Offshore wind power plant reactive power optimization distribution method based on fuzzy C-means clustering
Through the combination of fuzzy C-mean clustering and particle swarm algorithm, the reactive power distribution of wind farms is optimized, the problems of complex calculations and incoordinated control in the existing technology are solved, and the safe and economical operation and voltage of offshore wind farms are realized. Stablize.
Patent Information
- Application Number
- CN202410979245.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-12
AI Technical Summary
The existing technology reactive power optimization distribution strategy in wind farms fails to fully consider the reactive power optimization distribution inside the wind farm, resulting in slow calculation speed and complex control, and is not suitable for the application of large-scale wind farms.
The method based on fuzzy C-mean clustering is adopted to divide the fans in the offshore wind farm into different groups, and the reactive power adjustment ability of each group is determined based on the clustering results, and the particle swarm algorithm is used to distribute reactive power. Combined with the converter for closed-loop control, optimize the reactive distribution strategy.
It effectively reduces the difficulty of solving the reactive distribution volume of the unit, ensures the safe and economical operation of offshore wind farms, reduces system network loss and node voltage fluctuations, and improves the voltage control performance of the connection point.
Smart Images

Figure CN118944209A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of reactive power optimization of wind farms, and in particular to a reactive power optimization allocation method for offshore wind farms based on fuzzy C-means clustering. Background Art
[0002] In recent years, wind power technology has developed rapidly, and the capacity of wind turbines and the scale of wind farms have gradually increased. Modern power systems require wind farms to have a certain reactive power support capacity. At present, wind farms generally use a distribution strategy based on the reactive power margin of the unit to generate reactive power reference values for wind turbines. However, this distribution strategy only considers the output limit of the wind turbine, and does not consider the reactive power optimization distribution within the wind farm. How to fully optimize the reactive power regulation capability of wind turbines to improve the node voltage quality of wind farms and reduce system losses has become a current research hotspot.
[0003] Reactive power optimization of wind farms is generally aimed at the problem of active network loss or node voltage fluctuation in wind farms. Reactive power optimization models are established, and conventional mathematical optimization methods or intelligent optimization algorithms are used to obtain the optimal reactive power distribution scheme that meets the constraints, and determine the reactive power tasks of each unit. With the continuous increase in wind power penetration, the number of units in wind farms is also increasing. If the reactive power tasks of each unit are solved, it will lead to problems such as slow calculation speed, complex and uncoordinated control, and inconvenient management, which is not suitable for field applications. Some literature proposes to use an equivalent unit to equalize the wind farm, which can reduce the dimension of the model and facilitate calculation. The grid connection point voltage characteristics of the equivalent model are similar to the specific model, but it does not reflect the output characteristics of the units in the field, and the calculation accuracy of the reactive power distribution results is not high. Summary of the invention
[0004] In view of the problems existing in the prior art, the present invention provides a reactive power optimization allocation method for offshore wind farms based on fuzzy C-means clustering to reduce system losses and stabilize the voltage in the wind power grid-connected system, thereby ensuring the safe and economical operation of offshore wind farms.
[0005] To achieve the above technical objectives, the present invention adopts the following technical solution: a method for optimizing reactive power allocation in offshore wind farms based on fuzzy C-means clustering, which specifically includes the following steps:
[0006] Step 1: Obtain the wind speed at the location of each wind turbine in the offshore wind farm, and predict the active output of each wind turbine by combining the wind turbine dynamics model;
[0007] Step 2: The rotation speed of each wind turbine, the wind speed at the location and the predicted active output are used as clustering indicators, and the fuzzy C-means clustering method is used to cluster the wind turbines in the offshore wind farm;
[0008] Step 3: According to the clustering results, determine the lower limit of the maximum reactive power and the upper limit of the minimum reactive power in each group, and obtain the reactive power regulation capacity after grouping;
[0009] Step 4: The wind farm cluster AVC system calculates the reactive power demand reference value in the offshore wind farm according to the grid connection point voltage index issued by the power grid and the operation information of the measurement period before and after the grid connection point;
[0010] Step 5: If the reactive power demand reference value in the offshore wind farm exceeds the reactive power regulation capability that can be provided by all wind turbines in the offshore wind farm, the excess reactive power compensation shall be borne by SVG; otherwise, execute step 6;
[0011] Step 6: If the reactive power demand reference value in the offshore wind farm exceeds the reactive power regulation capability after grouping, the reactive power demand reference value in the offshore wind farm is distributed to each wind turbine according to the ratio of the reactive power upper limit provided by each wind turbine to the reactive power upper limit provided by all wind turbines in the offshore wind farm, and the converter is adjusted to perform closed-loop control on the reactive power of the wind turbine; otherwise, step 7 is executed;
[0012] Step 7: Construct a reactive power optimization model for offshore wind farms after clustering, use a particle swarm algorithm to determine the reactive power allocation of each cluster, evenly distribute the reactive power to the wind turbines in the cluster, and adjust the converter to perform closed-loop control on the reactive power of the wind turbines.
[0013] Furthermore, the prediction process of the active output of each wind turbine in step 1 is:
[0014]
[0015] Among them, P WT,a represents the active output of the ath wind turbine in the offshore wind farm, C p,a represents the wind energy utilization coefficient of the ath wind turbine in the offshore wind farm, R represents the blade radius of the wind turbine, ρ 空气 represents the air density, v w,a Indicates the wind speed at the location of the ath wind turbine in the offshore wind farm.
[0016] Furthermore, step 2 includes the following sub-steps:
[0017] Step 2.1, the rotation speed of each wind turbine in the offshore wind farm, the wind speed at the location and the predicted active output are combined into a cluster sample;
[0018] Step 2.2, set the number of clustering groups and initialize each cluster center;
[0019] Step 2.3, calculate the Euclidean distance between each clustered sample and each cluster center respectively, determine the membership of each clustered sample to each cluster according to the Euclidean distance, and calculate the sum of the weighted distances from each clustered sample to the cluster center;
[0020] Step 2.4: Update the cluster center according to the membership degree, and repeat step 2.3. If the difference between the absolute value of the sum of the weighted distances from each clustered sample to the cluster center and the absolute value of the sum of the weighted distances from each clustered sample to the cluster center last time is less than the preset accuracy, the optimal membership degree and cluster center are obtained.
[0021] Step 2.5: Cluster the samples according to the optimal membership degree and cluster center.
[0022] Furthermore, the calculation process of the weighted sum of square distances from each clustered sample to the cluster center in step 2.3 is:
[0023]
[0024] Among them, J(U,C) represents the weighted sum of square distances from each clustered sample to the cluster center, U represents the membership matrix, C represents the cluster center matrix, M represents the number of clusters, i represents the index of M, N represents the number of clustered samples, j represents the index of N, and ρ j represents the weight function of the jth cluster sample, x j represents the jth cluster sample, μ i (x j ) represents x j For the membership of the i-th cluster, b represents the fuzzy degree of the membership matrix, c i is the i-th cluster center, d(x j ,c i ) is x j With c i The Euclidean distance of .
[0025] Furthermore, the updating process of the cluster center in step 2.4 is:
[0026]
[0027] Furthermore, step 4 includes the following sub-steps:
[0028] Step 4.1 The wind farm cluster AVC system sets the grid connection point voltage measurement period T according to the grid connection point voltage issued by the power grid. m , divided into several sampling periods T s , obtain the grid point voltage measurement period T m Average internal grid voltage Where t is the grid voltage measurement period T m The number of voltage sampling times within, t = T m / T s ;
[0029] Step 4.2: Calculate the reactive power demand reference value of the offshore wind farm by combining the average grid voltage with the grid connection point voltage reference value and voltage reactive power change rate issued by the power grid. Among them, Q2 is the reactive power value in the current grid-connected point voltage measurement cycle, is the grid connection point voltage reference value issued by the power grid, k VQ It is the reactive power change rate of the grid connection point voltage issued by the power grid within two adjacent measurement cycles.
[0030] Furthermore, the reactive power regulation capability provided by all wind turbines in the offshore wind farm in step 5 Specifically:
[0031]
[0032] in, The upper limit of reactive power provided to all wind turbines in an offshore wind farm, Q max_j The upper limit of reactive power provided by the jth wind turbine, The lower limit of reactive power provided to all wind turbines in an offshore wind farm, Q min_j The lower limit of reactive power provided by the jth wind turbine.
[0033] Furthermore, step 7 includes the following sub-steps:
[0034] Step 7.1, according to the collection line loss, grid voltage deviation and wind turbine terminal voltage deviation in the offshore wind farm, establish the offshore wind farm reactive power optimization model, and use the power flow constraint, the upper and lower limit constraints of reactive power output of each wind turbine, the upper and lower limit constraints of reactive power regulation capacity of each group and the upper and lower limit constraints of reactive power output of reactive compensation device SVG as the constraint conditions of the offshore wind farm reactive power optimization model;
[0035] Step 7.2: Initialize the number of particle populations within the reactive power output range, use the vector composed of the reactive power distribution of each group as the particle position, and generate the particle velocity;
[0036] Step 7.3, each particle position in the particle population is substituted into the constraint conditions of the reactive power optimization model of the offshore wind farm, the flow information of each node of the offshore wind farm is calculated, the flow information is substituted into the reactive power optimization model of the offshore wind farm to calculate the fitness function value, and the particle position with the smallest fitness function value is retained;
[0037] Step 7.4, update the particle position and velocity, and repeat step 7.3;
[0038] Step 7.5, compare the currently retained fitness function value with the fitness function value retained last time, and retain the particle position with the smallest fitness function value;
[0039] Step 7.6, repeat steps 7.4-7.5 until the maximum number of iterations is reached, and the particle position with the smallest fitness function value finally retained is used as the reactive power allocation amount of each group;
[0040] Step 7.7: Evenly distribute the reactive power to the fans in the group according to the reactive power distribution amount of each group, and adjust the converter to perform closed-loop control on the reactive power of the fans.
[0041] Furthermore, the establishment process of the reactive power optimization model of the offshore wind farm in step 7.1 is as follows:
[0042] F(x)=ω1f1(x)+ω2f2(x)+ω3f3(x)
[0043] Among them, F(x) represents the objective function of the reactive power optimization model of offshore wind farms, f1(x) represents the objective function of the collection line loss in offshore wind farms, s represents the number of branches in the offshore wind farm, U u Indicates the voltage amplitude of the first node u of the branch, U w represents the voltage amplitude of the node w at the end of the branch, θ uw It represents the voltage phase difference between the first node u and the last node w of the branch, G uw represents the conductance between the first node u and the last node w of the branch, ω1 represents the weight of f1(x); f2(x) represents the objective function of the grid-connected voltage deviation, U PCC Indicates the grid-connected point voltage sampling value, is the grid connection point reference value issued by the power grid, ω2 represents the weight of f2(x); f3(x) represents the target function of the wind turbine terminal voltage deviation, N is the number of wind turbines in the offshore wind farm, U j and They represent the real-time voltage and reference voltage at the j-th wind turbine respectively, and ω3 represents the weight of f3(x).
[0044] Furthermore, the process of updating the particle velocity in step 7.4 is:
[0045]
[0046] in, represents the velocity of the dth element in the particle at the kth iteration, ω (k) represents the nonlinear adaptive time-varying inertia weight at the kth iteration, c1 represents the first learning factor, r1 represents the first random number, P best represents the minimum fitness function value in the current iteration, represents the velocity of the dth element in the particle at the kth iteration, c2 represents the second learning factor, r2 represents the second random number, Gbest Represents the global minimum fitness function value;
[0047] The process of updating the particle position is:
[0048]
[0049] Where α represents the local adaptation factor.
[0050] Compared with the prior art, the present invention has the following beneficial effects: the reactive power optimization distribution method for offshore wind farms based on fuzzy C-means clustering of the present invention utilizes the active power output of wind turbines in the offshore wind farm, and divides the wind turbines into different groups through fuzzy C-means clustering, thereby effectively reducing the difficulty of solving the reactive power distribution of the units; at the same time, different distribution strategies are selected according to the reactive power regulation capability of the wind farm to ensure the safe and economical operation of the offshore wind farm; when the reactive power demand reference value in the offshore wind farm is less than the reactive power regulation capability after grouping, the particle swarm algorithm is used to solve the reactive power distribution of each group, and the reactive power is evenly distributed within the group, so as to effectively utilize the reactive power regulation capability of the wind turbine itself to reduce the system network loss, reduce the node voltage fluctuation, and improve the voltage control performance of the wind farm grid connection point. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Figure 1 The present invention is a flow chart of the reactive power optimization allocation method for offshore wind farms based on fuzzy C-means clustering. DETAILED DESCRIPTION
[0052] The technical solution of the present invention is further explained below in conjunction with the accompanying drawings.
[0053] like Figure 1 The flowchart of the reactive power optimization allocation method for offshore wind farms based on fuzzy C-means clustering of the present invention is as follows:
[0054] Step 1: Obtain the wind speed at each wind turbine in the offshore wind farm, and predict the active output of each wind turbine by combining the wind turbine dynamics model; specifically,
[0055] Calculate the input power of the fan blades swept against the wind according to the fan dynamics model Where R represents the blade radius of the fan, ρ 空气 represents the air density, v w,a Indicates the wind speed at the location of the ath wind turbine in the offshore wind farm. Since the wind turbine cannot convert all wind energy into mechanical energy, it is necessary to introduce the wind energy utilization coefficient C p Correct the input power of the wind turbine blades swept against the wind to predict the active output of each wind turbine:
[0056]
[0057] Among them, P WT,a represents the active output of the ath wind turbine in the offshore wind farm, C p,a represents the wind energy utilization coefficient of the ath wind turbine in the offshore wind farm, C p,a It is determined by the tip speed ratio λ and pitch angle β of the wind turbine, that is, C p,a (λ,β), λ is the tip speed ratio of the fan, ω w is the angular velocity of the fan blades.
[0058] Step 2: The speed of each wind turbine, the wind speed at its location and the predicted active power output are used as grouping indicators, and the fuzzy C-means clustering method is used to cluster the wind turbines in the offshore wind farm. Wind turbines with similar operating characteristics are equivalent to several units for subsequent reactive power distribution control, thereby effectively reducing the difficulty of solving the aircraft reactive power distribution. The specific sub-steps include the following:
[0059] Step 2.1, the rotation speed of each wind turbine in the offshore wind farm, the wind speed at the location and the predicted active output are combined into a cluster sample;
[0060] Step 2.2, set the number of clustering groups and initialize each cluster center;
[0061] Step 2.3, calculate the Euclidean distance between each clustered sample and each cluster center respectively, determine the membership of each clustered sample to each cluster based on the Euclidean distance, and calculate the weighted sum of the distances from each clustered sample to the cluster center, taking into account the different contributions of the clustered sample indicators to the classification results:
[0062]
[0063] Among them, J(U,C) represents the weighted sum of square distances from each clustered sample to the cluster center, U represents the membership matrix, C represents the cluster center matrix, M represents the number of clusters, i represents the index of M, N represents the number of clustered samples, j represents the index of N, and ρ j represents the weight function of the jth cluster sample, x j represents the jth cluster sample, μ i (x j ) represents x j For the membership of the i-th cluster, b is a constant greater than 1, indicating the fuzziness of the membership matrix. c i is the i-th cluster center, d(x j ,c i ) is x j With c i The Euclidean distance of
[0064]
[0065] Step 2.4: Update the cluster center according to the membership degree, and repeat step 2.3. If the difference between the absolute value of the sum of the weighted distances from each clustered sample to the cluster center and the absolute value of the sum of the weighted distances from each clustered sample to the cluster center last time is less than the preset accuracy, the optimal membership degree and cluster center are obtained.
[0066] The updating process of the cluster center in the present invention is:
[0067] Step 2.5: Cluster the samples according to the optimal membership degree and cluster center.
[0068] Step 3: According to the clustering results, determine the lower limit and upper limit of the maximum reactive power in each group, and obtain the reactive power regulation capacity after grouping:
[0069]
[0070] in, Indicates the reactive power regulation capability after grouping. Indicates the lower limit of the reactive power regulation capability of the fan after grouping. m i Indicates the number of fans in each group, represents the lower limit of reactive power regulation capability of the ith group, Q min_j represents the lower limit of reactive power provided by the jth wind turbine; Indicates the upper limit of the reactive power regulation capability of the fan after grouping. represents the upper limit of reactive power regulation capability of the ith group, Q max_j The upper limit of reactive power provided by the jth wind turbine.
[0071] Step 4: Taking the grid connection point voltage as the control target, analyzing the reactive power demand of the offshore wind farm, and determining the reactive power adjustment of the wind turbine. Since the system voltage changes all the time, in order to avoid the frequent adjustment of reactive power output of the wind turbine, which may reduce the service life of the converter or even damage it, the average value of the grid connection point voltage in a certain period of time is considered as the basis for reactive power adjustment of the wind farm. Therefore, the wind farm cluster AVC system calculates the reactive power demand reference value in the offshore wind farm according to the grid connection point voltage index issued by the power grid and the operation information of the measurement period before and after the grid connection point. Specifically, it includes the following sub-steps:
[0072] Step 4.1 The wind farm cluster AVC system sets the grid connection point voltage measurement period T according to the grid connection point voltage issued by the power grid. m , divided into several sampling periods T s , obtain the grid point voltage measurement period Tm Average internal grid voltage Where t is the grid voltage measurement period T m The number of voltage sampling times within, t = T m / T s ,and It is the grid connection point voltage reference value issued by the power grid. is the control error allowed for the grid connection point voltage;
[0073] Step 4.2: Calculate the reactive power demand reference value of the offshore wind farm by combining the average grid voltage with the grid connection point voltage reference value and voltage reactive power change rate issued by the power grid. Among them, Q2 is the reactive power value in the current grid-connected point voltage measurement cycle, k VQ is the reactive power change rate of the grid connection point voltage issued by the power grid in two adjacent measurement cycles, U1 and Q1 are the grid connection point voltage and reactive power values of the previous measurement cycle respectively.
[0074] Step 5. Reactive equipment in wind farms generally includes wind turbines and reactive compensation devices such as SVG. SVG can respond quickly to reactive compensation instructions, but it is relatively expensive. In order to reduce the voltage regulation cost of wind farms, the reactive regulation capability of wind turbines is used first. Exceeding the reactive power regulation limit that all wind turbines in an offshore wind farm can provide The excess reactive power compensation task will be borne by SVG Otherwise, go to step 6;
[0075] The reactive power regulation capability provided by all wind turbines in the offshore wind farm of the present invention Specifically:
[0076]
[0077] in, The upper limit of reactive power provided to all wind turbines in an offshore wind farm, Q max_j The upper limit of reactive power provided by the jth wind turbine, The lower limit of reactive power provided to all wind turbines in an offshore wind farm, Q min_j The lower limit of reactive power provided by the jth wind turbine.
[0078] Step 6: If the reactive power limit provided by all wind turbines in the offshore wind farm before grouping Can meet the reactive power demand reference value of wind farm But the reactive power regulation capability after grouping Cannot meet the reactive power demand reference value of wind farm At this time, the wind farm does not meet the conditions for clustering, so the equal margin method is uniformly adopted. The reactive power demand reference value in the offshore wind farm is allocated to each wind turbine according to the ratio of the reactive power upper limit provided by each wind turbine to the reactive power upper limit provided by all wind turbines in the offshore wind farm, and the converter is adjusted to perform closed-loop control on the reactive power of the wind turbine. Otherwise, execute step 7;
[0079] Step 7. In order to be able to efficiently control wind turbines in large-scale offshore wind farms, it is necessary to group wind turbines and divide several units with similar operating characteristics into a cluster for unified control. The wind turbines in the cluster evenly distribute the reactive power command value of each group, avoiding the need to control each wind turbine individually, which is beneficial to reducing management and maintenance costs. There are a large number of units in offshore wind farms, and the reactive power output value of each wind turbine group is continuously adjustable. At the same time, changes in wind speed will bring about changes in the reactive power output limit of each group. This is a multi-variable, multi-constrained mixed nonlinear optimization problem. The particle swarm algorithm has a fast search speed and high efficiency, and has a good effect on solving optimization problems of continuous variables and is easy to converge to the optimal solution. If the reactive power regulation capability of the offshore wind farm after grouping Able to meet the reactive power demand reference value of wind farm Then, a reactive power optimization model for offshore wind farms after grouping is constructed, and the reactive power allocation of each group is determined by using the particle swarm algorithm, and the wind turbines in the group are evenly distributed, and the converter is adjusted to perform closed-loop control on the reactive power of the wind turbines. Specifically, the following sub-steps are included:
[0080] Step 7.1, according to the collection line loss, grid voltage deviation and wind turbine terminal voltage deviation in the offshore wind farm, establish the offshore wind farm reactive power optimization model, and use the power flow constraint, the upper and lower limit constraints of reactive power output of each wind turbine, the upper and lower limit constraints of reactive power regulation capacity of each group and the upper and lower limit constraints of reactive power output of reactive compensation device SVG as the constraint conditions of the offshore wind farm reactive power optimization model;
[0081] The process of establishing the reactive power optimization model of the offshore wind farm in the present invention is as follows:
[0082] F(x)=ω1f1(x)+ω2f2(x)+ω3f3(x)
[0083] Among them, F(x) represents the objective function of the reactive power optimization model of offshore wind farms, f1(x) represents the objective function of the collection line loss in offshore wind farms, s represents the number of branches in the offshore wind farm, U u Indicates the voltage amplitude of the first node u of the branch, U w represents the voltage amplitude of the node w at the end of the branch, θ uw It represents the voltage phase difference between the first node u and the last node w of the branch, G uwrepresents the conductance between the first node u and the last node w of the branch, ω1 represents the weight of f1(x); f2(x) represents the objective function of the grid-connected voltage deviation, U PCC Indicates the grid-connected point voltage sampling value, is the grid connection point reference value issued by the power grid, ω2 represents the weight of f2(x); f3(x) represents the target function of the wind turbine terminal voltage deviation, N is the number of wind turbines in the offshore wind farm, U j and They represent the real-time voltage and reference voltage at the j-th wind turbine respectively, and ω3 represents the weight of f3(x).
[0084] The power flow constraints in the present invention are:
[0085]
[0086] Among them, P u , Q u are the active power and reactive power injected into node u respectively.
[0087] The upper and lower limits of reactive power output of each fan in the present invention are: Q min_i ≤Q i ≤Q max_i ;
[0088] The upper and lower limits of reactive power regulation capability of each group in the present invention are:
[0089] The upper and lower limits of reactive power output of the reactive power compensation device SVG in the present invention are: SVGmin ≤Q SVG ≤Q SVGmax .
[0090] Step 7.2: Initialize the number of particle populations within the reactive power output range, use the vector composed of the reactive power distribution of each group as the particle position, and generate the particle velocity;
[0091] Step 7.3, substitute the position of each particle in the particle population into the constraint conditions of the offshore wind farm reactive power optimization model, calculate the flow information of each node of the offshore wind farm, substitute the flow information into the offshore wind farm reactive power optimization model to calculate the fitness function value, and judge whether the particle speed exceeds [v min ,v max ] range, if it exceeds this range, the particle speed is limited to this extreme value; at the same time, check whether the position of the particle is within the constraint range. If it is within the constraint range, the fitness value of each particle is calculated using the offshore wind farm reactive power optimization model. If it is not within the constraint range, the objective function value is set to this extreme value; retain the particle position with the smallest fitness function value;
[0092] Step 7.4, update the particle position and velocity, and repeat step 7.3;
[0093] The process of updating particle velocity is:
[0094]
[0095] in, represents the velocity of the dth element in the particle at the kth iteration, ω (k) represents the nonlinear adaptive time-varying inertia weight at the kth iteration, ω start is the initial value of inertia weight; ω end is the inertia weight termination value; c1 represents the first learning factor, c 1max 、c 1min are the maximum and minimum values of the first learning factor c1, K max represents the maximum number of iterations; r1 represents the first random number, P best represents the minimum fitness function value in the current iteration, represents the velocity of the dth element in the particle at the kth iteration, c2 represents the second learning factor, c 2max 、c 2min are the maximum and minimum values of the second learning factor c2, r2 represents the second random number, G best Represents the global minimum fitness function value;
[0096] The process of updating the particle position is:
[0097]
[0098] Wherein, α represents the local adaptation factor, α=r3×[r3+0.5], and r3 is a random number between [0,1].
[0099] Step 7.5, compare the currently retained fitness function value with the fitness function value retained last time, and retain the particle position with the smallest fitness function value;
[0100] Step 7.6, repeat steps 7.4-7.5 until the maximum number of iterations is reached, and the particle position with the smallest fitness function value finally retained is used as the reactive power allocation amount of each group;
[0101] Step 7.7, evenly distribute the reactive power to the fans in the group according to the reactive power distribution amount of each group, and adjust the converter to perform closed-loop control on the reactive power of the fans;
[0102] The rule for evenly distributing the fans in the group in the present invention is:
[0103]
[0104] The reactive power optimization allocation method for offshore wind farms based on fuzzy C-means clustering in the present invention selects different allocation strategies according to the reactive power regulation capability of the wind farm to reduce system losses and stabilize the voltage in the wind power grid-connected system, thereby ensuring the safe and economical operation of the offshore wind farm.
[0105] In a technical solution of the present invention, a computer-readable storage medium is also provided, storing a computer program, wherein the computer program enables a computer to execute the offshore wind farm reactive power optimization allocation method based on fuzzy C-means clustering.
[0106] In one technical solution of the present invention, an electronic device is also provided, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the offshore wind farm reactive power optimization allocation method based on fuzzy C-means clustering is implemented.
[0107] In the embodiments disclosed in the present application, the computer storage medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. The computer storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the above. More specific examples of computer storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.
[0108] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0109] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.
Claims
1. A reactive power optimization allocation method for offshore wind farms based on fuzzy C-means clustering, characterized in that: The specific steps include: Step 1: Obtain the wind speed at the location of each wind turbine in the offshore wind farm, and predict the active output of each wind turbine by combining the wind turbine dynamics model; Step 2: The rotation speed of each wind turbine, the wind speed at the location and the predicted active output are used as clustering indicators, and the fuzzy C-means clustering method is used to cluster the wind turbines in the offshore wind farm; Step 3: According to the clustering results, determine the lower limit of the maximum reactive power and the upper limit of the minimum reactive power in each group, and obtain the reactive power regulation capacity after grouping; Step 4: The wind farm cluster AVC system calculates the reactive power demand reference value in the offshore wind farm according to the grid connection point voltage index issued by the power grid and the operation information of the measurement period before and after the grid connection point; Step 5: If the reactive power demand reference value in the offshore wind farm exceeds the reactive power regulation capability that can be provided by all wind turbines in the offshore wind farm, the excess reactive power compensation shall be borne by SVG; Otherwise, go to step 6; Step 6: If the reactive power demand reference value in the offshore wind farm exceeds the reactive power regulation capability after grouping, the reactive power demand reference value in the offshore wind farm is distributed to each wind turbine according to the ratio of the reactive power upper limit provided by each wind turbine to the reactive power upper limit provided by all wind turbines in the offshore wind farm, and the converter is adjusted to perform closed-loop control on the reactive power of the wind turbine; otherwise, step 7 is executed; Step 7: Construct a reactive power optimization model for offshore wind farms after clustering, use a particle swarm algorithm to determine the reactive power allocation of each cluster, evenly distribute the reactive power to the wind turbines in the cluster, and adjust the converter to perform closed-loop control on the reactive power of the wind turbines.
2. The method for optimizing reactive power allocation of offshore wind farms based on fuzzy C-means clustering according to claim 1 is characterized in that: The prediction process of the active output of each wind turbine in step 1 is: Among them, P WT,a represents the active output of the ath wind turbine in the offshore wind farm, C p,a represents the wind energy utilization coefficient of the ath wind turbine in the offshore wind farm, R represents the blade radius of the wind turbine, ρ 空气 represents the air density, v w,a Indicates the wind speed at the location of the ath wind turbine in the offshore wind farm.
3. The method for optimizing reactive power allocation of offshore wind farms based on fuzzy C-means clustering according to claim 2 is characterized in that: Step 2 includes the following sub-steps: Step 2.1, the rotation speed of each wind turbine in the offshore wind farm, the wind speed at the location and the predicted active output are combined into a cluster sample; Step 2.2, set the number of clustering groups and initialize each cluster center; Step 2.3, calculate the Euclidean distance between each clustered sample and each cluster center respectively, determine the membership of each clustered sample to each cluster according to the Euclidean distance, and calculate the sum of the weighted distances from each clustered sample to the cluster center; Step 2.4: Update the cluster center according to the membership degree, and repeat step 2.
3. If the difference between the absolute value of the sum of the weighted distances from each clustered sample to the cluster center and the absolute value of the sum of the weighted distances from each clustered sample to the cluster center last time is less than the preset accuracy, the optimal membership degree and cluster center are obtained. Step 2.5: Cluster the samples according to the optimal membership degree and cluster center.
4. The method for optimizing reactive power allocation in offshore wind farms based on fuzzy C-means clustering according to claim 3 is characterized in that: The calculation process of the weighted sum of distances from each clustered sample to the cluster center in step 2.3 is: Among them, J(U,C) represents the weighted sum of square distances from each clustered sample to the cluster center, U represents the membership matrix, C represents the cluster center matrix, M represents the number of clusters, i represents the index of M, N represents the number of clustered samples, j represents the index of N, and ρ j represents the weight function of the jth cluster sample, x j represents the jth cluster sample, μ i (x j ) represents x j For the membership of the i-th cluster, b represents the fuzzy degree of the membership matrix, c i is the i-th cluster center, d(x j ,c i ) is x j With c i The Euclidean distance of .
5. The method for optimizing reactive power allocation of offshore wind farms based on fuzzy C-means clustering according to claim 4 is characterized in that: The updating process of cluster center in step 2.4 is:
6. The method for optimizing reactive power allocation in offshore wind farms based on fuzzy C-means clustering according to claim 4 is characterized in that: Step 4 includes the following sub-steps: Step 4.1 The wind farm cluster AVC system sets the grid connection point voltage measurement period T according to the grid connection point voltage issued by the power grid. m , divided into several sampling periods T s , obtain the grid point voltage measurement period T m Average internal grid voltage Where t is the grid voltage measurement period T m The number of voltage sampling times within, t = T m / T s ; Step 4.2: Calculate the reactive power demand reference value of the offshore wind farm by combining the average grid voltage with the grid connection point voltage reference value and voltage reactive power change rate issued by the power grid. Among them, Q2 is the reactive power value in the current grid-connected point voltage measurement cycle, is the grid connection point voltage reference value issued by the power grid, k VQ It is the reactive power change rate of the grid connection point voltage issued by the power grid within two adjacent measurement cycles.
7. The method for optimizing reactive power allocation in offshore wind farms based on fuzzy C-means clustering according to claim 6, characterized in that: The reactive power regulation capability provided by all wind turbines in the offshore wind farm in step 5 Specifically: in, The upper limit of reactive power provided to all wind turbines in an offshore wind farm, Q max_j The upper limit of reactive power provided by the jth wind turbine, The lower limit of reactive power provided to all wind turbines in an offshore wind farm, Q min_j The lower limit of reactive power provided by the jth wind turbine.
8. The method for optimizing reactive power allocation in offshore wind farms based on fuzzy C-means clustering according to claim 7, characterized in that: Step 7 includes the following sub-steps: Step 7.1, according to the collection line loss, grid voltage deviation and wind turbine terminal voltage deviation in the offshore wind farm, establish the offshore wind farm reactive power optimization model, and use the power flow constraint, the upper and lower limit constraints of reactive power output of each wind turbine, the upper and lower limit constraints of reactive power regulation capacity of each group and the upper and lower limit constraints of reactive power output of reactive compensation device SVG as the constraint conditions of the offshore wind farm reactive power optimization model; Step 7.2: Initialize the number of particle populations within the reactive power output range, use the vector composed of the reactive power distribution of each group as the particle position, and generate the particle velocity; Step 7.3, each particle position in the particle population is substituted into the constraint conditions of the reactive power optimization model of the offshore wind farm, the flow information of each node of the offshore wind farm is calculated, the flow information is substituted into the reactive power optimization model of the offshore wind farm to calculate the fitness function value, and the particle position with the smallest fitness function value is retained; Step 7.4, update the particle position and velocity, and repeat step 7.3; Step 7.5, compare the currently retained fitness function value with the fitness function value retained last time, and retain the particle position with the smallest fitness function value; Step 7.6, repeat steps 7.4-7.5 until the maximum number of iterations is reached, and the particle position with the smallest fitness function value finally retained is used as the reactive power allocation amount of each group; Step 7.7: Evenly distribute the reactive power to the fans in the group according to the reactive power distribution amount of each group, and adjust the converter to perform closed-loop control on the reactive power of the fans.
9. The method for optimizing reactive power allocation in offshore wind farms based on fuzzy C-means clustering according to claim 8, characterized in that: The process of establishing the reactive power optimization model of offshore wind farm in step 7.1 is as follows: F(x)=ω1f1(x)+ω2f2(x)+ω3f3(x) Among them, F(x) represents the objective function of the reactive power optimization model of offshore wind farms, f1(x) represents the objective function of the collection line loss in offshore wind farms, s represents the number of branches in the offshore wind farm, U u Indicates the voltage amplitude of the first node u of the branch, U w represents the voltage amplitude of the node w at the end of the branch, θ uw It represents the voltage phase difference between the first node u and the last node w of the branch, G uw represents the conductance between the first node u and the last node w of the branch, ω1 represents the weight of f1(x); f2(x) represents the objective function of the grid-connected voltage deviation, U PCC Indicates the grid-connected point voltage sampling value, is the grid connection point reference value issued by the power grid, ω2 represents the weight of f2(x); f3(x) represents the target function of the wind turbine terminal voltage deviation, N is the number of wind turbines in the offshore wind farm, U j and They represent the real-time voltage and reference voltage at the j-th wind turbine respectively, and ω3 represents the weight of f3(x).
10. The method for optimizing reactive power allocation in offshore wind farms based on fuzzy C-means clustering according to claim 9, characterized in that: The process of updating particle velocity in step 7.4 is: in, represents the velocity of the dth element in the particle at the kth iteration, ω (k) represents the nonlinear adaptive time-varying inertia weight at the kth iteration, c1 represents the first learning factor, r1 represents the first random number, P best represents the minimum fitness function value in the current iteration, represents the velocity of the dth element in the particle at the kth iteration, c2 represents the second learning factor, r2 represents the second random number, G best Represents the global minimum fitness function value; The process of updating the particle position is: Where α represents the local adaptation factor.
Citation Information
Cited By
Method for optimizing a spatial layout of a wind power plant
EP4752357A1
Method for optimizing a spatial layout of a wind power plant
WO2026115133A1
