CN109802383A - Distributed generation system equivalent modeling method based on clustering algorithm - Google Patents
Distributed generation system equivalent modeling method based on clustering algorithm Download PDFInfo
- Publication number
- CN109802383A CN109802383A CN201811557482.5A CN201811557482A CN109802383A CN 109802383 A CN109802383 A CN 109802383A CN 201811557482 A CN201811557482 A CN 201811557482A CN 109802383 A CN109802383 A CN 109802383A
- Authority
- CN
- China
- Prior art keywords
- distributed generation
- generation resource
- class
- equivalent
- central point
- 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
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a kind of distributed generation system equivalent modeling method based on clustering algorithm.The present invention includes: 1, chooses equivalent line impedance of the distributed generation resource at points of common connection PCC as one of clustering target, to obtain the clustering target data of n distributed generation resource;2, clustering target threshold value is set, n distributed generation resource is divided by c class by clustering algorithm.Wherein, classification number and initial cluster center point are obtained using the cover clustering algorithm first, then is further clustered by Fuzzy C-Means Cluster Algorithm and obtains final cluster centre point;3, of a sort distributed generation resource will be assigned to and merge into an equivalent distributed generation resource, each equivalent parameters of equivalent distributed generation resource are calculated, to obtain the Equivalent Model of distributed generation system.Emulation accuracy is ensured in the case where simulation scale is big according to the cluster Equivalent Model that the invention constructs, while reducing the complexity and simulation time of model.
Description
Technical field
The present invention relates to a kind of distributed generation system equivalent modeling method based on clustering algorithm, belongs to distributed energy
Technical field of power generation.
Background technique
In recent years, since fossil energy is exhausted, environmental degradation, the mankind constantly increase renewable energy demand, renewable
Electric power share shared by energy electricity generation system is continuously increased.Since there are hundreds and thousands of a parallel network reverses for distributed generation system
Device, control mode is complicated, and the time used in simulation analysis is long, and occupies larger calculating space.For the ease of studying distributed power generation
The characteristic of system, while avoiding establishing detailed model to each distributed generation resource, it is necessary to establish distributed generation system
Equivalent Model.The similar distributed generation resource of state is merged, so that simulation scale is quantitatively reduced, when reducing emulation
Between.
For the equivalent modeling of electricity generation system, has more domestic and international academic papers at present and analyzed and propose solution party
Case, such as:
" Yan Kai, Zhang Baohui, Qu Jiping wait the modeling of photovoltaic generating system transient state and equivalence [J] power train blanket insurance to document 1
Shield and control, 2015,43 (1): 1-8. "
Document 2 " Naik R, Mohan N, Rogers M, et al.A novel grid interface, ojtimized
for utility-scale ajjlications of jhotovoltaic,wind-electric,and fuel-cell
Systems [J] .IEEE Transactions on Jower Delivery, 2002,10 (4): 1920-1926. " it is (" a kind of new
Emerging grid interface is optimized for the photovoltaic, wind-powered electricity generation and fuel cell system of public utilities sizable application " --- 2002
Year IEEE periodical)
" Sheng Wanxing, Ji Yu, Wu Ming wait based on the region centralization photovoltaic hair for improving Fuzzy C-Means Cluster Algorithm to document 3
Electric system dynamic grouping modeling [J] electric power network technique, 2017 (10) "
Document 1 is equivalent at a photovoltaic generation unit by entire photovoltaic plant, however works as photovoltaic generation unit operating status
When differing larger, equivalence can generate large error at a photovoltaic generation unit.Document 2 is by inverter two sides element according to inversion
The requirement of device topological structure carries out abbreviation, and this method cannot reflect the dynamic characteristic of photovoltaic generating system each section comprehensively.Document 3
Equivalent modeling is carried out for region centralization photovoltaic generating system, but is not fully appropriate for distributed generation system.
Summary of the invention
The purpose of the present invention is being directed to the deficiency of above-mentioned background technique, a kind of distributed power generation based on clustering algorithm is provided
System equivalent modeling method: using equivalent line impedance of the distributed generation resource at points of common connection PCC as one of clustering target,
The cover clustering algorithm is used first, then is classified by Fuzzy C-Means Cluster Algorithm to distributed generation resource, by similar distribution
Formula power supply carries out equivalence, to construct the Equivalent Model of distributed generation system.
The object of the present invention is achieved like this, and the present invention provides a kind of distributed generation systems based on clustering algorithm
Equivalent modeling method, the distributed generation system include n distributed generation resource and h common bus, wherein h and n are positive
Integer, and h≤n, distributed generation resource and transformer series are followed by common bus, and common bus is tree topology, common bus
On branch point be node, all common bus converge at points of common connection PCC and access power grid, the method includes with
Lower step:
Step 1 calculates separately equivalent line impedance of the n distributed generation resource at points of common connection PCC, as point
One of the clustering target of cloth power supply, to obtain the clustering target data of n distributed generation resource;
Any one distributed generation resource in n distributed generation resource is denoted as i-th of distributed generation resource, 1≤i≤n, and sets
Node where i distributed generation resource is the m node layer of common bus where it, then i-th of distributed generation resource is to commonly connected
Equivalent line impedance Z at point PCCeqiCalculation formula are as follows:
Zeqi=(Zm×Im+Zm-1×Im-1+…+Z2×I2+Z1×I1)/In
Wherein, ZmFor the line impedance of i-th of distributed generation resource to m node layer, ImTo flow through i-th of distributed generation resource
To the line impedance Z of m node layermElectric current, Zm-1It is the m node layer that is connect with i-th of distributed generation resource to m-1 layers
The line impedance of node, Im-1Route to flow through the m node layer connecting with i-th of distributed generation resource to m-1 node layer hinders
Anti- Zm-1Electric current, Z2Line impedance for the 3rd node layer to the 2nd node layer being connect with i-th of distributed generation resource, I2To flow through
The line impedance Z for the 3rd node layer to the 2nd node layer being connect with i-th of distributed generation resource2Electric current, Z1To be distributed with i-th
The line impedance of the 2nd node layer to the 1st node layer of formula power supply connection, I1To flow through the connect with i-th of distributed generation resource the 2nd
Line impedance Z of the node layer to the 1st node layer1Electric current, InFor the output electric current of n-th of distributed generation resource;
Step 2, setting clustering target threshold value, obtain classification number c by clustering algorithm, i.e., n distributed generation resource are divided into c
A class, wherein 2≤c≤n;
Step 3 will assign to of a sort distributed generation resource and merge into an equivalent distributed generation resource, calculate equal Distribution values
Each equivalent parameters of formula power supply, to obtain the Equivalent Model of distributed generation system;
Any one class in c class is denoted as j-th of class, the cluster centre point of j-th of class is denoted as cluster centre point vj,
1≤j≤c;J-th of class is assigned to equipped with r distributed generation resource, any one distributed generation resource is denoted as distributed generation resource u, and 1
≤ r≤n, u=1,2...r;
Each equivalent parameters of the equivalent distributed generation resource of j-th of class are calculated according to following formula:
Leq=L/a
Ceq=aC
Cdc_eq=aCdc
Steq=aSt
Zteq=Zt/a
Kp1_eq=aKp1
Ki1_eq=aKi1
Kp2_eq=Kp2/a
Ki2_eq=Ki2/a
A=Sall/Scenter
Wherein, LeqFor the filter inductance in equivalent distributed generation resource, L is as cluster centre point vjDistributed generation resource in
Filter inductance;CeqFor the filter capacitor in equivalent distributed generation resource, C is as cluster centre point vjDistributed generation resource in
Filter capacitor;Cdc_eqFor the DC filter capacitor in equivalent distributed generation resource, CdcFor as cluster centre point vjDistributed electrical
DC filter capacitor in source;SteqThe rated capacity of step-up transformer, S are met by equivalent distributed generation resourcetFor as in cluster
Heart point vjThe connect step-up transformer of distributed generation resource rated capacity;ZteqStep-up transformer is connect by equivalent distributed generation resource
Impedance, ZtFor as cluster centre point vjThe connect step-up transformer of distributed generation resource impedance;Kp1_eqIt is equivalent distributed
Supply voltage controls the proportionality coefficient of outer ring, Kp1For as cluster centre point vjDistributed generation resource voltage control outer ring ratio
Coefficient, Ki1_eqThe integral coefficient of outer ring, K are controlled for equivalent distributed generation resource voltagei1For as cluster centre point vjDistribution
The integral coefficient of supply voltage control outer ring;Kp2_eqFor the proportionality coefficient of equivalent distributed generation resource current control inner ring, Kp2To make
For cluster centre point vjDistributed generation resource current control inner ring proportionality coefficient;Ki2_eqFor equivalent distributed generation resource current control
The integral coefficient of inner ring, Ki2For as cluster centre point vjDistributed generation resource current control inner ring integral coefficient;SallIt is
The total capacity of distributed generation resource, S in j classcenterFor as cluster centre point vjDistributed generation resource capacity;ZeqFor equivalence
Equivalent line impedance of the distributed generation resource at points of common connection PCC, SuFor the capacity of u-th of distributed generation resource in j-th of class,
ZequFor equivalent line impedance of u-th of the distributed generation resource in j-th of class at points of common connection PCC.
Preferably, in step 2, the clustering algorithm is to obtain classification number and initial clustering using the cover clustering algorithm first
Central point, then further clustered by Fuzzy C-Means Cluster Algorithm and obtain final cluster centre point, the specific steps are as follows:
Step 2.1 obtains data acquisition system X, X={ x according to the clustering target data of n distributed generation resource1,x2,…,
xi,…,xn},x1Indicate the clustering target data of the 1st distributed generation resource, x2Indicate the clustering target of the 2nd distributed generation resource
Data, xiIndicate the clustering target data of i-th of distributed generation resource, xnIndicate the clustering target data of n-th of distributed generation resource;
Step 2.2, setting first area threshold value T1 and second area threshold value T2, wherein T1 > T2, is clustered using the cover and is calculated
Method carries out initial clustering, obtains classification number c and initial cluster center point set V;By the 1st in initial cluster center point set V
The central point of class is denoted as central point v1, the central point of the 2nd class is denoted as central point v in initial cluster center point set V2, initial poly-
The central point of j-th of class is denoted as central point v in the point set V of class centerj, c-th of class central point in initial cluster center point set V
It is denoted as central point vc, then initial cluster center point set V={ v1,v2,…,vj,…,vc};
Step 2.2.1, initialization enables j=1, appoints from data acquisition system X and takes a clustering target data as initial clustering
The central point v of j-th of class in the point set V of centerj;
Step 2.2.2, the clustering target data and central point v in data acquisition system X are obtainedjThe distance between set, be denoted as
Distance set Dj, Dj={ d1j,d2j,…,dij,…,dnj},d1jIndicate the clustering target data x of the 1st distributed generation resource1With
Central point vjThe distance between, d2jIndicate the clustering target data x of the 2nd distributed generation resource2With central point vjThe distance between,
dijIndicate the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between, dnjIndicate n-th of distribution
The clustering target data x of power supplynWith central point vjThe distance between;
Step 2.2.3, initialization enables the i=1 in step 2.2.2;
Step 2.2.4, it makes the following judgment:
If the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between dijLess than first area threshold
Value T1, then by the clustering target data x of i-th of distributed generation resourceiIt is added in j-th of class;
If the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between dijLess than second area threshold
Value T2, then by the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between dijFrom distance set DjIn
It deletes, by the clustering target data x of i-th of distributed generation resourceiIt is deleted from data acquisition system X;
Step 2.2.5, i+1 is assigned to i, and judges whether i > n is true, if so, then obtain the distance set of update
Dj' and the data acquisition system X' that updates, and execute 2.2.6;Otherwise, return step 2.2.4 executes judgement;
Step 2.2.6, from the distance set D of updatej' in select maximum distance dpj, 1≤p≤n, then with p-th of distribution
The clustering target data x of power supplypAs+1 central point v of jthj+1;
Step 2.2.7, judge whether the data acquisition system X' updated is sky, if it is empty, then obtains initial cluster center point set
Close V={ v1,v2,…,vj,…,vcAnd classification number c;Otherwise, j+1 is assigned to j, and return step 2.2.2 is executed;
Step 2.3, using the classification number c obtained by the cover clustering algorithm and initial cluster center point set V as Fuzzy C
The primary condition of means clustering algorithm is further clustered using Fuzzy C-Means Cluster Algorithm, and obtains final cluster centre point set
Close V(b)={ v1(b),v2(b),…,vj(b),…,vc(b)};
Step 2.3.1, the number of iterations b is defined, and initializes and enables b=1, the initial cluster center point that step 2.2 is obtained
The c cluster centre point set V that set V is obtained as the b-1 times iteration(b-1), V(b-1)={ v1(b-1),v2(b-1),…,
vj(b-1),…,vc(b-1), v1(b-1)Indicate the central point for the 1st class that the b-1 times iteration obtains, v2(b-1)It indicates to change for the b-1 times
The central point for the 2nd class that generation obtains, vj(b-1)Indicate the central point for j-th of class that the b-1 times iteration obtains, vc(b-1)Indicate the
Stopping criterion for iteration ε is arranged in c-th of class central point that b-1 iteration obtains;
Step 2.3.2, be calculate by the following formula that i-th of distributed generation resource after the b times iteration belong to j-th of class is subordinate to angle value
uij(b), to obtain the n × c dimension square for being respectively subordinate to angle value composition that n distributed generation resource after the b times iteration is belonging respectively to c class
Battle array Un×c(b),
Wherein, | | xi-vj(b-1)||2For the clustering target data x of i-th of distributed generation resourceiIt is obtained with the b-1 times iteration
The central point v of j-th of classj(b-1)Between Euclidean distance, vk(b-1)Indicate the central point for k-th of class that the b-1 times iteration obtains, | |
xi-vk(b-1)||2For the clustering target data x of i-th of distributed generation resourceiThe central point of k-th of the class obtained with the b-1 times iteration
vk(b-1)Between Euclidean distance, m is fuzzy coefficient, takes m=2;
Step 2.3.3, the cluster centre point set V that the b times iteration obtains is obtained(b), V(b)={ v1(b),v2(b),…,
vj(b),…,vc(b), v1(b)Indicate the central point for the 1st class that the b times iteration obtains, v2(b)Indicate the b times iteration obtains
The central point of 2 classes, vj(b)Indicate the central point for j-th of class that the b times iteration obtains, vc(b)Indicate the b times iteration obtains
C class central point is calculate by the following formula the central point v for j-th of class that the b times iteration obtainsj(b):
Step 2.3.4, calculating target function J (U according to the following formulan×c(b),V(b)) value:
Wherein | | xi-vj(b)||2For the clustering target data x of i-th of distributed generation resourceiThe jth obtained with the b times iteration
The central point v of a classj(b)Between Euclidean distance;
If step 2.3.5, objective function J (Un×c(b),V(b)) value be less than stopping criterion for iteration ε, then algorithm terminate, obtain
Final cluster centre point set V(b)={ v1(b),v2(b),…,vj(b),…,vc(b)};Otherwise, it assigns the value of b+1 to b, and returns
Step 2.3.2 is executed.
Compared with the existing technology, beneficial effects of the present invention are as follows:
1, complicated for topological structure in distributed generation system, the characteristics of distributed generation resource enormous amount, propose will point
Equivalent line impedance of the cloth power supply at points of common connection PCC will arrive at points of common connection PCC etc. as one of clustering target
Distributed generation resource similar in value line impedance is polymerized to one kind.
2, cluster classification number and initial cluster center are obtained using the cover clustering algorithm first, then poly- by fuzzy C-mean algorithm
Class algorithm further clusters in detail, solves the problems, such as that Fuzzy C-Means Cluster Algorithm is overly dependent upon initial cluster center.
3, the complexity that detailed model is greatly reduced by the Equivalent Model that clustering algorithm obtains is meeting error requirements
In the case where substantially reduce simulation time.
Detailed description of the invention
Fig. 1 is that the present invention is based on the flow charts of the distributed generation system equivalent modeling method of clustering algorithm.
Fig. 2 is the distributed photovoltaic power generation system detailed model schematic diagram of one embodiment of the invention.
Fig. 3 is the flow chart that clustering algorithm is used in the present invention.
Fig. 4 is cluster result-subordinated-degree matrix signal of the distributed photovoltaic power generation system of one embodiment of the invention
Figure.
Fig. 5 is cluster result-classification results schematic diagram of the distributed photovoltaic power generation system of one embodiment of the invention.
Fig. 6 is the detailed model and Equivalent Model simulation result comparison diagram of one embodiment of the invention.
Specific embodiment
Distributed generation system of the present invention includes n distributed generation resource and h common bus, wherein h and n are
Positive integer, and h≤n, distributed generation resource and transformer series are followed by common bus, and common bus is tree topology, public mother
Branch point on line is node, and all common bus converge at points of common connection PCC and access power grid.Fig. 2 gives this
The distributed photovoltaic power generation system detailed model schematic diagram of invention one embodiment.It may be seen that in the present embodiment, it is distributed
Electricity generation system includes 140 distributed generation resources and 5 common bus.
Fig. 1 be the present invention is based on the flow chart of the distributed generation system equivalent modeling method of clustering algorithm, can by the figure
See, the distributed generation system equivalent modeling method based on clustering algorithm the following steps are included:
Step 1 calculates separately equivalent line impedance of the n distributed generation resource at points of common connection PCC, as point
One of the clustering target of cloth power supply, to obtain the clustering target data of n distributed generation resource;
Any one distributed generation resource in n distributed generation resource is denoted as i-th of distributed generation resource, 1≤i≤n, and sets
Node where i distributed generation resource is the m node layer of common bus where it, then i-th of distributed generation resource is to commonly connected
Equivalent line impedance Z at point PCCeqiCalculation formula are as follows:
Zeqi=(Zm×Im+Zm-1×Im-1+…+Z2×I2+Z1×I1)/In
Wherein, ZmFor the line impedance of i-th of distributed generation resource to m node layer, ImTo flow through i-th of distributed generation resource
To the line impedance Z of m node layermElectric current, Zm-1It is the m node layer that is connect with i-th of distributed generation resource to m-1 layers
The line impedance of node, Im-1Route to flow through the m node layer connecting with i-th of distributed generation resource to m-1 node layer hinders
Anti- Zm-1Electric current, Z2Line impedance for the 3rd node layer to the 2nd node layer being connect with i-th of distributed generation resource, I2To flow through
The line impedance Z for the 3rd node layer to the 2nd node layer being connect with i-th of distributed generation resource2Electric current, Z1To be distributed with i-th
The line impedance of the 2nd node layer to the 1st node layer of formula power supply connection, I1To flow through the connect with i-th of distributed generation resource the 2nd
Line impedance Z of the node layer to the 1st node layer1Electric current, InFor the output electric current of n-th of distributed generation resource;
Step 2, setting clustering target threshold value, obtain classification number c by clustering algorithm, i.e., n distributed generation resource are divided into c
A class, and c cluster centre point is obtained, wherein 2≤c≤n.
The clustering algorithm is to obtain classification number and initial cluster center point using the cover clustering algorithm first, then pass through mould
Paste C means clustering algorithm, which further clusters, obtains final cluster centre point, and Fig. 3 show the stream that clustering algorithm is used in the present invention
Cheng Tu, the specific steps are as follows:
Step 2.1 obtains data acquisition system X, X={ x according to the clustering target data of n distributed generation resource1,x2,…,
xi,…,xn},x1Indicate the clustering target data of the 1st distributed generation resource, x2Indicate the clustering target of the 2nd distributed generation resource
Data, xiIndicate the clustering target data of i-th of distributed generation resource, xnIndicate the clustering target data of n-th of distributed generation resource;
Step 2.2, setting first area threshold value T1 and second area threshold value T2, wherein T1 > T2, is clustered using the cover and is calculated
Method carries out initial clustering, obtains classification number c and initial cluster center point set V;By the 1st in initial cluster center point set V
The central point of class is denoted as central point v1, the central point of the 2nd class is denoted as central point v in initial cluster center point set V2, initial poly-
The central point of j-th of class is denoted as central point v in the point set V of class centerj, c-th of class central point in initial cluster center point set V
It is denoted as central point vc, then initial cluster center point set V={ v1,v2,…,vj,…,vc};
Step 2.2.1, initialization enables j=1, appoints from data acquisition system X and takes a clustering target data as initial clustering
The central point v of j-th of class in the point set V of centerj;
Step 2.2.2, the clustering target data and central point v in data acquisition system X are obtainedjThe distance between set, be denoted as
Distance set Dj, Dj={ d1j,d2j,…,dij,…,dnj},d1jIndicate the clustering target data x of the 1st distributed generation resource1With
Central point vjThe distance between, d2jIndicate the clustering target data x of the 2nd distributed generation resource2With central point vjThe distance between,
dijIndicate the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between, dnjIndicate n-th of distribution
The clustering target data x of power supplynWith central point vjThe distance between;
Step 2.2.3, initialization enables the i=1 in step 2.2.2;
Step 2.2.4, it makes the following judgment:
If the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between dijLess than first area threshold
Value T1, then by the clustering target data x of i-th of distributed generation resourceiIt is added in j-th of class;
If the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between dijLess than second area threshold
Value T2, then by the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between dijFrom distance set DjIn
It deletes, by the clustering target data x of i-th of distributed generation resourceiIt is deleted from data acquisition system X;
Step 2.2.5, i+1 is assigned to i, and judges whether i > n is true, if so, then obtain the distance set of update
Dj' and the data acquisition system X' that updates, and execute 2.2.6;Otherwise, return step 2.2.4 executes judgement;
Step 2.2.6, from the distance set D of updatej' in select maximum distance dpj, 1≤p≤n, then with p-th of distribution
The clustering target data x of power supplypAs+1 central point v of jthj+1;
Step 2.2.7, judge whether the data acquisition system X' updated is sky, if it is empty, then obtains initial cluster center point set
Close V={ v1,v2,…,vj,…,vcAnd classification number c;Otherwise, j+1 is assigned to j, and return step 2.2.2 is executed;
Step 2.3, using the classification number c obtained by the cover clustering algorithm and initial cluster center point set V as Fuzzy C
The primary condition of means clustering algorithm is further clustered using Fuzzy C-Means Cluster Algorithm, and obtains final cluster centre point set
Close V(b)={ v1(b),v2(b),…,vj(b),…,vc(b)};
Step 2.3.1, the number of iterations b is defined, and initializes and enables b=1, the initial cluster center point that step 2.2 is obtained
The c cluster centre point set V that set V is obtained as the b-1 times iteration(b-1), V(b-1)={ v1(b-1),v2(b-1),…,
vj(b-1),…,vc(b-1), v1(b-1)Indicate the central point for the 1st class that the b-1 times iteration obtains, v2(b-1)It indicates to change for the b-1 times
The central point for the 2nd class that generation obtains, vj(b-1)Indicate the central point for j-th of class that the b-1 times iteration obtains, vc(b-1)Indicate the
Stopping criterion for iteration ε is arranged in c-th of class central point that b-1 iteration obtains;
Step 2.3.2, be calculate by the following formula that i-th of distributed generation resource after the b times iteration belong to j-th of class is subordinate to angle value
uij(b), to obtain the n × c dimension square for being respectively subordinate to angle value composition that n distributed generation resource after the b times iteration is belonging respectively to c class
Battle array Un×c(b),
Wherein, | | xi-vj(b-1)||2For the clustering target data x of i-th of distributed generation resourceiIt is obtained with the b-1 times iteration
The central point v of j-th of classj(b-1)Between Euclidean distance, vk(b-1)Indicate the central point for k-th of class that the b-1 times iteration obtains, | |
xi-vk(b-1)||2For the clustering target data x of i-th of distributed generation resourceiThe central point of k-th of the class obtained with the b-1 times iteration
vk(b-1)Between Euclidean distance, m is fuzzy coefficient, takes m=2;
Step 2.3.3, the cluster centre point set V that the b times iteration obtains is obtained(b), V(b)={ v1(b),v2(b),…,
vj(b),…,vc(b), v1(b)Indicate the central point for the 1st class that the b times iteration obtains, v2(b)Indicate the b times iteration obtains
The central point of 2 classes, vj(b)Indicate the central point for j-th of class that the b times iteration obtains, vc(b)Indicate the b times iteration obtains
C class central point is calculate by the following formula the central point v for j-th of class that the b times iteration obtainsj(b):
Step 2.3.4, calculating target function J (U according to the following formulan×c(b),V(b)) value:
Wherein | | xi-vj(b)||2For the clustering target data x of i-th of distributed generation resourceiThe jth obtained with the b times iteration
The central point v of a classj(b)Between Euclidean distance;
If step 2.3.5, objective function J (Un×c(b),V(b)) value be less than stopping criterion for iteration ε, then algorithm terminate, obtain
Final cluster centre point set V(b)={ v1(b),v2(b),…,vj(b),…,vc(b)};Otherwise, it assigns the value of b+1 to b, and returns
Step 2.3.2 is executed.
Fig. 4, Fig. 5 are that the above cluster result being calculated is carried out to the distributed generation system in the embodiment of the present invention,
Fig. 4 is the subordinated-degree matrix that distributed generation resource last time iteration obtains, and Fig. 5 is what distributed generation resource last time iteration obtained
Classification results, it can be seen that 140 distributed generation resources are divided into 5 major class altogether.
Step 3 will assign to of a sort distributed generation resource and merge into an equivalent distributed generation resource, calculate equal Distribution values
Each equivalent parameters of formula power supply, to obtain the Equivalent Model of distributed generation system.
Any one class in c class is denoted as j-th of class, the cluster centre point of j-th of class is denoted as cluster centre point vj,
1≤j≤c;J-th of class is assigned to equipped with r distributed generation resource, any one distributed generation resource is denoted as distributed generation resource u, and 1
≤ r≤n, u=1,2...r;
Each equivalent parameters of the equivalent distributed generation resource of j-th of class are calculated according to following formula:
Leq=L/a
Ceq=aC
Cdc_eq=aCdc
Steq=aSt
Zteq=Zt/a
Kp1_eq=aKp1
Ki1_eq=aKi1
Kp2_eq=Kp2/a
Ki2_eq=Ki2/a
A=Sall/Scenter
Wherein, LeqFor the filter inductance in equivalent distributed generation resource, L is as cluster centre point vjDistributed generation resource in
Filter inductance;CeqFor the filter capacitor in equivalent distributed generation resource, C is as cluster centre point vjDistributed generation resource in
Filter capacitor;Cdc_eqFor the DC filter capacitor in equivalent distributed generation resource, CdcFor as cluster centre point vjDistributed electrical
DC filter capacitor in source;SteqThe rated capacity of step-up transformer, S are met by equivalent distributed generation resourcetFor as in cluster
Heart point vjThe connect step-up transformer of distributed generation resource rated capacity;ZteqStep-up transformer is connect by equivalent distributed generation resource
Impedance, ZtFor as cluster centre point vjThe connect step-up transformer of distributed generation resource impedance;Kp1_eqIt is equivalent distributed
Supply voltage controls the proportionality coefficient of outer ring, Kp1For as cluster centre point vjDistributed generation resource voltage control outer ring ratio
Coefficient, Ki1_eqThe integral coefficient of outer ring, K are controlled for equivalent distributed generation resource voltagei1For as cluster centre point vjDistribution
The integral coefficient of supply voltage control outer ring;Kp2_eqFor the proportionality coefficient of equivalent distributed generation resource current control inner ring, Kp2To make
For cluster centre point vjDistributed generation resource current control inner ring proportionality coefficient;Ki2_eqFor equivalent distributed generation resource current control
The integral coefficient of inner ring, Ki2For as cluster centre point vjDistributed generation resource current control inner ring integral coefficient;SallIt is
The total capacity of distributed generation resource, S in j classcenterFor as cluster centre point vjDistributed generation resource capacity;ZeqFor equivalence
Equivalent line impedance of the distributed generation resource at points of common connection PCC, SuFor the capacity of u-th of distributed generation resource in j-th of class,
ZequFor equivalent line impedance of u-th of the distributed generation resource in j-th of class at points of common connection PCC.
The present embodiment passes through the accuracy of simulating, verifying Equivalent Model, by sampling detailed model and Equivalent Model public
Error at tie point PCC between the simulation waveform of active power and two waveforms of calculating, defines error calculation formula Δ x=∫ (b
(i)-a (i)) dt/ ∫ b (i) dt, wherein Δ x represents the error between Equivalent Model and detailed model, and b (i) is Equivalent Model
Simulation waveform, a (i) are the simulation waveform of detailed model, and t is simulation time.Fig. 6 is detailed model and cluster of the present embodiment etc.
It is worth the model simulation waveform comparison diagram that active power changes at points of common connection PCC, while single machine Equivalent Model is set in public affairs
The simulation waveform of active power variation compares at tie point PCC altogether.The imitative of Equivalent Model is clustered by calculating, under stable situation
True error is 2.92%, and the phantom error of single machine Equivalent Model is 15.7%, is gathered under the current intelligence of active power generation step
The phantom error of class Equivalent Model is 2.95%, and the phantom error of single machine Equivalent Model is 15.72%, can be obtained by calculated result
The dynamic characteristic of detailed model can be tracked to cluster Equivalent Model, and greatly reduces error compared with single machine Equivalent Model.
Claims (2)
1. a kind of distributed generation system equivalent modeling method based on clustering algorithm, which is characterized in that distributed generation system
Include n distributed generation resource and h common bus, wherein h and n is positive integer, and h≤n, distributed generation resource and transformer
Series connection is followed by common bus, and common bus is tree topology, and the branch point on common bus is node, all common bus
Converge at points of common connection PCC and access power grid, the described method comprises the following steps:
Step 1 calculates separately equivalent line impedance of the n distributed generation resource at points of common connection PCC, as distribution
One of clustering target of power supply, to obtain the clustering target data of n distributed generation resource;
Any one distributed generation resource in n distributed generation resource is denoted as i-th of distributed generation resource, 1≤i≤n, and is set i-th
Node where distributed generation resource is the m node layer of common bus where it, then i-th of distributed generation resource to points of common connection
Equivalent line impedance Z at PCCeqiCalculation formula are as follows:
Zeqi=(Zm×Im+Zm-1×Im-1+…+Z2×I2+Z1×I1)/In
Wherein, ZmFor the line impedance of i-th of distributed generation resource to m node layer, ImTo flow through i-th of distributed generation resource to m
The line impedance Z of node layermElectric current, Zm-1For the m node layer that is connect with i-th of distributed generation resource to m-1 node layer
Line impedance, Im-1For flow through the m node layer being connect with i-th of distributed generation resource to m-1 node layer line impedance Zm-1
Electric current, Z2Line impedance for the 3rd node layer to the 2nd node layer being connect with i-th of distributed generation resource, I2To flow through and i-th
The line impedance Z of the 3rd node layer to the 2nd node layer of a distributed generation resource connection2Electric current, Z1For with i-th of distributed generation resource
The line impedance of the 2nd node layer to the 1st node layer of connection, I1To flow through the 2nd node layer connecting with i-th of distributed generation resource
To the line impedance Z of the 1st node layer1Electric current, InFor the output electric current of n-th of distributed generation resource;
Step 2, setting clustering target threshold value, obtain classification number c by clustering algorithm, i.e., n distributed generation resource are divided into c
Class, wherein 2≤c≤n;
Step 3 will assign to of a sort distributed generation resource and merge into an equivalent distributed generation resource, calculate equivalent distributed electrical
Each equivalent parameters in source, to obtain the Equivalent Model of distributed generation system;
Any one class in c class is denoted as j-th of class, the cluster centre point of j-th of class is denoted as cluster centre point vj, 1≤j
≤c;J-th of class is assigned to equipped with r distributed generation resource, any one distributed generation resource is denoted as distributed generation resource u, 1≤r≤
N, u=1,2...r;
Each equivalent parameters of the equivalent distributed generation resource of j-th of class are calculated according to following formula:
Leq=L/a
Ceq=aC
Cdc_eq=aCdc
Steq=aSt
Zteq=Zt/a
Kp1_eq=aKp1
Ki1_eq=aKi1
Kp2_eq=Kp2/a
Ki2_eq=Ki2/a
A=Sall/Scenter
Wherein, LeqFor the filter inductance in equivalent distributed generation resource, L is as cluster centre point vjDistributed generation resource in filter
Wave inductance;CeqFor the filter capacitor in equivalent distributed generation resource, C is as cluster centre point vjDistributed generation resource in filtering
Capacitor;Cdc_eqFor the DC filter capacitor in equivalent distributed generation resource, CdcFor as cluster centre point vjDistributed generation resource in
DC filter capacitor;SteqThe rated capacity of step-up transformer, S are met by equivalent distributed generation resourcetFor as cluster centre point
vjThe connect step-up transformer of distributed generation resource rated capacity;ZteqThe resistance of step-up transformer is connect by equivalent distributed generation resource
It is anti-, ZtFor as cluster centre point vjThe connect step-up transformer of distributed generation resource impedance;Kp1_eqFor equivalent distributed generation resource
Voltage controls the proportionality coefficient of outer ring, Kp1For as cluster centre point vjDistributed generation resource voltage control outer ring ratio system
Number, Ki1_eqThe integral coefficient of outer ring, K are controlled for equivalent distributed generation resource voltagei1For as cluster centre point vjDistributed electrical
The integral coefficient of source voltage control outer ring;Kp2_eqFor the proportionality coefficient of equivalent distributed generation resource current control inner ring, Kp2For conduct
Cluster centre point vjDistributed generation resource current control inner ring proportionality coefficient;Ki2_eqFor in equivalent distributed generation resource current control
The integral coefficient of ring, Ki2For as cluster centre point vjDistributed generation resource current control inner ring integral coefficient;SallFor jth
The total capacity of distributed generation resource, S in a classcenterFor as cluster centre point vjDistributed generation resource capacity;ZeqFor equivalence point
Equivalent line impedance of the cloth power supply at points of common connection PCC, SuFor the capacity of u-th of distributed generation resource in j-th of class,
ZequFor equivalent line impedance of u-th of the distributed generation resource in j-th of class at points of common connection PCC.
2. the distributed generation system equivalent modeling method based on clustering algorithm as described in claim 1, which is characterized in that step
In rapid 2, the clustering algorithm is to obtain classification number and initial cluster center point using the cover clustering algorithm first, then pass through fuzzy
C means clustering algorithm, which further clusters, obtains final cluster centre point, the specific steps are as follows:
Step 2.1 obtains data acquisition system X, X={ x according to the clustering target data of n distributed generation resource1,x2,…,xi,…,
xn},x1Indicate the clustering target data of the 1st distributed generation resource, x2Indicate the clustering target data of the 2nd distributed generation resource, xi
Indicate the clustering target data of i-th of distributed generation resource, xnIndicate the clustering target data of n-th of distributed generation resource;
Step 2.2, setting first area threshold value T1 and second area threshold value T2, wherein T1 > T2, using the cover clustering algorithm into
Row initial clustering obtains classification number c and initial cluster center point set V;By the 1st class in initial cluster center point set V
Central point is denoted as central point v1, the central point of the 2nd class is denoted as central point v in initial cluster center point set V2, in initial clustering
The central point of j-th of class is denoted as central point v in heart point set Vj, c-th of class central point is denoted as in initial cluster center point set V
Central point vc, then initial cluster center point set V={ v1,v2,…,vj,…,vc};
Step 2.2.1, initialization enables j=1, appoints from data acquisition system X and takes a clustering target data as initial cluster center
The central point v of j-th of class in point set Vj;
Step 2.2.2, the clustering target data and central point v in data acquisition system X are obtainedjThe distance between set, be denoted as distance set
Close Dj, Dj={ d1j,d2j,…,dij,…,dnj},d1jIndicate the clustering target data x of the 1st distributed generation resource1With central point
vjThe distance between, d2jIndicate the clustering target data x of the 2nd distributed generation resource2With central point vjThe distance between, dijIt indicates
The clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between, dnjIndicate the poly- of n-th of distributed generation resource
Class achievement data xnWith central point vjThe distance between;
Step 2.2.3, initialization enables the i=1 in step 2.2.2;
Step 2.2.4, it makes the following judgment:
If the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between dijLess than first area threshold value
T1, then by the clustering target data x of i-th of distributed generation resourceiIt is added in j-th of class;
If the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between dijLess than second area threshold value
T2, then by the clustering target data x of i-th of distributed generation resourceiWith central point vjThe distance between dijFrom distance set DjIn delete
It removes, by the clustering target data x of i-th of distributed generation resourceiIt is deleted from data acquisition system X;
Step 2.2.5, i+1 is assigned to i, and judges whether i > n is true, if so, then obtain the distance set D of updatej' and
The data acquisition system X' of update, and execute 2.2.6;Otherwise, return step 2.2.4 executes judgement;
Step 2.2.6, from the distance set D of updatej' in select maximum distance dpj, 1≤p≤n, then with p-th of distributed generation resource
Clustering target data xpAs+1 central point v of jthj+1;
Step 2.2.7, judge whether the data acquisition system X' updated is sky, if it is empty, then obtains initial cluster center point set V=
{v1, v2..., vj..., vcAnd classification number c;Otherwise, j+1 is assigned to j, and return step 2.2.2 is executed;
Step 2.3, using the classification number c obtained by the cover clustering algorithm and initial cluster center point set V as fuzzy C-mean algorithm
The primary condition of clustering algorithm is further clustered using Fuzzy C-Means Cluster Algorithm, and obtains final cluster centre point set
V(b)={ v1(b), v2(b)..., vj(b)..., vc(b)};
Step 2.3.1, the number of iterations b is defined, and initializes and enables b=1, the initial cluster center point set V that step 2.2 is obtained
The c cluster centre point set V obtained as the b-1 times iteration(b-1), V(b-1)={ v1(b-1), v2(b-1)..., vj(b-1)...,
vc(b-1), v1(b-1)Indicate the central point for the 1st class that the b-1 times iteration obtains, v2(b-1)Indicate the b-1 times iteration obtains
The central point of 2 classes, vj(b-1)Indicate the central point for j-th of class that the b-1 times iteration obtains, vc(b-1)Indicate the b-1 times iteration
Stopping criterion for iteration ε is arranged in c-th obtained of class central point;
Step 2.3.2, be calculate by the following formula that i-th of distributed generation resource after the b times iteration belong to j-th of class is subordinate to angle value
uij(b), to obtain the n × c dimension square for being respectively subordinate to angle value composition that n distributed generation resource after the b times iteration is belonging respectively to c class
Battle array Un×c(b),
Wherein, | | xi-vj(b-1)||2For the clustering target data x of i-th of distributed generation resourceiThe jth obtained with the b-1 times iteration
The central point v of a classj(b-1)Between Euclidean distance, vk(b-1)Indicate the central point for k-th of class that the b-1 times iteration obtains, | | xi-
vk(b-1)||2For the clustering target data x of i-th of distributed generation resourceiThe central point of k-th of the class obtained with the b-1 times iteration
vk(b-1)Between Euclidean distance, m is fuzzy coefficient, takes m=2;
Step 2.3.3, the cluster centre point set V that the b times iteration obtains is obtained(b), V(b)={ v1(b),v2(b),…,vj(b),…,
vc(b), v1(b)Indicate the central point for the 1st class that the b times iteration obtains, v2(b)Indicate the 2nd class that the b times iteration obtains
Central point, vj(b)Indicate the central point for j-th of class that the b times iteration obtains, vc(b)It indicates in c-th of class that the b times iteration obtains
Heart point is calculate by the following formula the central point v for j-th of class that the b times iteration obtainsj(b):
Step 2.3.4, calculating target function J (U according to the following formulan×c(b),V(b)) value:
Wherein | | xi-vj(b)||2For the clustering target data x of i-th of distributed generation resourceiJ-th of the class obtained with the b times iteration
Central point vj(b)Between Euclidean distance;
If step 2.3.5, objective function J (Un×c(b),V(b)) value be less than stopping criterion for iteration ε, then algorithm terminate, obtain most
Whole cluster centre point set V(b)={ v1(b),v2(b),…,vj(b),…,vc(b)};Otherwise, it assigns the value of b+1 to b, and returns to step
Rapid 2.3.2 is executed.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811557482.5A CN109802383B (en) | 2018-12-19 | 2018-12-19 | Clustering algorithm based equivalent modeling method for distributed power generation system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811557482.5A CN109802383B (en) | 2018-12-19 | 2018-12-19 | Clustering algorithm based equivalent modeling method for distributed power generation system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109802383A true CN109802383A (en) | 2019-05-24 |
CN109802383B CN109802383B (en) | 2020-07-07 |
Family
ID=66557274
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811557482.5A Active CN109802383B (en) | 2018-12-19 | 2018-12-19 | Clustering algorithm based equivalent modeling method for distributed power generation system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109802383B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111401412A (en) * | 2020-02-29 | 2020-07-10 | 同济大学 | Distributed soft clustering method based on average consensus algorithm in Internet of things environment |
CN117610402A (en) * | 2023-10-19 | 2024-02-27 | 国家电网有限公司华东分部 | Multi-machine clustering equivalent modeling method and device for distributed photovoltaic |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102541621A (en) * | 2011-11-30 | 2012-07-04 | 中国电力科学研究院 | Simulation equivalent method of wind-photovoltaics-energy storage joint power generation system |
CN105938578A (en) * | 2016-04-13 | 2016-09-14 | 合肥工业大学 | Large-scale photovoltaic power station equivalent modeling method based on clustering analysis |
-
2018
- 2018-12-19 CN CN201811557482.5A patent/CN109802383B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102541621A (en) * | 2011-11-30 | 2012-07-04 | 中国电力科学研究院 | Simulation equivalent method of wind-photovoltaics-energy storage joint power generation system |
CN105938578A (en) * | 2016-04-13 | 2016-09-14 | 合肥工业大学 | Large-scale photovoltaic power station equivalent modeling method based on clustering analysis |
Non-Patent Citations (2)
Title |
---|
FEI LI ; FAN WU ; XING ZHANG ; JIZHONG XI ; XUAN ZUO: "Research on polymerization equivalent modeling in large — Scale photovoltaic power plant", 《IECON 2017 - 43RD ANNUAL CONFERENCE OF THE IEEE INDUSTRIAL ELECTRONICS SOCIETY》 * |
王皓怀 等;: "风光储联合发电系统的组合建模与等值", 《中国电机工程学报》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111401412A (en) * | 2020-02-29 | 2020-07-10 | 同济大学 | Distributed soft clustering method based on average consensus algorithm in Internet of things environment |
CN111401412B (en) * | 2020-02-29 | 2022-06-14 | 同济大学 | Distributed soft clustering method based on average consensus algorithm in Internet of things environment |
CN117610402A (en) * | 2023-10-19 | 2024-02-27 | 国家电网有限公司华东分部 | Multi-machine clustering equivalent modeling method and device for distributed photovoltaic |
Also Published As
Publication number | Publication date |
---|---|
CN109802383B (en) | 2020-07-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Syahputra et al. | Optimal distribution network reconfiguration with penetration of distributed energy resources | |
WO2019056753A1 (en) | Dynamic equivalent modeling method for distributed photovoltaic power station cluster | |
CN109193649A (en) | Distributed photovoltaic cluster dynamic equivalent modeling method based on deepness belief network | |
CN107732957A (en) | A kind of high permeability distribution type renewable energy generating assemblage classification method | |
CN108304972B (en) | Active power distribution network frame planning method based on supply and demand interaction and DG (distributed generation) operation characteristics | |
CN107508280B (en) | A kind of reconstruction method of power distribution network and system | |
CN105069236B (en) | Consider the broad sense load joint probability modeling method of wind power plant node space correlation | |
CN109120009A (en) | The active distribution network fault recovery method that meter and distributed generation resource power output change at random | |
CN106651000B (en) | Black-start partitioning method based on semi-supervised spectral clustering | |
CN109802383A (en) | Distributed generation system equivalent modeling method based on clustering algorithm | |
CN114725982A (en) | Distributed photovoltaic cluster refined division and modeling method | |
CN110311372A (en) | Sub-area division method based on spectral clustering | |
CN110492470A (en) | Power distribution network multidimensional typical scene generation method based on load characteristics clustering and Equivalent Network | |
Penangsang et al. | Optimal placement and sizing of distributed generation in radial distribution system using K-means clustering method | |
Wu et al. | Equivalent modeling method for regional decentralized photovoltaic clusters based on cluster analysis | |
CN113131515A (en) | Intelligent microgrid AI fault detection and network optimization reconstruction system and method | |
CN109638892B (en) | Photovoltaic power station equivalent modeling method based on improved fuzzy clustering algorithm | |
CN110380407A (en) | A kind of power distribution network running optimizatin method considering the electric irrigation and drainage load of agricultural | |
CN114329857A (en) | Distributed power supply planning method based on improved whale algorithm | |
CN115021269A (en) | Two-stage optimal power flow solving method based on data driving | |
Chen et al. | Distributed photovoltaic power cluster partition based on DPC Clustering algorithm | |
CN114389261A (en) | Distributed photovoltaic power cluster division method considering grid-connected management mode | |
CN108233373B (en) | Probability harmonic analysis method considering weather scene for distributed photovoltaic access power distribution network | |
Syahputra et al. | An artificial immune system algorithm approach for reconfiguring distribution network | |
Liu et al. | Topology Identification Method of Low-voltage Station Area Based on Hierarchical Clustering Algorithm and WDDTW Distance |
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 |