Wind power plant equivalent modeling method and system for fitting dynamic characteristics of wind power plant, medium and processor

Abstract

The invention discloses a wind power plant equivalent modeling method and system for fitting the dynamic characteristics of a wind power plant, a medium and a processor, and relates to the technical field of equivalent modeling. Inputting the key clustering index into a K-means algorithm, and outputting an optimal clustering division result of the wind turbine generator; constructing a wind power plant clustering equivalent model according to the optimal clustering division result; constructing a wind power plant detailed model according to the related parameters of each wind turbine generator in the wind power plant; inputting a transient response error between the wind power plant detailed model and the wind power plant clustering equivalent model into the optimized BP neural network for training to obtain an error correction model; and adding and compensating an error compensation value output by the error correction model into the wind power plant clustering equivalent model to obtain a corrected wind power plant clustering equivalent model. According to the method, the error correction-clustering equivalent model fitting the dynamic characteristics of the wind power plant is constructed.

Description

Technical Field

[0001] The present invention relates to the technical field of equivalent modeling, and in particular to a wind farm equivalent modeling method, system, medium and processor for fitting dynamic characteristics of a wind farm. Background Art

[0002] With the continuous development of new energy, the proportion of renewable energy power generation is increasing. For example, the construction of large wind farms is rising and increasing. However, due to the intermittent and random nature of wind energy, the grid connection of large wind farms has a great impact on the safety of the power grid. Establishing an efficient and reliable large wind farm model is the basis for analyzing the stability of the power system caused by large wind farms, and is of great significance to the modeling research of wind farms.

[0003] At present, the research on wind farm equivalent modeling mainly focuses on the multi-machine equivalent method, which can flexibly solve the modeling problems in different scenarios. Existing research uses genetic algorithms, random approximation methods, and particle swarm algorithms to optimize the internal parameters of the equivalent machine. The clustering-based equivalent method ensures a certain accuracy of the wind farm model while reducing the time consumption of software simulation, ignoring the operating status and dynamic parameter differences within the wind farm. The wind farm is a complex, high-order, and strongly nonlinear system. The traditional modeling method cannot fit the dynamic characteristics of the wind farm well. Summary of the invention

[0004] In view of the problem that the modeling methods in the prior art cannot well fit the dynamic characteristics of wind farms, the present invention provides a wind farm equivalent modeling method, system, medium and processor that fit the dynamic characteristics of wind farms, and realizes the construction of an error correction-clustering equivalent model that fits the dynamic characteristics of wind farms. The specific technical solution is as follows:

[0005] A wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm comprises:

[0006] Obtain key clustering indicators of wind turbines selected according to the operating characteristics of the wind turbines;

[0007] The key clustering index is input into the K-means algorithm, the K-means algorithm is optimized by using a simulated annealing algorithm, and the optimal clustering division result of the wind turbine is output;

[0008] Constructing a wind farm clustering equivalent model according to the optimal clustering division result, and obtaining a transient response of the wind farm clustering equivalent model;

[0009] Construct a detailed model of the wind farm according to the relevant parameters of each wind turbine in the wind farm, and obtain the transient response of the detailed model of the wind farm;

[0010] Optimizing the hyperparameters of the BP neural network by using the Cheetah optimization algorithm, and inputting the transient response error between the wind farm detailed model and the wind farm cluster equivalent model into the optimized BP neural network for training to obtain an error correction model;

[0011] The error compensation value output by the error correction model is added to the wind farm cluster equivalent model to obtain a corrected wind farm cluster equivalent model.

[0012] Preferably, the step of inputting the key clustering index into a K-means algorithm, optimizing the K-means algorithm using a simulated annealing algorithm, and outputting an optimal clustering division result of the wind turbine generator set includes:

[0013] Obtaining preset wind turbine characteristic data;

[0014] Perform k-means clustering on the characteristic data of wind turbines to obtain clustering results;

[0015] The clustering result is used as the initial solution to calculate the objective function value of the sum of the distances from each sample to the corresponding cluster center;

[0016] The objective function value is used as the initialization temperature of the simulated annealing algorithm, and the annealing speed and the maximum annealing times are set;

[0017] Randomly change the cluster of a cluster sample in the clustering partition result and recalculate the new objective function value;

[0018] Determine whether the new objective function value is the optimal objective function value. If so, save the clustering division as the optimal clustering division result. If not, proceed to the next step.

[0019] Calculate the difference between the new objective function value and the initial objective function value, and determine whether a new solution is obtained;

[0020] Determine whether the maximum number of annealing times has been reached. If so, output the optimal clustering division; if not, continue to iterate until the maximum number of annealing times has been reached, and output the optimal clustering division result.

[0021] Preferably, constructing a wind farm clustering equivalent model according to the optimal clustering division result includes:

[0022] The equivalent parameters of each cluster are calculated according to the optimal clustering division result, and the wind farm clustering equivalent model is constructed according to the equivalent parameters.

[0023] Preferably, the method of optimizing the hyperparameters of the BP neural network by the Cheetah optimization algorithm includes:

[0024] The hyperparameters of BP neural network including initial learning rate, maximum number of iterations, and number of neurons in hidden layer are used as optimization objects of CO algorithm, and three-dimensional CO search algorithm dimension is established to initialize cheetah population.

[0025] Calculate the fitness value corresponding to each cheetah's position to determine the quality of the current solution;

[0026] Update the cheetah's position based on its pursuit behavior;

[0027] Check whether the current solution is better than the global best solution. If so, update the global best solution.

[0028] Determine whether the maximum number of iterations has been reached. If so, output the optimal cheetah position to obtain the BP neural network hyperparameter value. If not, continue iterating until the optimal cheetah position is output to obtain the BP neural network hyperparameter value.

[0029] Preferably, the step of inputting the transient response error between the wind farm detailed model and the wind farm cluster equivalent model into the optimized BP neural network for training to obtain the error correction model comprises:

[0030] Obtaining a first output active power and a first bus voltage according to a transient response of the wind farm detailed model;

[0031] Obtaining a second output active power and a second bus voltage according to a transient response of the wind farm clustering equivalent model;

[0032] Calculating a power difference between the first output active power and the second output active power, and a voltage difference between the first bus voltage and the second bus voltage;

[0033] The power difference and the voltage difference are used as mapping targets of the optimized BP neural network, and a training set and a verification set are selected and input into the optimized BP neural network for training to obtain an error correction model.

[0034] Preferably, the step of obtaining key clustering indicators of wind turbines selected according to the operating characteristics of the wind turbines includes:

[0035] Acquire a plurality of state variables of the wind turbine generator set selected according to the operating characteristics of the wind turbine generator set, and use the state variables as initial clustering indicators;

[0036] The factor analysis method is used to extract the key clustering indicators of the wind turbine generator set from the initial clustering indicators.

[0037] A wind farm equivalent modeling system for fitting the dynamic characteristics of a wind farm, applied to the aforementioned wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm, comprises:

[0038] A key clustering index acquisition unit, used to acquire key clustering indexes of wind turbines selected according to the operating characteristics of the wind turbines;

[0039] The optimal clustering division output unit is used to input the key clustering index into the K-means algorithm, optimize the K-means algorithm by using the simulated annealing algorithm, and output the optimal clustering division result of the wind turbine generator set;

[0040] A first transient response calculation unit, used for constructing a wind farm clustering equivalent model according to the optimal clustering division result, and obtaining a transient response of the wind farm clustering equivalent model;

[0041] The second transient response calculation unit is used to construct a detailed model of the wind farm according to relevant parameters of each wind turbine in the wind farm, and obtain the transient response of the detailed model of the wind farm;

[0042] A correction model building unit is used to optimize the hyperparameters of the BP neural network through the Cheetah optimization algorithm, and input the transient response error between the wind farm detailed model and the wind farm cluster equivalent model into the optimized BP neural network for training to obtain an error correction model;

[0043] The equivalent model correction unit is used to add the error compensation value output by the error correction model to the wind farm cluster equivalent model to obtain a corrected wind farm cluster equivalent model.

[0044] A computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the aforementioned wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm.

[0045] A processor is used to run a program, wherein when the program is run, the aforementioned wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm is executed.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] A wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm of the present invention obtains key clustering indicators of wind turbines selected according to the operating characteristics of the wind turbines; inputs the key clustering indicators into a K-means algorithm, optimizes the K-means algorithm using a simulated annealing algorithm, and outputs the optimal clustering division result of the wind turbines; constructs a wind farm clustering equivalent model according to the optimal clustering division result, and obtains the transient response of the wind farm clustering equivalent model; constructs a wind farm detailed model according to relevant parameters of each wind turbine in the wind farm, and obtains the transient response of the wind farm detailed model; optimizes the hyperparameters of a BP neural network by a cheetah optimization algorithm, and inputs the transient response error between the wind farm detailed model and the wind farm clustering equivalent model into the optimized BP neural network for training to obtain an error correction model; and adds the error compensation value output by the error correction model to the wind farm clustering equivalent model to obtain a corrected wind farm clustering equivalent model. The present invention adopts SA-K-means algorithm to cluster wind turbines and construct a cluster equivalent model of wind farm; utilizes CO-BP neural network, takes transient response errors of detailed model and cluster equivalent model as input, trains to obtain error correction model, adds output compensation of network to cluster equivalent model, and realizes construction of error correction-cluster equivalent model that fits dynamic characteristics of wind farm. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0049] Figure 1 The present invention is a flow chart of a wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm.

[0050] Figure 2 The present invention is a flow chart of an embodiment of a method for wind farm equivalent modeling for fitting the dynamic characteristics of a wind farm.

[0051] Figure 3 This is a flow chart of another embodiment of a wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm according to the present invention.

[0052] Figure 4 This is a flow chart of another embodiment of a wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm according to the present invention.

[0053] Figure 5 This is a flow chart of another embodiment of a wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm according to the present invention.

[0054] Figure 6 This is a schematic diagram of a wind farm equivalent modeling system for fitting the dynamic characteristics of a wind farm according to the present invention. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0057] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0058] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0059] See the following examples Figures 1 to 6 .

[0060] The embodiment of the present application provides a wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm, including:

[0061] Step S01, obtaining key clustering indicators of wind turbines selected according to operating characteristics of the wind turbines;

[0062] According to the operating characteristics of wind turbines, factor analysis is used to process the redundancy and correlation between state variables and screen out the key clustering indicators of wind turbines.

[0063] Specifically, in a preferred embodiment of the present application, the obtaining of key clustering indicators of wind turbines selected according to the operating characteristics of the wind turbines includes:

[0064] Step S11, obtaining a plurality of state variables of the wind turbine generator set selected according to the operating characteristics of the wind turbine generator set, and using the state variables as initial clustering indicators;

[0065] Based on the mechanical and electrical characteristics of the wind turbine, in this embodiment, 13 state variables including pitch angle, rotation speed, electromagnetic torque, mechanical torque, wind turbine torque, rotor current d and q axis components, stator current d and q axis components, rotor voltage d and q axis components, and potential d and q axis components are selected as initial clustering indicators;

[0066] In the two-phase synchronous rotating coordinate system, the flux and voltage equations of the generator are:

[0067]

[0068]

[0069] Where: R s , R r is the stator and rotor resistance; i sd 、i sq 、i rd 、i rq is the d-axis and q-axis components of the stator and rotor current; u sd 、u sq 、u rd 、u rq is the stator and rotor d and q axis components of the voltage; ψ sd , sq , rd , rq is the stator and rotor d and q axis components of the magnetic flux; L ss =L m +L s , L rr =L m +L r ; L s , L r is the stator and rotor leakage inductance; L m is the mutual inductance; s is the slip rate.

[0070] The mechanical power of a wind turbine is expressed as the wind energy obtained by the wind turbine:

[0071]

[0072] Where: C P is the power coefficient; ρ is the air density; A is the blade swept area; ν is the wind speed; λ is the tip speed ratio; β is the pitch angle.

[0073] Ignoring the transient process of the stator circuit, the differential expression of the potential Ed and Eq is:

[0074]

[0075]

[0076] The differential expression between the wind turbine and the mechanical transmission during the transient process is:

[0077]

[0078] Where: T T is the mechanical torque; T e is the electromagnetic torque; T M is the wind turbine torque; T d , T j are the time constants respectively.

[0079] By solving the algebraic equation and differential equation of the wind turbine, we can get s, β, T T , T M , T e 、i sd 、i sq 、i rd 、i rq 、u rd 、u rq 、E d 、E q The 13 initial state variables are reduced to a lower order using factor analysis to obtain the clustering index of wind turbines, which is used as the initial clustering index.

[0080] Step S12: extracting key clustering indicators of wind turbines from the initial clustering indicators using factor analysis.

[0081] Factor analysis can be used to reduce dimension and extract common factors. In order to effectively remove the adverse effects of reverse indicators and dimensions and retain as much original information as possible, it is necessary to positively and standardize multiple related indicators during the analysis process.

[0082]

[0083]

[0084]

[0085] Where x ij is the jth original parameter of the i-th indicator; μ i is the average parameter of the ith indicator; Si is the standard deviation of the ith indicator; S i is the variance of the ith indicator; is the index x ij Standardized parameters of is the maximum value among the i-th index; is the maximum value among the i-th index, that is; is the index x ij The forward parameter of .

[0086] (1) Construct a factor model, assuming X = (X 1 ,X 2 ,…,X p ) T is a common factor random variable, and the factor mathematical model is: i =α i1 F 1 +α i2 F 2 +…+α im F m +ε i , i=1,2,…,p,F m is a common factor, ε i is a special factor.

[0087] (2) Solve the factor loading matrix. The factor loading matrix is ​​obtained by the principal component method. The solution process is as follows

[0088] Calculate the covariance matrix of the original variables, variable X i With F j The covariance of is:

[0089]

[0090] Calculate the characteristic roots and unit eigenvectors of the covariance matrix and solve the matrix A, that is:

[0091]

[0092] where λ p , T p is the characteristic root and the eigenvector corresponding to the characteristic root one by one.

[0093] (3) Perform factor rotation on the loading matrix. Factor rotation can be performed by orthogonal rotation or oblique rotation to differentiate the values ​​of the loading matrix into two levels: 0 and 1, which is convenient for naming the common factors.

[0094] (4) Using Thompson regression method to estimate factor scores, W is the factor score coefficient matrix.

[0095] According to the factor scores, the variables containing the most information are selected as potential factors. The potential factors mainly include torque, stator and rotor current q-axis component, rotor current d-axis component, rotor voltage dq-axis component, and potential dq-axis state variables. Under the condition of ignoring mechanical loss and copper loss and iron loss, T M =T T=T e , Ed and Eq can be represented by the q-axis component of the stator current, urd, urq, and ird can be represented by the torque and the q-axis component of the rotor current, so the three final grouping indicators are obtained, which are the stator and rotor q-axis current and electromagnetic torque of the generator.

[0096] The characteristic data of each unit in the wind farm are: Where P i =(p i1 ,p i2 ,p i3 ), where M is the number of wind turbines, (p i1 ,p i2 ,p i3 ) represent the q-axis current and electromagnetic torque of the stator and rotor of the generator of the i-th wind turbine group respectively.

[0097] Step S02: input the key clustering index into the K-means algorithm, optimize the K-means algorithm using a simulated annealing algorithm, and output the optimal clustering division result of the wind turbine group;

[0098] Specifically, in a preferred implementation of the present application, the key clustering index is input into the K-means algorithm, the K-means algorithm is optimized by using a simulated annealing algorithm, and the optimal clustering division result of the wind turbine is outputted, including:

[0099] Step S21, obtaining preset wind turbine characteristic data;

[0100] Assume that the characteristic data of each unit in the wind farm is Where P i =(p i1 ,p i2 ,p i3 ), respectively represent the stator and rotor q-axis current and electromagnetic torque of the generator of the i-th wind turbine.

[0101] Step S22, performing k-means clustering on the wind turbine characteristic data P to obtain a clustering division result;

[0102] Step S23, using the clustering result as the initial solution to calculate the objective function value of the sum of distances from each sample to the corresponding cluster center;

[0103] The clustering result is used as the initial solution w to calculate the objective function value:

[0104]

[0105] Where w is the clustering partition, is the center of the ith cluster, is the distance from the sample to the corresponding cluster center; J wis the sum of the distances from each sample to the corresponding cluster center.

[0106] Step S24, using the objective function value as the initialization temperature of the simulated annealing algorithm, and setting the annealing speed and the maximum annealing times;

[0107] Initialization temperature T 0 , let T 0 =J w , initialize the annealing speed a and the maximum number of annealing times

[0108] Step S25, randomly changing the cluster of a cluster sample in the clustering division result, and recalculating a new objective function value;

[0109] Random perturbation generates a new clustering partition w′, that is, randomly changing the current cluster to which a cluster sample belongs, and calculating the new objective function value J w′ ;

[0110] Step S26, determining whether the new objective function value is the optimal objective function value, if so, saving the clustering division as the optimal clustering division result, if not, executing the next step;

[0111] Determine the new objective function value J w′ Is it the optimal objective function value? If it is, save the clustering partition w′ as the optimal clustering partition, J w′ is the optimal objective function value; otherwise, jump to the next step S27;

[0112] Step S27, calculating the difference between the new objective function value and the initial objective function value, and determining whether a new solution is obtained;

[0113] Calculate the difference ΔJ between the new objective function value and the current objective function value, and determine whether ΔJ is less than 0: If ΔJ < 0, accept the new solution and take it as the current solution; if ΔJ ≥ 0, accept the new solution with probability q (q = e ΔJ / Kt ) accepts the new solution. K is a constant and t is the current temperature.

[0114] Step S28, determine whether the maximum annealing times are reached, if so, output the optimal clustering division; if not, continue to iterate until the maximum annealing times are reached, and output the optimal clustering division result.

[0115] When the maximum number of annealing times is reached, the algorithm ends and the optimal clustering division is output; when the maximum number of annealing times is reached, according to the annealing formula T(t) = T 0 ·a t Annealing is performed at temperature t, a is an adjustable parameter and is set to 0.99, and the process returns to step S25 to continue iteration.

[0116] The simulated annealing algorithm (SA) is used to optimize the K-means algorithm to avoid the clustering process falling into the local optimal solution.

[0117] Step S03: constructing a wind farm clustering equivalent model according to the optimal clustering division result, and obtaining a transient response of the wind farm clustering equivalent model;

[0118] Specifically, in a preferred implementation of the present application, constructing a wind farm clustering equivalent model according to the optimal clustering division result includes:

[0119] The equivalent parameters of each cluster are calculated according to the optimal clustering division result, and the wind farm clustering equivalent model is constructed according to the equivalent parameters.

[0120] The wind turbine cluster division results are obtained according to the SA-K-means algorithm, the homology equivalent method is used to calculate the operating status of the wind turbine, and the weighted average method is used to solve the equivalent parameters of the wind turbine.

[0121] Wind turbine parameters equivalent wind speed v eq and equivalent speed ω eq The calculations are:

[0122]

[0123]

[0124] Where: v wi ,ω wi are the wind speed and rotation speed of the i-th wind turbine.

[0125] Wind turbine equivalent parameters

[0126]

[0127] Where: S eq , P eq , Q eq are the apparent power, active power and reactive power of the unit respectively; R eq , L eq , C eq They are resistance, inductance and capacitance respectively; H eq , K eq , D eq They are the shaft system inertia time constant, stiffness coefficient and damping coefficient respectively.

[0128] In addition to the various model parameters, the collector line also has a certain impact on the grid-connected operation of the wind farm. The impedance of the equivalent line is obtained by weighted average:

[0129]

[0130] Where: Z eq is the equivalent impedance; P zaFor flow through Z eq The total power of Z li is the line impedance of each unit; P zi For flow through Z li total power.

[0131] Step S04: constructing a detailed model of the wind farm according to relevant parameters of each wind turbine in the wind farm, and obtaining a transient response of the detailed model of the wind farm;

[0132] Step S05, optimizing the hyperparameters of the BP neural network by using the Cheetah optimization algorithm, and inputting the transient response error between the wind farm detailed model and the wind farm cluster equivalent model into the optimized BP neural network for training to obtain an error correction model;

[0133] Specifically, in a preferred embodiment of the present application, the optimization of the hyperparameters of the BP neural network by the Cheetah optimization algorithm includes:

[0134] Step S51, taking the hyper parameters of the BP neural network including the initial learning rate, the maximum number of iterations, and the number of neurons in the hidden layer as the CO algorithm optimization object, establishing a three-dimensional CO search algorithm dimension, and initializing the cheetah population;

[0135] Step S52, calculating the fitness value corresponding to the position of each cheetah to determine the quality of the current solution;

[0136] Step S53, updating the position of the cheetah according to its chasing behavior;

[0137] The cheetah's speed and direction are determined by its current optimal position and the location of its prey:

[0138]

[0139] in, is the speed and position of the i-th cheetah in the t-th iteration, is the current global best position,

[0140] is the position of the prey, r 1 、r 2 are two random numbers;

[0141] Step S54: Check whether the current solution is better than the global optimal solution. If so, update the global optimal solution.

[0142] Step S55, determine whether the maximum number of iterations has been reached. If so, output the optimal cheetah optimal position and obtain the BP neural network hyperparameter value. If not, return to step S52 and continue iterating until the optimal cheetah optimal position is output and the BP neural network hyperparameter value is obtained.

[0143] The Cheetah Optimization (CO) algorithm is used to optimize the hyperparameters of the BP neural network, avoiding the problems of poor generalization ability and high randomness caused by traditional hyperparameter selection methods.

[0144] Specifically, in a preferred embodiment of the present application, the transient response error between the wind farm detailed model and the wind farm cluster equivalent model is input into the optimized BP neural network for training to obtain the error correction model, which includes:

[0145] Transient simulations are performed on the detailed model of the wind farm and the equivalent model of the wind farm respectively, and their transient response expressions are shown below.

[0146] Step S501, obtaining a first output active power and a first bus voltage according to a transient response of the wind farm detailed model;

[0147] The transient simulation of the detailed model of the wind farm is carried out as follows:

[0148] The transient response of the detailed wind farm model is expressed as follows:

[0149] The first output active power:

[0150] P d (t) = [p d1 (t),p d2 (t),p d3 (t),...,p dn (t)]

[0151] The first bus voltage:

[0152] U d (t)=[u d1 (t),u d2 (t),u d3 (t),...,u dn (t)]

[0153] In the above formula: p dn (t) represents the transient response curve of the nth group of output active power of the detailed model of the wind farm; u dn (t) represents the transient response curve of the nth group of bus voltages of the detailed model of the wind farm. d (t) and U d Each item in (t) is a time series of length τ:

[0154] p di (t) = [p di (1),p di (2),p di (3),...,p di (τ)],i=1,2,...,n

[0155] u di (t)=[u di (1),u di (2),u di (3),...,u di (τ)],i=1,2,...,n

[0156] In the above formula, p di (τ) represents the output active power value of the i-th group of data of the wind farm detailed model at the τth moment; u di (τ) represents the bus voltage value at the τth moment of the i-th group of data in the detailed model of the wind farm.

[0157] Step S502: obtaining a second output active power and a second bus voltage according to the transient response of the wind farm clustering equivalent model;

[0158] The transient simulation of the wind farm equivalent model is as follows:

[0159] The transient response curve of the wind farm clustering equivalent model is as follows.

[0160] Second output active power:

[0161] P c (t) = [p c1 (t),p c2 (t),p c3 (t),...,p cn (t)]

[0162] Second bus voltage:

[0163] U c (t)=[u c1 (t),u c2 (t),u c3 (t),...,u cn (t)]

[0164] in:

[0165] p di (t) = [p di (1),p di (2),p di (3),...,p di (τ)],i=1,2,...,n

[0166] u di (t)=[u di (1),u di (2),u di (3),...,u di(τ)],i=1,2,...,n

[0167] Step S503, calculating a power difference between the first output active power and the second output active power, and a voltage difference between the first bus voltage and the second bus voltage;

[0168] The power difference is calculated as follows:

[0169] P e (t) = P d (t)-P c (t) = [p d1 (t)-p c1 (t),p d2 (t)-p c2 (t),...,p dn (t)-p cn (t)]

[0170] The voltage difference is calculated as follows:

[0171] U e (t) = U d (t)-U c (t)=[u d1 (t)-u c1 (t),u d2 (t)-u c2 (t),...,u dn (t)-u cn (t)]

[0172] Step S504: Use the power difference and voltage difference as mapping targets of the optimized BP neural network, and select a training set and a verification set to input into the optimized BP neural network for training to obtain an error correction model.

[0173] The transient response errors (power difference and voltage difference) of the wind farm detailed model and the wind farm cluster equivalent model are taken as the mapping targets of the neural network. The first 4 / 5 of the data are taken as the training set, and the last 1 / 5 of the data are taken as the validation set to train the CEM model (error correction model).

[0174] The equivalent model of a wind farm is actually a compromise solution for modeling. Equivalent modeling reduces the accuracy of wind turbine modeling to a certain extent, but reduces the simulation time of the model to a certain extent. Modeling using the clustering method cannot overcome the inherent defects of wind farms. The accuracy will still be significantly reduced in transient response simulation. By adding the error compensation value output by the CEM model to the cluster equivalent model, a cluster equivalent model that is corrected and close to the detailed model can be obtained, realizing the construction of an error correction-cluster equivalent model that fits the dynamic characteristics of a wind farm.

[0175] Step S06: adding the error compensation value output by the error correction model to the wind farm clustering equivalent model to obtain a corrected wind farm clustering equivalent model.

[0176] The output of CEM is close to the error value P of a specific scene. e (t) and U e (t), finally, the output result is added and compensated to the cluster equivalent model based on the SA-K-means algorithm, and the cluster equivalent model that is corrected and close to the detailed model can be obtained.

[0177] A wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm of the present invention obtains key clustering indicators of wind turbines selected according to the operating characteristics of the wind turbines; inputs the key clustering indicators into a K-means algorithm, optimizes the K-means algorithm using a simulated annealing algorithm, and outputs the optimal clustering division result of the wind turbines; constructs a wind farm clustering equivalent model according to the optimal clustering division result, and obtains the transient response of the wind farm clustering equivalent model; constructs a wind farm detailed model according to relevant parameters of each wind turbine in the wind farm, and obtains the transient response of the wind farm detailed model; optimizes the hyperparameters of a BP neural network by a cheetah optimization algorithm, and inputs the transient response error between the wind farm detailed model and the wind farm clustering equivalent model into the optimized BP neural network for training to obtain an error correction model; and adds the error compensation value output by the error correction model to the wind farm clustering equivalent model to obtain a corrected wind farm clustering equivalent model. The present invention adopts SA-K-means algorithm to cluster wind turbines and construct a cluster equivalent model of wind farm; utilizes CO-BP neural network, takes transient response errors of detailed model and cluster equivalent model as input, trains to obtain error correction model, adds output compensation of network to cluster equivalent model, and realizes construction of error correction-cluster equivalent model that fits dynamic characteristics of wind farm.

[0178] The embodiment of the present application further provides a wind farm equivalent modeling system for fitting the dynamic characteristics of a wind farm, which is applied to the aforementioned wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm, comprising:

[0179] A key clustering index acquisition unit, used to acquire key clustering indexes of wind turbines selected according to the operating characteristics of the wind turbines;

[0180] The optimal clustering division output unit is used to input the key clustering index into the K-means algorithm, optimize the K-means algorithm by using the simulated annealing algorithm, and output the optimal clustering division result of the wind turbine generator set;

[0181] A first transient response calculation unit, used for constructing a wind farm clustering equivalent model according to the optimal clustering division result, and obtaining a transient response of the wind farm clustering equivalent model;

[0182] The second transient response calculation unit is used to construct a detailed model of the wind farm according to relevant parameters of each wind turbine in the wind farm, and obtain the transient response of the detailed model of the wind farm;

[0183] A correction model building unit is used to optimize the hyperparameters of the BP neural network through the Cheetah optimization algorithm, and input the transient response error between the wind farm detailed model and the wind farm cluster equivalent model into the optimized BP neural network for training to obtain an error correction model;

[0184] The equivalent model correction unit is used to add the error compensation value output by the error correction model to the wind farm cluster equivalent model to obtain a corrected wind farm cluster equivalent model.

[0185] In this embodiment, the functional explanation corresponding to each unit is the same as the explanation of the aforementioned wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm, and the technical effect is also the same, which will not be repeated here.

[0186] An embodiment of the present application also provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the aforementioned wind farm equivalent modeling method for fitting the dynamic characteristics of the wind farm.

[0187] The technical effect of this embodiment is the same as the technical effect of a wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm in an embodiment, and will not be repeated here.

[0188] The present invention can be used in many general or special computer system environments or configurations, such as personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like.

[0189] An embodiment of the present application further provides a processor, which is used to run a program, wherein when the program is run, the aforementioned wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm is executed.

[0190] The technical effect of this embodiment is the same as the technical effect of the method for wind farm equivalent modeling for fitting dynamic characteristics of a wind farm in the embodiment, and will not be repeated here.

[0191] The processor in this embodiment may be a central processing unit (CPU), a controller, a microcontroller, or other data processing chips.

[0192] Those of ordinary skill in the art will appreciate that the units of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. 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 the present invention.

[0193] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0194] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0195] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nlyMemory), random access memory (RAM, RandomAccessMemory), mobile hard disk, magnetic disk or optical disk, etc., which can store program code.

[0196] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm, characterized in that: include: Obtain key clustering indicators of wind turbines selected according to the operating characteristics of the wind turbines; The key clustering index is input into the K-means algorithm, the K-means algorithm is optimized by using a simulated annealing algorithm, and the optimal clustering division result of the wind turbine is output; Constructing a wind farm clustering equivalent model according to the optimal clustering division result, and obtaining a transient response of the wind farm clustering equivalent model; Construct a detailed model of the wind farm according to the relevant parameters of each wind turbine in the wind farm, and obtain the transient response of the detailed model of the wind farm; The hyperparameters of the BP neural network are optimized by the Cheetah optimization algorithm, and the transient response error between the wind farm detailed model and the wind farm cluster equivalent model is input into the optimized BP neural network for training to obtain an error correction model; the error compensation value output by the error correction model is added and compensated to the wind farm cluster equivalent model to obtain a corrected wind farm cluster equivalent model.

2. A wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm according to claim 1, characterized in that: The key clustering index is input into the K-means algorithm, the K-means algorithm is optimized by using the simulated annealing algorithm, and the optimal clustering division result of the wind turbine group is outputted, including: Obtaining preset wind turbine characteristic data; Perform k-means clustering on the characteristic data of wind turbines to obtain clustering results; The clustering result is used as the initial solution to calculate the objective function value of the sum of the distances from each sample to the corresponding cluster center; The objective function value is used as the initialization temperature of the simulated annealing algorithm, and the annealing speed and the maximum annealing times are set; Randomly change the cluster of a cluster sample in the clustering partition result and recalculate the new objective function value; Determine whether the new objective function value is the optimal objective function value. If so, save the clustering division as the optimal clustering division result. If not, proceed to the next step. Calculate the difference between the new objective function value and the initial objective function value, and determine whether a new solution is obtained; Determine whether the maximum number of annealing times has been reached. If so, output the optimal clustering division; if not, continue to iterate until the maximum number of annealing times has been reached, and output the optimal clustering division result.

3. A wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm according to claim 1, characterized in that: The step of constructing a wind farm clustering equivalent model according to the optimal clustering division result comprises: The equivalent parameters of each cluster are calculated according to the optimal clustering division result, and the wind farm clustering equivalent model is constructed according to the equivalent parameters.

4. A wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm according to claim 1, characterized in that: The hyper parameters of the BP neural network optimized by the Cheetah Optimization Algorithm include: The hyperparameters of BP neural network including initial learning rate, maximum number of iterations, and number of neurons in hidden layer are used as optimization objects of CO algorithm, and three-dimensional CO search algorithm dimension is established to initialize cheetah population. Calculate the fitness value corresponding to each cheetah's position; Update the cheetah's position based on its pursuit behavior; Check whether the current solution is better than the global best solution. If so, update the global best solution. Determine whether the maximum number of iterations has been reached. If so, output the optimal cheetah's optimal position and obtain the BP neural network hyperparameter value. If not, continue iterating until the optimal cheetah's optimal position is output and the BP neural network hyperparameter value is obtained.

5. A wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm according to claim 3, characterized in that: The step of inputting the transient response error between the wind farm detailed model and the wind farm cluster equivalent model into the optimized BP neural network for training to obtain the error correction model comprises: Obtaining a first output active power and a first bus voltage according to a transient response of the wind farm detailed model; Obtaining a second output active power and a second bus voltage according to a transient response of the wind farm clustering equivalent model; Calculating a power difference between the first output active power and the second output active power, and a voltage difference between the first bus voltage and the second bus voltage; The power difference and the voltage difference are used as mapping targets of the optimized BP neural network, and a training set and a verification set are selected and input into the optimized BP neural network for training to obtain an error correction model.

6. A wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm according to claim 1, characterized in that: The key clustering indexes of the wind turbines selected according to the operating characteristics of the wind turbines are obtained, including: Acquire a plurality of state variables of the wind turbine generator set selected according to the operating characteristics of the wind turbine generator set, and use the state variables as initial clustering indicators; The factor analysis method is used to extract the key clustering indicators of the wind turbine generator set from the initial clustering indicators.

7. A wind farm equivalent modeling system for fitting the dynamic characteristics of a wind farm, characterized in that: A wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm as described in any one of claims 1 to 6, comprising: A key clustering index acquisition unit, used to acquire key clustering indexes of wind turbines selected according to the operating characteristics of the wind turbines; The optimal clustering division output unit is used to input the key clustering index into the K-means algorithm, optimize the K-means algorithm by using the simulated annealing algorithm, and output the optimal clustering division result of the wind turbine generator set; A first transient response calculation unit, used for constructing a wind farm clustering equivalent model according to the optimal clustering division result, and obtaining a transient response of the wind farm clustering equivalent model; The second transient response calculation unit is used to construct a detailed model of the wind farm according to relevant parameters of each wind turbine in the wind farm, and obtain the transient response of the detailed model of the wind farm; A correction model building unit is used to optimize the hyperparameters of the BP neural network through the Cheetah optimization algorithm, and input the transient response error between the wind farm detailed model and the wind farm cluster equivalent model into the optimized BP neural network for training to obtain an error correction model; The equivalent model correction unit is used to add the error compensation value output by the error correction model to the wind farm cluster equivalent model to obtain a corrected wind farm cluster equivalent model.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute a wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm as described in any one of claims 1 to 6.

9. A processor, characterized in that: The processor is used to run a program, wherein when the program is run, a wind farm equivalent modeling method for fitting the dynamic characteristics of a wind farm as described in any one of claims 1 to 6 is executed.

Images