A reactive power partitioning and control method for a wind farm based on cluster analysis

Abstract

The application discloses a kind of wind farm reactive partition and control method based on cluster analysis belonging to reactive voltage control technical field of wind farm grid connection, comprising: analyzing the reactive power ability of double-fed unit in wind farm and its influencing factors, combining the historical operation data stored by wind farm AVC system, clustering analysis is carried out to wind turbine generator, the reactive power regulation ability of partition and wind farm is calculated;After determining the reactive power demand by grid dispatching instruction and grid connection point voltage information, the corresponding control strategy is given by the reactive power regulation ability of each level, the wind farm is no longer equivalent to one or several units for control, the voltage fluctuation of internal node of wind farm is considered;Make full use of the reactive power ability of double-fed wind turbine, reduce the fluctuation of machine terminal voltage, improve the voltage stability of unit, improve the safety of wind turbine generator operation.The application is easy to implement, and has strong operability.

Description

Technical Field

[0001] This invention belongs to the field of reactive power and voltage control technology for grid-connected wind farms, and specifically relates to a reactive power zoning and control method for wind farms based on cluster analysis. Background Technology

[0002] In recent years, with the increasing prominence of energy and environmental issues, the promotion of new energy power generation has become urgent. As a major component of new energy power generation, wind power's grid connection rate is continuously increasing, and its impact on voltage cannot be ignored. Large wind farms, due to the terrain and topography of their installation sites, have wide spatial spans, large wind speed differences, and long internal power collection lines. Without reactive power compensation, under certain external disturbances, some turbine terminals may experience voltage exceeding limits, thus affecting the safe operation of the system. As the mainstream model in current wind farms, the double-fed induction generator (DFIG) inherently possesses a certain reactive power compensation capability. Compared to other reactive power compensation devices, such as SVC and SVG, using DFIG's reactive power control capability is more cost-effective, and its reactive power control capability should be fully utilized.

[0003] Existing research on grid-connected reactive power and voltage control in wind farms has yielded many effective results. However, most past studies have treated wind farms as equivalent to one or a few turbines for grid-connected reactive power and voltage control, keeping the grid connection point voltage within a reasonable range. This does not reflect the voltage situation at each node within the wind farm, and the turbine terminal voltage may still exceed the limit. Many studies have also considered the coordinated control of all reactive power equipment, including all wind turbines, with the optimization objectives of minimizing node deviation, minimizing grid loss, and maximizing dynamic reactive power reserve. However, large wind farms often have hundreds or thousands of turbines. If each turbine is considered, the curse of dimensionality will occur, which is not conducive to real-time control. Summary of the Invention

[0004] The purpose of this invention is to propose a method for reactive power zoning and control of wind farms based on cluster analysis, characterized by the following steps:

[0005] Step 1: Obtain power grid data and wind farm operation data;

[0006] Step 2: Perform power flow calculations on the entire power grid to obtain the reactive voltage characteristics of each node in the wind farm, analyze the reactive power regulation capability of the doubly-fed induction generator (DFIG) to determine its influencing factors, and combine the relevant data from Step 1 to divide all units in the wind farm using a clustering algorithm.

[0007] Step 3: Combine the wind farm operation data in Step 1 to obtain the reactive power and voltage control capability of the units in each region, and take the unit with the smallest reactive power control capability in the region as the representative of the reactive power output of all units in the region, and calculate the reactive power control capability of each region and wind farm.

[0008] Step 4: Combining the power grid data and grid connection point voltage information from Step 1, calculate the reactive power demand at the current moment, and provide corresponding control strategies based on the reactive power regulation capabilities at each level.

[0009] In step 1, the power grid data includes power grid voltage and power grid dispatch instructions; the wind farm operation data includes wind farm topology, line parameters, transformer parameters, dynamic reactive power compensation device capacity, unit output power, and historical active power output data of the units.

[0010] In step 2, after determining the influencing factors of the unit's reactive power control capability, relevant historical data is obtained, normalized, and a clustering algorithm is used to divide the region.

[0011] Step 2, which uses a clustering algorithm to divide all the units in the wind farm, includes the following steps:

[0012] Step 21: Analysis of the doubly fed generator unit shows that active power output and stator voltage are the influencing factors of the unit's reactive power control capability. Assuming that the stator voltage remains constant, the active power output of the unit is used as the source data for cluster analysis.

[0013] Step 22: Normalize the historical active power output data;

[0014] Step 23, apply the Kuyama & Sugeno judgment criterion (V FS Determine the number of categories;

[0015]

[0016] in, V represents the membership degree from the j-th data point to the i-th class center. FS The minimum point corresponds to the optimal number of clusters;

[0017] Step 24: Divide all units in the wind farm into zones using fuzzy c-means clustering.

[0018] In step 3, the reactive power regulation capability of the unit is:

[0019]

[0020] Among them, Q max Q min Q represents the upper and lower limits of the unit's reactive power output; smax Q sminQ represents the upper and lower limits of the reactive power output of the stator-side converter; cmax Q cmin Q represents the upper and lower limits of the reactive power output of the grid-side converter; limit This refers to the reactive power control capability of the generator unit.

[0021] The reactive power regulation capability of the clustered wind farm is as follows:

[0022]

[0023] Where K is the total number of areas within the wind farm, and c i Let be the number of units in region i. The upper and lower limits of reactive power output of wind farms after zoning; These are the upper and lower limits of reactive power output of all units within region i; It refers to the reactive power regulation capability of wind farms after zoning.

[0024] The reactive power regulation capability of a wind farm is:

[0025]

[0026] Where N is the number of feeders and M is the number of generators connected to each feeder. This refers to the reactive power control capability of the wind farm.

[0027] In step 4, the specific reactive power control strategy is as follows:

[0028] When the reactive power control capability of a zone can meet the reactive power demand, the reactive power compensation task for each zone is obtained through an optimization algorithm, and the reactive power is allocated proportionally within each zone. When the reactive power control capability of a zone cannot meet the reactive power demand, but the reactive power control capability of the wind farm can, the reactive power is allocated with equal margin according to the reactive power capacity of each wind turbine.

[0029]

[0030] Among them, Q ref For reactive power demand;

[0031] When the reactive power regulation capacity of a wind farm cannot meet the reactive power demand, the wind farm will operate at full capacity within a reasonable range, with reactive power capacity allocated equally to each wind turbine. Any shortfall will be compensated by a reactive power compensation device.

[0032]

[0033] Q SVG =Q ref -Q DFIG

[0034] In the formula: Q DFIG For the compensation task assigned to DFIG; Q limitThe current reactive power limit of the wind farm is at this moment; Q SVG The compensation task allocated to SCG.

[0035] The optimization objective function includes a node voltage objective that minimizes node voltage deviation and a network loss objective that minimizes active power loss within the system. The node voltage objective is...

[0036]

[0037] Among them, U pcc U is the actual voltage at the grid connection point. ref N represents the voltage command issued by the power grid, N represents the total number of feeders in the wind farm, and M represents the total number of generators on each feeder.

[0038] The network loss target is:

[0039]

[0040] Among them, G ij Let θ be the admittance between nodes i and j. ij Let be the phase angle difference between nodes i and j.

[0041] In summary, the objective function for hierarchical partitioning optimization is:

[0042] F = α1F1 + α2F2

[0043] Among them, α1 and α2 are the weight coefficients of each sub-objective function.

[0044] The beneficial effects of this invention are that the reactive power zoning and control method for wind farms provided by this invention no longer treats the wind farm as equivalent to one or a few turbine units for control, but considers the voltage fluctuations of nodes within the wind farm; and utilizes fuzzy c-means clustering to divide the turbine units within the wind farm into different regions. When performing reactive power compensation, compared to the traditional method of allocating reactive power based on the reactive power capacity of each turbine unit, it can more effectively utilize the reactive power capacity of the turbine units, reduce voltage fluctuations at the turbine terminals within the wind farm, and improve the operational safety of the wind turbine units. This invention is easy to implement and highly operable. Attached Figure Description

[0045] Figure 1 A schematic diagram of the overall framework for reactive power zone control in a wind farm;

[0046] Figure 2 A flowchart for reactive power zone control in a wind farm;

[0047] Figure 3 The reactive power limit diagram for a doubly-fed induction generator (DFIG) is shown.

[0048] Figure 4 For the reactive power regulation capability of zoning and wind farms

[0049] Figure 5 The power output curve of a wind farm over a single day;

[0050] Figure 6 A comparison chart of voltage fluctuations within a day for the equal margin allocation method. Detailed Implementation

[0051] This invention proposes a reactive power zoning and control method for wind farms based on cluster analysis. The invention will be further described below with reference to the accompanying drawings and embodiments.

[0052] like Figure 1 The diagram shown illustrates the overall framework of reactive power zone control for a wind farm. Figure 2 The diagram shown is a flowchart of reactive power zoning control in a wind farm. The method includes the following steps:

[0053] Step 1: Obtain power grid data and wind farm operation data.

[0054] Step 2: Perform power flow calculations on the entire power grid to obtain the reactive voltage characteristics of each node in the wind farm, analyze the reactive power regulation capability of the doubly-fed induction generator (DFIG) to determine its influencing factors, and combine the relevant data from Step 1 to divide all units in the wind farm using a clustering algorithm.

[0055] Step 3: Combining the wind farm operation data from Step 1, obtain the reactive power and voltage control capability of the units in each region, and use the unit with the smallest reactive power control capability in the region as the representative of the reactive power output of all units in the region, and calculate the reactive power control capability of each region and wind farm.

[0056] Step 4: Combining the power grid data and grid connection point voltage information from Step 1, calculate the reactive power demand at the current moment, and provide corresponding control strategies based on the reactive power regulation capabilities at each level.

[0057] Typical wind farm collector system topology, such as Figure 1 As shown, there are 6 feeders in the wind farm, with 11 wind turbine units on each feeder. The end unit is numbered 1, and the first unit is numbered 11. Each turbine is equipped with a box-type transformer and a dynamic reactive power compensation device is installed at the grid connection point.

[0058] Step 1: Obtain power grid data and wind farm operation data.

[0059] Specifically, it obtains grid voltage and grid dispatch instructions; wind farm operation data includes wind farm topology, line parameters, transformer parameters, dynamic reactive power compensation device capacity, unit output power, and historical active power output data of the units.

[0060] Step 2: Perform power flow calculations on the entire power grid to obtain the reactive voltage characteristics of each node in the wind farm, analyze the reactive power regulation capability of the doubly-fed induction generator (DFIG) to determine its influencing factors, and combine the relevant data from Step 1 to divide all units in the wind farm using a clustering algorithm.

[0061] Specifically, a grid-connected model of the wind farm is built using the grid parameters obtained in step 1, and power flow calculation is performed using the Newton-Raphson method. The reactive power output of each node is changed, and the voltage changes of each node are observed to obtain the reactive voltage characteristics of each node.

[0062] Specifically, the reactive power regulation capability of a doubly-fed wind turbine is determined by the stator-side and grid-side converters:

[0063]

[0064]

[0065] Among them, Q smax Q smin Q represents the upper and lower limits of the stator-side reactive power output. cmax Q cmin These represent the upper and lower limits of reactive power output of the grid-side converter. Considering the reactive power limits of both the stator-side and grid-side converters, the reactive power limit of the doubly-fed induction generator (DFIG) wind turbine can be obtained, i.e., the reactive power regulation capability of the turbine. Figure 3 As shown:

[0066]

[0067] Among them, Q max Q min Q represents the upper and lower limits of the unit's reactive power output; limit This refers to the reactive power control capability of the generator unit.

[0068] Specifically, in step 2, the division of all units within the wind farm is completed through the following steps:

[0069] Step 21: Analysis of the doubly fed generator unit shows that active power output and stator voltage are the influencing factors of the unit's reactive power control capability. Assuming that the stator voltage remains constant, the active power output of the unit is used as the source data for cluster analysis.

[0070] Step 22: Normalize the historical active power output data;

[0071] Step 23, apply the Kuyama & Sugeno judgment criterion (V FS Determine the number of categories;

[0072]

[0073] in, V represents the membership degree from the j-th data point to the i-th class center. FS The minimum point corresponds to the optimal number of clusters.

[0074] Step 24: All units in the wind farm are partitioned using fuzzy c-means clustering. The partitioning results are shown in Table 1.

[0075] Table 1. Unit Numbers in Each Zone of the Wind Farm

[0076] serial number Zone 1 14 20 21 24 25 28 40 49 50 Zone 2 1 2 7 9 10 13 26 30 31 32 45 46 48 51 52 55 61 Zone 3 8 11 15 17 18 23 29 33 34 35 39 44 47 53 54 62 66 District 4 3 4 5 6 12 16 19 22 57 58 District 5 27 56 60 65

[0077] Step 3: Combining the wind farm operation data from Step 1, obtain the reactive power and voltage control capability of the units in each region, and use the unit with the smallest reactive power control capability in the region as the representative of the reactive power output of all units in the region, and calculate the reactive power control capability of each region and wind farm.

[0078] Specifically, in step 3, the reactive power regulation capability of the clustered wind farm is:

[0079]

[0080] Where K is the total number of areas within the wind farm, and c i Let be the number of units in region i. The upper and lower limits of reactive power output of wind farms after zoning; These are the upper and lower limits of reactive power output of all units within region i; It refers to the reactive power regulation capability of wind farms after zoning.

[0081] The reactive power regulation capability of a wind farm is:

[0082]

[0083] Where N is the number of feeders and M is the number of generators connected to each feeder. This refers to the reactive power regulation capability of the wind farm. The reactive power regulation capability of different zones and wind farms is as follows: Figure 4 As shown.

[0084] Step 4: Combining the power grid data and grid connection point voltage information from Step 1, calculate the reactive power demand at the current moment, and provide corresponding control strategies based on the reactive power regulation capabilities at each level.

[0085] Specifically, reactive power demand can be determined by the voltage, reactive power value, and voltage command at the current grid connection point:

[0086] Q ref =Q2+(U ref -U2) / K

[0087] Where Q2 is the reactive power value at the current grid connection point, U refU2 is the voltage command, U2 is the current grid connection point voltage, and K is the reactive voltage sensitivity of the grid connection point.

[0088] Furthermore, in step 4, the specific reactive power control strategy is as follows: when the reactive power regulation capability of a zone can meet the reactive power demand, the reactive power compensation task for each region is obtained through an optimization algorithm, and the reactive power is allocated proportionally within each region; when the reactive power regulation capability of a zone cannot meet the reactive power demand, but the reactive power regulation capability of the wind farm can, the reactive power is allocated according to the equal margin of each wind turbine's reactive power capacity.

[0089]

[0090] Among them, Q ref This is for reactive power demand.

[0091] When the reactive power regulation capacity of a wind farm cannot meet the reactive power demand, the wind farm will operate at full capacity within a reasonable range, with reactive power capacity allocated equally to each wind turbine. Any shortfall will be compensated by a reactive power compensation device.

[0092]

[0093] Q SVG =Q ref -Q DFIG

[0094] In the formula: Q DFIG For the compensation task assigned to DFIG; Q limit The current reactive power limit of the wind farm is at this moment; Q SVG The compensation task allocated to SCG.

[0095] To compare the differences between the two methods, a voltage equalization index is introduced:

[0096]

[0097] Figure 5 This represents the daily power output of a certain wind farm. Figure 6 This is a comparison chart of voltage fluctuations within a day between the method described in this invention and the equal margin allocation method. As can be seen from the voltage fluctuation areas in the chart, the voltage fluctuations at the wind farm turbine terminals are effectively reduced after adopting the reactive voltage zoning control described in this invention.

Claims

1. A method for reactive power zoning and control of wind farms based on cluster analysis, characterized in that, Includes the following steps: Step 1: Obtain power grid data and wind farm operation data; Step 2: Perform power flow calculations on the entire power grid to obtain the reactive voltage characteristics of each node in the wind farm, analyze the reactive power regulation capability of the doubly-fed induction generator (DFIG) to determine its influencing factors, and combine the power grid data and wind farm operation data from Step 1 to divide all units in the wind farm using a clustering algorithm. Step 3: Combine the wind farm operation data in Step 1 to obtain the reactive power control capability of the units in each region, and take the unit with the smallest reactive power control capability in the region as the representative of the reactive power output of all units in the region, and calculate the reactive power control capability of each region and wind farm. Step 4: Combining the power grid data and grid connection point voltage information from Step 1, calculate the reactive power demand at the current moment, and provide corresponding control strategies based on the reactive power regulation capabilities at each level. In step 4, the specific reactive power control strategy is as follows: When the reactive power control capability of a zone can meet the reactive power demand, the reactive power compensation task for each zone is obtained through an optimization algorithm, and the reactive power is allocated proportionally within each zone. When the reactive power control capability of a zone cannot meet the reactive power demand, but the reactive power control capability of the wind farm can, the reactive power is allocated with equal margin according to the reactive power capacity of each wind turbine. ， in, For reactive power demand; For the region The number of units, It refers to the reactive power regulation capability of the wind farm after zoning. For the reactive power control capability of the wind farm; This refers to the reactive power control capability of the generating unit; When the reactive power regulation capacity of a wind farm cannot meet the reactive power demand, the wind farm will operate at full capacity within a reasonable range, with reactive power capacity allocated equally to each wind turbine. Any shortfall will be compensated by a reactive power compensation device. ， ， In the formula: For the compensation tasks assigned to DFIG; For the reactive power control capability of the wind farm; For compensation tasks assigned to SVG.

2. The wind farm reactive power zoning and control method based on cluster analysis according to claim 1, characterized in that, In step 1, the power grid data includes power grid voltage and power grid dispatch instructions; the wind farm operation data includes wind farm topology, line parameters, transformer parameters, dynamic reactive power compensation device capacity, unit output power, and historical active power output data of the units.

3. The wind farm reactive power zoning and control method based on cluster analysis according to claim 1, characterized in that, In step 2, after determining the influencing factors of the unit's reactive power control capability, relevant historical data is obtained, normalized, and a clustering algorithm is used to divide the region.

4. The wind farm reactive power zoning and control method based on cluster analysis according to claim 1, characterized in that, Step 2, which uses a clustering algorithm to divide all the units in the wind farm, includes the following steps: Step 21: Analysis of the doubly fed generator unit shows that active power output and stator voltage are the influencing factors of the unit's reactive power control capability. Assuming that the stator voltage remains constant, the active power output of the unit is used as the source data for cluster analysis. Step 22: Normalize the historical active power output data; Step 23: Determine the number of categories using the Kuyama & Sugeno criterion; ， in, , This represents the membership degree of the j-th data point to the i-th class center. The minimum point corresponds to the optimal number of clusters; Step 24: Divide all units in the wind farm into zones using fuzzy c-means clustering.

5. The wind farm reactive power zoning and control method based on cluster analysis according to claim 1, characterized in that, In step 3, the reactive power regulation capability of the unit is ， in, , These are the upper and lower limits of the unit's reactive power output; , These are the upper and lower limits of the reactive power output of the stator-side converter. , The upper and lower limits of the reactive power output of the grid-side converter; This refers to the reactive power control capability of the generator unit.

6. The wind farm reactive power zoning and control method based on cluster analysis according to claim 1, characterized in that, The reactive power regulation capability of the wind farm after clustering is: ， in, This represents the total number of areas within the wind farm. For the region The number of units, , The upper and lower limits of reactive power output of wind farms after zoning; , It is a region Upper and lower limits of reactive power output of all generating units within the facility; It refers to the reactive power regulation capability of wind farms after zoning.

7. The wind farm reactive power zoning and control method based on cluster analysis according to claim 6, characterized in that, The reactive power regulation capability of a wind farm is: ， Where N is the number of feeders, and M is the number of generators connected to each feeder; This refers to the reactive power control capability of the wind farm.

8. The wind farm reactive power zoning and control method based on cluster analysis according to claim 1, characterized in that, The optimization objective function includes both the node voltage objective (minimizing node voltage deviation) and the network loss objective (minimizing active power loss within the system). The node voltage objective is... ， in, The actual voltage at the grid connection point. This refers to the voltage command issued by the power grid. This represents the total number of feeders within the wind farm. The total number of units on each feeder line; The network loss target is: ， in, For nodes , Admittance between For nodes , The phase angle difference between them; In summary, the objective function for hierarchical partitioning optimization is: ， in, , The weight coefficients of each sub-objective function.

Images