CN110766032A - Power distribution network data clustering integration method based on hierarchical progressive strategy - Google Patents

Power distribution network data clustering integration method based on hierarchical progressive strategy Download PDF

Info

Publication number
CN110766032A
CN110766032A CN201810842519.2A CN201810842519A CN110766032A CN 110766032 A CN110766032 A CN 110766032A CN 201810842519 A CN201810842519 A CN 201810842519A CN 110766032 A CN110766032 A CN 110766032A
Authority
CN
China
Prior art keywords
cluster
theta
clustering
distribution network
power distribution
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201810842519.2A
Other languages
Chinese (zh)
Inventor
王希
罗海珠
邵平珍
袁璐
王广生
查四平
黄纪佳
陈立敏
涂筠
曾春
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Corp of China SGCC
State Grid Jiangxi Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
State Grid Jiangxi Electric Power Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by State Grid Corp of China SGCC, State Grid Jiangxi Electric Power Co Ltd filed Critical State Grid Corp of China SGCC
Priority to CN201810842519.2A priority Critical patent/CN110766032A/en
Publication of CN110766032A publication Critical patent/CN110766032A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/231—Hierarchical techniques, i.e. dividing or merging pattern sets so as to obtain a dendrogram
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Economics (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Water Supply & Treatment (AREA)
  • Public Health (AREA)
  • Primary Health Care (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention is suitable for the technical field of power grid data analysis and decision research, and provides a power distribution network data clustering integration method, which comprises the following steps: generating an alternative cluster set of the power distribution network data; screening the candidate cluster set based on a hierarchical progressive local weight algorithm to obtain a basic cluster set; and integrating the basic clustering set by a hierarchical clustering method to obtain final integrated clustering. Therefore, the power distribution network data clustering integration method based on the hierarchical progressive strategy can effectively improve the clustering effect of the power distribution network data.

Description

Power distribution network data clustering integration method based on hierarchical progressive strategy
Technical Field
The invention relates to the technical field of power grid data analysis and decision research, in particular to a power distribution network data clustering integration method.
Background
The clustering integration method is an effective power distribution network line data clustering analysis method, wherein an integration strategy based on clustering weight information entropy is an effective novel integration clustering scheme, but the method has the defects of large calculation amount, sensitivity to basic clustering and the like.
In summary, the clustering method for power distribution network data in the prior art obviously has inconvenience and defects in practical use, so that improvement is needed.
Disclosure of Invention
In view of the above defects, the present invention provides a power distribution network data clustering integration method, which can effectively improve the clustering effect of power distribution network data.
In order to achieve the above object, the present invention provides a power distribution network data clustering integration method based on a hierarchical progressive strategy, which includes:
A. generating an alternative cluster set of the power distribution network data;
B. screening the candidate cluster set based on a hierarchical progressive local weight algorithm to obtain a basic cluster set;
C. and integrating the basic clustering set by a hierarchical clustering method to obtain final integrated clustering.
According to the power distribution network data clustering integration method, in the step A, the alternative clustering set is generated through a K-means method.
According to the power distribution network data clustering integration method, the step of generating the alternative cluster set through a K-means method comprises the following steps:
a1, initial data set S ═ S for distribution network data1,s2,...,snThe element number | S | is n, and the number of the alternative clusters of the alternative cluster set to be generated is M;
a2, setting a first control parameter j and setting an initial value to be 1, and defining a parameter k;
a3, if j is less than M, the slave data set S ═ S1,s2,...,snRandomly selecting K elements as initial cluster centers to form K clusters;
a4, setting a second control parameter i to be 1, if i is smaller than n-k, jumping to the step A5, otherwise, jumping to the step A7;
a5, respectively calculating Euclidean distances from the ith element to k cluster centers in the remaining n-k elements in the data set S, and selecting the cluster closest to the cluster center to join;
a6, adding 1 to the second control parameter i, and going to step A5;
a7, selecting 1 cluster center from the k clusters again to form k new cluster centers, calculating the Euclidean distance from the element to the new cluster center, if the new cluster center is the same as the previous cluster center, jumping to the step A8, otherwise, jumping to the step A4;
a8, obtaining the candidate cluster thetajAnd the mark is determined clustering and added with an alternative clustering set theta;
a9, adding 1 to the first control parameter j, and jumping to the step A3;
a10, obtaining the candidate cluster set as Θ ═ θ1，θ2，...，θM}。
According to the power distribution network data clustering and integrating method, the step B comprises the following steps:
b1, using the candidate cluster set Θ { Θ ═ θ generated by the K-means method1，θ2，....θ，θMSetting an initial value of a control parameter M as 1, and setting a cycle time limit M;
b2, judging whether the control parameter M is less than or equal to M, if so, executing the next step, otherwise, turning to the step 9;
b3, calculating candidate cluster set theta ═ { theta ═ theta1，θ2，...，θMThe m-th cluster in (theta)mAny one of the clusters of
Figure BDA0001745917190000021
θmE Θ, entropy of uncertainty information with respect to all said candidate clusters in Θ
Figure BDA0001745917190000022
B4, calculating theta obtained in the step B3mOf each clusterSum of (θ)m)；
B5 converting the sum value Σ (θ) using a normalized weight conversion methodm) Entropy of uncertain information of clusters in each of said candidate clusters
Figure BDA0001745917190000024
Said sum Σ (θ)m) After normalization by the normalization weight conversion method, the weight W (Sigma (theta)) of the corresponding candidate cluster is obtainedm) Make the value interval of the weight be (0, 1)]；
B6, setting a first threshold α and a second threshold β, and 0< ═ α < β < ═ 1;
b7, W (theta) of all the candidate clusters calculated in the step B5m) ) is compared to a first threshold α and a second threshold β if the candidate cluster θ ismE theta satisfies the condition α < W (theta)m) β and/or β W (Σ)(θm) 1), the candidate cluster is labeled as determined. If the candidate cluster is thetamE theta is left with 0< W (theta)m) If α, marking the alternative cluster as deletion;
b8, adding 1 to the control parameter m, and returning to the step B2.
B9, theta ═ theta for the candidate cluster set1，θ2，....θ，θMAnd extracting all the candidate clusters marked as determined from the theta, and re-marking the number of the candidate clusters marked as determined as M to obtain a basic cluster set theta ═ theta { (theta)1，θ2，...，θM}。
According to the power distribution network data clustering integration method, in the step B3, the uncertain information entropy is
Figure BDA0001745917190000031
The calculation formula of (2) is as follows:
Figure BDA0001745917190000032
wherein M is more than or equal to 1 and less than or equal to M, N is more than or equal to 1 and less than or equal to NM，
Figure BDA0001745917190000033
1≤μ≤M，1≤j≤NM，
Figure BDA0001745917190000034
| is the number of elements of the set.
According to the power distribution network data clustering integration method, in the step B4, the sum value sigma (theta)m) The calculation formula of (2) is as follows:
Figure BDA0001745917190000035
wherein N ism＝|θm|。
According to the power distribution network data clustering integration method, in the step B5, the calculation formula of the normalized weight conversion method is as follows:
W(*)＝e-*。
according to the power distribution network data clustering and integrating method, the step C comprises the following steps:
c1, calculating an initial data set S ═ S of the distribution network data based on the weight information of the clusters in the basic cluster obtained in the step B1,s2,...,snThe times of occurrence of any two elements in the same cluster of each basic cluster in the basic cluster set and the product of the times and the cluster weight serve as weighted integration distance Dis(s) between any two elementsi,sj)。
C2, initial data set S ═ { S } obtained based on step C11,s2,...,sxThe weighted integration distance Dis(s) between any two elementsi,sj) And as the clustering distance among elements in the hierarchical clustering method, hierarchical clustering is carried out on the initial data set S, and the final integrated clustering output is obtained.
According to the power distribution network data clustering integration method, the initial data set S ═ S1,s2,...,sxThe weighted integration distance Dis(s) between any two elementsi,sj) The calculation formula of (2) is as follows:
Figure BDA0001745917190000036
wherein s isi∈S,sjE S and Si≠sj,，Cn mClustering θ for the basismMiddle SiThe cluster is marked as si∈Cn m，Cn m∈θm,n∈[1,Nm]，wi mIs equal to
Figure BDA0001745917190000037
If s isi∈Cn mWhen s isjAlso belong to the basis cluster thetamCluster C in (1)n mTime phiij mIf s is 1i∈Cn mWhen s isjNot belonging to said basic cluster θmCluster C in (1)n mTime phiij m＝0。
Aiming at the characteristics of high complexity degree, huge data volume and the like of power grid distribution network data, a power distribution network data clustering integration method based on a hierarchical progressive strategy is used by introducing a hierarchical progressive thought in an uncertain theory, and firstly, an alternative clustering set of the power distribution network data is generated; then, screening the candidate cluster set based on a hierarchical progressive local weight algorithm to obtain a basic cluster set; and finally, integrating the basic clustering set by a hierarchical clustering method to obtain the final integrated cluster. Therefore, the power distribution network data clustering integration method based on the hierarchical progressive strategy can effectively improve the clustering effect of the power distribution network data.
Drawings
FIG. 1 is a flow chart of a power distribution network data clustering integration method based on a hierarchical progressive strategy according to the present invention;
FIG. 2 is a flowchart of a preferred embodiment of a method for clustering and integrating power distribution network data based on a hierarchical progressive strategy according to the present invention;
FIG. 3 is a flowchart of method steps of a preferred embodiment of a method for clustering and integrating power distribution network data based on a hierarchical progressive strategy according to the present invention;
FIG. 4 is one of the algorithm flow diagrams of the method steps corresponding to the preferred embodiment of the method for power distribution network data clustering integration based on the hierarchical progressive strategy of the present invention;
FIG. 5 is a flowchart of method steps of a preferred embodiment of a method for clustering and integrating power distribution network data based on a hierarchical progressive strategy according to the present invention;
fig. 6 is one of the algorithm flow charts of the method for power distribution network data clustering integration based on the hierarchical progressive strategy according to the present invention, which corresponds to the method steps of the preferred embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, in an embodiment of the present invention, a method for clustering and integrating power distribution network data based on a hierarchical progressive policy is provided, including:
step S101, generating an alternative cluster set of power distribution network data;
s102, screening the candidate cluster set based on a hierarchical progressive local weight algorithm to obtain a basic cluster set;
and S103, integrating the basic cluster set by a hierarchical clustering method to obtain the final integrated cluster.
In the embodiment, a research object is a data set of power distribution network data, aiming at the characteristics of high complexity degree, huge data volume and the like of the power distribution network data and the defects of the existing integrated clustering technology, a power distribution network data clustering integration method based on a hierarchical progressive strategy is used by introducing a hierarchical progressive thought in an uncertain theory, and firstly, an alternative clustering set of the power distribution network data is generated; then, screening the candidate cluster set based on a hierarchical progressive local weight algorithm to obtain a basic cluster set; and finally, integrating the basic clustering set by a hierarchical clustering method to obtain the final integrated cluster. Therefore, the anti-interference performance and robustness of the power distribution network data clustering and the final effect of the integrated clustering can be effectively improved, and a more powerful reference basis is provided for the power data clustering.
Referring to fig. 2, in an embodiment of the present invention, in the step S101, the set of candidate clusters is generated by a K-means method. Specifically, M candidate clusters (all labeled as definite clusters) are produced by a K-means clustering method on the data set S, wherein each candidate cluster has K clusters, and the candidate cluster set is composed of theta ═ theta1，θ2，...，θM}; the specific steps of generating the candidate cluster set by the K-means method comprise:
in step S201, S ═ S for the initial data set of the distribution network data1,s2,...,snThe element number | S | is n, and the number of the alternative clusters of the alternative cluster set to be generated is M;
step S202, setting a first control parameter j and setting an initial value to be 1, and defining a parameter k;
in step S203, if j is smaller than M, the slave data set S ═ S1,s2,...,snRandomly selecting K elements as initial cluster centers to form K clusters;
step S204, setting a second control parameter i to be 1, if i is smaller than n-k, jumping to step S205, otherwise, jumping to step S207;
step S205, selecting the added clusters according to Euclidean distance: respectively calculating Euclidean distances from the ith element to k cluster centers in the remaining n-k elements in the data set S, and selecting the cluster closest to the cluster center to join, namely selecting the cluster where the cluster center closest to the selected cluster center is located according to the Euclidean distances to join;
step S206, adding 1 to the second control parameter i, and going to step S205;
step S207, respectively selecting 1 cluster center from the k clusters again to form k new cluster centers, calculating the Euclidean distance from the element to the new cluster center, if the new cluster center is the same as the previous cluster center, jumping to step S208, otherwise, jumping to step S204;
step S208, obtaining the alternative cluster thetajAnd the mark is determined clustering and added with an alternative clustering set theta;
step S209, adding 1 to the first control parameter j, and jumping to step S203;
step S210, obtaining the candidate cluster set as Θ ═ θ1，θ2，...，θM}。
Referring to fig. 3 to 4, fig. 3 is a flowchart of method steps of the present invention, fig. 4 is an algorithm flowchart corresponding to the method steps of this embodiment, in an embodiment of the present invention, basic clusters are screened based on a local weight algorithm of hierarchical progression, iterative computation of a hierarchical progression idea is performed based on two decision strategies in an uncertain theory and a local information entropy computation weight, and finally, basic clusters with higher quality and fewer number are screened. The step S102 includes:
step S301 of using the candidate cluster set Θ { θ } generated by the K-means method1，θ2，...，θMSetting an initial value of a control parameter M as 1, and setting a cycle time limit M;
step S302, judging whether the control parameter M is less than or equal to M, if so, executing the next step, otherwise, turning to the step S309;
step S303, calculating a candidate cluster set Θ ═ θ1，θ2，...，θMThe m-th cluster in (theta)mAny one of the clusters ofθmE Θ, entropy of uncertainty information with respect to all said candidate clusters in Θ
Figure BDA0001745917190000066
In particular, the uncertain information entropy
Figure BDA0001745917190000067
The calculation formula of (2) is as follows:
Figure BDA0001745917190000061
wherein M is more than or equal to 1 and less than or equal to M, N is more than or equal to 1 and less than or equal to NM，
Figure BDA0001745917190000062
1≤μ≤M，1≤j≤NM，
Figure BDA0001745917190000063
| is the number of elements of the set.
Step S304, calculating theta obtained in step S303mEntropy of uncertain information per cluster in
Figure BDA0001745917190000068
Sum of (θ)m) (ii) a In particular, the sum Σ, (θm) The calculation formula of (2) is as follows:
Figure BDA0001745917190000064
wherein N ism＝|θm|。
Step S305, converting the sum value sigma (theta) by using a normalized weight conversion methodm) Entropy encoding the uncertain information of each cluster in the candidate cluster
Figure BDA0001745917190000069
Said sum Σ (θ)m) After normalization by the normalization weight conversion method, the weight W (Sigma (theta)) of the corresponding candidate cluster is obtainedm) Make the value interval of the weight be (0, 1)](ii) a Specifically, the calculation formula of the normalized weight conversion method is as follows:
W(*)＝e-*。
step S306, setting a first threshold α and a second threshold β, and 0< ═ α < β < ═ 1;
step S307, the W (theta) of all the candidate clusters calculated in step S305 is calculatedm) Compare the candidate cluster to the first threshold α and the second threshold β if θ is the candidate clustermE theta satisfies the condition α < W (theta)m) β and/or β W (theta)m) And if so, marking the candidate cluster as determined (namely the candidate cluster has relatively good effect and can be used as effective candidate cluster data of the power distribution network).
If the candidate cluster is thetamE theta is left with 0< W (theta)m) If the value is less than α, marking the alternative cluster as deleted (namely the alternative cluster is relatively poor and cannot be used as effective alternative cluster data of the power distribution network);
in step S308, the control parameter m is incremented by 1, and the process returns to step S302.
Step S309, for the candidate cluster set Θ ═ θ1，θ2，...，θMExtracting all the candidate clusters marked as determined from the theta, and extracting all the candidate clusters marked as determinedAnd the number of the alternative clusters is recorded as M again, and a basic cluster set theta is obtained1，θ2，...，θM}。
Referring to fig. 5 to 6, fig. 5 is a flowchart of method steps of the present invention, fig. 6 is a flowchart of an algorithm corresponding to the method steps of the present embodiment, and in one embodiment of the present invention, the step S103 includes:
step S501, calculating an initial data set S ═ S of the power distribution network data based on the weight information of the clusters in the basic cluster obtained in the step B1,s2,...,snThe times of occurrence of any two elements in the same cluster of each basic cluster in the basic cluster set and the product of the times and the cluster weight serve as weighted integration distance Dis(s) between any two elementsi,sj) In particular, the initial data set S ═ S1,s2,...,sxThe weighted integration distance Dis(s) between any two elementsi,sj) The calculation formula of (2) is as follows:
Figure BDA0001745917190000071
wherein s isi∈S,sjE S and Si≠sj,，Cn mClustering θ for the basismMiddle SiThe cluster is marked as si∈Cn m，Cn m∈θm,n∈[1,Nm]，wi mIs equal to
Figure BDA0001745917190000072
If s isi∈Cn mWhen s isjAlso belong to the basis cluster thetamCluster C in (1)n mTime phiij mIf s is 1i∈Cn mWhen s isjNot belonging to said basic cluster θmCluster C in (1)n mTime phi ij m＝0。
Step S502, based on the initial value obtained in step S501Data set S ═ S1,s2,...,sxThe weighted integration distance Dis(s) between any two elementsi,sj) And as the clustering distance among elements in the hierarchical clustering method, hierarchical clustering is carried out on the initial data set S by adopting a classical hierarchical clustering mode, and the final integrated clustering output is obtained.
In summary, the invention uses a power distribution network data clustering integration method based on a hierarchical progressive strategy by introducing a hierarchical progressive idea aiming at the characteristics of high complexity degree, huge data volume and the like of power distribution network data, and firstly, generates an alternative clustering set of the power distribution network data; then, screening the candidate cluster set based on a hierarchical progressive local weight algorithm to obtain a basic cluster set; and finally, integrating the basic clustering set by a hierarchical clustering method to obtain the final integrated cluster. Therefore, the power distribution network data clustering integration method based on the hierarchical progressive strategy can effectively improve the anti-interference performance and robustness of the power distribution network data clustering and the final effect of the integrated clustering, and provides a more powerful reference basis for the power data clustering.
The present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof, and it should be understood that various changes and modifications can be effected therein by one skilled in the art without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (9)

1. A power distribution network data clustering integration method based on a hierarchical progressive strategy is characterized by comprising the following steps:
A. generating an alternative cluster set of the power distribution network data;
B. screening the candidate cluster set based on a hierarchical progressive local weight algorithm to obtain a basic cluster set;
C. and integrating the basic clustering set by a hierarchical clustering method to obtain final integrated clustering.
2. The power distribution network data clustering integration method according to claim 1, wherein in the step a, the candidate cluster set is generated by a K-means method.
3. The power distribution network data clustering integration method according to claim 2, wherein the step of generating the candidate cluster set by a K-means method comprises:
a1, initial data set S ═ S for distribution network data1,s2,...,snThe element number | S | is n, and the number of the alternative clusters of the alternative cluster set to be generated is M;
a2, setting a first control parameter j and setting an initial value to be 1, and defining a parameter k;
a3, if j is less than M, the slave data set S ═ S1,s2,...,snRandomly selecting K elements as initial cluster centers to form K clusters;
a4, setting a second control parameter i to be 1, if i is smaller than n-k, jumping to the step A5, otherwise, jumping to the step A7;
a5, respectively calculating Euclidean distances from the ith element to k cluster centers in the remaining n-k elements in the data set S, and selecting the cluster closest to the cluster center to join;
a6, adding 1 to the second control parameter i, and going to step A5;
a7, selecting 1 cluster center from the k clusters again to form k new cluster centers, calculating the Euclidean distance from the element to the new cluster center, if the new cluster center is the same as the previous cluster center, jumping to the step A8, otherwise, jumping to the step A4;
a8, obtaining the candidate cluster thetajAnd the mark is determined clustering and added with an alternative clustering set theta;
a9, adding 1 to the first control parameter j, and jumping to the step A3;
a10, obtaining the candidate cluster set as Θ ═ θ1，θ2，...，θM}。
4. The power distribution network data clustering and integrating method according to claim 3, wherein the step B comprises the steps of:
b1, using the candidate cluster set Θ { Θ ═ θ generated by the K-means method1，θ2，...，θMSetting an initial value of a control parameter M as 1, and setting a cycle time limit M;
b2, judging whether the control parameter M is less than or equal to M, if so, executing the next step, otherwise, turning to
Step 9;
b3, calculating candidate cluster set theta ═ { theta ═ theta1，θ2，...，θMThe m-th cluster in (theta)mAny one of the clusters of
Figure FDA0001745917180000021
Entropy of uncertain information with respect to all said candidate clusters in Θ
Figure FDA0001745917180000022
B4, calculating theta obtained in the step B3mOf each cluster
Figure FDA0001745917180000023
Sum of (θ)m)；
B5 converting the sum value Σ (θ) using a normalized weight conversion methodm) Entropy of uncertain information of clusters in each of said candidate clusters
Figure FDA0001745917180000024
Said sum Σ (θ)m) After normalization by the normalization weight conversion method, the weight W (Sigma (theta)) of the corresponding candidate cluster is obtainedm) Make the value interval of the weight be (0, 1)]；
B6, setting a first threshold α and a second threshold β, and 0< ═ α < β < ═ 1;
b7, W (theta) of all the candidate clusters calculated in the step B5m) ) is compared to a first threshold α and a second threshold β if the candidate cluster θ ismE.g. theta satisfiesCondition α < W (Theta)m) β and/or β W (theta)m) 1), the candidate cluster is labeled as determined. If the candidate cluster is thetamE theta is left with 0< W (theta)m) If α, marking the alternative cluster as deletion;
b8, adding 1 to the control parameter m, and returning to the step B2.
B9, theta ═ theta for the candidate cluster set1，θ2，...，θMAnd extracting all the candidate clusters marked as determined from the theta, and re-marking the number of the candidate clusters marked as determined as M to obtain a basic cluster set theta ═ theta { (theta)1，θ2，...，θM}。
5. The power distribution network data clustering integration method according to claim 4, wherein in the step B3, the uncertain information entropy is determined
Figure FDA0001745917180000025
The calculation formula of (2) is as follows:
Figure FDA0001745917180000026
wherein M is more than or equal to 1 and less than or equal to M, N is more than or equal to 1 and less than or equal to NM，1≤μ≤M，1≤j≤NM，
Figure FDA0001745917180000028
| is the number of elements of the set.
6. The power distribution network data clustering integration method according to claim 4, wherein in the step B4, the sum value Σ (θ)m) The calculation formula of (2) is as follows:
Figure FDA0001745917180000029
wherein N ism＝θm|。
7. The method for clustering and integrating the data of the power distribution network according to claim 4, wherein in the step B5, the calculation formula of the normalized weight conversion method is as follows:
W(*)＝e-*。
8. the power distribution network data clustering integration method according to claim 1, wherein the step C comprises:
c1, calculating an initial data set S ═ S of the distribution network data based on the weight information of the clusters in the basic cluster obtained in the step B1,s2,...,snThe times of occurrence of any two elements in the same cluster of each basic cluster in the basic cluster set and the product of the times and the cluster weight serve as weighted integration distance Dis(s) between any two elementsi,sj)。
C2, initial data set S ═ { S } obtained based on step C11,s2,...,sxThe weighted integration distance Dis(s) between any two elementsi,sj) And as the clustering distance among elements in the hierarchical clustering method, hierarchical clustering is carried out on the initial data set S, and the final integrated clustering output is obtained.
9. The power distribution network data clustering integration method according to claim 1, wherein the initial data set S ═ S1,s2,...,sxThe weighted integration distance Dis(s) between any two elementsi,sj) The calculation formula of (2) is as follows:
wherein s isi∈S,sjE S and Si≠sj,，Cn mIs the base polymerClass thetamMiddle SiThe cluster is marked as si∈Cn m，Cn m∈θm,n∈[1,Nm]，wi mIs equal to
Figure FDA0001745917180000032
If s isi∈Cn mWhen s isjAlso belong to the basis cluster thetamCluster C in (1)n mTime phiij mIf s is 1i∈Cn mWhen s isjNot belonging to said basic cluster θmCluster C in (1)n mTime phiij m＝0。
CN201810842519.2A 2018-07-27 2018-07-27 Power distribution network data clustering integration method based on hierarchical progressive strategy Pending CN110766032A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810842519.2A CN110766032A (en) 2018-07-27 2018-07-27 Power distribution network data clustering integration method based on hierarchical progressive strategy

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810842519.2A CN110766032A (en) 2018-07-27 2018-07-27 Power distribution network data clustering integration method based on hierarchical progressive strategy

Publications (1)

Publication Number Publication Date
CN110766032A true CN110766032A (en) 2020-02-07

Family

ID=69327674

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810842519.2A Pending CN110766032A (en) 2018-07-27 2018-07-27 Power distribution network data clustering integration method based on hierarchical progressive strategy

Country Status (1)

Country Link
CN (1) CN110766032A (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2763469A1 (en) * 2013-02-04 2014-08-06 Commissariat à l'Énergie Atomique et aux Énergies Alternatives Method for modifying the status of local access points in a cellular network
CN105844303A (en) * 2016-04-08 2016-08-10 云南大学 Sampling type clustering integration method based on local and global information
CN106096805A (en) * 2016-05-10 2016-11-09 华北电力大学 A kind of residential electricity consumption load classification method based on entropy assessment feature selection
CN106485262A (en) * 2016-09-09 2017-03-08 国网山西省电力公司晋城供电公司 A kind of bus load Forecasting Methodology
CN107169511A (en) * 2017-04-27 2017-09-15 华南理工大学 Clustering ensemble method based on mixing clustering ensemble selection strategy
CN107480694A (en) * 2017-07-06 2017-12-15 重庆邮电大学 Three clustering methods are integrated using the weighting selection evaluated twice based on Spark platforms

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2763469A1 (en) * 2013-02-04 2014-08-06 Commissariat à l'Énergie Atomique et aux Énergies Alternatives Method for modifying the status of local access points in a cellular network
CN105844303A (en) * 2016-04-08 2016-08-10 云南大学 Sampling type clustering integration method based on local and global information
CN106096805A (en) * 2016-05-10 2016-11-09 华北电力大学 A kind of residential electricity consumption load classification method based on entropy assessment feature selection
CN106485262A (en) * 2016-09-09 2017-03-08 国网山西省电力公司晋城供电公司 A kind of bus load Forecasting Methodology
CN107169511A (en) * 2017-04-27 2017-09-15 华南理工大学 Clustering ensemble method based on mixing clustering ensemble selection strategy
CN107480694A (en) * 2017-07-06 2017-12-15 重庆邮电大学 Three clustering methods are integrated using the weighting selection evaluated twice based on Spark platforms

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
DONG HUANG,CHANG-DONG WANG,JIAN-HUANG LAI,: "Locally Weighted Ensemble Clustering", 《IEEE TRANSACTIONS ON CYBERNETICS》 *
桂云苗，朱金福: "一种用信息熵确定聚类权重的方法", 《统计与决策》 *

Similar Documents

Publication Publication Date Title
CN110880019B (en) Method for adaptively training target domain classification model through unsupervised domain
Kingrani et al. Estimating the number of clusters using diversity
CN110674865B (en) Rule learning classifier integration method oriented to software defect class distribution unbalance
CN113780344B (en) Hierarchical clustering-based federal learning method and hierarchical clustering-based federal learning system
CN109766710B (en) Differential privacy protection method of associated social network data
Zhang et al. Projective ranking: A transferable evasion attack method on graph neural networks
CN115358487A (en) Federal learning aggregation optimization system and method for power data sharing
Cheng et al. An elitism strategy based genetic algorithm for streaming pattern discovery in wireless sensor networks
CN112800231A (en) Power data verification method and device, computer equipment and storage medium
CN101414365B (en) Vector code quantizer based on particle group
CN116340646A (en) Recommendation method for optimizing multi-element user representation based on hypergraph motif
Gao et al. A deep learning framework with spatial-temporal attention mechanism for cellular traffic prediction
CN108614932B (en) Edge graph-based linear flow overlapping community discovery method, system and storage medium
Lee et al. NAS-TasNet: Neural architecture search for time-domain speech separation
JP5518757B2 (en) Document classification learning control apparatus, document classification apparatus, and computer program
CN117676385A (en) Charging station load prediction system and method under horizontal clustering LSTM federal learning
Li et al. A game-theoretic method for designing distortion function in spatial steganography
CN110766032A (en) Power distribution network data clustering integration method based on hierarchical progressive strategy
CN112348041A (en) Log classification and log classification training method and device, equipment and storage medium
Shen et al. Pattern-coupled sparse Bayesian learning for recovery of block-sparse signals
CN116228014A (en) DC power distribution network benefit evaluation system and method for infrastructure network access
Ke et al. Spark-based feature selection algorithm of network traffic classification
Purnawansyah et al. K-Means clustering implementation in network traffic activities
CN115481415A (en) Communication cost optimization method, system, device and medium based on longitudinal federal learning
CN115587187A (en) Knowledge graph complementing method based on small sample

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20200207