, X is corresponding characteristic vector, and matrix A acts on its feature
On vectorial X, only make it that X length is changed, scaling is exactly corresponding eigenvalue λ.When A is n rank invertible matrix
When, A and PTAP is similar, and similar matrix has identical characteristic value.Especially, when A is symmetrical matrix, A singular value is equal to A
Characteristic value, orthogonal matrix P, P be presentTAP be A eigenvalue cluster into diagonal matrix, to A carry out singular value decomposition with regard to that can obtain
All characteristic values and P matrixes.
Step 5: pass through PTX, the data Y after dimensionality reduction is obtained, and have λi=E (yi 2), X=PY=p1y1+p2y2+...+
pmym, wherein p1,p2,...,pmFor A characteristic vector, i.e. principal component load, represent related to corresponding wind farm grid-connected voltage
Coefficient, coefficient correlation corresponding to first, second principal component is down to 2 dimensions as horizontal, ordinate, so far N-dimensional voltage signal before extraction,
Wherein E represents to seek formula of variance；
The theoretical foundation explanation of this case：1. p in step 51,p2,...,pmIt is actual special corresponding to all characteristic values of P matrixes
Sign vector, characteristic value are descending arrangements, the first two characteristic value and having been over all characteristic value sums exhausted big
Part.We take characteristic vector corresponding to the first two characteristic value, and wind power plant voltage signal so is reduced into 2 dimension spaces from n dimensions；
2. processing generally mathematically is exactly that original P index is done into linear combination, as new overall target.But this linear group
Conjunction can have a lot if without restriction.If it is that first overall target is designated as F1 by first linear combination, it would be desirable that
F1 reflects the information of original index as much as possible, and method here is exactly to be expressed with F1 variance, i.e. VAR (F1) is bigger,
Represent that the information that F1 is included is more.Therefore, F1 selected in all linear combination should be variance maximum, therefore claim F1
For first principal component.
When first principal component is not enough to represent the information of original P index, considers further that and choose F2, i.e. Second principal component, F1
Existing information need not be appeared in F2 again；
Step 6: each magnitude of voltage under the two-dimensional coordinate in the step 5 is believed by k-means (k averages) clustering algorithms
Number coefficient correlation be divided into k classes, i.e., m wind farm group is divided into k group；
The theoretical foundation k-mean of this case algorithm flow explanation arbitrarily selects k object to make from m data object first
For initial cluster center；And for remaining other objects, then according to their similarities (distance) with these cluster centres, point
(cluster centre representated by) cluster most like with it is not assigned these to；Then calculate again each obtain newly cluster it is poly-
Class center (averages of all objects in the cluster)；This process is constantly repeated untill canonical measure function starts convergence.
Typically all had the characteristics that using mean square deviation as .k cluster of canonical measure function：Each cluster is compact as far as possible in itself,
It is and separated as far as possible between respectively clustering.
Step 7: calculate and obtain m point in two-dimensional space to the central point of k group in the step 6 Euclidean away from
From m point is incorporated into closest classification respectively；
Step 8: according to cluster result, the arithmetic average of all elements Euclidean distance in each group is recalculated simultaneously
Obtain k center；
Step 9：Repeating said steps seven are constant up to cluster result, draw the similar wind farm group of transient behavior, individually
Electric field in groups is off-grid electric field.
Fig. 2 is two-dimensional principal component analysis explanation figure, is described as follows.
The theoretical foundation principal component analysis of this case is also referred to as principal component analysis, it is intended to using the thought of dimensionality reduction, multi objective is turned
A small number of overall targets (i.e. principal component) are turned to, wherein each principal component can reflect the most information of original variable, and institute
Do not repeated mutually containing information.When analyzing higher-dimension multivariate sample data, it is necessary to which the key issue solved is the data to observation inconvenience
Carry out simplifying processing.Such as have p member conceptual datas, artificially therefrom extract n sample unit and analyzed, that is, it is empty to obtain p dimensions
Between n point, p × n initial data altogether.If every unitary is all incoherent in this p member totality, can be problem equivalent
Analyzed into p single index data.But possess certain correlation unavoidably between this p stochastic variable under normal circumstances, and
And in higher dimensional space visual representation, the relation between p × n raw data points is difficult to directly observe, if but can be p dimension spaces
Interior point is mapped to two dimension or three-dimensional compared with being showed in lower dimensional space, and retains the prevailing relationship between Data In High-dimensional Spaces
Can high degree Simplified analysis.
As shown in Fig. 2 n sample point presses its X1、X2Two observational variables are spread in ellipse：
According to rotation transformation formula：
U is orthogonal matrix, that is, is had：
U '=U-1, U ' U=E (3)
Conversion causes n sample point in Z1Dispersion degree is maximum on direction of principal axis, i.e. Z1Variance it is maximum, variable Z1Represent
Most information of initial data, claim Z1For first principal component.Therefore, can be by Z1Analysis learn the spy of n sample point
Different information.
Embodiment：Fig. 3 is the geographical wiring diagram of western Liaoning Province Fuxin power network, and region 220kV transmission systems access 9 altogether
Wind power plant, total installation of generating capacity about 950MW, the wind farm group of certain scale is preliminarily formed, and concentrated and be incorporated to the peaceful change in 500kV north.
In addition, the region also includes 3 thermal power plants and steam power plant, balance nodes are located at Eastern Liaoning, China Pu He changes.
During wind power plant normal operation, experimental simulation first time electric network fault, failure solution after duration 0.625s, 0.2s
Remove, the transient voltage of 9 wind power plants is analyzed, Fig. 4 is each wind power plant transient voltage track, and Fig. 5 is PCA analysis knots
Fruit, grouping result are as follows：
Fig. 4 is understood, after failure occurs, drop in various degree occurs for each wind-powered electricity generation field voltage, wherein distance fault position compared with
Near bundle orchid mountain breeze field, peaceful ditch wind field voltage falling situation are the most serious, fall to below 0.4pu, remaining wind field position because residing for
It is mutually variant to put different voltage falling situations.Numbering 1-10 respectively correspondingly manages Long Wan, Fu Xi, Wang Ying in wiring diagram in Fig. 5
Sub, peaceful ditch, Zha Lanshan, Tang Fang, post horse pond, Fu Bei and Fuxin wind power plant, by comparing Fig. 3 and Fig. 5, No. 1 Long Wan and 6
Number pool mill wind power plant No. 3 sub- wind fields of Wang Ying with positioned at the southwest of Fuxin power network, has pole with the change of 220kV Black Hills is connected to
Independent to form a group for similar same tone, No. 7 post horse ponds, with 220kV A Jin changes are connected to, also possess with No. 9 Fuxin wind power plants
Obvious same tone.Farther out, drop lesser extent, the relatively serious wind power plant of voltage falling for above wind power plant distance fault position
The Dong Liang being in close to faulty line is concentrated to become and soughing of the wind in the pines change area, bundle orchid mountain breeze field the most serious, peaceful ditch wind field difference
Corresponding 4, No. 5 positions, become with the 220kV Dong Liang in faulty line one end are connected to, No. 2 mound No. 8 Fu Beifengchang in west are tight because being connected to
The 220kV soughings of the wind in the pines that adjacent Dong Liang becomes become, and slightly above blue mountain breeze field, peaceful ditch wind field are pricked after voltage falling.As shown in Figure 5, above-mentioned 4
Wind power plant is because having similar same tone close to abort situation.Thus illustrate, system grid structure is to influence wind power plant by failure
One of principal element of behavioral characteristics after disturbance.Therefore under present case, it may be considered that according to the grouping result of table 1 to each wind
Electric field coordinates control, according to air-blower control strategy and dynamic reactive compensation device capacity etc., reasonable distribution (1,6,3) (7,9) (2,
8,5,4) in group between reactive power compensation amount and group without the distribution of work, and then realize unified management.
Second of electric network fault of experimental simulation, voltage decline, after 0.4s occurs for failure, the voltage that detects further under
Drop, compares figure 5, Fig. 7 understand, each wind-powered electricity generation field voltage is secondary fall after, PCA algorithms are projected to the simulation number of two-dimensional space
The position at strong point changes, and illustrates that the disturbed track similitude of each field grid entry point voltage accordingly changes.It can be seen from table 2, first
The southwest that secondary failure Central Plains is located at Fuxin power network has the 1 of same tone, 3, No. 6 wind field, because of No. 3 wind in second of failure
Electric field off-the-line off-grid, voltage falling degree is maximum, and same tone is no longer kept with 1, No. 6 wind field, and PCA algorithms can effectively know it
Do not come out, it was demonstrated that this paper institutes extracting methods has the ability that off-grid identifies.Such as following table：
Experimental simulation third time electric network fault, the transient voltage track of 9 wind power plants is as shown in figure 8, Fig. 9 analyzes for PCA
As a result, time 46ms is calculated.As shown in Figure 8, after fault degree aggravates, No. 3 sub- wind fields of Wang Ying are caused to system in 0.4s off-grids
Secondary pulse, cause No. 2 Fu Xi, No. 9 Fuxin wind power plants because Voltage Drop is less than its LVRT minimum operation magnitude of voltage, in 0.6s
System voltage falls again after carving off-grid simultaneously.Comparison diagram 7, Fig. 9 are understood, fall the position of the data point of rear two-dimensional space three times
Change again, No. 2 wind power plants no longer have close same tone with No. 8 wind power plants, observe No. 7 wind power plants and No. 9 wind-powered electricity generations
It can also draw same conclusion, it was demonstrated that this paper institutes extracting method can effectively identify the dynamic characteristic of wind power plant.
The foregoing is only a preferred embodiment of the present invention, but protection scope of the present invention be not limited thereto,
Any one skilled in the art the invention discloses technical scope in, the change or replacement that can readily occur in,
It should all be included within the scope of the present invention.Therefore, protection scope of the present invention should be with scope of the claims
It is defined.
Claims (3)
- A kind of 1. wind farm group transient voltage clustering recognition method based on principal component analysis, it is characterised in that including following step Suddenly：Step 1: control room, which gathers each wind power plant output voltage values Data Concurrent, gives electrical power services device, electrical power services device will The magnitude of voltage and default standard voltage value received is compared；Step 2: electrical power services device recognizes the magnitude of voltage lower than standard voltage value 10% of reception, whole wind field grid entry points are recorded Voltage value signal S (n)=[s1(n),s2(n),...,sm(n)]T, n=1,2 ..., N,Wherein m is the quantity of wind power plant, n is each wind field voltage signal quantity and N >=2, T representing matrixs transposition, S (n) are Matrix that magnitude of voltage is formed, sm(n) n-th of voltage value signal of m-th of wind park is represented；Step 3: it is U=[u to obtain m wind power plant voltage signal average1,u2,...,um]T, obtain each wind field voltage value signal Result S (the n)-U=X for going equalization to handle,Wherein X is the voltage value matrix after equalizationThat is U is wind power plant voltage signal average, T Representing matrix transposition；Step 4: obtain X covariance matrix X XT=A, passes through PTAP=diag [λ1,λ2,...,λm], obtain orthogonal matrix P, wherein λ1≥λ2≥...≥λm>=0 be matrix A m characteristic value, PTIt is right for P transposed matrix, wherein diag expressions one Angular moment battle array, λ are referred to as A characteristic value；Step 5: pass through PTX, the data Y after dimensionality reduction is obtained, and have λi=E (yi 2), X=PY=p1y1+p2y2+...+pmym, Wherein p1,p2,...,pmFor A characteristic vector, i.e. principal component load, expression and corresponding wind farm grid-connected voltage coefficient correlation, Coefficient correlation corresponding to first, second principal component is down to 2 dimensions, wherein E as horizontal, ordinate, so far N-dimensional voltage signal before extraction Formula of variance is sought in expression；Step 6: by k-means clustering algorithms by the phase relation of each voltage value signal under the two-dimensional coordinate in the step 5 Number is divided into k classes, i.e., m wind farm group is divided into k group；Step 7: calculate and obtain m point in two-dimensional space to the Euclidean distance of the central point of k group in the step 6, M point is incorporated into closest classification respectively；Step 8: according to cluster result, the arithmetic average of all elements Euclidean distance in each group is recalculated and obtains k Individual center；Step 9：Repeating said steps seven are constant up to cluster result, draw the similar wind farm group of transient behavior, individually in groups Electric field be off-grid electric field.
- 2. a kind of wind farm group transient voltage clustering recognition method based on principal component analysis according to claim 1, institute State Step 6: by k-means clustering algorithms by each voltage signal under the two-dimensional coordinate in the step 5 by coefficient correlation point For 3 classes, i.e., m wind power plant is divided into 3 groups.
- 3. a kind of wind farm group transient voltage clustering recognition method based on principal component analysis according to claim 1, institute State Step 6: being divided the coefficient correlation of each voltage signal under the two-dimensional coordinate in the step 5 by k-means clustering algorithms For 4 classes, i.e., m wind power plant is divided into 4 groups.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710834264.0A CN107482683B (en) | 2017-09-15 | 2017-09-15 | Wind power plant group transient voltage cluster identification method based on principal component analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710834264.0A CN107482683B (en) | 2017-09-15 | 2017-09-15 | Wind power plant group transient voltage cluster identification method based on principal component analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107482683A true CN107482683A (en) | 2017-12-15 |
| CN107482683B CN107482683B (en) | 2019-12-10 |
Family
ID=60585260
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710834264.0A Active CN107482683B (en) | 2017-09-15 | 2017-09-15 | Wind power plant group transient voltage cluster identification method based on principal component analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107482683B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109740641A (en) * | 2018-12-18 | 2019-05-10 | 清华四川能源互联网研究院 | A kind of area user identification zone method clustered based on principal component analysis and k-means |
| CN109815975A (en) * | 2018-12-14 | 2019-05-28 | 中国科学院深圳先进技术研究院 | A kind of objective classification method and relevant apparatus based on robot |
| CN111768066A (en) * | 2020-05-09 | 2020-10-13 | 国网江苏省电力有限公司无锡供电分公司 | Park electric heating load coupling relation analysis method and device based on fusion characteristics |
| CN111985815A (en) * | 2020-08-21 | 2020-11-24 | 国网能源研究院有限公司 | Method and device for screening energy and power operation evaluation indexes |
| CN112114186A (en) * | 2020-08-14 | 2020-12-22 | 广西电网有限责任公司电力科学研究院 | Voltage sag source identification method and device based on distance discriminant analysis |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105303468A (en) * | 2015-11-20 | 2016-02-03 | 国网天津市电力公司 | Comprehensive evaluation method of smart power grid construction based on principal component cluster analysis |
| CN105741188A (en) * | 2016-01-26 | 2016-07-06 | 重庆大学 | Method for dividing electromagnetic transient coherent generator groups of double-feed wind generator groups |
| CN106226572A (en) * | 2016-07-13 | 2016-12-14 | 国家电网公司 | Household loads recognition methods based on transient characteristic cluster |
| CN106451527A (en) * | 2016-07-15 | 2017-02-22 | 华北电力大学 | Doubly-fed wind farm group aggregating method and doubly-fed wind farm group aggregating system based on dynamic characteristics of rotor current |
| CN106992545A (en) * | 2017-05-02 | 2017-07-28 | 贵州大学 | The machine-electricity transient model and modeling method of weak consistency wind speed profile mountain region wind power plant |
-
2017
- 2017-09-15 CN CN201710834264.0A patent/CN107482683B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105303468A (en) * | 2015-11-20 | 2016-02-03 | 国网天津市电力公司 | Comprehensive evaluation method of smart power grid construction based on principal component cluster analysis |
| CN105741188A (en) * | 2016-01-26 | 2016-07-06 | 重庆大学 | Method for dividing electromagnetic transient coherent generator groups of double-feed wind generator groups |
| CN106226572A (en) * | 2016-07-13 | 2016-12-14 | 国家电网公司 | Household loads recognition methods based on transient characteristic cluster |
| CN106451527A (en) * | 2016-07-15 | 2017-02-22 | 华北电力大学 | Doubly-fed wind farm group aggregating method and doubly-fed wind farm group aggregating system based on dynamic characteristics of rotor current |
| CN106992545A (en) * | 2017-05-02 | 2017-07-28 | 贵州大学 | The machine-electricity transient model and modeling method of weak consistency wind speed profile mountain region wind power plant |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109815975A (en) * | 2018-12-14 | 2019-05-28 | 中国科学院深圳先进技术研究院 | A kind of objective classification method and relevant apparatus based on robot |
| CN109740641A (en) * | 2018-12-18 | 2019-05-10 | 清华四川能源互联网研究院 | A kind of area user identification zone method clustered based on principal component analysis and k-means |
| CN111768066A (en) * | 2020-05-09 | 2020-10-13 | 国网江苏省电力有限公司无锡供电分公司 | Park electric heating load coupling relation analysis method and device based on fusion characteristics |
| CN111768066B (en) * | 2020-05-09 | 2022-09-09 | 国网江苏省电力有限公司无锡供电分公司 | Park electric heating load coupling relation analysis method and device based on fusion characteristics |
| CN112114186A (en) * | 2020-08-14 | 2020-12-22 | 广西电网有限责任公司电力科学研究院 | Voltage sag source identification method and device based on distance discriminant analysis |
| CN111985815A (en) * | 2020-08-21 | 2020-11-24 | 国网能源研究院有限公司 | Method and device for screening energy and power operation evaluation indexes |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107482683B (en) | 2019-12-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107482683A (en) | A kind of wind farm group transient voltage clustering recognition method based on principal component analysis | |
| Singh et al. | A review on distributed generation planning | |
| Wang et al. | Toward the prediction level of situation awareness for electric power systems using CNN-LSTM network | |
| CN105403811B (en) | Wind power plant power grid fault diagnosis method and device | |
| CN103870923B (en) | The wind energy turbine set group of planes polymerization of coagulation type hierarchical clustering algorithm based on comentropy | |
| CN103761349B (en) | A kind of wind energy turbine set equivalent modeling method based on Wind turbines probability same tone | |
| Ali et al. | Probabilistic clustering of wind generators | |
| CN104036073B (en) | Double-fed wind power plant dynamic equivalence modeling method suitable for active power characteristic analysis | |
| CN107732957A (en) | A kind of high permeability distribution type renewable energy generating assemblage classification method | |
| CN112818491A (en) | Wind power plant aggregation equivalent modeling method based on principal component analysis and clustering algorithm | |
| CN107732958B (en) | Independent micro-grid black start optimization method considering source-load cooperative recovery | |
| CN106953379A (en) | A kind of balance control method and device of energy-storage battery state-of-charge in parallel | |
| CN110224385A (en) | The Optimal Configuration Method of superconductive failure current limiter in a kind of active distribution network | |
| Yan et al. | Deep learning based total transfer capability calculation model | |
| CN112083270A (en) | Wind power plant current collection line single-phase earth fault line selection method based on correlation coefficient | |
| Chen et al. | Fault diagnosis and classification for photovoltaic arrays based on principal component analysis and support vector machine | |
| CN108460228B (en) | Wind power plant equivalence method based on multi-objective optimization algorithm | |
| Tang et al. | Self-adaptive protection strategies for distribution system with DGs and FCLs based on data mining and neural network | |
| Lin et al. | Clustering wind turbines for a large wind farm using spectral clustering approach based on diffusion mapping theory | |
| Xu et al. | Optimization of emergency load shedding based on cultural particle swarm optimization algorithm | |
| CN108416459A (en) | A kind of site selecting method of battery energy storage power station | |
| Sun et al. | A preliminary study on the Thevenin equivalent parameters of DFIG wind farms | |
| Akwasi et al. | Observer controller-based structure for a modified flower pollination algorithm for wind power generation | |
| Hatziargyriou et al. | On-line dynamic security assessment of power systems in large islands with high wind power penetration | |
| CN108964134A (en) | The probability analysis method of distributed generation resource planning based on area gray relational decision-making |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |