Wind power-load scene construction method and device based on time sequence correlation clustering fusion
Through the method based on time-series related clustering fusion, a typical wind power-load scenario is constructed, which solves the problem of ignoring the impact of time series in the existing technology, reduces the calculation error of wind power and improves the typicality of scene construction, and optimizes the strategy and The model provides the basis.
Patent Information
- Application Number
- CN202410970394.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-19
- Publication Date
- 2024-11-08
AI Technical Summary
The prior art ignores the impact of time series on the output power of the wind farm when building wind power-load scenarios, resulting in large errors in wind power calculations and inconsistent scenario construction.
Using the method based on time-series-related clustering fusion, by collecting wind speed data and load data, a time series-based wind speed-wind power conversion function is constructed, and noise reduction preprocessing is performed. The orderly clustering and K-means clustering methods are used to segment and Divide the data, determine the optimal number of segmentation and clustering, and build a typical wind power-load scenario.
The wind power calculation error is reduced, the typicality of scene construction is improved, and the typical scenario characteristics of the source charge output characteristics of the wind power access power grid are explored, laying the foundation for subsequent reactive power optimization strategies and models.
Smart Images

Figure CN118916720A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of power systems and automation technologies thereof, and in particular to a method and device for constructing a wind power-load scenario based on time series related clustering fusion. Background Art
[0002] As wind power is continuously connected to the power system network, the randomness and uncontrollability of its output has a great impact on the voltage control of the power grid.
[0003] Technical research and case verification in the field of power systems and their automation rely on verification implementation in typical scenarios, so the construction of typical wind power-load scenarios will directly affect subsequent research. However, most current studies convert wind speed and wind power into a one-to-one function when converting wind speed-wind power in wind farms. By comparing the current wind speed with the cut-in wind speed, rated wind speed, and cut-out wind speed, the wind power corresponding to the current wind speed is obtained, ignoring the impact of time series on the output power of wind farms.
[0004] Therefore, it is necessary to propose a method for constructing a typical scenario that takes into account the impact of timing on the output power of wind farms, so as to reduce the error in wind power calculation and improve the typicality of scenario construction. Summary of the invention
[0005] The purpose of the present invention is to provide a method and device for constructing a wind power-load scenario based on time series correlation clustering fusion, which is used to solve the problems in the prior art that the impact of time series on the output power of wind farms is ignored and the construction of wind power-load scenarios is not typical enough, so as to reduce the error in wind power calculation and improve the typicality of scenario construction.
[0006] In order to achieve the above objectives, in a first aspect, the present invention provides a method for constructing a wind power-load scenario based on time series related clustering fusion, comprising: Step 1: Collect wind speed data and load data of the wind farm, construct a wind speed-wind power conversion function based on time series, and obtain the initial wind power-load data set of the wind farm; Step 2: Perform noise reduction preprocessing on the initial wind power-load data set to obtain a wind power-load data set; Step 3: Segment the load data set in the wind power-load data set based on the ordered clustering method, determine the optimal number of segmentations and construct a first-order typical load scenario; Step 4: According to the first-order typical load scenario and the optimal number of splits, the corresponding wind power data set is divided into multiple scenario sets. The range of the number of clusters is set according to the number of load data in each scenario set, and the wind power data in each scenario set is clustered. The silhouette coefficients under different numbers of clusters are calculated, and the number of clusters corresponding to the maximum silhouette coefficient is selected as the optimal number of clusters to construct a second-order typical wind power scenario.
[0007] According to a wind power-load scenario construction method based on time series related clustering fusion provided by the present invention, in step 1, the wind speed-wind power conversion function based on the time series is: when or hour,
[0008] when hour,
[0009] when hour,
[0010] In the formula, For wind farms Wind power at the time; For wind farms Wind speed at the moment; The cut-in wind speed set for the wind farm; The cut-out wind speed set for the wind farm; is the rated wind speed of the wind farm; is the density of air, A is the swept area of the wind turbines in the wind farm; is the rated power of the wind farm; is the wind power of the wind farm at the last moment; and They are the low recovery wind speed and high recovery wind speed set for the wind farm respectively. According to a wind power-load scenario construction method based on time series related clustering fusion provided by the present invention, the swept area of the wind turbines in the wind farm is expressed as: A= 2 πr 2 In the formula, A is the swept area of the wind turbines in the wind farm, r is the radius of the wind turbine in the wind farm. According to a wind power-load scenario construction method based on time series correlation clustering fusion provided by the present invention, in step 2, the DBSCAN density clustering algorithm is used to perform noise reduction preprocessing on the wind power-load initial data set.
[0011] According to a wind power-load scenario construction method based on time series related clustering fusion provided by the present invention, step 3 specifically includes: Determine the objective function; Genetic algorithm is used to optimize the number of segmentation points and the silhouette coefficient of ordered clustering under different numbers of segmentation points is calculated; The number of segmentation points corresponding to the maximum silhouette coefficient is selected as the optimal number of segmentation points to construct a first-order typical load scenario. According to a wind power-load scenario construction method based on time series related clustering fusion provided by the present invention, the objective function is to minimize the sum of squared deviations. According to a wind power-load scenario construction method based on time series related clustering fusion provided by the present invention, the range of the number of segmentation points is: 1≤p≤15 Among them, p is the number of segmentation points.
[0012] According to a wind power-load scenario construction method based on time series related clustering fusion provided by the present invention, the wind power data includes wind power, and step 4 specifically includes: Divide the wind power into p+1 scene sets; K-means clustering is performed on the wind power in each scenario set. The range of the number of clusters is set according to the number of load data in each scenario set. The silhouette coefficients under different numbers of clusters are calculated. The number of clusters corresponding to the maximum silhouette coefficient is selected as the optimal number of clusters to construct a second-order typical wind power scenario. According to a wind power-load scenario construction method based on time series related clustering fusion provided by the present invention, the range of the number of clusters is: 2≤N≤
[0013] in, f is the number of load data, and N is the number of clusters.
[0014] In a second aspect, the present invention further provides a wind power-load scenario construction device based on time series related clustering fusion, comprising: A collection unit is used to collect wind speed data and load data of the wind farm, construct a wind speed-wind power conversion function based on a time series, and obtain an initial wind power-load data set of the wind farm; A preprocessing unit, used for performing noise reduction preprocessing on the initial wind power-load data set to obtain a wind power-load data set; A first construction unit is used to segment the load data set in the wind power-load data set based on an ordered clustering method, determine an optimal number of segmentations, and construct a first-order typical load scenario; The second construction unit is used to divide the corresponding wind power data set into multiple scenario sets according to the first-order typical load scenario and the optimal number of splits, set the range of the number of clusters according to the number of load data in each scenario set, cluster the wind power data in each scenario set, calculate the silhouette coefficients under different numbers of clusters, select the number of clusters corresponding to the maximum silhouette coefficient as the optimal number of clusters, and construct a second-order typical wind power scenario.
[0015] The present invention provides a method and device for constructing a wind power-load scenario based on time series related clustering fusion. The method and device construct a wind power-load initial data set based on the wind speed-wind power conversion function of the time series, perform noise reduction preprocessing on the initial wind power-load data set, and then perform first-order ordered clustering. The wind power data is divided according to the time series segmentation results of the typical load scenario, and the divided wind power data is subjected to second-order clustering, and finally the construction of the typical wind power-load scenario is completed. The present invention mines the typical scenario characteristics of the source-load output characteristics of the power grid to which wind power is connected, reduces the error in wind power calculation, and improves the typicality of scenario construction; and can lay a foundation for the subsequent solution of the comprehensive reactive power optimization strategy set under different typical wind power-load scenarios, and construct reactive power optimization models for each scenario, so as to improve the convergence, accuracy and scenario generalization capabilities of the model. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0017] In the attached picture: Figure 1 A flow chart of a wind power-load scenario construction method based on time series related clustering fusion provided by the present invention; Figure 2a and Figure 2b They are respectively the schematic diagram of the wind power per unit value distribution of the wind farm throughout the year and the corresponding schematic diagram of the load data distribution throughout the year; Figure 3a and Figure 3b They are respectively the schematic diagram of the initial data distribution of wind power-load and the schematic diagram of the noise point distribution after density noise reduction; Figure 4 It is a curve graph of the silhouette coefficient values for different numbers of segmentation points based on the ordered clustering method; Figure 5 A schematic diagram showing the division of load data into seven typical load scenarios; Figure 6 The 24-hour load distribution curve diagram for the period included in typical load scenario 2; Figure 7 It is a curve graph of the silhouette coefficient value under different numbers of clusters based on the K-means clustering method; Figure 8 for Schematic diagram of the results after clustering wind power scene samples. DETAILED DESCRIPTION
[0018] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0019] Some embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.
[0020] See also Figure 1 The present invention provides a method for constructing a wind power-load scenario based on time series related clustering fusion, comprising the following steps: Step 1: Collect wind speed data and load data of the wind farm, construct a wind speed-wind power conversion function based on time series, and obtain the initial wind power-load data set of the wind farm; It should be noted that since the output data of wind farms (referred to as wind farms) are directly affected by factors such as weather and geographical environment, the output data of wind farms in different regions are mostly different. Therefore, the present invention proposes the wind power output of wind farms. The local wind speed There is a direct impact and functional conversion relationship:
[0021] In the formula, is the wind power output of the wind farm; is the density of air, A is the swept area of the wind turbines in the wind farm; is the actual wind speed of the wind farm; The cut-in wind speed set for the wind farm; The cut-out wind speed set for the wind farm; is the rated wind speed of the wind farm; is the rated power output of the wind farm.
[0022] The relationship between wind power and wind speed in actual wind farms is somewhat different from the traditional analysis method. For example, when the wind speed is too high, far exceeding the cut-out wind speed set by the wind farm, When the wind speed is too low, the wind farm maintenance personnel will adjust the wind power to 0 in a short time. When the wind speed is too low, the influence of the wind farm's minimum recovery wind speed must be considered. Therefore, there is no one-to-one correspondence between the wind power output of the wind farm and the wind speed. The influence of the wind speed at the previous moment must be considered at the current moment. Therefore, the wind speed-wind power conversion function considering the time series established by the present invention is as follows: when or hour:
[0023] when hour:
[0024] when hour:
[0025] In the formula, For wind farms Wind power at the time; For wind farms Wind speed at the moment; The cut-in wind speed set for the wind farm; The cut-out wind speed set for the wind farm; is the rated wind speed of the wind farm; is the density of air, A is the swept area of the wind turbines in the wind farm; is the rated power of the wind farm; is the wind power of the wind farm at the last moment; and They are the low recovery wind speed and high recovery wind speed set for the wind farm respectively.
[0026] Among them, the expression of the swept area of the wind turbine in the wind farm is: A= 2 πr 2 In the formula, A is the swept area of the wind turbines in the wind farm, r is the radius of the wind turbine in the wind farm.
[0027] Step 2: Perform noise reduction preprocessing on the initial wind power-load data set to remove data concentration, abnormal data and low-probability scenario data to obtain a wind power-load data set; In some embodiments, the DBSCAN density clustering algorithm is used to perform noise reduction preprocessing on the initial wind power-load data set to eliminate abnormal data and low-probability wind power-load data, reduce the difficulty of constructing subsequent wind power-load typical scenarios, and improve typical characteristics.
[0028] Step 3: Segment the load data set in the wind power-load data set based on the ordered clustering method, determine the optimal number of segmentations and construct a first-order typical load scenario; In some embodiments, step 3 specifically includes: Determine the objective function, which is to minimize the sum of squared deviations; A genetic algorithm is used to optimize the number of segmentation points and calculate the silhouette coefficient of ordered clustering under different numbers of segmentation points; wherein the range of the number of segmentation points p can be: 1≤p≤15.
[0029] The number of segmentation points corresponding to the maximum silhouette coefficient is selected as the optimal number of segmentation points to construct a first-order typical load scenario.
[0030] Step 4: According to the first-order typical load scenario and the optimal number of splits, the corresponding wind power data set is divided into multiple scenario sets. The range of the number of clusters is set according to the number of load data in each scenario set, and the wind power data in each scenario set is clustered. The silhouette coefficients under different numbers of clusters are calculated, and the number of clusters corresponding to the maximum silhouette coefficient is selected as the optimal number of clusters to construct a second-order typical wind power scenario.
[0031] In some embodiments, the wind power data includes wind power, and step 4 specifically includes: Divide the wind power into p+1 scene sets V ,Right now ; K-means clustering is performed on the wind power in each scene set. For example, the unit of load data is 24 hours. The Euclidean distance is used to measure the distance between the load data center and each load data, that is, the Euclidean distance is used to measure the distance between sample vectors. The range of clustering number is set according to the number of load data in each scene set. For example, assuming the scenario set The number of internal load data is f , set the number of clusters N to 2≤N≤ The silhouette coefficients under different numbers of clusters are calculated, the number of clusters corresponding to the maximum silhouette coefficient is selected as the optimal number of clusters, and a second-order typical wind power scenario is constructed.
[0032] Based on the same inventive concept, the present invention also provides a wind power-load scenario construction device based on time series related clustering fusion, comprising: A collection unit, used to collect wind speed data and load data of the wind farm, construct a wind speed-wind power conversion function based on a time series, and obtain an initial wind power-load data set of the wind farm; A preprocessing unit, configured to perform noise reduction preprocessing on the wind power-load initial data set to obtain a wind power-load data set; A first construction unit is used to segment the load data set in the wind power-load data set based on an ordered clustering method, determine an optimal number of segmentations, and construct a first-order typical load scenario; The second construction unit is used to divide the corresponding wind power data set into multiple scenario sets according to the first-order typical load scenario and the optimal number of partitions, set the range of the number of clusters according to the number of load data in each scenario set, cluster the wind power data in each scenario set, calculate the silhouette coefficients under different numbers of clusters, select the number of clusters corresponding to the maximum silhouette coefficient as the optimal number of clusters, and construct a second-order typical wind power scenario.
[0033] The aforementioned wind power-load scenario construction device and method based on time series related clustering fusion correspond to each other and will not be elaborated here.
[0034] The following is a specific embodiment provided by the present invention.
[0035] The annual operating wind speed data and load data of a wind farm in a certain region in 2020 are collected as data samples. The interval of data samples is 1 hour. The wind power of the wind farm is solved based on the wind speed-wind power conversion function considering the time series, and the wind power is normalized to construct the initial wind power-load data set, as shown in Figure 2.
[0036] The wind power-load initial data set is preprocessed based on the DBSCAN density clustering denoising method to remove data concentration, abnormal data and low-probability scene data. The effect after preprocessing and denoising is shown in Figure 3.
[0037] After preprocessing and noise reduction of the initial wind power-load data set, 32 noise points are obtained. The time series distribution of the noise points, wind power per unit value and load data information are shown in Table 1.
[0038] After removing 32 noise points from the initial wind power-load data set, the wind power-load data set after noise reduction preprocessing is obtained. The wind power-load data set does not contain data concentration, abnormal data, and low-probability scenario data, which reduces the difficulty of subsequent load scenario construction and wind power scenario construction, making the constructed load scenario and wind power scenario more typical.
[0039] Table 1. Wind power-load initial data noise point information
[0040] For the load data in the denoised wind power-load dataset, a typical load scenario is constructed based on the ordered clustering method. The objective function is to minimize the sum of squared deviations. A genetic algorithm is used to optimize the number of splits in the range of [1,15]. The values of the silhouette coefficients for different numbers of splits are as follows: Figure 4 shown.
[0041] Depend on Figure 4 It can be seen that when the number of splits is in the range of [1,3], the corresponding silhouette coefficient value shows an upward trend. When the number of splits is in the range of [3,6], the corresponding silhouette coefficient value shows a short drop and then continues to rise. When the number of splits is 6, the silhouette coefficient value reaches the maximum. Therefore, the optimal number of splits is 6, that is, the load data is divided into 7 typical scenarios, such as Figure 5 The time series segmentation of the seven typical scenarios after load data division and the distribution information of the scenario proportion probability are shown in Table 2.
[0042] Table 2 Time series probability distribution of typical scenarios of load ordered clustering
[0043] Depend on Figure 5 From the division of typical scenarios, it can be seen that after the noise reduction and orderly clustering of the annual load data, the waveform changes of the seven typical scenarios are consistent with the time series change law of the load curve before clustering, and the load curves in 24-hour units in each typical load scenario also have similar change characteristics. Taking the typical load scenario 2 in the typical scenario as an example, the 24-hour distribution curve of the load in the period included in the typical load scenario 2 is as follows: Figure 6 shown.
[0044] Based on the ordered clustering method, the load data set of the pre-processed and denoised wind power-load data set is optimally segmented. When the segmentation point is 6, the corresponding contour coefficient value for evaluating the clustering effect is the largest, so the load data is divided into 7 typical load scenarios. The wind power data set is divided by the 6 segmentation points obtained by the ordered clustering of the load data, and 7 wind power data scene sets V are obtained, namely .
[0045] by Taking the wind power scene sample as an example, we cluster it based on the K-means method and use the Euclidean distance to measure the distance between sample vectors. The silhouette coefficients under different numbers of clusters are as follows: Figure 7 shown.
[0046] Depend on Figure 7It can be seen that when the number of K-means clusters is in the range of [3,7], the curve fluctuates greatly, showing an obvious upward or downward trend. When the number of clusters is in the range of [7,10], the corresponding silhouette coefficient value has been on an upward trend. When the number of clusters is greater than 10, the silhouette coefficient shows a downward trend. Therefore, when the number of clusters is 10, the corresponding silhouette coefficient value is the largest, then The optimal number of clusters for wind power scenario samples is 10. The results after clustering the wind power scene samples are as follows: Figure 8 shown.
[0047] Depend on Figure 8 It can be seen that The wind power scene samples are clustered into 10 typical wind power scenes, and the 24-hour wind power output curve change rules in each typical wind power scene are similar. In general, the overall change rules of different typical wind power scenes are similar to the morphological characteristics of the 24-hour load change curve in the corresponding typical load scene 2. Therefore, the denoised wind power-load scene clustering results based on ordered and K-means clustering provided by the present invention are reasonable, and the typical wind power scene construction results after clustering can reflect the output characteristics of various types.
[0048] In summary, the wind power-load scenario construction method and device based on time series related clustering fusion provided by the present invention fully considers the output of the previous moment in the wind farm and the influence of high and low recovery wind speeds, and establishes a wind speed-wind power conversion function based on the time series, thereby reducing the wind power calculation error caused by the time series problem. The wind power-load initial data set is constructed based on the wind speed-wind power conversion function of the time series, and the DBSCAN density clustering method is used to perform noise reduction preprocessing on the wind power-load initial data set, and then first-order ordered clustering is performed, and the wind power data is divided according to the time series segmentation results of the typical load scenario, and the divided wind power data is subjected to second-order K-means clustering, and finally the construction of the typical wind power-load scenario is completed. The present invention mines the typical scenario characteristics of the source-load output characteristics of the power grid connected to wind power, reduces the wind power calculation error, and improves the typicality of scenario construction; and can be used for the subsequent solution of the comprehensive reactive optimization strategy set under different typical wind power-load scenarios, and the reactive optimization model of each scenario is constructed to improve the convergence, accuracy and scenario generalization ability of the model. Lay the foundation.
[0049] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the embodiments disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed by the present invention. It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.
Claims
1. A wind power-load scenario construction method based on time series correlation clustering fusion, It is characterized in that include: Step 1: Collect wind speed data and load data of the wind farm, construct a wind speed-wind power conversion function based on a time series, and obtain an initial wind power-load data set of the wind farm; Step 2: performing noise reduction preprocessing on the wind power-load initial data set to obtain a wind power-load data set; Step 3: segment the load data set in the wind power-load data set based on an ordered clustering method, determine the optimal number of segmentations, and construct a first-order typical load scenario; Step 4: According to the first-order typical load scenario and the optimal number of splits, the corresponding wind power data set is divided into multiple scenario sets, the range of the number of clusters is set according to the number of load data in each scenario set, and the wind power data in each scenario set is clustered, and the silhouette coefficients under different numbers of clusters are calculated. The number of clusters corresponding to the maximum silhouette coefficient is selected as the optimal number of clusters to construct a second-order typical wind power scenario.
2. According to claim 1, the wind power-load scenario construction method based on time series related clustering fusion, It is characterized in that In step 1, the wind speed-wind power conversion function based on time series is: when or hour, when hour, when hour, In the formula, For wind farms Wind power at the time; For wind farms Wind speed at the moment; The cut-in wind speed set for the wind farm; The cut-out wind speed set for the wind farm; is the rated wind speed of the wind farm; is the density of air, A is the swept area of the wind turbines in the wind farm; is the rated power of the wind farm; is the wind power of the wind farm at the last moment; and They are the low recovery wind speed and high recovery wind speed set for the wind farm respectively.
3. According to claim 2, the wind power-load scenario construction method based on time series related clustering fusion, It is characterized in that The expression of the swept area of the wind turbines in the wind farm is: A= 2 πr 2 In the formula, A is the swept area of the wind turbines in the wind farm, r is the radius of the wind turbine in the wind farm.
4. The wind power-load scenario construction method based on time series related clustering fusion according to claim 1, It is characterized in that In the step 2, the wind power-load initial data set is subjected to denoising preprocessing using the DBSCAN density clustering algorithm.
5. The wind power-load scenario construction method based on time series related clustering fusion according to claim 1, It is characterized in that The step 3 specifically includes: Determine the objective function; Genetic algorithm is used to optimize the number of segmentation points and the silhouette coefficient of ordered clustering under different numbers of segmentation points is calculated; The number of segmentation points corresponding to the maximum silhouette coefficient is selected as the optimal number of segmentation points to construct a first-order typical load scenario.
6. The wind power-load scenario construction method based on time series related clustering fusion according to claim 5, It is characterized in that The objective function is to minimize the sum of squared deviations.
7. The wind power-load scenario construction method based on time series related clustering fusion according to claim 6, It is characterized in that The range of the number of segmentation points is: 1≤p≤15 Among them, p is the number of segmentation points.
8. The wind power-load scenario construction method based on time series related clustering fusion according to claim 7, It is characterized in that The wind power data includes wind power, and step 4 specifically includes: Divide the wind power into p+1 scene sets; K-means clustering is performed on the wind power in each scenario set. The range of the number of clusters is set according to the number of load data in each scenario set. The silhouette coefficients under different numbers of clusters are calculated. The number of clusters corresponding to the maximum silhouette coefficient is selected as the optimal number of clusters to construct a second-order typical wind power scenario.
9. The wind power-load scenario construction method based on time series related clustering fusion according to claim 8, It is characterized in that The range of the number of clusters is: 2≤N≤ in, f is the number of load data, and N is the number of clusters.
10. A wind power-load scenario construction device based on time series related clustering fusion, It is characterized in that include: A collection unit, used to collect wind speed data and load data of the wind farm, construct a wind speed-wind power conversion function based on a time series, and obtain an initial wind power-load data set of the wind farm; A preprocessing unit, configured to perform noise reduction preprocessing on the wind power-load initial data set to obtain a wind power-load data set; A first construction unit is used to segment the load data set in the wind power-load data set based on an ordered clustering method, determine an optimal number of segmentations, and construct a first-order typical load scenario; The second construction unit is used to divide the corresponding wind power data set into multiple scenario sets according to the first-order typical load scenario and the optimal number of partitions, set the range of the number of clusters according to the number of load data in each scenario set, cluster the wind power data in each scenario set, calculate the silhouette coefficients under different numbers of clusters, select the number of clusters corresponding to the maximum silhouette coefficient as the optimal number of clusters, and construct a second-order typical wind power scenario.
Citation Information
Patent Citations
Wind power prediction method based on time sequence and neural network method
CN102880810A
Power distribution network distributed photovoltaic and energy storage power station investment decision-making method
CN113972694A
Short-term wind power prediction method considering fluctuation process division
CN115423174A
New energy unit investment and construction strategy optimization method and system considering capacity compensation
CN116191568A
Method and system for optimizing cut-in / cut-out wind speed of wind turbine unit, device, and medium
WO2023020524A1
