CN115146742A - Offshore wind farm unit grouping and flagship machine selection method suitable for farm group control - Google Patents
Offshore wind farm unit grouping and flagship machine selection method suitable for farm group control Download PDFInfo
- Publication number
- CN115146742A CN115146742A CN202210930213.9A CN202210930213A CN115146742A CN 115146742 A CN115146742 A CN 115146742A CN 202210930213 A CN202210930213 A CN 202210930213A CN 115146742 A CN115146742 A CN 115146742A
- Authority
- CN
- China
- Prior art keywords
- wind
- flagship
- clustering
- farm
- grouping
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010187 selection method Methods 0.000 title claims abstract description 8
- 238000000034 method Methods 0.000 claims abstract description 48
- 238000010219 correlation analysis Methods 0.000 claims abstract description 23
- 238000005259 measurement Methods 0.000 claims abstract description 12
- 239000011159 matrix material Substances 0.000 claims description 25
- 239000013598 vector Substances 0.000 claims description 20
- 230000000694 effects Effects 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 13
- 230000008878 coupling Effects 0.000 claims description 7
- 238000010168 coupling process Methods 0.000 claims description 7
- 238000005859 coupling reaction Methods 0.000 claims description 7
- 239000006185 dispersion Substances 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 7
- 238000004458 analytical method Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 4
- 239000000654 additive Substances 0.000 claims description 3
- 230000000996 additive effect Effects 0.000 claims description 3
- 230000001808 coupling effect Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000011524 similarity measure Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 2
- 238000012795 verification Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
Images
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
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Public Health (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Wind Motors (AREA)
Abstract
The invention discloses an offshore wind farm unit grouping and flagship machine selection method suitable for farm group control, and mainly relates to the technical field of wind power; the method comprises the following steps: s1, extracting characteristics of a wind turbine generator; s2, clustering and grouping the wind power plants based on an improved Gaussian density distance clustering algorithm; s3, selecting a flagship set based on correlation analysis; the method utilizes an improved Gauss density distance algorithm to group the wind turbine generators; by adopting a correlation analysis method, the flagship set with representativeness and full coverage is selected according to the edge priority and redundancy principle, the problem that the input variable measurement of the wind turbine set of the offshore wind farm is inaccurate is solved in a targeted manner, and more accurate wind farm scheduling control is facilitated.
Description
Technical Field
The invention relates to the technical field of wind power, in particular to a method for grouping offshore wind power plant sets and selecting flagship machines, which is suitable for field group control.
Background
The large wind power plant system generally has the characteristics of high latitude, strong time variation and strong nonlinearity, and is a highly complex hybrid system. In order to reduce the problem of dimension increase caused by the number of wind turbines, grouping the wind turbines is an effective method. Grouping can greatly reduce the size and complexity of the optimization problem, but also can cause the precision to be reduced, thereby causing the optimization efficiency of the final target to be reduced or even reversely optimized. Therefore, the grouping of wind turbines requires a compromise between accuracy and complexity.
The existing grouping methods for wind turbines can be divided into the following three categories:
the first method is based on the input characteristics of the wind turbine generator and mainly studies the influence of input wind speed on the grouping of the wind turbine generator, but the method is not enough for considering other factors and has uncertainty;
the second method is based on the output characteristics of the wind turbine generator, the main idea is to group the wind turbine generator according to the output power of the wind turbine generator, the principle of the method is simple and easy to realize, the grouping result is usually determined, but the defects that real-time control cannot be realized and the like exist theoretically;
the third grouping method comprehensively considers the first two factors, performs comprehensive analysis on the input and output characteristics of the wind turbine generator and then performs grouping, and has the characteristics of accurate grouping and high reliability.
The traditional grouping method of the offshore wind farm comprises a grouping method based on a current collection line, a grouping method based on arrangement geographic positions, a grouping method based on mechanical control system characteristics and the like, and the problems that the application range is limited and the wake effect is ignored exist.
Disclosure of Invention
The invention aims to solve the problems in the prior art, provides a method for grouping offshore wind power plant units and selecting flagships, which is suitable for farm group control, comprehensively considers the wind inlet speed, the output power, the relative position of the wind power plant units and the wake effect between the wind power plant units, and utilizes an improved Gauss density distance algorithm to group the wind power plant units; by adopting a correlation analysis method, the flagship unit with representativeness and full coverage is selected according to the edge priority and redundancy principle, the problem that the input variable measurement of the wind turbine of the offshore wind farm is inaccurate is solved in a targeted manner, and more accurate wind farm scheduling control is facilitated.
In order to achieve the purpose, the invention is realized by the following technical scheme:
the grouping and flagship machine selecting method for the offshore wind farm machine set suitable for the farm group control comprises the following steps:
s1, extracting characteristics of a wind turbine generator;
s2, clustering and grouping the wind power plants based on an improved Gaussian density distance clustering algorithm;
and S3, selecting the flagship unit based on correlation analysis.
Preferably, in the step S1, the wind speed, the output power, and the relative position of the unit are selected as the clustering input.
Preferably, the step S2 includes the steps of:
s21, data preprocessing, wherein the data preprocessing comprises screening and dimension reduction;
s22, calculating related data of a clustering algorithm, wherein the related data comprises an input matrix, a mean value and deviation vector of the input matrix, a Gaussian matrix, an additive Gaussian vector, a Gaussian average vector, a Gaussian deviation vector, a distance matrix, a distance average vector and a distance deviation vector;
s23, determining a clustering center point, searching the maximum value of the GPM from the non-clustered data, and designating a point with the highest density as the clustering center;
s24, after the clustering center is determined, searching all clustering members of the cluster;
s25, defining cluster member adjacent points and updating threshold values by using an adjacent sample list and an adjacent search list;
and S26, evaluating the grouping quality.
Preference is given toIf S is between two wind turbine groups in step S24 overlap >0, i.e. there is a coupling, S overlap The calculation formula of (c) is:
wherein r is 1 For the wake radius, r, of the upwind turbine at the downwind turbine 2 The diameter of the wind wheel of the down-wind turbine generator set is shown, and d is the distance between two circle centers.
Preferably, the step S25 includes:
the neighbor sample list is formed by using all samples in the space, and the neighbor search list only contains the samples in the non-clustering list;
initially, the neighbor search list only stores the central point of all data, and the adjacent sample list stores the samples of each neighbor search list member, the samples are located in a circle with a radius of a fixed distance threshold, and there is no coupling effect between any samples, the fixed distance threshold uses the euclidean distance, and the calculation formula is:
determining gradient density and distance change on the area by the adjacent sample list through a gradient distance threshold and a Gaussian gradient density threshold;
the gradient distance threshold is calculated as:
where μ (GPD) is the average of all GPDs in the dataset;
the gaussian gradient density threshold is calculated as:
GGDT＝σ(SPLvariances)；
excluding a sample point if the GDTs of the data of the neighborhood sample are too scattered;
introducing clustering criteria including density criteria and distance criteria, and checking whether the Gaussian value of the sample is greater than or equal to the difference value of FGDT and GGDT in the density criteria; in the distance criterion, checking whether the Euclidean distance of one sample is smaller than or equal to the sum of FDT and GDT; if the density criterion and the distance criterion are satisfied, the sample is included in the cluster;
the neighbor search list is then updated in a recursive manner.
Preferably, in step S26: the verification of the number of the packets comprises verifying the number of the packets based on the outline coefficient and verifying the number of the packets based on Davies-Bouldin Index
The method for verifying the number of the groups based on the contour coefficient comprises the following steps:
assuming that the number of the wind turbines is n, the profile coefficient expression is as follows:
wherein, the sample profile coefficient of the ith wind turbine generator set is as follows:
assuming that the wind turbine generator i is divided into c groups after being clustered, wherein a (i) represents the average distance between the wind turbine generator i and all other wind turbine generators belonging to the c groups, and b (i) represents the minimum value of the average distance between all the wind turbine generators in each group of the wind turbine generator i and the non-c groups;
wherein s is i ∈[-1,1]When s is i When the cluster is closer to 1, the cluster of the wind turbine generator is more reasonable; when s is i When the clustering result is smaller or even negative, the clustering result of the wind turbine generator is unreasonable;
the method for verifying the number of the packets based on the Davies-Bouldin Index comprises the following steps:
defining a value S representing the degree of dispersion i And represents the dispersion degree of the data points in the ith class:
wherein, T i Representing the number of wind turbines in group i, X j Denotes the jth wind turbine generator set in the ith group, A i Represents the center of the ith group; the value of q is 1 or 2, when q is 1, the mean value of clustering from each point to the center is represented, and when q is 2, the standard deviation of the distance from each point to the center is represented;
defining a distance value M i,j And represents the distance between the ith class and the jth class:
defining a similarity measurement index which represents the similarity between the ith class and the jth class:
for different cluster groups, taking the maximum value D of similarity measure index i ，D i ＝max j≠i R i,j ；
Finally, the DBI index is obtained by taking the mean of the maximum similarity of all clusters:
preferably, the step S3 includes the steps of:
s31, formulating a flagship set selection principle, wherein the selection of the flagship set needs to meet a representative principle, an edge priority principle, a full coverage principle and a redundancy principle;
s32, clustering and grouping the main wind directions, and selecting flagship units with the number of N through correlation analysis;
s33, clustering and grouping other wind directions, wherein the number of the wind directions is M;
s34, comprehensively selecting the flagship set, and clustering and grouping other wind directions by taking the flagship set as a clustering center when M = N; and when M is less than N, performing correlation analysis on the N flagship units in the wind direction sector, and then selecting the front M flagship units as clustering centers to perform clustering grouping under the principle of edge priority.
Preferably, in step S31, the flagship aircraft is selected by using a correlation analysis method based on Pearson correlation coefficients, and the calculation formula is as follows:
wherein n is the number of samples,
and
are respectively x i And y i The mean value of (a);
the representative principle refers to that the flagship set is the most representative set in the group, and the representative is defined by using an average Pearson coefficient; the edge priority principle means that a wind turbine at the edge of a wind power plant is superior to a wind turbine in the wind power plant, because fans at the edge of the wind power plant are more likely not to be influenced by a wake effect under different wind directions, and the measured data is more reliable; the full coverage principle means that the effect of the flagship aircraft needs to cover all wind directions; the redundancy principle is that the number of the flagship machines is larger than or equal to the number of the clusters, and at least one flagship machine exists in each cluster, so that the flagship machines exist in all the clusters of the whole wind power plant.
Preferably, in step S32, according to historical wind measurement data of the wind farm, clustering and grouping wind turbines in a wind direction with the highest wind frequency, performing correlation analysis on the turbines in each group, sorting the correlation analysis, and selecting the first three wind turbines with the highest correlation coefficient, wherein if an edge wind turbine exists in the three wind turbines, the edge wind turbine is selected as a flagship turbine; if not, selecting the wind power generator set with the highest correlation coefficient as the flagship machine; the selected flagship unit meets the representative principle and the edge priority principle.
Preferably, in the step S33, for the full coverage principle and the redundancy principle, the number of flagship aircraft selected in the main wind direction is defined as N, and the optimal clustering number is obtained by performing cluster number analysis on other wind direction sectors in the range of [2, N ] through the contour coefficient and the DBI in the step S26.
Compared with the prior art, the invention has the beneficial effects that:
1. the method comprehensively considers the wind inlet speed, the output power and the relative position of the wind generation set, maintains the operating characteristics of the wind power plant to the maximum extent, simultaneously takes the wake effect as the coupling factor of the wind generation set, reduces the influence of the wake effect in the grouping process, and groups the wind generation set by using the improved Gaussian density distance clustering algorithm.
2. Aiming at the problem of inaccurate measurement of the input variable of the fan in the offshore wind farm, the invention selects the flagship unit which can fully cover the wind direction and has representativeness by adopting a correlation analysis method and following representativeness, edge priority, full coverage and redundancy.
3. The method overcomes the defects that the traditional grouping method only considers the input characteristic or the output characteristic and does not consider the influence caused by the wake effect, the input and output characteristics and the wake effect of the wind turbine generator are comprehensively considered, the method has the characteristics of accurate grouping and high reliability, simultaneously, the concept of the flagship generator is introduced, the wind turbine generator with the most representativeness is selected as the flagship generator, the problem of inaccurate measurement of the input variable of the wind turbine generator of the offshore wind farm is solved in a targeted manner, and the method is favorable for more accurate scheduling control of the wind farm.
Drawings
FIG. 1 is a flow chart of the present invention
FIG. 2 is a flow chart of the present invention for clustering groups of wind farms using an improved Gaussian density distance clustering algorithm;
FIG. 3 is a schematic wind farm wake relationship diagram;
fig. 4 is a flow chart of the present invention for flagship aircraft selection using correlation analysis.
Detailed Description
The invention will be further illustrated with reference to the following specific examples. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. Further, it should be understood that various changes or modifications of the present invention may be made by those skilled in the art after reading the teaching of the present invention, and these equivalents also fall within the scope of the present application.
The wind turbine generator group is required to be accurately measured, but at the present stage, the wind turbine generator measurement still has a large error, and the flagship machine can effectively solve the problem of inaccurate measurement. The flagship machine is characterized in that when a wind power plant is built, a large amount of high-precision low-delay equipment such as a laser radar, a communication device and the like is additionally arranged on a part of wind power sets in the wind power plant so as to ensure the accuracy of measured data and the timeliness of data transmission. Due to the huge economic cost of the flagship machine, only a small number of flagship machine sets can be installed.
Example (b): as shown in the attached drawings 1-4, the invention relates to a method for grouping offshore wind turbine generators and selecting flagship machines, which is suitable for farm group control, wherein the wind turbine generators are grouped by utilizing an improved Gauss density distance algorithm by comprehensively considering the wind inlet speed, the output power, the relative position and the wake effect between the wind turbine generators; and selecting the flagship set with representativeness and full coverage by adopting a correlation analysis method according to the edge priority and redundancy principle.
The method comprises the following steps:
s1, extracting characteristics of the wind turbine generator.
The method comprises the steps of extracting and selecting the most representative features from original data, taking the extracted and selected features as input of a clustering algorithm, considering wake effect and keeping the operating characteristics of a wind power plant to the maximum extent, and selecting inflow wind speed u, v, output power P and position coordinates (x, y) of a wind turbine generator as input of clustering. That is, the matrix X = [ u, v, P, X, y ].
S2, clustering and grouping the offshore wind power plants based on an improved Gaussian density distance clustering algorithm, and specifically comprises the following steps:
s21, preprocessing data;
the tasks performed in the pre-processing stage are as follows:
(1) Screening: scanning all data in the data set, and deleting the data from the data set if any one of the attribute information of the data is missing;
(2) And (3) reducing the dimensionality: if the variance of a dimension vector is zero, then the dimension is removed from the input dataset.
S22, calculating related data of a clustering algorithm;
in the calculation process, various data structures such as matrixes and vectors are used, and the data structures are as follows:
(1) Input matrix X n×d ：
Where n represents the number of samples in the dataset and d represents the attribute dimension of the sample data. Specifically, in this embodiment, n represents the number of wind turbines in the offshore wind farm, and d represents the dimension of the clustering input matrix X.
(2) Mean value mu of the input matrix d And a deviation vector σ d
For input matrix X n×d Each column of (1) calculates mu d And σ d In the subsequent pre-processing of the input dataBoth processing and computation of the gaussian distribution require them.
(3) Gauss Matrix (Gauss Matrix, GM)
GM n×n The method comprises the discrete reciprocal effect of every two points in the data set, which is a non-negative and symmetrical matrix, the diagonal line of the matrix comprises the maximum Gaussian value of each point in the data set, and the multidimensional Gaussian value calculation mode is as follows:
(4) Additive Gaussian vector (AG), AG n Is to the matrix GM n×n The row sum of (a) results in:
(5) Gaussian Mean vector (GPM), GPM n Is prepared by mixing GM n×n The summed values of the rows in the matrix, except for the diagonal elements, are divided by (n-1) to yield:
(6) Gaussian Point Deviation vector (GPD) by using GPM calculated in equation (3.4) j Value, calculating GM n×n The standard deviation of each row of the matrix is used to obtain GPD n 。
(7) Distance Matrix (DM), the elements of which are the euclidean distances between points, the matrix being a non-negative, zero diagonal symmetric matrix, and the formula:
(8) Distance Point Mean vector (DPM), DPM n Is obtained by mixing DM n×n The summed values of the rows in the matrix, except for the diagonal elements, are divided by (n-1) to yield:
(9) Distance Point Deviation vector (DPD) by using DPM j Value, calculate DM n×n The standard deviation of each row of the matrix is used to obtain the DPD n 。
And S23, determining a clustering center point, and designating a point with the highest density as a clustering center by searching the maximum value of the GPM from the non-clustered data.
S24, after the clustering center is determined, searching all clustering members of the cluster; the method comprises the following specific steps:
it is determined whether a coupling phenomenon exists between the member and the cluster center and the member. The coupling between the wind turbine sets is represented by a wake effect, and when the wind turbine sets are in the wake influence range of the wind turbine sets above the wind turbine sets, the coupling between the two wind turbine sets is shown. As shown in figure 3, if S between two wind turbine units overlap >0, there is a coupling effect. The calculation formula is as follows:
wherein r is 1 Is the wake radius r generated by the upwind wind turbine at the downwind wind turbine 2 The diameter of the wind wheel of the down-wind turbine generator set is shown, and d is the distance between two circle centers.
S25, defining cluster member adjacent points and an updating threshold by using a sample in Proximity List (SPL) and a Neighbor Search List (NSL); the method comprises the following specific steps:
the SPL is formed using all samples in the space, while the NSL contains only the samples in the non-clustered list. Initially, the NSL only holds the center point of all data, while the SPL holds samples for each NSL member that lie within a circle of Fixed Distance Threshold (FDT) radius with no coupling between any samples. FDT adopts Euclidean distance, and the calculation formula is as follows:
SPL determines the gradient density and distance change over the area by GDT and GGDT. The GDT and GGDT calculations are:
GGDT＝σ(SPLvariances)；
where μ (GPD) is the average of all GPDs in the data set.
FDTs are used to determine clusters and cluster membership. When a sample is located in the neighborhood of a cluster, the Gaussian Threshold equation and the Gradient Distance Threshold (GDT) should be satisfied, and the GDT and Gaussian Gradient Density Threshold (GGDT) represent the variance change. If the values of the neighborhood samples satisfy the threshold condition, the neighborhood samples are kept in a cluster using these threshold equations.
Since each point in the neighborhood acts as a cluster center, a relative dynamic threshold is needed to measure the convergence or divergence of the neighborhood samples. If the GDT of the data for the neighborhood samples is too scattered, then the sample point is excluded.
Clusters may have different shapes and connectivity. Therefore, a gradient threshold equation is introduced to represent the clustering of these arbitrary shape density variations. The calculation of the gradient threshold comprises two steps:
1) Calculating the variance of the current sample in the neighborhood;
2) The mean and deviation of the previously calculated variances are calculated.
Each cluster scan, the samples are added to the cluster using these dynamic values, completing the inclusion of the neighborhood. The GDD defines which sample belongs to which cluster. Two criteria must be met in this process: a density criterion (criterion a) and a distance criterion (criterion B). In the density criterion, it is checked whether the gaussian value of the sample is greater than or equal to the difference between FGDT and GGDT. In the distance criterion, it is checked whether the euclidean distance of one sample is equal to or less than the sum of FDT and GDT. If criteria A and B are met, the sample is included in the cluster. Criteria a and B are as follows:
criterion A: (centrr., s) is not less than FGDT-GGDT;
criterion B: (centr, s) is less than or equal to FDT + GDT;
the NSL is updated in a recursive manner. As in the first loop, only the center points are clustered into one cluster and saved to the NSL. In the second cycle, neighborhood samples for the center point are added to the NSL, clusters are updated, and so on. The samples newly added to the NSL form an integral with the original point in the NSL as the center point for the next step, so that the neighborhood is expanded until there are no satisfied samples. When all the samples are clustered, the algorithm automatically stops.
S26, evaluating the grouping quality;
the number of packets is verified based on the contour coefficients. Assuming that the number of the wind turbines is n, the profile coefficient expression is as follows:
wherein, the sample profile coefficient of the ith wind turbine generator set is as follows:
assuming that the wind turbine i is divided into c groups after being clustered, a (i) represents the average distance between the wind turbine i and all other wind turbines belonging to the c groups, and b (i) represents the minimum value of the average distance between all the wind turbines in each group of the wind turbine group i and the non-c groups. Wherein s is i ∈[-1,1]When s is i When the cluster is closer to 1, the cluster of the wind turbine generator is more reasonable; when s is i And when the number is smaller or even negative, the clustering result of the wind turbine generator is unreasonable.
The number of packets is verified based on the Davies-Bouldin Index. Defining a value S representing the degree of dispersion i And represents the dispersion degree of the data points in the ith class:
wherein, T i Representing the number of wind turbines in group i, X j Denotes the jth wind turbine generator set in the ith group, A i Indicating the center of the ith group. The value of q is 1 or 2, when q is 1, the mean value of clustering from each point to the center is represented, and when q is 2, the standard deviation of the distance from each point to the center is represented, which can be used for measuring the dispersion degree, and in this embodiment, q =2 is taken.
Defining a distance value M i,j And represents the distance between the ith class and the jth class:
defining a similarity measurement index which represents the similarity between the ith class and the jth class:
for different cluster groups, taking the maximum value D of similarity measure index i ，D i ＝max j≠i R i,j . I.e. the largest value of the similarity of the ith group to the other groups. Finally, the DBI index is obtained by averaging the maximum similarity of all clusters:
S3, selecting a flagship machine based on the clustering result; the method comprises the following specific steps:
s31, selecting a flagship set principle;
the invention adopts a correlation analysis method based on Pearson correlation coefficient to select the flagship aircraft, and the calculation formula is as follows:
wherein n is the number of samples,
and
are respectively x i And y i Is measured.
The selection requirement of the flagship aircraft meets the requirement:
(1) Representative principles: the flagship set is the most representative set in the group, and the representativeness is defined by an average Pearson coefficient;
(2) Edge precedence principle: the wind turbine at the edge of the wind power plant is superior to the wind turbine in the wind power plant, because the fans at the edge of the wind power plant are more likely not to be influenced by the wake effect under different wind directions, and the measured data is more reliable;
(3) The full coverage principle is as follows: the flagship aircraft needs to cover all wind directions;
(4) Redundancy principle: the number of the flagship machines is more than or equal to the number of the clusters, and at least one flagship machine group exists in each cluster, so that the flagship machines exist in all the clusters of the whole wind power plant.
S32, clustering and grouping the main wind directions, and selecting flagship units with the number of N through correlation analysis;
clustering and grouping wind turbines in the wind direction (main wind direction) with the highest wind frequency according to historical wind measurement data of the wind power plant, performing correlation analysis on the turbines in each group, sequencing the correlation analysis, and selecting the first three wind turbines with the highest correlation coefficient, wherein if an edge wind turbine exists in the three wind turbines, the edge wind turbine is selected as a flagship turbine; and if not, selecting the wind power generator set with the highest correlation coefficient as the flagship aircraft. The selected flagship unit meets the principle (1) representative principle and the principle (2) edge priority principle.
S33, clustering and grouping other wind directions, wherein the number of the wind directions is M;
the offshore wind farm can obtain a wind direction rose diagram according to historical wind measuring data, the wind direction rose diagram divides the wind direction into a plurality of sectors according to the sectors, and each sector comprises a wind direction angle in the same range. For the principle (3) of full coverage and the principle (4) of redundancy, the number of the flagship aircraft selected in the main wind direction is defined as N, and the optimal clustering number is obtained by performing clustering number analysis on other wind direction sectors in the range of [2, N ] through the contour coefficient and the DBI in S26.
S34, comprehensively selecting a flagship set;
when M = N, clustering and grouping other wind directions by taking the flagship machine as a clustering center; and when M is less than N, performing correlation analysis on the N flagship units under the wind direction sector, and then selecting the front M flagship units as clustering centers to perform clustering grouping under the principle of edge priority.
Claims (10)
1. The method for grouping the offshore wind power plant units and selecting the flagship machines is suitable for field group control, and is characterized by comprising the following steps of:
s1, extracting characteristics of a wind turbine generator;
s2, clustering and grouping the wind power plants based on an improved Gaussian density distance clustering algorithm;
and S3, selecting the flagship unit based on correlation analysis.
2. The method for grouping the offshore wind farm units and selecting the flagship aircraft suitable for the farm group control according to claim 1, wherein in the step S1, the wind inlet speed, the output power and the relative position of the units are selected as clustering input.
3. The offshore wind farm unit grouping and flagship aircraft selection method suitable for farm group control according to claim 1, wherein the step S2 comprises the steps of:
s21, data preprocessing, wherein the data preprocessing comprises screening and dimension reduction;
s22, calculating related data of a clustering algorithm, wherein the related data comprises an input matrix, a mean value and deviation vector of the input matrix, a Gaussian matrix, an additive Gaussian vector, a Gaussian average vector, a Gaussian deviation vector, a distance matrix, a distance average vector and a distance deviation vector;
s23, determining a clustering center point, searching the maximum value of the GPM from the non-clustered data, and designating a point with the highest density as the clustering center;
s24, after the clustering center is determined, searching all clustering members of the cluster;
s25, defining cluster member adjacent points and an updating threshold value by using an adjacent sample list and an adjacent search list;
and S26, evaluating the grouping quality.
4. The offshore wind farm unit grouping and flagship machine selection method suitable for farm group control according to claim 3, wherein in step S24, if S between two wind farm units is determined overlap >0, i.e. there is a coupling，S overlap The calculation formula of (c) is:
wherein r is 1 Is the wake radius r generated by the upwind wind turbine at the downwind wind turbine 2 The diameter of the wind wheel of the down-wind turbine generator set is shown, and d is the distance between two circle centers.
5. The offshore wind farm unit grouping and flagship aircraft selection method suitable for farm group control according to claim 3, wherein the step S25 comprises:
the neighbor sample list is formed by using all samples in the space, and the neighbor search list only contains the samples in the non-clustering list;
initially, the neighbor search list only stores the central point of all data, and the adjacent sample list stores the samples of each neighbor search list member, the samples are located in a circle with a radius of a fixed distance threshold, and there is no coupling effect between any samples, the fixed distance threshold uses the euclidean distance, and the calculation formula is:
determining gradient density and distance change on the area by the adjacent sample list through a gradient distance threshold and a Gaussian gradient density threshold;
the gradient distance threshold is calculated as:
where μ (GPD) is the average of all GPDs in the dataset;
the gaussian gradient density threshold is calculated as:
GGDT＝σ(SPLvariances)；
excluding a sample point if the GDTs of the data of the neighborhood sample are too scattered;
introducing clustering criteria including density criteria and distance criteria, and checking whether the Gaussian value of the sample is greater than or equal to the difference value of FGDT and GGDT in the density criteria; in the distance criterion, checking whether the Euclidean distance of one sample is smaller than or equal to the sum of the FDT and the GDT; if the density criterion and the distance criterion are satisfied, the sample is included in the cluster;
the neighbor search list is then updated in a recursive manner.
6. The offshore wind farm unit grouping and flagship aircraft selection method suitable for farm group control according to claim 3, wherein in step S26: the verification of the packet number comprises verifying the packet number based on the outline coefficient and verifying the packet number based on the Davies-Bouldin Index
The method for verifying the number of the groups based on the contour coefficient comprises the following steps:
assuming that the number of the wind turbines is n, the profile coefficient expression is as follows:
wherein, the sample profile coefficient of the ith wind turbine generator set is as follows:
the clustering method comprises the steps that a, a (i) represents the average distance between the wind turbine i and all other wind turbines belonging to the c group, and b (i) represents the minimum value of the average distance between all wind turbines in each group of the wind turbine group i and the non-c group;
wherein s is i ∈[-1,1]When s is i When the cluster is closer to 1, the cluster of the wind turbine generator is more reasonable; when s i When the clustering result is smaller or even negative, the clustering result of the wind turbine generator is unreasonable;
the method for verifying the number of the packets based on the Davies-Bouldin Index comprises the following steps:
defining a value S representing the degree of dispersion i And represents the dispersion degree of the data points in the ith class:
wherein, T i Representing the number of wind turbines in group i, X j Denotes the jth wind turbine generator set in the ith group, A i Represents the center of the ith group; the value of q is 1 or 2, when q is 1, the mean value of clustering from each point to the center is represented, and when q is 2, the standard deviation of the distance from each point to the center is represented;
defining a distance value M i,j And represents the distance between the ith class and the jth class:
defining a similarity measurement index which represents the similarity between the ith class and the jth class:
for different cluster groups, taking the maximum value D of similarity measure index i ，D i ＝max j≠i R i,j ；
Finally, the DBI index is obtained by taking the mean of the maximum similarity of all clusters:
7. the offshore wind farm unit grouping and flagship aircraft selection method suitable for farm group control according to claim 6, wherein the step S3 comprises the steps of:
s31, establishing a flagship set selection principle, wherein the selection of the flagship set needs to meet a representative principle, an edge priority principle, a full coverage principle and a redundancy principle;
s32, clustering and grouping the main wind directions, and selecting flagship units with the number of N through correlation analysis;
s33, clustering and grouping other wind directions, wherein the number of the wind directions is M;
s34, comprehensively selecting a flagship set, and clustering and grouping other wind directions by taking the flagship set as a clustering center when M = N; and when M is less than N, performing correlation analysis on the N flagship units under the wind direction sector, and then selecting the front M flagship units as clustering centers to perform clustering grouping under the principle of edge priority.
8. The method for selecting the group of offshore wind farm units and the flagship aircraft suitable for the farm group control according to claim 7, wherein in the step S31, the flagship aircraft is selected by a correlation analysis method based on Pearson correlation coefficients, and the calculation formula is as follows:
wherein n is the number of samples,
and
are respectively x i And y i The mean value of (a);
the representative principle means that the flagship set is the most representative set in the group, and the representative is defined by average Pearson coefficient; the edge priority principle means that a wind turbine at the edge of a wind power plant is superior to a wind turbine in the wind power plant, because fans at the edge of the wind power plant are more likely not to be influenced by a wake effect under different wind directions, and measured data are more reliable; the full coverage principle means that the effect of the flagship aircraft needs to cover all wind directions; the redundancy principle is that the number of the flagship machines is larger than or equal to the number of the clusters, and at least one flagship machine exists in each cluster, so that the flagship machines exist in all the clusters of the whole wind power plant.
9. The method for selecting the flag ship machine and the grouping of the offshore wind farm units suitable for the farm group control according to claim 7, wherein in the step S32, the wind farm units are clustered and grouped in the wind direction with the highest wind frequency according to historical wind measurement data of the wind farm, correlation analysis is performed on the units in each group, the correlation analysis is sequenced, the first three wind farm units with the highest correlation coefficient are selected, and if an edge wind farm unit exists in the three wind farm units, the edge wind farm unit is selected as the flag ship unit; if not, selecting the wind power generator set with the highest correlation coefficient as the flagship machine; the selected flagship unit meets the representative principle and the edge priority principle.
10. The method for grouping offshore wind farm units and selecting flagship machines suitable for farm group control according to claim 7, wherein in step S33, for the full coverage principle and the redundancy principle, the number of flagship machines selected in the main wind direction is defined as N, and in other wind direction sectors, the optimal clustering number is obtained by performing clustering number analysis in the range of [2, N ] on the contour coefficient and DBI in step S26.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210930213.9A CN115146742A (en) | 2022-08-03 | 2022-08-03 | Offshore wind farm unit grouping and flagship machine selection method suitable for farm group control |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210930213.9A CN115146742A (en) | 2022-08-03 | 2022-08-03 | Offshore wind farm unit grouping and flagship machine selection method suitable for farm group control |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115146742A true CN115146742A (en) | 2022-10-04 |
Family
ID=83414203
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210930213.9A Pending CN115146742A (en) | 2022-08-03 | 2022-08-03 | Offshore wind farm unit grouping and flagship machine selection method suitable for farm group control |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115146742A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116993026A (en) * | 2023-09-26 | 2023-11-03 | 无锡九方科技有限公司 | Large-scale wind power plant unit operation parameter optimization method |
-
2022
- 2022-08-03 CN CN202210930213.9A patent/CN115146742A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116993026A (en) * | 2023-09-26 | 2023-11-03 | 无锡九方科技有限公司 | Large-scale wind power plant unit operation parameter optimization method |
| CN116993026B (en) * | 2023-09-26 | 2023-12-19 | 无锡九方科技有限公司 | Large-scale wind power plant unit operation parameter optimization method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111369070B (en) | Multimode fusion photovoltaic power prediction method based on envelope clustering | |
| CN109670213B (en) | Multi-machine equivalent modeling method and device for wind power plant | |
| CN103916820A (en) | Wireless indoor locating method based on access point stability degree | |
| CN110264002B (en) | Wind power plant micro-siting scheme evaluation method based on cluster analysis | |
| CN111050282A (en) | Multi-time fuzzy inference weighted KNN positioning method | |
| CN106875037A (en) | Wind-force Forecasting Methodology and device | |
| CN110263834B (en) | Method for detecting abnormal value of new energy power quality | |
| CN114444378A (en) | Short-term power prediction method for regional wind power cluster | |
| CN112801332B (en) | Short-term wind speed prediction method based on gray level co-occurrence matrix | |
| CN108734359B (en) | Wind power prediction data preprocessing method | |
| CN114021483A (en) | Ultra-short-term wind power prediction method based on time domain characteristics and XGboost | |
| CN112149883A (en) | Photovoltaic power prediction method based on FWA-BP neural network | |
| CN115146742A (en) | Offshore wind farm unit grouping and flagship machine selection method suitable for farm group control | |
| CN114138012A (en) | Fan blade inspection method, device, equipment and storage medium | |
| CN115712845A (en) | Lightning space-time clustering method and system based on radius parameter optimization | |
| CN111985690B (en) | Site selection method for booster station of wind power base | |
| CN110009035B (en) | Wind measuring station cluster space clustering method based on image matching | |
| CN114429233A (en) | Cluster algorithm-based station-level optimal power control method and system | |
| CN111062509A (en) | Wind power medium-short term prediction method and system based on improved neural network | |
| CN115442887B (en) | Indoor positioning method based on RSSI of cellular network | |
| CN116070458A (en) | New wind power plant scene generation method based on RAC-GAN | |
| CN116258972A (en) | Deep learning-based rock high-steep slope structural surface extraction method | |
| CN116029440A (en) | Ultra-short-term power prediction method and device for photovoltaic power station | |
| CN109977797A (en) | The optimization method of single order object detector based on sequence loss function | |
| CN115619605A (en) | Traditional residential environment quality evaluation method based on semantic information |
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 |