Wind Farm Clustering System and Method Based on Pearson Correlation Coefficient

Abstract

A wind farm clustering system based on the Pearson correlation coefficient, comprising: a historical power generation data acquisition module, a wind farm active power output time series vector construction module, a Pearson correlation coefficient calculation module, a wind farm clustering module and a clustering analysis module; the clustering analysis module performs clustering analysis based on the data provided by the above modules and then outputs the total number of wind farm group divisions M , the wind farms included in each subgroup and the number of wind farms K , the coincidence rate of each wind farm group s results. In this way, there is no need to preset the number of subgroups in advance, but the wind farms are automatically clustered into a certain number of wind farm groups according to the historical data of the wind farm output, which can make the aggregation result more in line with the actual situation. The present invention also provides a wind farm clustering method based on the Pearson correlation coefficient.

Description

Technical Field:

[0001] The present invention relates to the technical field of wind farm monitoring in power systems, and particularly relates to a wind farm clustering system and method based on the Pearson correlation coefficient. Background Art:

[0002] In China, the proportion of thermal power installed capacity in the energy structure has been rapidly decreasing, and there is also a differentiation within renewable energy power generation. Among them, the growth rates of wind power and solar energy are significantly faster than those of hydropower and nuclear power, and solar energy is even faster than wind power. Solar energy is expected to replace hydropower as the second largest power source.

[0003] However, with the rapid increase in the penetration rate of new energy, the imbalance between the output of new energy and that of conventional units has intensified, and the frequency and voltage problems caused by power imbalance are relatively prominent. New energy has large randomness and volatility, making it difficult to be effectively incorporated into power balance. During peak load periods, the peak shaving capacity is severely insufficient, the overall power balance margin is continuously decreasing, and the pressure of power supply guarantee is continuously increasing. The simultaneity rate of new energy power generation reflects the comprehensive capacity utilization of new energy power plants in a region. For wind farms in the same region, there are certain differences in wind resources in different parts of the region, and there are also certain differences in the simultaneity rates of power generation of each wind farm. The outputs of different wind farms are superimposed and offset each other. Therefore, it is very necessary to analyze the historical data of wind farm power generation, find out the power generation fluctuation laws of wind farms, and cluster and partition wind farms with similar fluctuation laws to accurately calculate the simultaneity rate of wind farm output. Summary of the Invention:

[0004] In view of this, it is necessary to provide a wind farm clustering system based on the Pearson correlation coefficient to calculate the simultaneity rate of the output of a certain type of wind farm through this system.

[0005] It is also necessary to provide a wind farm clustering method based on the Pearson correlation coefficient.

[0006] A wind farm clustering system based on the Pearson correlation coefficient includes: a historical power generation data acquisition module, a wind farm active power output time series vector construction module, a Pearson correlation coefficient calculation module, a wind farm clustering module, and a clustering analysis module;

[0007] The historical power generation data acquisition module is used to obtain the historical power generation data of the wind farm within a predetermined period of time, where the historical power generation data refers to the actual output and the maximum output of the recorded wind farm;

[0008] The wind farm active power output time series vector construction module is used to construct the active power output time series vector P of each wind farm according to the obtained historical power generation data i ={P i1 , P i2 , Pi3 , …, P in};

[0009] The Pearson correlation coefficient calculation module is used to randomly select a wind farm that has not been clustered into a group as the first wind farm, and use formula one to calculate the Pearson correlation coefficients between the first wind farm and all the remaining wind farms that have not been clustered into a group one by one. Among them, the wind farm selected from all the remaining wind farms that have not been clustered into a group is used as the second wind farm:

[0010]

[0011] where k represents the subscript of the time series, respectively represent the expected values of the active power output time series vectors of the first wind farm and the second wind farm;

[0012] The wind farm clustering module is used to compare each calculated Pearson correlation coefficient with a preset Pearson correlation coefficient threshold value. When it is compared that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient are clustered into a group; when it is compared that the Pearson correlation coefficient is not greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm cannot be clustered into a group;

[0013] The clustering analysis module is used to judge whether there are still wind farms that have not been clustered into a group. When it is judged that there are no wind farms that have not been clustered into a group, the clustering analysis results of the wind farms are output, including: the total number M of wind farm group divisions, the wind farms and the number K of wind farms included in each subgroup, and the simultaneity rate s of each wind farm group. Among them, the simultaneity rate s of the mth wind farm group m is obtained by calculating through the following formula two:

[0014]

[0015] where P ic is the installed capacity of wind farm i.

[0016] A wind farm clustering system method based on Pearson correlation coefficient includes the following steps:

[0017] Obtain the historical power generation data of the wind farms within a predetermined period of time. Among them, the historical power generation data refers to the actual output and the maximum output of the recorded wind farms;

[0018] Construct the active power output time series vector P of each wind farm according to the obtained historical power generation data i ={P i1 , P i2 , Pi3 , …, P in};

[0019] Randomly select a wind farm that has not been clustered into a group as the first wind farm, and use formula one to calculate the Pearson correlation coefficient between all the remaining wind farms that have not been clustered into a group and the first wind farm. Among them, the wind farm selected from all the remaining wind farms that have not been clustered into a group is used as the second wind farm:

[0020]

[0021] Among them, k represents the subscript of the time series, respectively represent the expected values of the active power output time series vectors of the first wind farm and the second wind farm;

[0022] Compare each calculated Pearson correlation coefficient with a preset Pearson correlation coefficient threshold value. When it is compared that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient are clustered into a group; when it is compared that the Pearson correlation coefficient is not greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm cannot be clustered into a group;

[0023] Judge whether there are still wind farms that have not been clustered into a group. When it is judged that there are no wind farms that have not been clustered into a group, output the clustering analysis results of the wind farms, including: the total number M of wind farm group divisions, the wind farms and the number K of wind farms included in each sub-group, and the simultaneity rate s of each wind farm group. Among them, the simultaneity rate s of the mth wind farm group m is obtained by calculating through the following formula two:

[0024]

[0025] Among them, P ic is the installed capacity of wind farm i.

[0026] In the above wind farm clustering system and method based on the Pearson correlation coefficient, by obtaining the historical power generation data of the wind farms within a predetermined period of time, and constructing the active power output time series vector P of each wind farm according to the obtained historical power generation data i ={P i1 , P i2 , P i3 , …, P in}; Randomly select a wind farm that has not been clustered into a group as the first wind farm, and use the formula to calculate the Pearson correlation coefficient between the first wind farm and all the remaining wind farms that have not been clustered into a group one by one. Compare each calculated Pearson correlation coefficient with the preset Pearson correlation coefficient threshold value. When it is found that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, cluster the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient into one group; when it is found that the Pearson correlation coefficient is not greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm cannot be clustered into a group. When it is determined that there are no wind farms that have not been clustered into a group, output the clustering analysis result of the wind farms. The clustering analysis result includes: the total number M of wind farm group divisions, the wind farms and the number K of wind farms included in each sub-group, and the coincidence rate s of each wind farm group. In this way, there is no need to preset the number of sub-groups in advance, but the wind farms are automatically clustered into a certain number of wind farm groups according to the historical data of wind farm output, which can make the aggregation result more in line with the actual situation. BRIEF DESCRIPTION OF THE DRAWINGS:

[0027] Figure 1 It is a schematic diagram of the functional modules of the wind farm clustering system based on Pearson correlation coefficient of the present invention.

[0028] Figure 2 It is a flowchart of the wind farm clustering method based on Pearson correlation coefficient of the present invention. In the figure: the wind farm clustering system 10 based on Pearson correlation coefficient, the historical power generation data acquisition module 20, the wind farm active power output time series vector construction module 30, the Pearson correlation coefficient calculation module 40, the wind farm clustering module 50, the clustering analysis module 60, the correlation analysis module 70, the storage module 80, and the steps S101 to S109 of the wind farm clustering method based on Pearson correlation coefficient. DETAILED DESCRIPTION OF THE EMBODIMENTS:

[0029] The present invention discloses a wind farm clustering analysis solution based on Pearson correlation coefficient, which can be suitable for the clustering analysis of wind farms in large power grids at the provincial level and above, lay a foundation for accurately calculating the coincidence rate of wind farm output, and provide technical support for further improving the refinement level of annual power system operation planning.

[0030] Please also refer to Figure 1, The wind farm clustering system 10 based on the Pearson correlation coefficient is introduced in detail below. This system includes a historical power generation data acquisition module 20, a wind farm active power output time series vector construction module 30, a Pearson correlation coefficient calculation module 40, a wind farm clustering module 50, a clustering analysis module 60, and a storage module 80. Among them, the storage module 80 is used to store the data involved in the technical solution of this application. The historical power generation data acquisition module 20, the wind farm active power output time series vector construction module 30, the Pearson correlation coefficient calculation module 40, the wind farm clustering module 50, and the clustering analysis module 60 obtain the required data or store the processed data from the storage module 80. The historical power generation data acquisition module 20, the wind farm active power output time series vector construction module 30, the Pearson correlation coefficient calculation module 40, the wind farm clustering module 50, and the clustering analysis module 60 are all data processing devices with data processing capabilities. After each module in the wind farm clustering system 10 based on the Pearson correlation coefficient runs the corresponding computer program code, data exchange between them is realized and the following functions are provided:

[0031] The historical power generation data acquisition module 20 is used to obtain the historical power generation data of the wind farm within a predetermined period of time. Among them, the historical power generation data refers to the recorded actual output and maximum output of the wind farm;

[0032] The wind farm active power output time series vector construction module 30 is used to construct the active power output time series vector P of each wind farm according to the obtained historical power generation data i ={P i1 , P i2 , P i3 , …, P in};

[0033] The Pearson correlation coefficient calculation module 40 is used to randomly select a wind farm that has not been clustered into a group as the first wind farm, and use formula one to traverse and calculate the Pearson correlation coefficient between the remaining all wind farms that have not been clustered into a group and the first wind farm. Among them, the wind farm selected from the remaining all wind farms that have not been clustered into a group is used as the second wind farm:

[0034]

[0035] where k represents the subscript of the time series, respectively represent the expected values of the active power output time series vectors of the first wind farm and the second wind farm;

[0036] The wind farm clustering module 50 is used to compare each calculated Pearson correlation coefficient with a preset Pearson correlation coefficient threshold value. When it is compared that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient are clustered into a group; when it is compared that the Pearson correlation coefficient is not greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm cannot be clustered into a group; for example, the preset Pearson correlation coefficient threshold value is 0.2. Among them, the wind farm clustering module 50 makes a clustering completion flag for the clustered first wind farm and second wind farm, and the Pearson correlation coefficient calculation module 40 judges whether the selected wind farms are clustered according to the clustering completion flag.

[0037] The clustering analysis module 60 is used to judge whether there are wind farms that have not been clustered into groups. When it is judged that there are no wind farms that have not been clustered into groups, the clustering analysis result of the wind farms is output, including: the total number M of wind farm group divisions, the wind farms and the number K of wind farms included in each sub-group, and the simultaneity rate s of each wind farm group. Among them, the simultaneity rate s of the mth wind farm group m is obtained by calculating through the following formula two:

[0038]

[0039] Among them, P ic is the installed capacity of wind farm i.

[0040] Furthermore, the clustering analysis module 60 is also used to output a preset execution message to the Pearson correlation coefficient calculation module 40 when it is judged that there are wind farms that have not been clustered into groups, so that the Pearson correlation coefficient calculation module 40 works.

[0041] Furthermore, it also includes a correlation analysis module 70. The correlation analysis module 70 analyzes the correlation between the first wind farm and the second wind farm corresponding to the calculated Pearson correlation coefficient, and outputs the corresponding result message to the wind farm clustering module 50 according to the analysis result of the correlation between the first wind farm and the second wind farm, so that the wind farm clustering module 50 starts the clustering operation. Among them, the correlation rule is: the Pearson correlation coefficient greater than zero indicates that P i and P j are positively correlated, and less than zero indicates that P i and P j are negatively correlated; the absolute value of the Pearson correlation coefficient is between (0.8, 1.0] indicates that P i and Pj Very strongly correlated, between (0.6, 0.8] indicates P i and P j Strongly correlated, between (0.4, 0.6] indicates P i and P j Moderately correlated, between (0.2, 0.4] indicates P i and P j Weakly correlated, between [0.0, 0.2] indicates P i and P j Very weakly correlated or uncorrelated;

[0042] When the clustering analysis module 60 compares and finds that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, it outputs a start information to the correlation analysis module 70. In response to the start information, the correlation analysis module 70 analyzes the correlation between the first wind farm and the second wind farm according to the rules of the correlation and the calculated Pearson correlation coefficient. When the correlation analysis module 70 analyzes and determines that the first wind farm and the second wind farm are any one of the results of weakly correlated, strongly correlated, and very strongly correlated, it outputs a result information to the clustering analysis module 60. The clustering analysis module 60 clusters the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient into a group according to the result information.

[0043] The following combines specific examples to further describe the wind farm clustering system 10 based on the Pearson correlation coefficient. Taking 14 wind farms in a certain area as an example for analysis, with a 15-minute time interval, the wind farm clustering system 10 based on the Pearson correlation coefficient takes the active power output time series vectors of the wind farms at 365×96 time points throughout the year. The calculation results of the Pearson correlation coefficients of the wind farms are shown in Table 1, and the clustering results are shown in Table 2:

[0044] Table 1 Pearson Correlation Coefficients of Wind Farms

[0045] wind field 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 1.00 0.31 -0.01 0.00 0.20 0.05 0.18 -0.03 -0.08 -0.04 0.27 0.28 0.25 -0.02 2 0.31 1.00 0.06 0.09 0.19 -0.12 0.03 0.15 0.10 0.11 0.07 0.31 0.36 -0.07 3 -0.01 0.06 1.00 0.71 0.61 0.59 0.69 0.58 0.65 0.59 0.38 0.23 0.31 0.53 4 0.00 0.09 0.71 1.00 0.72 0.53 0.63 0.62 0.67 0.68 0.42 0.30 0.28 0.47 5 0.20 0.19 0.61 0.72 1.00 0.38 0.56 0.31 0.36 0.37 0.34 0.34 0.44 0.27 6 0.05 -0.12 0.59 0.53 0.38 1.00 0.75 0.41 0.49 0.43 0.64 0.47 0.04 0.79 7 0.18 0.03 0.69 0.63 0.56 0.75 1.00 0.49 0.57 0.51 0.61 0.47 0.25 0.59 8 -0.03 0.15 0.58 0.62 0.31 0.41 0.49 1.00 0.92 0.82 0.37 0.32 0.17 0.58 9 -0.08 0.10 0.65 0.67 0.36 0.49 0.57 0.92 1.00 0.79 0.38 0.30 0.27 0.60 10 -0.04 0.11 0.59 0.68 0.37 0.43 0.51 0.82 0.79 1.00 0.39 0.32 0.19 0.57 11 0.27 0.07 0.38 0.42 0.34 0.64 0.61 0.37 0.38 0.39 1.00 0.54 0.31 0.51 12 0.28 0.31 0.23 0.30 0.34 0.47 0.47 0.32 0.30 0.32 0.54 1.00 0.25 0.45 13 0.25 0.36 0.31 0.28 0.44 0.04 0.25 0.17 0.27 0.19 0.31 0.25 1.00 -0.06 14 -0.02 -0.07 0.53 0.47 0.27 0.79 0.59 0.58 0.60 0.57 0.51 0.45 -0.06 1.00

[0046] Table 2 Wind Farm Clustering Results

[0047] wind farm cluster wind farm composition number of wind farms coincidence rate 1 1、2、5、11、12、13 6 26.7％ 2 3、4、6、7、8、9、10、14 8 43.2％

[0048] A wind farm clustering method provided by the present invention includes the following steps:

[0049] Step S101, obtaining historical power generation data of a wind farm within a predetermined period of time, where the historical power generation data refers to the actual output and the maximum output of the recorded wind farm;

[0050] Step S103: Construct the active power output time series vector \(P\) for each wind farm based on the obtained historical power generation data i =\(\{P\) i1 , \(P\) i2 , \(P\) i3 , \(\cdots\), \(P\) in \};

[0051] Step S105: Randomly select a wind farm that has not been clustered as the first wind farm, and use Equation (1) to calculate the Pearson correlation coefficient between the first wind farm and all the remaining wind farms that have not been clustered one by one. Among them, the wind farm selected from all the remaining wind farms that have not been clustered is used as the second wind farm:

[0052]

[0053] where \(k\) represents the subscript of the time series, respectively represent the expected values of the active power output time series vectors of the first wind farm and the second wind farm;

[0054] Step S107: Compare each calculated Pearson correlation coefficient with the preset Pearson correlation coefficient threshold. When it is found that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold, cluster the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient into one group; when it is found that the Pearson correlation coefficient is not greater than the preset Pearson correlation coefficient threshold, the first wind farm and the second wind farm cannot be clustered into one group. For example, the preset Pearson correlation coefficient threshold is 0.2.

[0055] Step S109: Determine whether there are still wind farms that have not been clustered. When it is determined that there are no wind farms that have not been clustered, output the clustering analysis results of the wind farms, including: the total number \(M\) of wind farm group divisions, the wind farms and the number \(K\) of wind farms included in each subgroup, and the simultaneity rate \(s\) of each wind farm group. Among them, the simultaneity rate \(s\) of the \(m\)-th wind farm group m is obtained by calculating with Equation (2) as follows:

[0056]

[0057] where \(P\) ic is the installed capacity of wind farm \(i\).

[0058] Furthermore, when it is determined that there are still wind farms that have not been clustered, execute the step of "randomly select a wind farm that has not been clustered as the first wind farm, and use Equation (1) to calculate the Pearson correlation coefficient between the first wind farm and all the remaining wind farms that have not been clustered one by one".

[0059] Further, the method further includes the following steps: analyzing the correlation between the first wind farm and the second wind farm corresponding to the calculated Pearson correlation coefficient according to the Pearson correlation coefficient, and determining whether to start clustering the first wind farm and the second wind farm into a group according to the result of the analyzed correlation between the first wind farm and the second wind farm. Wherein, the rule of correlation is: Pearson correlation coefficient greater than zero indicates that P i and P j are positively correlated, and less than zero indicates that P i and P j are negatively correlated; the absolute value of the Pearson correlation coefficient between (0.8, 1.0] indicates that P i and P j are extremely strongly correlated, between (0.6, 0.8] indicates that P i and P j are strongly correlated, between (0.4, 0.6] indicates that P i and P j are moderately correlated, between (0.2, 0.4] indicates that P i and P j are weakly correlated, between [0.0, 0.2] indicates that P i and P j are extremely weakly correlated or uncorrelated;

[0060] Among them, the step of "when it is compared that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, then clustering the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient into a group" in step S107 is specifically: when it is compared that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, analyzing the correlation between the first wind farm and the second wind farm according to the rule of the correlation and the calculated Pearson correlation coefficient, and when it is analyzed and determined that the first wind farm and the second wind farm are any one of the results of weak correlation, strong correlation, and extremely strong correlation, then clustering the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient into a group.

[0061] In the above wind farm clustering system and method based on the Pearson correlation coefficient, by obtaining the historical power generation data of the wind farm within a predetermined period of time, constructing the active power output time series vector P of each wind farm according to the obtained historical power generation data i ={P i1 , P i2 , P i3 , …, P in}; Randomly select a wind farm that has not been clustered into a group as the first wind farm, and use the formula to calculate the Pearson correlation coefficient between all the remaining wind farms that have not been clustered into a group and the first wind farm one by one. Compare each calculated Pearson correlation coefficient with the preset Pearson correlation coefficient threshold value. When it is compared that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, cluster the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient into a group; when it is compared that the Pearson correlation coefficient is not greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm cannot be clustered into a group; when it is judged that there is no wind farm that has not been clustered into a group, output the clustering analysis result of the wind farms. The clustering analysis result includes: the total number of wind farm group divisions M, the wind farms and the number of wind farms K included in each subgroup, and the simultaneity rate s of each wind farm group. In this way, there is no need to preset the number of subgroups in advance, but automatically cluster the wind farms into a certain number of wind farm groups according to the historical data of wind farm output, which can make the aggregation result more in line with the actual situation.

Claims

1. A wind farm clustering system based on the Pearson correlation coefficient, characterized in that Including: A historical power generation data acquisition module, a wind farm active power output time series vector construction module, a Pearson correlation coefficient calculation module, a wind farm clustering module, and a clustering analysis module; The historical power generation data acquisition module is used to obtain the historical power generation data of a wind farm within a predetermined period of time. Herein, the historical power generation data refers to the recorded actual output and maximum output of the wind farm; The active power output time series vector construction module of the wind farm is used to construct the active power output time series vector P of each wind farm according to the obtained historical power generation data i ={P i1 , P i2 , P i3 , …, P in}; The Pearson correlation coefficient calculation module is used to randomly select a wind farm that has not been clustered into a group as the first wind farm, and use Formula 1 to calculate the Pearson correlation coefficients between the first wind farm and all the remaining wind farms that have not been clustered into a group. Herein, a wind farm selected from all the remaining wind farms that have not been clustered into a group is used as the second wind farm: where k represents the subscript of the time series, respectively representing the expected values of the active power output time series vectors of the first and second wind farms; The wind farm clustering module is used to compare each calculated Pearson correlation coefficient with a preset Pearson correlation coefficient threshold value. When it is compared that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient are clustered into a group; when it is compared that the Pearson correlation coefficient is not greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm cannot be clustered into a group; The clustering analysis module is used to determine whether there are wind farms that have not been clustered into groups. When it is determined that there are no wind farms that have not been clustered into groups, the clustering analysis results of the wind farms are output, including: the total number M of wind farm group divisions, the wind farms and the number K of wind farms included in each subgroup, and the simultaneity rate s of each wind farm group. Among them, the simultaneity rate s of the m-th wind farm group m is obtained by calculating through the following formula two: Among them, P ic is the installed capacity of wind farm i.

2. The wind farm clustering system based on the Pearson correlation coefficient according to claim 1, wherein: The preset Pearson correlation coefficient threshold value is 0.

2.

3. The wind farm clustering system based on the Pearson correlation coefficient according to claim 1, characterized in that: The clustering analysis module is further used to output a preset execution message to the Pearson correlation coefficient calculation module when it is determined that there are still wind farms that have not been clustered into a group, so that the Pearson correlation coefficient calculation module can work.

4. The Pearson correlation coefficient-based wind farm clustering system according to claim 1 or 3, characterized in that: It further includes a correlation analysis module. The correlation analysis module analyzes the correlation between the first wind farm and the second wind farm corresponding to the calculated Pearson correlation coefficient, and outputs corresponding result information to the wind farm clustering module according to the analysis result of the correlation between the first wind farm and the second wind farm, so that the wind farm clustering module can start the clustering operation.

5. The wind farm clustering system based on Pearson correlation coefficient according to claim 4, characterized in that: The rule of correlation is: Pearson correlation coefficient Greater than zero indicates that P i and P j are positively correlated. Less than zero indicates that P i and P j are negatively correlated; The absolute value of the Pearson correlation coefficient between (0.8, 1.0] indicates that P i and P j are extremely strongly correlated. Between (0.6, 0.8] indicates that P i and P j are strongly correlated. Between (0.4, 0.6] indicates that P i and P j are moderately correlated. Between (0.2, 0.4] indicates that P i and P j are weakly correlated. Between [0.0, 0.2] indicates that P i and P j are extremely weakly correlated or uncorrelated; When the clustering analysis module compares that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, it outputs a start message to the correlation analysis module. The correlation analysis module responds to the start message and analyzes the correlation between the first wind farm and the second wind farm according to the correlation rule and the calculated Pearson correlation coefficient. When the correlation analysis module analyzes and determines that the first wind farm and the second wind farm are any one of weakly correlated, strongly correlated, and extremely strongly correlated, it outputs result information to the clustering analysis module. The clustering analysis module clusters the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient into a group according to the result information.

6. A wind farm clustering system method based on Pearson correlation coefficient, including the following steps: Obtain the historical power generation data of a wind farm within a predetermined period of time. Herein, the historical power generation data refers to the recorded actual output and maximum output of the wind farm; Construct the active power output time series vector \(P\) for each wind farm based on the obtained historical power generation data i =\(\{P\) i1 , \(P\) i2 , \(P\) i3 , \(\cdots\), \(P\) in \}; Randomly select a wind farm that has not been clustered into a group as the first wind farm, and use the formula to calculate the Pearson correlation coefficient between the first wind farm and all the remaining wind farms that have not been clustered into a group one by one. Among them, the wind farm selected from all the remaining wind farms that have not been clustered into a group is used as the second wind farm: where k represents the subscript of the time series, respectively representing the expected values of the active power output time series vectors of the first and second wind farms; Compare each calculated Pearson correlation coefficient with the preset Pearson correlation coefficient threshold value. When it is compared that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient are clustered into a group; when it is compared that the Pearson correlation coefficient is not greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm cannot be clustered into a group; Determine whether there are wind farms that have not been clustered into groups. When it is determined that there are no wind farms that have not been clustered into groups, output the clustering analysis results of the wind farms, including: the total number of wind farm group divisions M, the wind farms and the number of wind farms K included in each subgroup, and the simultaneity rate s of each wind farm group. Among them, the simultaneity rate s of the mth wind farm group m is obtained by calculating through the following formula two: Among them, P ic is the installed capacity of wind farm i.

7. The method of the wind farm clustering system based on the Pearson correlation coefficient according to claim 6, characterized in that: The preset Pearson correlation coefficient threshold value is 0.

2.

8. The method of the wind farm clustering system based on the Pearson correlation coefficient according to claim 6, characterized in that: When it is judged that there are still wind farms that have not been clustered into a group, execute the step of "randomly select a wind farm that has not been clustered into a group as the first wind farm, and use the formula to calculate the Pearson correlation coefficient between the first wind farm and all the remaining wind farms that have not been clustered into a group one by one".

9. The method of the wind farm clustering system based on the Pearson correlation coefficient according to claim 6 or 8, characterized in that, It also includes the following steps: It also includes the following steps: Analyze the correlation between the first wind farm and the second wind farm corresponding to the calculated Pearson correlation coefficient, and determine whether to start the clustering operation for the first wind farm and the second wind farm according to the analysis results of the correlation between the first wind farm and the second wind farm.

10. The method of the wind farm clustering system based on the Pearson correlation coefficient according to claim 9, characterized in that, The rule of correlation is: Pearson correlation coefficient Greater than zero indicates that P i and P j are positively correlated. Less than zero indicates that P i and P j are negatively correlated; The absolute value of the Pearson correlation coefficient between (0.8, 1.0] indicates that P i and P j are extremely strongly correlated. Between (0.6, 0.8] indicates that P i and P j are strongly correlated. Between (0.4, 0.6] indicates that P i and P j are moderately correlated. Between (0.2, 0.4] indicates that P i and P j are weakly correlated. Between [0.0, 0.2] indicates that P i and P j are extremely weakly correlated or uncorrelated; The step of "when it is compared that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient are clustered into a group" is specifically: when it is compared that the Pearson correlation coefficient is greater than the preset Pearson correlation coefficient threshold value, analyze the correlation between the first wind farm and the second wind farm according to the correlation rules and the calculated Pearson correlation coefficient. When it is analyzed and judged that the first wind farm and the second wind farm are any one of weak correlation, strong correlation, and extremely strong correlation, the first wind farm and the second wind farm corresponding to each Pearson correlation coefficient are clustered into a group.

Images