CN104850612A - Enhanced cohesion hierarchical clustering-based distribution network user load feature classifying method - Google Patents
Enhanced cohesion hierarchical clustering-based distribution network user load feature classifying method Download PDFInfo
- Publication number
- CN104850612A CN104850612A CN201510241173.7A CN201510241173A CN104850612A CN 104850612 A CN104850612 A CN 104850612A CN 201510241173 A CN201510241173 A CN 201510241173A CN 104850612 A CN104850612 A CN 104850612A
- Authority
- CN
- China
- Prior art keywords
- similarity coefficient
- hierarchical clustering
- characteristic quantity
- calculate
- daily load
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/285—Clustering or classification
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The invention relates to an enhanced cohesion hierarchical clustering-based distribution network user load feature classifying method, which is characterized by comprising steps: calculating a daily load curve characteristic quantity according to an active power curve and a reactive power curve of users; obtaining a daily load characteristic quantity set (see the specification) of N users, an enhanced damping coefficient gamma and a similar coefficient matrix P(X) among all points; forming all groups of merging routes into a merging route set Sg(s), and calculating a hierarchical clustering cohesion process by using a value iteration algorithm; obtaining a group of routes with a minimal similar coefficient value weight sum in the merging route set Sg(s). By using the clustering enhanced cohesion hierarchical clustering algorithm for the characteristic quantities, return values of all results of each layer of clustering are calculated, the cohesion merging route with the maximal return value is selected, the accuracy of the clustering algorithm is improved, defects such as a sensitive initial value, occurrence of a continuous error, and integral deviation of the result of the hierarchical clustering are avoided, and certain measures are adopted to prevent the influence of a singular value on the results.
Description
Technical field
The present invention relates to a kind of distribution customer charge tagsort method, being specifically related to a kind of distribution customer charge tagsort method based on strengthening Agglomerative Hierarchical Clustering.
Background technology
Along with the raising with electrical power services consciousness of improving of electricity market, scientific classification is carried out to power distribution network customer charge characteristic and sums up user's typical load curve, be conducive to grasping distribution network load moving law in whole region, reasonable arrangement power distribution network is dispatched, and improves the scientific validity of the work such as load prediction, dsm, tou power price, peak-clipping and valley-filling.For power distribution network safe and reliable operation and raising economic and social benefits create conditions.
Power distribution network customer charge custom and the future development of part throttle characteristics towards polynary differentiation, the factor affecting customer charge characteristic is also comparatively complicated, agricultural load is generally comparatively large by region and seasonal effect, industrial type load according to its industry characteristic, effectiveness of operation and the basic production equipment that uses different.The change of Commercial Load is comparatively large by affecting festivals or holidays and market situation, presents the bouble-bow shape of comparatively rule generally according to time variations.Residential electricity consumption part throttle characteristics is by the impact of the factor such as seasonal variations and household electrical appliance, but wherein income level is the key determining its load custom.The current classification to power consumer be all based on user when applying to install to the electricity consumption type that power supply department provides, do not reflect its sequential variation characteristic according to annual load classification is meticulous not yet, the analysis foundation of science can not be provided for power distribution network.Distribution network structure user distribution complicated and changeable is multi-point and wide-ranging, and real-time load variations noise has very large randomness more, and the method for magnanimity metric data complicate statistics is also difficult to sum up its typical case.
Summary of the invention
For the deficiencies in the prior art, the invention provides a kind of distribution customer charge tagsort method based on strengthening Agglomerative Hierarchical Clustering, reactive power load curve and active power load curve is drawn according to continuous data in marketing management system, obtain load curve and extract characteristic quantity, and cluster enhancing Agglomerative Hierarchical Clustering algorithm is carried out to characteristic quantity, setting Reward Program and transition probability, calculate the resultful return value of every strata class, choose the cohesion merge way that return value is maximum, improve the accuracy of clustering algorithm, avoid hierarchical clustering responsive to initial value, there is continuity mistake, the shortcomings such as result globality departs from, take certain measure to prevent singular value on the impact of result simultaneously.
The object of the invention is to adopt following technical proposals to realize:
Based on the distribution customer charge tagsort method strengthening Agglomerative Hierarchical Clustering, its improvements are, comprising:
Daily load curve characteristic quantity X is calculated according to the active power curves of user and reactive power curve
t, wherein T is time period window;
Obtain the daily load characteristic quantity set of N number of user
and strengthen ratio of damping γ and calculate its a little between similarity coefficient matrix P (X);
Each group merge way is formed merge way S set g (s), and utilizes Iteration algorithm to calculate the coacervation process of hierarchical clustering;
Obtain one group of path that in merge way S set g (s), similarity coefficient value weighted sum is minimum.
Preferably, the described active power curves according to user and reactive power curve calculate daily load curve characteristic quantity X
tcomprise:
Cut based on time period window T by daily load curve, the frequency acquisition of continuous data is
, then meritorious the and idle bidimensional characteristic quantity of described time period window T
for:
And calculate all period windows in a day
form described daily load curve characteristic quantity X
t; Wherein, t is the number of time period window, and Dt is the time of one day.
Preferably, the daily load characteristic quantity set of the N number of user of described acquisition
and strengthen ratio of damping γ and calculate its a little between similarity coefficient matrix P (X) comprising:
Calculate the daily load characteristic quantity set of described N number of user
a little between similarity coefficient
wherein i, j ∈ N, and represent with described matrix P (X);
Wherein,
for
with
between Manhattan distance, its computing formula is:
Wherein, n is
with
dimension,
for
a middle kth object,
for
a middle kth object.
Preferably, described by each group merge way formation merge way S set g (s), and the coacervation process utilizing Iteration algorithm to calculate hierarchical clustering comprises:
All choose the different scheme of M kind in the merging process each time of described hierarchical clustering, wherein M=sp/N, sp are the merging number of times obtaining final cluster result; Sg (s) is merging phase set of paths, comprises M
spgroup state path; Calculate the similarity coefficient weighted sum V of each group state path s
ms the formula of () is:
V
M(s)=E[R
1(P(X))+γR
2(P(X))+γ
2R
3(P(X))+...γ
sp-1R
sp(P(X))] (4)
Wherein, γ is for strengthening ratio of damping, the daily load characteristic quantity set that P (X) is N number of user
a little between similarity coefficient matrix, R
sp(P (X)) represents when merging for the sp time, chooses M value in described matrix P (X) from small to large successively;
Obtain described similarity coefficient weighted sum V
msimilarity coefficient weighted sum V maximum in (s)
*(s):
Further, described by each group merge way formation merge way S set g (s), and the coacervation process utilizing Iteration algorithm to calculate hierarchical clustering comprises:
The daily load characteristic quantity set of N number of user in selection matrix P (X)
a little between similarity coefficient
in maximum two classes merge, namely meet
two classes of i ≠ j merge;
Delete i-th of P (X), j is capable, i-th, j row, insert new row and column simultaneously, new ranks are the similarity coefficient between the new class that merges and every other cluster, and the similarity coefficient between described class and other classes equals the average similarity coefficient between two classes.
Further, described by each group merge way formation merge way S set g (s), and the coacervation process utilizing Iteration algorithm to calculate hierarchical clustering also comprises:
To V
*s () carries out value iteration until V (s) converges on V
*(s), formula is:
Wherein, under R (P (X)) represents current state, choose M value in described matrix P (X) from small to large successively; S' is upper one group of path of current path.
The beneficial effect that the present invention has:
Compared with prior art, the invention provides a kind of distribution customer charge tagsort method based on strengthening Agglomerative Hierarchical Clustering, by to the burden with power curve of user and the extraction of load or burden without work curvilinear characteristic amount, accurately can reflect the real change feature of load point, simultaneously, Agglomerative Hierarchical Clustering algorithm is strengthened by carrying out cluster to characteristic quantity, efficiently and not too much can not affect result precision by factor data noise, power distribution network typical load curve can be summed up without supervision again based on magnanimity power distribution network metric data, hold the integral load structure in region, be conducive to power department to adjust accordingly.
Accompanying drawing explanation
Fig. 1 is the method flow diagram of a kind of distribution customer charge tagsort method based on enhancing Agglomerative Hierarchical Clustering provided by the invention;
Fig. 2 is the method flow diagram of the renewal P (X) of a kind of distribution customer charge tagsort method based on enhancing Agglomerative Hierarchical Clustering provided by the invention.
Embodiment
Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in further detail.
For making the object of the embodiment of the present invention, technical scheme and advantage clearly, below in conjunction with the accompanying drawing in the embodiment of the present invention, technical scheme in the embodiment of the present invention is clearly and completely described, obviously, described embodiment is the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art, not making other embodiments all obtained under creative work prerequisite, belong to the scope of protection of the invention.
Based on the distribution customer charge tagsort method strengthening Agglomerative Hierarchical Clustering, as shown in Figure 1, comprising:
101, daily load curve characteristic quantity X is calculated according to the active power curves of user and reactive power curve
t, wherein T is time period window;
Concrete, described 101 comprise: cut based on time period window T by daily load curve, and the frequency acquisition of continuous data is
, then meritorious the and idle bidimensional characteristic quantity of described time period window T
for:
And calculate all period windows in a day
form described daily load curve characteristic quantity X
t; Wherein, t is the number of time period window, and Dt is the time of one day; This characteristic quantity cuts based on period window daily load curve, so cluster process fully can reflect the rule of a period of time internal loading time to time change.
102, the daily load characteristic quantity set of N number of user is obtained
and strengthen ratio of damping γ and calculate its a little between similarity coefficient matrix P (X);
Described 102 comprise: the daily load characteristic quantity set calculating described N number of user
a little between similarity coefficient
wherein i, j ∈ N, and represent with described matrix P (X);
Wherein,
for
with
between Manhattan distance, its computing formula is:
Wherein, n is
with
dimension,
for
a middle kth object,
for
a middle kth object; Because difference does not have squared, so the effect of outlier is weakened.
103, each group merge way is formed merge way S set g (s), and utilize Iteration algorithm to calculate the coacervation process of hierarchical clustering;
Further, described 103 comprise:
All choose the different scheme of M kind in the merging process each time of described hierarchical clustering, wherein M=sp/N, sp are the merging number of times obtaining final cluster result; Sg (s) is merging phase set of paths, comprises M
spgroup state path; Calculate the similarity coefficient weighted sum V of each group state path s
ms the formula of () is:
V
M(s)=E[R
1(P(X))+γR
2(P(X))+γ
2R
3(P(X))+...γ
sp-1R
sp(P(X))] (4)
Wherein, γ is for strengthening ratio of damping, the daily load characteristic quantity set that P (X) is N number of user
a little between similarity coefficient matrix, R
sp(P (X)) represents when merging for the sp time, chooses M value in described matrix P (X) from small to large successively;
Obtain described similarity coefficient weighted sum V
msimilarity coefficient weighted sum V maximum in (s)
*(s):
Can find out in the similarity coefficient weighted sum impact of the current nearer state of whole merge way middle distance on path larger, in the process of cluster, circulation upgrades P (X) according to the following steps, as shown in Figure 2:
The daily load characteristic quantity set of N number of user in a, selection matrix P (X)
a little between similarity coefficient
in maximum two classes merge, namely meet
two classes of i ≠ j merge;
B, delete P (X) i-th, j is capable, the i-th, j row, insert new row and column, new ranks are the similarity coefficient between the class of new merging and every other cluster, and the similarity coefficient between described class and other classes equals the average similarity coefficient between two classes simultaneously.
To V
*s () carries out value iteration until V (s) converges on V
*(s), formula is:
Wherein, under R (P (X)) represents current state, choose M value in described matrix P (X) from small to large successively; S' is upper one group of path of current path.
104, one group of path that in merge way S set g (s), similarity coefficient value weighted sum is minimum is obtained.
Finally should be noted that: above embodiment is only in order to illustrate that technical scheme of the present invention is not intended to limit; although with reference to above-described embodiment to invention has been detailed description; those of ordinary skill in the field are to be understood that: still can modify to the specific embodiment of the present invention or equivalent replacement; and not departing from any amendment of spirit and scope of the invention or equivalent replacement, it all should be encompassed within claims of the present invention.
Claims (6)
1., based on the distribution customer charge tagsort method strengthening Agglomerative Hierarchical Clustering, it is characterized in that, comprising:
Daily load curve characteristic quantity X is calculated according to the active power curves of user and reactive power curve
t, wherein T is time period window;
Obtain the daily load characteristic quantity set of N number of user
and strengthen ratio of damping γ and calculate its a little between similarity coefficient matrix P (X);
Each group merge way is formed merge way S set g (s), and utilizes Iteration algorithm to calculate the coacervation process of hierarchical clustering;
Obtain one group of path that in merge way S set g (s), similarity coefficient value weighted sum is minimum.
2. the method for claim 1, is characterized in that, the described active power curves according to user and reactive power curve calculate daily load curve characteristic quantity X
tcomprise:
Cut based on time period window T by daily load curve, the frequency acquisition of continuous data is
, then meritorious the and idle bidimensional characteristic quantity of described time period window T
for:
And calculate all period windows in a day
form described daily load curve characteristic quantity X
t; Wherein, t is the number of time period window, and Dt is the time of one day.
3. the method for claim 1, is characterized in that, the daily load characteristic quantity set of the N number of user of described acquisition
and strengthen ratio of damping γ and calculate its a little between similarity coefficient matrix P (X) comprising:
Calculate the daily load characteristic quantity set of described N number of user
a little between similarity coefficient
wherein i, j ∈ N, and represent with described matrix P (X);
Wherein,
for
with
between Manhattan distance, its computing formula is:
Wherein, n is
with
dimension,
for
a middle kth object,
for
a middle kth object.
4. the method for claim 1, is characterized in that, described by each group merge way formation merge way S set g (s), and the coacervation process utilizing Iteration algorithm to calculate hierarchical clustering comprises:
All choose the different scheme of M kind in the merging process each time of described hierarchical clustering, wherein M=sp/N, sp are the merging number of times obtaining final cluster result; Sg (s) is merging phase set of paths, comprises M
spgroup state path; Calculate the similarity coefficient weighted sum V of each group state path s
ms the formula of () is:
V
M(s)=E[R
1(P(X))+γR
2(P(X))+γ
2R
3(P(X))+...γ
sp-1R
sp(P(X))] (4)
Wherein, γ is for strengthening ratio of damping, the daily load characteristic quantity set that P (X) is N number of user
a little between similarity coefficient matrix, R
sp(P (X)) represents when merging for the sp time, chooses M value in described matrix P (X) from small to large successively;
Obtain described similarity coefficient weighted sum V
msimilarity coefficient weighted sum V maximum in (s)
*(s):
5. method as claimed in claim 4, is characterized in that, described by each group merge way formation merge way S set g (s), and the coacervation process utilizing Iteration algorithm to calculate hierarchical clustering comprises:
The daily load characteristic quantity set of N number of user in selection matrix P (X)
a little between similarity coefficient
in maximum two classes merge, namely meet
Two classes merge;
Delete i-th of P (X), j is capable, i-th, j row, insert new row and column simultaneously, new ranks are the similarity coefficient between the new class that merges and every other cluster, and the similarity coefficient between described class and other classes equals the average similarity coefficient between two classes.
6. method as claimed in claim 5, is characterized in that, described by each group merge way formation merge way S set g (s), and the coacervation process utilizing Iteration algorithm to calculate hierarchical clustering also comprises:
To V
*s () carries out value iteration until V (s) converges on V
*(s), formula is:
Wherein, under R (P (X)) represents current state, choose M value in described matrix P (X) from small to large successively; S' is upper one group of path of current path.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510241173.7A CN104850612B (en) | 2015-05-13 | 2015-05-13 | Distribution network user load characteristic classification method based on enhanced aggregation hierarchical clustering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510241173.7A CN104850612B (en) | 2015-05-13 | 2015-05-13 | Distribution network user load characteristic classification method based on enhanced aggregation hierarchical clustering |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104850612A true CN104850612A (en) | 2015-08-19 |
CN104850612B CN104850612B (en) | 2020-08-04 |
Family
ID=53850257
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510241173.7A Active CN104850612B (en) | 2015-05-13 | 2015-05-13 | Distribution network user load characteristic classification method based on enhanced aggregation hierarchical clustering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104850612B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105489216A (en) * | 2016-01-19 | 2016-04-13 | 百度在线网络技术(北京)有限公司 | Voice synthesis system optimization method and device |
CN107403247A (en) * | 2016-05-18 | 2017-11-28 | 中国电力科学研究院 | Based on the adaptive load classification polymerization analysis method for finding cluster core algorithm |
CN107657542A (en) * | 2016-07-25 | 2018-02-02 | 上海交通大学 | Public affairs become the anti-electricity-theft detecting and tracking method of taiwan area user |
CN109800766A (en) * | 2018-12-11 | 2019-05-24 | 湖北工业大学 | A kind of Novel Iterative Reconstruction Method based on cohesion hierarchical tree |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103325071A (en) * | 2013-06-09 | 2013-09-25 | 浙江大学 | User typical load curve construction method based on density clustering |
CN104200275A (en) * | 2014-06-24 | 2014-12-10 | 国家电网公司 | Power utilization mode classification and control method based on user behavior characteristics |
-
2015
- 2015-05-13 CN CN201510241173.7A patent/CN104850612B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103325071A (en) * | 2013-06-09 | 2013-09-25 | 浙江大学 | User typical load curve construction method based on density clustering |
CN104200275A (en) * | 2014-06-24 | 2014-12-10 | 国家电网公司 | Power utilization mode classification and control method based on user behavior characteristics |
Non-Patent Citations (2)
Title |
---|
LI XINWU: "Research on Text Clustering Algorithm Based on Agglomerative Hierarchical Clustering", 《第四届全国信息检索与安全内容学术会议论文集(下)》 * |
郭景峰等: "基于改进的凝聚性和分离性的层次聚类算法", 《计算机研究与发展》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105489216A (en) * | 2016-01-19 | 2016-04-13 | 百度在线网络技术(北京)有限公司 | Voice synthesis system optimization method and device |
CN105489216B (en) * | 2016-01-19 | 2020-03-03 | 百度在线网络技术(北京)有限公司 | Method and device for optimizing speech synthesis system |
CN107403247A (en) * | 2016-05-18 | 2017-11-28 | 中国电力科学研究院 | Based on the adaptive load classification polymerization analysis method for finding cluster core algorithm |
CN107657542A (en) * | 2016-07-25 | 2018-02-02 | 上海交通大学 | Public affairs become the anti-electricity-theft detecting and tracking method of taiwan area user |
CN109800766A (en) * | 2018-12-11 | 2019-05-24 | 湖北工业大学 | A kind of Novel Iterative Reconstruction Method based on cohesion hierarchical tree |
Also Published As
Publication number | Publication date |
---|---|
CN104850612B (en) | 2020-08-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhou et al. | Methodology that improves water utilization and hydropower generation without increasing flood risk in mega cascade reservoirs | |
Zhou et al. | Multi-energy net load forecasting for integrated local energy systems with heterogeneous prosumers | |
CN103729550B (en) | Multiple-model integration Flood Forecasting Method based on propagation time cluster analysis | |
CN105868853B (en) | Method for predicting short-term wind power combination probability | |
CN104715292A (en) | City short-term water consumption prediction method based on least square support vector machine model | |
CN107394809B (en) | Wind power integration system risk assessment method based on wind speed time period characteristics | |
CN105260803A (en) | Power consumption prediction method for system | |
CN106951980A (en) | A kind of multi-reservoir adaptability dispatching method based on RCP scenes | |
CN106022614A (en) | Data mining method of neural network based on nearest neighbor clustering | |
CN104376502A (en) | Electric power customer credit comprehensive evaluation method based on grey relational degree | |
CN104850612A (en) | Enhanced cohesion hierarchical clustering-based distribution network user load feature classifying method | |
CN110263984A (en) | Ultra-short term net load prediction technique based on phase space reconfiguration and deep neural network | |
CN103400204A (en) | Forecasting method for solar photovoltaic electricity generation amount based on SVM (support vector machine) - Markov combination method | |
CN103679288B (en) | Method for short-term prediction of generated output of runoff small hydropower station cluster and prediction system | |
CN104331840A (en) | Optimal electricity purchasing method for load retail dealer under electricity market environment | |
CN105005708A (en) | Generalized load characteristic clustering method based on AP clustering algorithm | |
CN106505624A (en) | Determine regulator control system and the method for power distribution network distributed power source optimum ability to arrange jobs | |
Dang et al. | A novel multivariate grey model for forecasting periodic oscillation time series | |
CN103353895A (en) | Pre-processing method of power distribution network line loss data | |
Li et al. | Optimal design and operation method of integrated energy system based on stochastic scenario iteration and energy storage attribute evaluation | |
CN105205572A (en) | Photovoltaic power generation output forecasting method based on EMD (Empirical Mode Decomposition) and Elman algorithm | |
CN108694475A (en) | Short-term time scale photovoltaic cell capable of generating power amount prediction technique based on mixed model | |
CN104182854A (en) | Mixed energy consumption measuring method for enterprise energy management system | |
CN107330548A (en) | Residential electricity consumption Forecasting Methodology based on time series models | |
CN110084430A (en) | A method of considering space-time characterisation design distributed photovoltaic power output prediction model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |