CN109861211A - A kind of power distribution network dynamic reconfiguration method based on data-driven - Google Patents
A kind of power distribution network dynamic reconfiguration method based on data-driven Download PDFInfo
- Publication number
- CN109861211A CN109861211A CN201910103428.1A CN201910103428A CN109861211A CN 109861211 A CN109861211 A CN 109861211A CN 201910103428 A CN201910103428 A CN 201910103428A CN 109861211 A CN109861211 A CN 109861211A
- Authority
- CN
- China
- Prior art keywords
- sequence
- feature
- load
- candidate
- 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.)
- Granted
Links
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of power distribution network dynamic reconfiguration method based on data-driven, extracts the symbolic feature of daily load curve, and slightly matches several historical load curves as the first candidate load sequence；Using the Time segments division method of the historical load curve most like with current daily load curve as the Time segments division method of current daily load curve；Several historical load sequences are slightly matched as candidate load sequence as the second candidate load sequence；To primitive character dimensionality reduction, the weight of feature after dimensionality reduction is determined, the second candidate load sequence is carefully matched, each period fits go out a candidate load sequence most like with current sequence；Use its static reconfiguration method as each period of current daily load curve static reconfiguration method.The present invention solves conventional method and needs artificial parameter preset and be easy the deficiency of convergence locally optimal solution, and can greatly reduce the calculating time of reconstruct, the Real time optimal dispatch especially suitable for power distribution network.
Description
Technical field
The present invention relates to distribution network reconfiguration technique field, specifically a kind of power distribution network based on data-driven is dynamic
State reconstructing method.
Background technique
Power distribution network reconfiguration is to improve power distribution network operation level by changing the open and-shut mode of interconnection switch and block switch
Effective ways, can be effectively reduced system active loss and improve node voltage to improve power quality.
Power distribution network reconfiguration is divided into static reconfiguration and two kinds of dynamic restructuring.Wherein, static reconfiguration is discontinuity surfaces at some
The state of upper optimization interconnection switch and block switch.Currently, the method for static reconfiguration mainly has Newton method, quadratic programming, people
Artificial neural networks method, interior point method, heuristic search algorithm etc..Wherein, quadratic programming is asked there are computationally intensive, poor astringency
Topic, though Newton method can fast convergence can not handle a large amount of inequality constraints in idle work optimization, artificial neural network method with
Distribution net work structure complication convergence rate it is slack-off, and artificial neural network method excessively relies on sample, and interior point method is being handled
In optimization process the problem of infeasible solution on there are obstacle, heuritic approach has that calculating speed is slow, is easy precocious defect.And
And since static reconfiguration has ignored the conditions such as load variations, switch operation constraint, it is difficult to directly apply in practical projects.
In contrast, dynamic restructuring then considers the situation of change of day part internal loading.Existing research is mainly according to one
Load curve is segmented by the situation of change of its internal loading, converts several static reconfiguration problems for dynamic restructuring problem.
In general, insufficient there are following 2 points: 1) existing dynamic restructuring segmentation method requires default segments or threshold
The parameters such as value, there are certain subjectivities, and reconstruct the restrictive conditions such as number or switch motion number and make these parameters
Reasonable value is more difficult.2) although the method for having segmentation has certain reasonability, not can guarantee is current loads water
Optimal segmentation strategy under flat.
Summary of the invention
In view of the deficiencies of the prior art, the present invention provides a kind of power distribution network dynamic reconfiguration method based on data-driven, fills
Divide and obtain the optimal control policy under current loads level using the data of history, makes up conventional method and need artificial parameter preset
Deficiency, reduce the time for determining optimal dynamic reconstruction method of power distribution network, and what can be guaranteed is current loads level
Under optimal control policy.
Present invention technical solution used for the above purpose is:
A kind of power distribution network dynamic reconfiguration method based on data-driven, comprising the following steps:
Step 1: extracting the symbolic feature of the daily load curve of power distribution network, construct classification tree, and slightly match from database
Several historical load curves are as the first candidate load sequence；
Step 2: the numerical characteristics of current daily load curve and the first candidate load sequence are extracted with stage feeding polymerization approximation method,
And the first candidate load sequence is carefully matched, obtain the historical load curve most like with current daily load curve；
Step 3: using the Time segments division method of the historical load curve most like with current daily load curve as working as the day before yesterday
The Time segments division method of load curve；
Step 4: the load average value of each node in each period is calculated, and classification tree is constructed according to symbolic feature,
Several historical load sequences are slightly matched from database as candidate load sequence as the second candidate load sequence；
Step 5: using the load average value of each node in each period as primitive character, with Principal Component Analysis pair
Primitive character dimensionality reduction determines the weight of feature after dimensionality reduction using entropy assessment, the second candidate load sequence is carefully matched, each
Period fits go out a candidate load sequence most like with current sequence；
Step 6: static reconfiguration method gone out with each period fits and the most like candidate load sequence of current sequence
Static reconfiguration method.
The classification tree is the tree structure of the symbolic feature comprising load curve.
The database is the historical data comprising several days, wherein including: symbolic feature in daily historical data；Match
The daily load curve of power grid and corresponding Time segments division method；The load average value of each node in each period, i.e., each
The primitive character of period；Each period corresponding static reconfiguration method.
The symbolic feature includes festivals or holidays, weather, season and topological structure.
The thick matching are as follows: deleted from several historical series of database and select sequence identical with current sequence symbolic feature
Column, as the first candidate load sequence.
The numerical characteristics that current daily load curve is extracted with stage feeding polymerization approximation method, comprising:
Original loads sequence is X={ x1,x2,…,xn, it is by the numerical characteristics that stage feeding polymerization approximation method is extracted: Y=
{y1,y2,...,ym}；Wherein, n represents the number of element in original loads sequence, and m indicates the number of feature in numerical characteristics, warp
M is less than n after crossing the processing of stage feeding polymerization approximation method, and n can be divided exactly by m；It enablesThen element can indicate in Y are as follows:Wherein yiIndicate i-th of the element of feature extracted；xjIndicate j-th yuan of original loads sequence
Element.
The thin matching are as follows: calculate the dynamic time warpings of the numerical characteristics of current loads sequence and candidate load sequence away from
From the corresponding candidate load sequence of the minimum dynamic time warping distance of output.
Wherein, the thin matching of step 2 are as follows: calculate the dynamic of the numerical characteristics of current loads sequence and the first candidate load sequence
State Time Warp distance exports the corresponding candidate load sequence of minimum dynamic time warping distance.The thin matching of step 5 are as follows: meter
The dynamic time warping distance for calculating the numerical characteristics of current loads sequence and the second candidate load sequence, exports minimum dynamic time
The corresponding candidate load sequence of deflection distance.
It is described with Principal Component Analysis to primitive character dimensionality reduction, comprising:
Calculate the correlation coefficient r between feature i and feature jij, the related coefficient between p feature constitutes related coefficient square
Battle array R=(rij)p×p, wherein correlation coefficient rijCalculation formula it is as follows:
Wherein,It is the mean value of feature i,It is the mean value of feature i, N is the sum of the second candidate load sequence, x'kiTable
Show i-th of element in k-th second candidate load sequences.x'kjIndicate j-th of element in k-th second candidate load sequences, p
It is the number of feature before dimensionality reduction in the second candidate load sequence.
Find out the eigenvalue λ of correlation matrixi(i=1,2...p), by each characteristic value according to sequence from big to small
Arrangement, and calculate the corresponding feature vector a of each characteristic valuei(i=1,2...p)；
The accumulation contribution rate of M principal component before calculating:
Wherein, αMThe accumulation contribution rate of M principal component before indicating；λiIt is the ith feature by after drained sequence from big to small
Value, λkIt is by after drained sequence from big to small, k-th of characteristic value, p is the number of element in the second candidate load sequence, i.e., former
The number of beginning feature；If preceding Nx- 1 accumulation contribution rate is less than 85%, and preceding NxWhen a accumulation contribution rate is greater than or equal to 85%,
Then the number of new feature is exactly Nx；
I-th of element z in j-th second candidate load sequencesi(i-th of element of new feature) can indicate are as follows:
zi=ai1x'j1+ai2x'j2+…aipx'jp,1≤i≤Nx
Wherein, aipIndicate that p-th of element of ith feature value character pair vector, p are dropped in the second candidate load sequence
The number of feature, x' before tieing upjpIt is p-th of element in j-th second candidate load sequences.
The weight that feature after dimensionality reduction is determined using entropy assessment, comprising:
The standardized method of feature is as follows:
The comentropy of each feature are as follows:
The weight w of j-th of featurejIt indicates are as follows:
Wherein, zijIt is that j-th of element, N are the second candidate load sequences in i-th second candidate load sequences after dimensionality reduction
Sum, NxIt is the number of new feature.
The invention has the following beneficial effects and advantage:
The present invention makes full use of the data of history to handle, and finds out the control for meeting the history power distribution network reconfiguration of present case
The operation of scheme commander current power distribution network processed solves conventional method and needs artificial parameter preset and be easy convergence locally optimal solution
Deficiency, and can greatly reduce the calculating time of reconstruct, the Real time optimal dispatch especially suitable for power distribution network.
Detailed description of the invention
Fig. 1 is power distribution network dynamic restructuring flow chart of the invention；
Fig. 2 is thick matching and thin matched flow chart；
Fig. 3 is the concrete structure diagram of classification tree；
Fig. 4 is embodiment power distribution network topology diagram；
Fig. 5 is emulation Time segments division result figure.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and embodiments.
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing to the present invention
Specific embodiment be described in detail.Many details are explained in the following description in order to fully understand this hair
It is bright.But the present invention can be implemented in many other ways than those described herein, and those skilled in the art can not disobey
Similar improvement is done in the case where back invention intension, therefore the present invention is not limited to the specific embodiments disclosed below.
Unless otherwise defined, all technical and scientific terms used herein and belong to technical field of the invention
The normally understood meaning of technical staff is identical.It is specific that description is intended merely in the term used in the description of invention herein
Embodiment purpose, it is not intended that in limitation the present invention.
It is as shown in Figure 1 power distribution network dynamic restructuring flow chart.
A kind of power distribution network dynamic reconfiguration method based on data-driven, comprising the following steps:
Step 1: extracting the symbolic feature of the daily load curve of power distribution network, the symbolic feature of this patent includes festivals or holidays, day
Gas, season and topological structure construct classification tree as shown in figure 3, and slightly matching several historical load curve conducts from database
First candidate load sequence；The power output of load is influenced by factors such as season, festivals or holidays and weather, under same scene
Load power output has certain similitude.Symbolic feature includes: festivals or holidays, weather, season and topological structure.As at the beginning of classification tree
Step selects the historical load sequence number similar with current loads sequence, this number is often larger, calculates to further decrease
Amount, carries out postsearch screening by given threshold.By given threshold come postsearch screening candidate's load sequence.Characteristic value is fallen in fair
Perhaps the candidate load sequence outside range excludes, to further reduce the number of candidate load sequence.
Database should contain n days accounts of the history, include daily in have:
1) symbolic feature: festivals or holidays, weather, the particular content in season and topological structure.
2) strategy of the daily load curve of power distribution network and corresponding Time segments division.
3) in each period each node load average value, i.e., the primitive character of each period.
4) corresponding static reconfiguration scheme of each period.
Step 2: the number of the current daily load curve of power distribution network and the first candidate load sequence is extracted with stage feeding polymerization approximation method
Value tag, comprising:
If original loads sequence is X={ x1,x2,…,xn, it is by the numerical characteristics that stage feeding polymerization approximation method is extracted: Y
={ y1,y2,...,ym}.Wherein, n represents the number of element in original loads sequence, and m indicates the number of feature in numerical characteristics,
M is bound to be less than n after the processing of stage feeding polymerization approximation method, and n can be divided exactly by m.It enablesThen element can indicate in Y
Are as follows:The original loads sequence transitions for being n length are realized by stage feeding polymerization approximation method
The numerical characteristics for being m at length, realize the process of dimensionality reduction.
Wherein, the current daily load curve of power distribution network is pre- using power distribution network history daily load curve data progress day preload
It measures.Then, the first candidate load sequence is carefully matched, is obtained negative with the most like history of current daily load curve
Lotus curve；The numerical characteristics of current daily load curve are extracted with stage feeding polymerization approximation method, every section of curve can indicate are as follows:And most like historical load curve is carefully matched from candidate load sequence.Thick matching
It is as shown in Figure 2 with thin matched flow chart.
Thick matching are as follows: assuming that a total of n historical series in database, wherein having the festivals or holidays of m historical series, day
Gas, season and topological structure are identical with current sequence, then this m historical series is just chosen as " the first candidate load sequence
Column ", this process are known as thick matching.
Thin matching are as follows: calculate separately the dynamic time warpings of the numerical characteristics of current loads sequence and candidate load sequence away from
From that corresponding candidate load sequence of the minimum dynamic time warping distance of output.
Step 3: using the Time segments division strategy with the most like historical load curve of current daily load curve, this period
Partition strategy is in current daily load curve.Wherein, the Time segments division policy-source of historical load curve is in database.It uses
Time segments division scheme of the dynamic bending apart from the smallest historical load curve.Assuming that the feature of two load sequences be respectively X and
Y, characteristic length are successively m and n, and two features are sorted according to time location, construct the distance matrix of m row n, it may be assumed that
d(xi,yj) indicate different characteristic sequence data object between element Euclidean distance, in distance matrix, often
The collection of one group of adjacent matrix elements is collectively referred to as crooked route, can be denoted as W=w1,w2...wk….Dynamic time warping distance is
Keep crooked route total length minimum, can be expressed asMost from point (1,1) point (m, n)
Short path can be acquired by dynamic programming.
Step 4: finding out the load average value of each node in each period, the number (load of element in load sequence
The length of sequence) it is equal to the number of power distribution network node.And several historical loads are slightly matched from database according to symbolic feature
Sequence is as candidate load sequence as the second candidate load sequence.Wherein, here symbolic feature and load characteristic above
It is consistent, it is still desirable to construct classification tree.
Step 5: finding out the average value of the load of each node in each period as primitive character, use principal component analysis
Method determines the weight of feature after dimensionality reduction using entropy assessment to primitive character dimensionality reduction.By the second candidate load sequence and current loads
After the thin matching operation of sequence progress and export a historical series.False current daily load curve is divided into n sections by step 3,
Each period can export a historical series, that is, export n historical series.It is illustrated in figure 5 emulation Time segments division result figure.
Firstly, using the load of each node as primitive character, with Principal Component Analysis to Feature Dimension Reduction.It specifically includes:
Calculate the correlation coefficient r between feature i and feature jij, the related coefficient between p feature constitutes correlation matrix R=
(rij)p×p, wherein correlation coefficient rijCalculation formula it is as follows:
Finding eigenvalue and eigenvector.Find out the eigenvalue λ of correlation matrixi(i=1,2...p) by each characteristic value
According to sequence arrangement from big to small, and calculate the corresponding feature vector a of each characteristic valuei(i=1,2...p).Calculate accumulation
Contribution rate and new feature.In order to guarantee that prime information loss is few as far as possible, the principal component number of selection should to accumulate contribution rate
Greater than 85%, the accumulation contribution rate calculation formula of preceding n principal component is as follows:αnN principal component before indicating
Accumulation contribution rate.I-th of new feature can be expressed as follows: zi=ai1x′1+ai2x′2+…aipx′p。
Secondly, determining the weight of feature using entropy assessment.Assuming that after Principal Component Analysis dimensionality reduction, m historical load
N new features of sequence constitute evaluations matrix, and the standardized method of feature is as follows:The comentropy of each feature
Are as follows:Greatly, weight also should be bigger.The weight w of j-th of featurejIt can indicate are as follows:Weight is bigger to indicate that this feature is bigger to the effect for calculating similitude, intuitively and effectively reacts
The difference degree of each feature.Finally, carefully being matched and the most like candidate load of current loads sequence from candidate load sequence
Sequence.
Step 6: the static reconfiguration scheme of historical series has record in the database.At n of current daily load curve
The corresponding static reconfiguration scheme of n historical series that the static reconfiguration scheme of section is just chosen using step 5.Using dynamic bending away from
Control strategy of the static reconfiguration scheme as each period from the smallest load sequence.
The present invention is further illustrated by taking IEEE33 Node power distribution system as an example, topological structure is shown in Fig. 4.The example thus of table 1
Load data, the other data of system are identical as IEEE33 node example, repeat no more.If the action frequency of each switch is most
It is 3 times, total action frequency of all switches is up to 15 times, and switch operating cost is 7 yuan/time, and electricity price is 0.7 yuan/kWh.
Finally it is divided into 6 sections by optimizing the load curve, segmentation result is shown in Fig. 5, substantially conforms to as can be seen from the results
The variation tendency of curve.
In order to verify based on load similitude matching Time segments division reasonability and correctness, using following 4 kinds of schemes into
Row emulation.Option A: without reconstruct.Option b: the dynamic restructuring based on load curve monotonicity Time segments division.Scheme C: it is based on
The dynamic restructuring of information on load entropy Time segments division.Scheme D: the dynamic restructuring based on data-driven.
Tables 1 and 2 is IEEE33 node calculated result.
Table 1
Table 2
| Algorithm | Network loss mean value/kW | Time-consuming/second | Accuracy rate |
| The present invention | 139.5 | 0.3 | 97% |
| Genetic algorithm | 143.0 | 41.3 | 65% |
| Particle swarm algorithm | 146.4 | 46.7 | 45% |
| Artificial bee colony algorithm | 144.6 | 58.9 | 64% |
| Simulated annealing | 148.5 | 51.2 | 33% |
Calculated result shows the optimizing by the way of random search due to traditional algorithm, causes to calculate overlong time, it is difficult to
Meet the requirement calculated in real time.Lacking for local optimum is easily trapped into addition, can be seen that from the accuracy rate of each traditional algorithm
It falls into.In contrast, the calculating time based on the matched static reconfiguration of load similitude is far smaller than traditional algorithm, can not only use
In offline, the requirement in line computation real-time can also be met.The accuracy rate of calculating is also higher than traditional algorithm, this illustrates this hair
Bright correctness and validity.
Claims (9)
1. a kind of power distribution network dynamic reconfiguration method based on data-driven, which comprises the following steps:
Step 1: extracting the symbolic feature of the daily load curve of power distribution network, construct classification tree, and slightly match from database several
Historical load curve is as the first candidate load sequence；
Step 2: the numerical characteristics of current daily load curve and the first candidate load sequence are extracted with stage feeding polymerization approximation method, and right
First candidate load sequence is carefully matched, and the historical load curve most like with current daily load curve is obtained；
Step 3: using the Time segments division method of the historical load curve most like with current daily load curve as current daily load
The Time segments division method of curve；
Step 4: calculating the load average value of each node in each period, and classification tree is constructed according to symbolic feature, from number
It is used as candidate load sequence as the second candidate load sequence according to several historical load sequences are slightly matched in library；
Step 5: using the load average value of each node in each period as primitive character, with Principal Component Analysis to original
Feature Dimension Reduction determines the weight of feature after dimensionality reduction using entropy assessment, the second candidate load sequence is carefully matched, each period
Match a candidate load sequence most like with current sequence；
Step 6: use each period fits go out and the most like candidate load sequence of current sequence static reconfiguration method as
The static reconfiguration method of each period of current daily load curve.
2. the power distribution network dynamic reconfiguration method according to claim 1 based on data-driven, it is characterised in that: the classification
Tree is the tree structure of the symbolic feature comprising load curve.
3. the power distribution network dynamic reconfiguration method according to claim 1 based on data-driven, it is characterised in that: the data
Library is the historical data comprising several days, wherein including: symbolic feature in daily historical data；The daily load curve of power distribution network
And corresponding Time segments division method；The load average value of each node in each period, i.e., the primitive character of each period；Often
A period corresponding static reconfiguration method.
4. described in any item power distribution network dynamic reconfiguration methods based on data-driven, feature exist according to claim 1~3
In: the symbolic feature includes festivals or holidays, weather, season and topological structure.
5. the power distribution network dynamic reconfiguration method according to claim 1 based on data-driven, it is characterised in that: described thick
Match are as follows: deleted from several historical series of database and select sequence identical with current sequence symbolic feature, as the first candidate
Load sequence.
6. the power distribution network dynamic reconfiguration method according to claim 1 based on data-driven, it is characterised in that: described use is divided
Duan Juhe approximation method extracts the numerical characteristics of current daily load curve, comprising:
Original loads sequence is X={ x1,x2,…,xn, it is by the numerical characteristics that stage feeding polymerization approximation method is extracted: Y={ y1,
y2,…,ym}；Wherein, n represents the number of element in original loads sequence, and m indicates the number of feature in numerical characteristics, through excessive
M is less than n after the processing of Duan Juhe approximation method, and n can be divided exactly by m；It enablesThen element can indicate in Y are as follows:Wherein yiIndicate i-th of the element of feature extracted；xjIndicate j-th yuan of original loads sequence
Element.
7. the power distribution network dynamic reconfiguration method according to claim 1 based on data-driven, it is characterised in that: described thin
Match are as follows: the dynamic time warping distance for calculating the numerical characteristics of current loads sequence and candidate load sequence exports minimum dynamic
Time Warp is apart from corresponding candidate load sequence.
8. the power distribution network dynamic reconfiguration method according to claim 1 based on data-driven, it is characterised in that: described with master
Componential analysis is to primitive character dimensionality reduction, comprising:
Calculate the correlation coefficient r between feature i and feature jij, the related coefficient between p feature constitutes correlation matrix R=
(rij)p×p, wherein correlation coefficient rijCalculation formula it is as follows:
Wherein,It is the mean value of feature i,It is the mean value of feature i, N is the sum of the second candidate load sequence, x'kiIndicate kth
I-th of element in a second candidate load sequence, x'kjIndicate j-th of element in k-th second candidate load sequences, p is second
In candidate load sequence before dimensionality reduction feature number；
Find out the eigenvalue λ of correlation matrixi(i=1,2 ... p), and each characteristic value is arranged according to sequence from big to small,
And calculate the corresponding feature vector a of each characteristic valuei(i=1,2 ... p)；
The accumulation contribution rate of M principal component before calculating:
Wherein, αMThe accumulation contribution rate of M principal component before indicating；λiIt is the ith feature value by after drained sequence from big to small, λk
It is by after drained sequence from big to small, k-th of characteristic value, p is the number of element in the second candidate load sequence, i.e. primitive character
Number；If preceding Nx- 1 accumulation contribution rate is less than 85%, and preceding NxIt is when a accumulation contribution rate is greater than or equal to 85%, then new special
The number of sign is exactly Nx；
I-th of element z in j-th second candidate load sequencesiIt indicates are as follows:
zi=ai1x'j1+ai2x'j2+…aipx'jp,1≤i≤Nx
Wherein, aipIndicate p-th of element of ith feature value character pair vector, p is in the second candidate load sequence before dimensionality reduction
The number of feature, x'jpIt is p-th of element in j-th second candidate load sequences.
9. the power distribution network dynamic reconfiguration method according to claim 1 based on data-driven, it is characterised in that: the utilization
Entropy assessment determines the weight of feature after dimensionality reduction, comprising:
The standardized method of feature is as follows:
The comentropy of each feature are as follows:
The weight w of j-th of featurejIt indicates are as follows:
Wherein, zijIt is that j-th of element, N are the total of the second candidate load sequence in i-th second candidate load sequences after dimensionality reduction
Number, NxIt is the number of new feature.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910103428.1A CN109861211B (en) | 2019-02-01 | 2019-02-01 | Dynamic reconfiguration method of power distribution network based on data driving |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910103428.1A CN109861211B (en) | 2019-02-01 | 2019-02-01 | Dynamic reconfiguration method of power distribution network based on data driving |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109861211A true CN109861211A (en) | 2019-06-07 |
| CN109861211B CN109861211B (en) | 2020-08-14 |
Family
ID=66897440
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910103428.1A Active CN109861211B (en) | 2019-02-01 | 2019-02-01 | Dynamic reconfiguration method of power distribution network based on data driving |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109861211B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111026741A (en) * | 2019-12-04 | 2020-04-17 | 国网天津市电力公司电力科学研究院 | Data cleaning method and device based on time series similarity |
| CN111144643A (en) * | 2019-12-24 | 2020-05-12 | 天津相和电气科技有限公司 | Day-ahead power load prediction method and device based on double-end automatic coding |
| CN111337742A (en) * | 2020-02-25 | 2020-06-26 | 广东电网有限责任公司 | Distribution network short-term load prediction data acquisition equipment |
| CN113447740A (en) * | 2021-06-21 | 2021-09-28 | 天津大学 | Non-invasive load event global optimization matching method and system |
| CN113761023A (en) * | 2021-08-24 | 2021-12-07 | 国网甘肃省电力公司 | Photovoltaic power generation short-term power prediction method based on improved generalized neural network |
| CN116339153A (en) * | 2023-05-22 | 2023-06-27 | 科大智能物联技术股份有限公司 | Lime kiln control method based on particle swarm optimization |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012103904A2 (en) * | 2011-02-05 | 2012-08-09 | Maschinenfabrik Reinhausen Gmbh | Method and device for determining the load flow in a power supply network |
| CN105719048A (en) * | 2016-01-05 | 2016-06-29 | 国网上海市电力公司 | Intermediate-voltage distribution operation state fuzzy integrated evaluation method based on principle component analysis method and entropy weight method |
| CN105990833A (en) * | 2015-01-29 | 2016-10-05 | 国家电网公司 | Power distribution network dynamic reconstruction method based on operation scene matching |
| CN107359614A (en) * | 2017-09-04 | 2017-11-17 | 天津大学 | Power distribution network safe operation dynamic reconfiguration method under the conditions of high permeability regenerative resource |
-
2019
- 2019-02-01 CN CN201910103428.1A patent/CN109861211B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012103904A2 (en) * | 2011-02-05 | 2012-08-09 | Maschinenfabrik Reinhausen Gmbh | Method and device for determining the load flow in a power supply network |
| CN105990833A (en) * | 2015-01-29 | 2016-10-05 | 国家电网公司 | Power distribution network dynamic reconstruction method based on operation scene matching |
| CN105719048A (en) * | 2016-01-05 | 2016-06-29 | 国网上海市电力公司 | Intermediate-voltage distribution operation state fuzzy integrated evaluation method based on principle component analysis method and entropy weight method |
| CN107359614A (en) * | 2017-09-04 | 2017-11-17 | 天津大学 | Power distribution network safe operation dynamic reconfiguration method under the conditions of high permeability regenerative resource |
Non-Patent Citations (2)
| Title |
|---|
| YANG HUPING: "Gradual Approaching Method for Distribution Network Dynamic", 《2008 WORKSHOP ON POWER ELECTRONICS AND INTELLIGENT TRANSPORTATION SYSTEM》 * |
| 刘吉刚 等: "计及风电光伏相关性的多目标配电网重构", 《国网技术学院学报》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111026741A (en) * | 2019-12-04 | 2020-04-17 | 国网天津市电力公司电力科学研究院 | Data cleaning method and device based on time series similarity |
| CN111144643A (en) * | 2019-12-24 | 2020-05-12 | 天津相和电气科技有限公司 | Day-ahead power load prediction method and device based on double-end automatic coding |
| CN111337742A (en) * | 2020-02-25 | 2020-06-26 | 广东电网有限责任公司 | Distribution network short-term load prediction data acquisition equipment |
| CN113447740A (en) * | 2021-06-21 | 2021-09-28 | 天津大学 | Non-invasive load event global optimization matching method and system |
| CN113761023A (en) * | 2021-08-24 | 2021-12-07 | 国网甘肃省电力公司 | Photovoltaic power generation short-term power prediction method based on improved generalized neural network |
| CN116339153A (en) * | 2023-05-22 | 2023-06-27 | 科大智能物联技术股份有限公司 | Lime kiln control method based on particle swarm optimization |
| CN116339153B (en) * | 2023-05-22 | 2023-09-01 | 科大智能物联技术股份有限公司 | Lime Kiln Control Method Based on Particle Swarm Optimization |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109861211B (en) | 2020-08-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109861211A (en) | A kind of power distribution network dynamic reconfiguration method based on data-driven | |
| CN109376772B (en) | Power load combination prediction method based on neural network model | |
| Huang et al. | A clustering based grouping method of nearly zero energy buildings for performance improvements | |
| CN106251001A (en) | A kind of based on the photovoltaic power Forecasting Methodology improving fuzzy clustering algorithm | |
| CN104809658B (en) | A kind of rapid analysis method of low-voltage distribution network taiwan area line loss | |
| CN110837915B (en) | Low-voltage load point prediction and probability prediction method for power system based on hybrid integrated deep learning | |
| CN111525587B (en) | Reactive load situation-based power grid reactive voltage control method and system | |
| CN110163429A (en) | A kind of short-term load forecasting method based on similar day optimal screening | |
| CN109146252A (en) | Based on the user power utilization behavior clustering method and system for adjusting potentiality index | |
| CN108564592A (en) | Based on a variety of image partition methods for being clustered to differential evolution algorithm of dynamic | |
| CN111461921B (en) | Load modeling typical user database updating method based on machine learning | |
| CN113762387B (en) | Multi-element load prediction method for data center station based on hybrid model prediction | |
| Wang et al. | Big data analytics for price forecasting in smart grids | |
| Wang et al. | Short-term load forecasting with LSTM based ensemble learning | |
| Varghese et al. | Smart grid consumer behavioral model using machine learning | |
| Sheng et al. | An optimized prediction algorithm based on XGBoost | |
| Prakash et al. | Short-Term Load Forcasting for Smart Power Systems Using Swarm Intelligence Algorithm | |
| CN111724064A (en) | Energy-storage-containing power distribution network planning method based on improved immune algorithm | |
| CN111652459A (en) | Power grid evaluation method considering new energy consumption multi-index connotation | |
| CN116415732A (en) | User side power load data processing method based on improved ARNN | |
| CN113657687B (en) | Power load prediction method based on feature engineering and multipath deep learning | |
| CN113837486A (en) | RNN-RBM-based distribution network feeder long-term load prediction method | |
| Liu et al. | Short-Term Power Load Forecasting via Recurrent Neural Network with Similar Day Selection | |
| CN114064794A (en) | Business expansion file mining and analyzing method based on big data technology | |
| Gao et al. | Daily power load curves analysis based on grey wolf optimization clustering algorithm |
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 |