CN109861211B - Dynamic reconfiguration method of power distribution network based on data driving - Google Patents
Dynamic reconfiguration method of power distribution network based on data driving Download PDFInfo
- Publication number
- CN109861211B CN109861211B CN201910103428.1A CN201910103428A CN109861211B CN 109861211 B CN109861211 B CN 109861211B CN 201910103428 A CN201910103428 A CN 201910103428A CN 109861211 B CN109861211 B CN 109861211B
- Authority
- CN
- China
- Prior art keywords
- load
- sequence
- candidate
- power distribution
- distribution network
- 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.)
- Active
Links
Images
Abstract
The invention relates to a dynamic reconstruction method of a power distribution network based on data driving, which comprises the steps of extracting symbolic characteristics of daily load curves, and roughly matching a plurality of historical load curves to serve as a first candidate load sequence; taking the time interval division method of the historical load curve most similar to the current daily load curve as the time interval division method of the current daily load curve; roughly matching a plurality of historical load sequences as candidate load sequences as second candidate load sequences; reducing the dimension of the original features, determining the weight of the features after the dimension reduction, performing fine matching on the second candidate load sequence, and matching a candidate load sequence which is most similar to the current sequence in each time period; and using the static reconstruction method as the static reconstruction method of each time interval of the current daily load curve. The method overcomes the defects that the traditional method needs manual parameter presetting and is easy to converge the local optimal solution, can greatly shorten the calculation time of reconstruction, and is particularly suitable for real-time optimization scheduling of the power distribution network.
Description
Technical Field
The invention relates to the technical field of power distribution network reconstruction, in particular to a data-driven power distribution network dynamic reconstruction method.
Background
The power distribution network reconstruction is an effective method for improving the operation level of the power distribution network by changing the opening and closing states of the interconnection switch and the section switch, and can effectively reduce the active loss of a system and improve the node voltage so as to improve the power quality.
The power distribution network reconstruction comprises static reconstruction and dynamic reconstruction. Wherein, the static reconfiguration is to optimize the states of the tie switch and the section switch on a certain time section. At present, the static reconstruction methods mainly include a newton method, a quadratic programming method, an artificial neural network method, an interior point method, a heuristic search algorithm, and the like. The quadratic programming method has the problems of large calculated amount and poor convergence, the Newton method can quickly converge but cannot process a large amount of inequality constraints in reactive power optimization, the convergence speed of the artificial neural network method becomes slow along with the complexity of the structure of the power distribution network, the artificial neural network method excessively depends on samples, the inner point method has obstacles on the problem that the inner point method cannot be solved in the process of processing optimization, and the heuristic algorithm has the defects of low calculation speed and easiness in precocity. Moreover, since the static reconfiguration ignores conditions such as load change and switch operation constraint, the method is difficult to be directly applied to actual engineering.
In contrast, the dynamic reconfiguration takes into account the changing conditions of the load in each time interval. The existing research mainly segments the load curve according to the change condition of the load in one day, and converts the dynamic reconstruction problem into a plurality of static reconstruction problems.
Overall, there are the following 2-point deficiencies: 1) the existing dynamic reconstruction segmentation method needs to preset parameters such as the number of segments or a threshold value, has certain subjectivity, and the reasonable values of the parameters are more difficult due to the limitation conditions such as the reconstruction times or the switching action times. 2) Although the existing segmentation method has certain rationality, the existing segmentation method cannot ensure the optimal segmentation strategy under the current load level.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a data-driven power distribution network dynamic reconstruction method, which fully utilizes historical data to obtain an optimal control strategy under the current load level, makes up the defect that the traditional method needs manual preset parameters, reduces the time for determining the optimal dynamic power distribution network reconstruction method, and can ensure that the optimal control strategy under the current load level is obtained.
The technical scheme adopted by the invention for realizing the purpose is as follows:
a dynamic reconstruction method of a power distribution network based on data driving comprises the following steps:
step 1: extracting the symbolic features of daily load curves of the power distribution network, constructing a classification tree, and roughly matching a plurality of historical load curves from a database to serve as a first candidate load sequence;
step 2: extracting numerical characteristics of the current daily load curve and the first candidate load sequence by using a piecewise aggregation approximation method, and performing fine matching on the first candidate load sequence to obtain a historical load curve most similar to the current daily load curve;
and step 3: taking the time interval division method of the historical load curve most similar to the current daily load curve as the time interval division method of the current daily load curve;
and 4, step 4: calculating the load average value of each node in each time interval, constructing a classification tree according to the symbolic features, and roughly matching a plurality of historical load sequences from a database to serve as candidate load sequences to serve as second candidate load sequences;
and 5: taking the load average value of each node in each time interval as an original feature, reducing the dimension of the original feature by using a principal component analysis method, determining the weight of the feature after dimension reduction by using an entropy weight method, and performing fine matching on a second candidate load sequence, wherein a candidate load sequence most similar to the current sequence is matched in each time interval;
step 6: and using the static reconstruction method of the candidate load sequence which is matched in each time interval and is most similar to the current sequence as the static reconstruction method of each time interval of the current daily load curve.
The classification tree is a tree structure containing symbolic features of load curves.
The database comprises historical data of a plurality of days, wherein the historical data of each day comprises: a symbolic feature; daily load curves of the power distribution network and a corresponding time interval division method; the average load value of each node in each time interval, namely the original characteristic of each time interval; and a static reconstruction method corresponding to each time interval.
The symbolic features include holidays, weather, seasons, and topology.
The coarse matching is as follows: and deleting a sequence with the same symbol characteristics as the current sequence from a plurality of historical sequences of the database to serve as a first candidate load sequence.
The method for extracting the numerical characteristics of the current daily load curve by using the piecewise aggregation approximation method comprises the following steps:
the original payload sequence is X ═ X1,x2,…,xnThe numerical characteristics extracted by the piecewise aggregation approximation method are as follows: y ═ Y1,y2,...,ym}; wherein n represents the number of elements in the original load sequence, m represents the number of features in the numerical characteristics, m is smaller than n after being processed by a piecewise aggregation approximation method, and n can be divided by m; order to
Then the elements in Y can be represented as:
wherein y isiThe ith element representing the extracted feature; x is the number ofjRepresenting the jth element of the original payload sequence.
The fine matching is as follows: and calculating the dynamic time bending distance of the numerical characteristics of the current load sequence and the candidate load sequence, and outputting the candidate load sequence corresponding to the minimum dynamic time bending distance.
Wherein, the fine matching of the step 2 is as follows: and calculating the dynamic time bending distance of the numerical characteristics of the current load sequence and the first candidate load sequence, and outputting the candidate load sequence corresponding to the minimum dynamic time bending distance. The fine matching of step 5 is: and calculating the dynamic time bending distance of the numerical characteristics of the current load sequence and the second candidate load sequence, and outputting the candidate load sequence corresponding to the minimum dynamic time bending distance.
The method for reducing the dimension of the original features by using a principal component analysis method comprises the following steps:
calculating a correlation coefficient r between the feature i and the feature jijThe correlation coefficients between the p features form a correlation coefficient matrix R ═ (R)ij)p×pWherein the correlation coefficient rijThe calculation formula of (a) is as follows:
wherein the content of the first and second substances,
is the mean value of the features i and,
is the mean of feature i, N is the total number of second candidate payload sequences, x'kiRepresenting the ith element in the kth second candidate payload sequence. x'kjAnd p is the number of the features before dimensionality reduction in the second candidate load sequence.
Determining the eigenvalues lambda of the correlation coefficient matrixi(i-1, 2.. p), arranging the characteristic values in descending order, and calculating the characteristic vector a corresponding to each characteristic valuei(i＝1,2...p)；
Calculating the cumulative contribution rate of the first M principal components:
wherein, αMRepresenting the cumulative contribution rate of the first M principal components; lambda [ alpha ]iAfter finishing the ranking from big to small, the ith characteristic value, lambdakAfter the order is arranged from big to small, the kth characteristic value is obtained, and p is the number of elements in the second candidate load sequence, namely the number of original characteristics; if the first Nx-1 cumulative contribution rate less than 85%, and the first NxWhen the cumulative contribution rate is greater than or equal to 85%, the number of the new features is Nx；
Ith element z in jth second candidate payload sequencei(new feature ith element) can be expressed as:
zi＝ai1x'j1+ai2x'j2+…aipx'jp,1≤i≤Nx
wherein, aipP is the number of the features before dimensionality reduction in the second candidate load sequence, x'jpIs the p-th element in the jth second candidate payload sequence.
The method for determining the weight of the feature after dimensionality reduction by using the entropy weight method comprises the following steps:
the normalization method of the features is as follows:
the information entropy of each feature is:
weight w of jth featurejExpressed as:
wherein z isijIs the jth element in the ith second candidate load sequence after dimensionality reduction, N is the total number of the second candidate load sequences, NxIs the number of new features.
The invention has the following beneficial effects and advantages:
the method and the device fully utilize historical data to process, find out the control scheme of the reconstruction of the historical power distribution network according with the current situation to command the operation of the current power distribution network, solve the defects that the traditional method needs manual preset parameters and is easy to converge local optimal solutions, can greatly shorten the calculation time of reconstruction, and are particularly suitable for real-time optimal scheduling of the power distribution network.
Drawings
FIG. 1 is a flow chart of the dynamic reconfiguration of a power distribution network of the present invention;
FIG. 2 is a flow chart of coarse and fine matching;
FIG. 3 is a detailed block diagram of a classification tree;
FIG. 4 is a diagram of an embodiment distribution network topology;
fig. 5 is a simulation time period division result diagram.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein, but rather should be construed as modified in the spirit and scope of the present invention as set forth in the appended claims.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
Fig. 1 shows a flow chart of dynamic reconfiguration of a power distribution network.
A dynamic reconstruction method of a power distribution network based on data driving comprises the following steps:
step 1: extracting symbolic features of daily load curves of the power distribution network, wherein the symbolic features of the method comprise holidays, weather, seasons and topological structures, constructing a classification tree as shown in figure 3, and roughly matching a plurality of historical load curves from a database to serve as a first candidate load sequence; the load output is influenced by factors such as seasons, festivals and holidays, weather and the like, and the load output has certain similarity under the same scene. The symbolic features include: holidays, weather, seasons, and topology. And (3) preliminarily selecting the historical load sequence number similar to the current load sequence from the classification tree, wherein the number is usually larger, and in order to further reduce the calculated amount, secondary screening is carried out by setting a threshold value. And secondarily screening candidate load sequences by setting a threshold value. And excluding the candidate load sequences with the characteristic values falling out of the allowable range, thereby further reducing the number of the candidate load sequences.
The database should contain n days of history, including for each day:
1) the symbol characteristics are as follows: holidays, weather, seasons and topological structure.
2) Daily load curve of the power distribution network and corresponding strategy of time interval division.
3) The average value of the loads of the nodes in each time interval is the original characteristic of each time interval.
4) A static reconstruction scheme for each time interval.
Step 2: extracting the numerical characteristics of the current daily load curve and the first candidate load sequence of the power distribution network by using a piecewise aggregation approximation method, wherein the numerical characteristics comprise the following steps:
let the original load sequence be X ═ X1,x2,…,xnThe numerical characteristics extracted by the piecewise aggregation approximation method are as follows: y ═ Y1,y2,...,ym}. Wherein n represents the number of elements in the original load sequence, m represents the number of features in the numerical features, m is smaller than n after being processed by the piecewise aggregation approximation method, and n can be divided by m. Order to
Then the elements in Y can be represented as:
the original load sequence with the length of n is converted into the numerical characteristic with the length of m by a segmentation aggregation approximation method, and the dimension reduction process is realized.
The current daily load curve of the power distribution network is obtained by using the historical daily load curve data of the power distribution network to predict the day-ahead load. Then, performing fine matching on the first candidate load sequence to obtain a historical load curve most similar to the current daily load curve; extracting the numerical characteristics of the current daily load curve by using a piecewise aggregation approximation method, wherein each section of curve can be expressed as:
and the most similar historical load curve is matched in a detailed mode from the candidate load sequence. The flow chart for coarse and fine matching is shown in fig. 2.
The rough matching is as follows: assuming that there are n historical sequences in the database in total, wherein m historical sequences have the same holiday, weather, season and topology as the current sequence, the m historical sequences are selected as the "first candidate load sequence", and the process is called coarse matching.
The fine matching is as follows: and respectively calculating the dynamic time bending distance of the numerical characteristics of the current load sequence and the candidate load sequence, and outputting the candidate load sequence corresponding to the minimum dynamic time bending distance.
And step 3: and a time interval division strategy of a historical load curve which is most similar to the current daily load curve is used in the current daily load curve. Wherein, the time interval division strategy of the historical load curve is derived from the database. A time-slicing scheme using a historical load curve with a minimum dynamic bending distance. Assuming that the features of the two load sequences are X and Y respectively, the feature lengths are m and n in turn, sorting the two features according to time positions, and constructing a distance matrix of m rows and n, namely:
d(xi,yj) Representing the euclidean distance of elements between different signature data objects, in a distance matrix the set of each set of adjacent matrix elements is called a curved path, which can be written as W ═ W1,w2...wk… are provided. The dynamic time warping distance is such as to minimize the total length of the warped path, which can be expressed as
The shortest path from point (1,1) to point (m, n) can be found by dynamic programming.
And 4, step 4: and (4) calculating the average value of the loads of all the nodes in each time interval, wherein the number of elements in the load sequence (the length of the load sequence) is equal to the number of the nodes of the power distribution network. And roughly matching a plurality of historical load sequences from the database as candidate load sequences as second candidate load sequences according to the symbolic features. Wherein the symbolic features here and the load features in the foregoing are consistent, it is still necessary to construct a classification tree.
And 5: and calculating the average value of the loads of all nodes in each time interval as an original characteristic, reducing the dimension of the original characteristic by using a principal component analysis method, and determining the weight of the characteristic after dimension reduction by using an entropy weight method. And performing fine matching operation on the second candidate load sequence and the current load sequence and outputting a historical sequence. The pseudo current daily load curve is divided into n segments through step 3, and each segment outputs a history sequence, namely n history sequences. Fig. 5 is a diagram showing the simulation time division result.
Firstly, the load of each node is used as an original characteristic, and the dimension of the characteristic is reduced by a principal component analysis method. The method specifically comprises the following steps: calculating a correlation coefficient r between the feature i and the feature jijThe correlation coefficients between the p features form a correlation coefficient matrix R ═ (R)ij)p×pWherein the correlation coefficient rijThe calculation formula of (a) is as follows:
and solving the eigenvalue and the eigenvector. Determining the eigenvalues lambda of the correlation coefficient matrixi(i-1, 2.. p) arranging the characteristic values in descending order, and calculating the characteristic vector a corresponding to each characteristic valuei(i ═ 1,2.. p). The cumulative contribution rate is calculated as well as the new features. In order to ensure that the original information loss is as small as possible, the number of the selected principal components should be such that the cumulative contribution rate is greater than 85%, and the cumulative contribution rate calculation formula of the first n principal components is as follows:
αnrepresenting the cumulative contribution of the first n principal components. The ith new feature can be expressed as follows: z is a radical ofi＝ai1x′1+ai2x′2+…aipx′p。
Second, the weight of the feature is determined using an entropy weight method. Assuming that n new features of m historical load sequences form an evaluation matrix after dimensionality reduction by a principal component analysis method, a feature standardization method is as follows:
the information entropy of each feature is:
large, the larger the weight should be. Weight w of jth featurejCan be expressed as:
the larger the weight is, the larger the effect of the characteristic on the calculation similarity is, and the difference degree of each characteristic is intuitively and effectively reflected. And finally, the candidate load sequence which is most similar to the current load sequence is finely matched from the candidate load sequences.
Step 6: the static reconstruction scheme of the historical sequence is recorded in the database. The static reconstruction schemes of the n periods of the current daily load curve use the static reconstruction schemes corresponding to the n history sequences selected in the step 5. And using a static reconstruction scheme of the load sequence with the minimum dynamic bending distance as a control strategy of each time interval.
The invention is further explained by taking an IEEE33 node power distribution system as an example, and the topological structure is shown in figure 4. Table 1 shows the load data of this example, and other data of the system are the same as those of the IEEE33 node example, and are not repeated. The action times of each switch are set to be 3 times at most, the total action times of all switches are set to be 15 times at most, the switch operation cost is 7 yuan/time, and the electricity price is 0.7 yuan/kWh.
The load curve is finally divided into 6 segments through optimization, the segmentation result is shown in figure 5, and the result shows that the load curve basically conforms to the variation trend of the curve.
In order to verify the rationality and correctness of the time interval division based on load similarity matching, the following 4 schemes are adopted for simulation. Scheme A: no reconstruction is performed. Scheme B: and dynamic reconstruction based on load curve monotonicity time interval division. Scheme C: and dynamically reconstructing based on load information entropy time interval division. Scheme D: dynamic reconstruction based on data driving.
Tables 1 and 2 show the results of IEEE33 node calculations.
TABLE 1
TABLE 2
| Algorithm | Loss per average power/kW | Time/second | Rate of accuracy |
| The 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 method | 148.5 | 51.2 | 33％ |
The calculation result shows that the traditional algorithm adopts a random search mode for optimization, so that the calculation time is too long, and the requirement of real-time calculation is difficult to meet. In addition, the accuracy of each conventional algorithm shows that the algorithm is easy to fall into local optimization. In contrast, the calculation time of the static reconstruction based on load similarity matching is far shorter than that of the traditional algorithm, and the method can be used for offline and can meet the requirement of online calculation real-time. The accuracy of the calculation is higher than that of the traditional algorithm, which shows the correctness and effectiveness of the invention.
Claims (9)
1. A dynamic reconstruction method of a power distribution network based on data driving is characterized by comprising the following steps:
step 1: extracting the symbolic features of daily load curves of the power distribution network, constructing a classification tree, and roughly matching a plurality of historical load curves from a database to serve as a first candidate load sequence;
step 2: extracting numerical characteristics of the current daily load curve and the first candidate load sequence by using a piecewise aggregation approximation method, and performing fine matching on the first candidate load sequence to obtain a historical load curve most similar to the current daily load curve;
and step 3: taking the time interval division method of the historical load curve most similar to the current daily load curve as the time interval division method of the current daily load curve;
and 4, step 4: calculating the load average value of each node in each time interval, constructing a classification tree according to the symbolic features, and roughly matching a plurality of historical load sequences from a database to serve as candidate load sequences to serve as second candidate load sequences;
and 5: taking the load average value of each node in each time interval as an original feature, reducing the dimension of the original feature by using a principal component analysis method, determining the weight of the feature after dimension reduction by using an entropy weight method, and performing fine matching on a second candidate load sequence, wherein a candidate load sequence most similar to the current sequence is matched in each time interval;
step 6: and using the static reconstruction method of the candidate load sequence which is matched in each time interval and is most similar to the current sequence as the static reconstruction method of each time interval of the current daily load curve.
2. The dynamic reconstruction method based on the data-driven power distribution network according to claim 1, characterized in that: the classification tree is a tree structure containing symbolic features of load curves.
3. The dynamic reconstruction method based on the data-driven power distribution network according to claim 1, characterized in that: the database comprises historical data of a plurality of days, wherein the historical data of each day comprises: a symbolic feature; daily load curves of the power distribution network and a corresponding time interval division method; the average load value of each node in each time interval, namely the original characteristic of each time interval; and a static reconstruction method corresponding to each time interval.
4. The dynamic reconstruction method for the power distribution network based on the data driving according to any one of claims 1 to 3, characterized in that: the symbolic features include holidays, weather, seasons, and topology.
5. The dynamic reconstruction method based on the data-driven power distribution network according to claim 1, characterized in that: the coarse matching is as follows: and deleting a sequence with the same symbol characteristics as the current sequence from a plurality of historical sequences of the database to serve as a first candidate load sequence.
6. The dynamic reconstruction method based on the data-driven power distribution network according to claim 1, characterized in that: the method for extracting the numerical characteristics of the current daily load curve by using the piecewise aggregation approximation method comprises the following steps:
the original payload sequence is X ═ X1,x2,...,xnThe numerical characteristics extracted by the piecewise aggregation approximation method are as follows: y ═ Y1,y2,...,ym}; wherein n represents the number of elements in the original load sequence, m represents the number of features in the numerical characteristics, m is smaller than n after being processed by a piecewise aggregation approximation method, and n can be divided by m; order to
Then the elements in Y can be represented as:
i is more than or equal to 1 and less than or equal to m; wherein y isiThe ith element representing the extracted feature; x is the number ofjRepresenting the jth element of the original payload sequence.
7. The dynamic reconstruction method based on the data-driven power distribution network according to claim 1, characterized in that: the fine matching is as follows: and calculating the dynamic time bending distance of the numerical characteristics of the current load sequence and the candidate load sequence, and outputting the candidate load sequence corresponding to the minimum dynamic time bending distance.
8. The dynamic reconstruction method based on the data-driven power distribution network according to claim 1, characterized in that: the method for reducing the dimension of the original features by using a principal component analysis method comprises the following steps:
calculating a correlation coefficient r between the feature i and the feature jijThe correlation coefficients between the p features form a correlation coefficient matrix R ═ (R)ij)p×pWherein the correlation coefficient rijThe calculation formula of (a) is as follows:
wherein the content of the first and second substances,
is the mean value of the features i and,
is the mean of the feature i, and N is the sum of the second candidate payload sequencesNumber, x'kiRepresents the ith element, x 'in the kth second candidate payload sequence'kjRepresenting the jth element in the kth second candidate load sequence, wherein p is the number of the features before dimensionality reduction in the second candidate load sequence;
determining the eigenvalues lambda of the correlation coefficient matrixi(i-1, 2.. p), arranging the characteristic values in descending order, and calculating the characteristic vector a corresponding to each characteristic valuei(i＝1,2...p)；
Calculating the cumulative contribution rate of the first M principal components:
wherein, αMRepresenting the cumulative contribution rate of the first M principal components; lambda [ alpha ]iAfter finishing the ranking from big to small, the ith characteristic value, lambdakAfter the order is arranged from big to small, the kth characteristic value is obtained, and p is the number of elements in the second candidate load sequence, namely the number of original characteristics; if the first Nx-1 cumulative contribution rate less than 85%, and the first NxWhen the cumulative contribution rate is greater than or equal to 85%, the number of the new features is Nx；
Ith element z in jth second candidate payload sequenceiExpressed as:
zi＝ai1x'j1+ai2x'j2+…aipx'jp,1≤i≤Nx
wherein, aipP is the number of the features before dimensionality reduction in the second candidate load sequence, x'jpIs the p-th element in the jth second candidate payload sequence.
9. The dynamic reconstruction method based on the data-driven power distribution network according to claim 1, characterized in that: the method for determining the weight of the feature after dimensionality reduction by using the entropy weight method comprises the following steps:
the normalization method of the features is as follows:
the information entropy of each feature is:
weight w of jth featurejExpressed as:
wherein, PijIs a matrix of features, zijIs the jth element in the ith second candidate load sequence after dimensionality reduction, N is the total number of the second candidate load sequences, NxIs the number of new features.
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 CN109861211A (en) | 2019-06-07 |
| CN109861211B true 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) |
Families Citing this family (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 |
| CN111337742B (en) * | 2020-02-25 | 2023-01-20 | 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 |
| 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 |
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 |
|---|
| Gradual Approaching Method for Distribution Network Dynamic;Yang HuPing;《2008 Workshop on Power Electronics and Intelligent Transportation System》;20080926;全文 * |
| 计及风电光伏相关性的多目标配电网重构;刘吉刚 等;《国网技术学院学报》;20171231;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109861211A (en) | 2019-06-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109861211B (en) | Dynamic reconfiguration method of power distribution network based on data driving | |
| US11409347B2 (en) | Method, system and storage medium for predicting power load probability density based on deep learning | |
| CN107516170B (en) | Difference self-healing control method based on equipment failure probability and power grid operation risk | |
| CN109768573B (en) | Power distribution network reactive power optimization method based on multi-target differential gray wolf algorithm | |
| CN107506865B (en) | Load prediction method and system based on LSSVM optimization | |
| CN107609667B (en) | Heat supply load prediction method and system based on Box _ cox transformation and UFCNN | |
| CN109345027B (en) | Micro-grid short-term load prediction method based on independent component analysis and support vector machine | |
| CN109615124B (en) | SCADA master station load prediction method based on deep learning | |
| CN112508279A (en) | Regional distributed photovoltaic prediction method and system based on spatial correlation | |
| CN113393339A (en) | Power distribution network short-term load prediction method based on deep neural network combination model | |
| CN115115113A (en) | Equipment fault prediction method and system based on graph attention network relation embedding | |
| Zhang et al. | A deep learning based real-time load forecasting method in electricity spot market | |
| CN116667369A (en) | Distributed photovoltaic voltage control method based on graph convolution neural network | |
| CN115481788B (en) | Phase change energy storage system load prediction method and system | |
| Bi et al. | Self-adaptive Teaching-learning-based Optimizer with Improved RBF and Sparse Autoencoder for Complex Optimization Problems | |
| Xu et al. | Short-term electricity consumption forecasting method for residential users based on cluster classification and backpropagation neural network | |
| CN117114160A (en) | Short-term photovoltaic power prediction method | |
| CN114707713A (en) | Low-voltage distribution network load prediction method considering distributed power supply influence | |
| CN105334730A (en) | Heating furnace oxygen content IGA optimization T-S fuzzy ARX modeling method | |
| CN111626465A (en) | New energy power short-term interval prediction method and system | |
| Di et al. | Online gaussian process regression for short-term probabilistic interval load prediction | |
| Chen et al. | Short-term power load model based on combined optimization of cuckoo algorithm and lightGBM | |
| Fang et al. | Research on Short-Term and Medium-Term Power Load Forecasting Based on STL-LightGBM | |
| Xiao et al. | GMDH based auto-regressive model for China's energy consumption prediction | |
| Zou et al. | Mid-term Forecast of Electricity Consumption of Enterprises Based on Bi-Directional LSTM and Nonpooling CNN |
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 |