A method for assessing the capacity value of wind farm clusters based on the Johnson distribution system

Abstract

This invention provides a method for assessing the capacity value of wind farm clusters based on the Johnson distribution system. The method comprises the following steps: Step 1: Establishing a wind farm cluster fluctuation model through power output analysis using a first-order Markov chain algorithm; Step 2: Establishing relevant models for different wind farm clusters through time-series power output analysis using the Johnson distribution system; Step 3: Conducting a reliability assessment of the power system based on the given load level and wind farm capacity, according to the power adequacy index; Step 4: Using a bisection method to adjust the load of the wind farm system to search for the reliable capacity of the wind farm cluster. This invention is designed for large-scale wind turbine integration into the power system and can scientifically and accurately assess its capacity value.

Description

Technical Field

[0001] This invention belongs to the technical fields of power balance analysis, power adequacy assessment, and power planning for wind farm clusters, and particularly relates to a method for assessing the capacity value of wind farm clusters using the Johnson distribution system. Background technology:

[0002] With the increasing maturity of wind power technology and the effective incentives from national policies for the development of clean and environmentally friendly new energy sources, my country's abundant wind energy resources are gradually being developed, and the installed capacity of wind power continues to grow. Wind farm output is characterized by randomness and intermittency, and its contribution to the power system's capacity after grid connection differs significantly from that of conventional turbines. Therefore, objectively evaluating the capacity value of wind farms is of great significance for long-term planning, optimized operation, and adequacy assessment of the power generation system.

[0003] Many experts and scholars at home and abroad have conducted in-depth research on wind power capacity valuation methods. Currently, there are the following valuation methods for wind farm capacity: (1) Using multi-state units to represent the stochastic characteristics of wind power output, and conducting wind farm capacity valuation based on stochastic production simulation; (2) Establishing a wind power output model considering wind speed changes based on the semi-invariant method, and analyzing the correlation between wind power output and the load connected to the system and its impact on wind power capacity value; (3) Considering the equivalent load curve of the wind farm's outage capacity, and obtaining the wind farm capacity value by calculating the effective load capacity of the wind farm; (4) Introducing the sequence operation theory into the wind power capacity valuation problem, using the sequence method to describe the stochasticity of conventional units, wind turbine units and loads, and realizing the rapid calculation of wind power capacity value. However, current capacity valuation of wind farms only focuses on the capacity value of a single wind farm. With the further development of wind resources, multiple wind farms will be connected to the system in the same wind area at the same time. The output of wind farms that are spatially close to each other will have a certain correlation, which will have a significant impact on the operation and planning of wind farms. Therefore, it is necessary to study the correlation of output between wind farm groups.

[0004] Currently, existing methods for analyzing correlations mainly fall into two categories: constructing a joint probability distribution of multivariate variables and obtaining correlation information using the Johnson distribution system. Methods for handling wind speed correlations in wind farm capacity valuation mainly include: matrix transformation, intelligent algorithms, Nataf transformation, and Copula functions. Among these, matrix transformation, when generating correlated samples, is theoretically only applicable to normally distributed variables; intelligent algorithms cannot guarantee the accuracy of the results and are generally time-consuming; Nataf transformation first generates independent samples conforming to a normal distribution, then uses matrix transformation to generate correlated samples, and finally uses inverse Nataf transformation to convert the samples into samples conforming to the original distribution function. This method overcomes the limitation of matrix transformation being only applicable to normally distributed variables, but requires knowledge of the marginal distribution functions of the variables; the Copula function method synthesizes multidimensional marginal distribution functions into a joint distribution function with a specific correlation coefficient, and then obtains samples with a certain degree of correlation through conditional sampling, which can generate correlated samples more accurately, but the process of generating high-dimensional joint distribution functions is relatively complex. This invention proposes a multidimensional random variable modeling method based on the Johnson distribution system. This method can more accurately simulate the marginal distribution of variables and the correlation between multidimensional variables, given the historical data of the variables.

[0005] The core of reliable capacity assessment for wind turbines lies in the reliability assessment of the power system. Traditional power system reliability studies assume a sufficient and stable primary energy supply. However, with the large-scale integration of intermittent renewable energy sources such as wind and solar power into the grid, this assumption is no longer valid. After the integration of large-scale wind farm clusters, their fluctuations are significant, and the improvement in load carrying capacity is probabilistic, necessitating reliable capacity assessments for wind farm clusters. However, existing studies on reliable capacity assessment of large wind and solar farms rely solely on historical data for wind farm output, failing to simulate the randomness of large-scale wind farm output. Reliability calculation modeling is often simplistic, not fully considering the output fluctuations and load fluctuations of large-scale wind farms. Monte Carlo sampling is a commonly used reliability calculation method, but existing studies typically select a fixed value as the Monte Carlo convergence criterion, failing to balance convergence speed and accuracy. Summary of the Invention

[0006] To address the technical problems existing in current technologies, this invention proposes a method for assessing the capacity value of wind farm clusters based on the Johnson distribution system. First, based on historical wind speed records from NASA, the fluctuation characteristics of wind farm output are extracted using a first-order Markov chain. Second, the correlation of output between multi-dimensional wind farms is established based on the Johnson distribution system. Third, for a given load level and wind farm capacity, system reliability is assessed using power adequacy as the evaluation index. This invention is designed for large-scale wind turbine integration into power systems, enabling a scientific and accurate assessment of their capacity value.

[0007] The core content of this invention can be summarized as follows:

[0008] A method for assessing the capacity value of wind farm clusters based on the Johnson distribution system, characterized by the following steps:

[0009] Step 1: Establish a wind farm cluster fluctuation model by analyzing the power output of the wind farm cluster using the first-order Markov chain algorithm;

[0010] Step 2: Establish relevant models for different wind farm groups by analyzing the time-series power output of wind farm groups using the Johnson distribution system;

[0011]

[0012] Where v is the wind speed, v in v out and v r These are the cut-in wind speed, cut-out wind speed, and rated wind speed of the wind turbine unit, respectively.

[0013] Let be the rated output power of the i-th fan;

[0014] Step 3: Based on the given load level and wind farm capacity, conduct a reliability assessment of the power system according to the power adequacy index.

[0015] Step 4: Use the binary search method to adjust the load of the wind farm system to search for the reliable capacity of the wind farm group.

[0016] Furthermore, Step 2 involves establishing relevant models for different wind farm clusters through time-series power output analysis using the Johnson distribution system; this includes:

[0017] 201. Calculate the wind speed state transition matrix for a specific month in the wind farm cluster fluctuation model according to the year requirements;

[0018] 202. Obtain the correlation matrix of different wind farms by processing the wind speed state transition matrix of a certain month using the Pearson correlation coefficient; wherein: the Pearson correlation coefficient is used to calculate the Pearson correlation coefficient of the wind speed of the i-th and j-th wind farms by denoting the historical time-series wind speed data of N wind farms in a certain region as V=[V1,V2,V3,…,VN]T.

[0019]

[0020] Among them, Cov(V i V j ) represents the covariance, σ i and σ j denoted as the variances of the wind speed values ​​for the i-th and j-th wind farms, respectively;

[0021] 203. Establish a time-series wind speed normal distribution model for each wind farm based on random number sampling;

[0022]

[0023] S L z = γ + ηln(x - ζ) x > ζ

[0024] S U :z=γ+ηarcsinh((x-ζ) / λ)

[0025] S N z = (x - μ) / σ

[0026] Where: γ, η, ζ, and λ are the parameters of the normal transformation, and μ and σ represent the mean and standard deviation of the variable x, respectively; S B S represents a family of bounded distribution functions. L S represents a family of log-normal distribution functions. U S represents a family of unbounded distribution functions. N Represents the normal distribution function;

[0027] 204. Generate a correlated time-series wind speed vector and power output vector for each wind farm based on the Johnson distribution system. Further, the process of generating the time-series wind speed vector for each wind farm in step 204 includes:

[0028] Test whether the sample data of the variable conforms to a normal distribution. If it does, calculate the mean and variance and express them as a linear function of the standard normal distribution; otherwise, proceed to the following steps.

[0029] Choose a real number y greater than 0, and find the probability values ​​p corresponding to -3y, -y, y, and 3y according to the cumulative distribution function of the standard normal distribution. -3y p -y py and p 3y ;

[0030] Find the corresponding quantile x in the input sample of time-series wind speed data x. -3y x -y x y and x 3y ;

[0031] Let m = x 3y -x y n = x -y -x -3y p = x y -x -y And calculate k = mn / p 2 Value:

[0032] If k < 1, choose S. B Distribution; if k=1, choose S L Distribution; if k>1, choose S U Distribution, in which:

[0033] S U distributed:

[0034]

[0035] S B distributed:

[0036]

[0037] S L distributed

[0038] Furthermore, the process of generating the wind farm output vector in step 204 includes:

[0039] The correlation coefficient matrix ρ is calculated according to the following formula. V Decomposition to obtain the lower triangular matrix

[0040]

[0041] The correlation coefficient matrix ρ is calculated according to the following formula. V Standard normal vector Z

[0042] Z = LV n

[0043] Wind speed data model for wind farms, based on the standard normal distribution variable Z and the correlation of the time-series wind speed normalization model for each wind farm:

[0044]

[0045]

[0046]

[0047] S N :x=μ+σz.

[0048] Furthermore, Step 4 employs a binary search method to adjust the load of the wind farm system and search for the reliable capacity of the wind farm group, including:

[0049] The capacity of conventional generating units in the power system is set as Cap. con The load level is L0, and its reliability level can be expressed as Re{Cap con ,L0}, when the capacity is Cap wind The wind turbines are connected to the power transmission system, improving the power supply reliability level to Re{Cap. con +Cap wind The capacity value of newly connected wind turbine units can be represented by the amount of additional load they carry.

[0050] When the load level increases to L0+ΔL, if the equal reliability criterion holds, then the load increase ΔL is equal to the access capacity Cap. wind The reliable capacity of distributed wind and solar turbine units; where:

[0051] The steps for implementing trusted capacity search based on the binary search method are as follows:

[0052] 401. Without connecting to the wind turbine cluster, maintain the system load level L0 unchanged, and assess the current reliability level Re{Cap}. con ,L0};

[0053] 402, access capacities are respectively Cap wind For wind power, maintaining a constant system load level L0 while improving system reliability, denoted as Re{Cap wind +Cap con ,L0}, set the initial value l1 = 1;

[0054] 403. Increase the load level to L0+l1*L' and assess the reliability level Re{Cap. wind +Cap con ,L0+l1*L'};

[0055] Determine if Re{Cap} is satisfied. wind +Cap con ,L0+l1*L'}>Re{Cap con If not, proceed to step 404; if yes, proceed to step 405.

[0056] 404. Let l1 = l1 + 1, then return to step 403.

[0057] 405. The load increase corresponding to the reliable capacity is considered to be between [L0+(l1-1)*L',L0+l1*L']. Based on the bisection method, the reliable capacity of the wind turbine is searched until the load level L0+ΔL (ΔL∈[L0+(l1-1)*L',L0+l1*L']) corresponds to Re{Cap wind +Cap con If L0+ΔL} satisfies the equal reliability criterion shown in equation (x), then ΔL obtained in this case is the reliable capacity of the wind turbine; that is:

[0058] Re{Cap con ,L0}=Re{Cap con +Cap wind ,L0+ΔL}.

[0059] Beneficial effects

[0060] This invention proposes a method for assessing the capacity value of wind farm clusters based on the Johnson distribution system. Its key features are: ① using a first-order Markov chain to preserve the volatility of wind farm output; ② considering the correlation of output between different wind farms based on the Johnson distribution system; and ③ using the increment of the system's supplied load as the reliable capacity of the wind farm cluster based on the principle of equal power supply reliability. This method overcomes the blind spots of the proportional coefficient method, yields capacity value assessment results based on power supply reliability, and provides auxiliary support information for power system power balance analysis and power planning. Attached Figure Description

[0061] Figure 1 This is a flowchart of the wind farm cluster capacity value assessment method of the present invention.

[0062] Figure 2 This is a schematic diagram of the wind farm cluster capacity value search method based on equal reliability according to the present invention. Detailed Implementation

[0063] The following is in conjunction with the appendix Figure 1-2 The present invention is described as follows:

[0064] This invention proposes a method for assessing the capacity value of wind farm clusters based on the Johnson distribution system. This method considers the fluctuations in wind farm output based on a first-order Markov chain and models the correlation of output between different wind farms using the Johnson distribution system. By fully taking into account the output correlation between different wind farms, it assesses the reliable capacity of the wind farm cluster.

[0065] like Figure 1As shown: This invention will complete the assessment of the capacity value of a wind farm cluster based on the Johnson distribution system through the following steps:

[0066] Step 1: Statistical Analysis of Wind Farm Output Fluctuation Based on First-Order Markov Chain

[0067] Step 2: Time-series power output simulation of wind farm clusters based on the Johnson distribution system

[0068] Step 3: Reliability assessment of power systems including wind farm clusters based on power adequacy indicators

[0069] Step 4: Trusted Capacity Search of Wind Farm Clusters Based on Equal Power Supply Reliability Criteria

[0070] This invention proposes a method for assessing the capacity value of wind farm clusters based on the Johnson distribution system. The following is a detailed explanation in conjunction with the appendix. Figure 1 The implementation process of this invention will be described in further detail.

[0071] This invention proposes a method for assessing the capacity value of wind farm clusters based on the Johnson distribution system. The following is a detailed explanation in conjunction with the appendix. Figure 1 The implementation process of this invention patent will be described in further detail.

[0072] 1. Statistical Analysis of Wind Farm Output Fluctuation Based on First-Order Markov Chain

[0073] 1.1 Obtaining historical wind speed data

[0074] Download historical time-series wind speed data for the wind farm construction site from The POWER Project module on the NASA Power website, using an hourly time scale to download 8760 hours of continuous time-series wind speed data for 20 consecutive years.

[0075] 1.2 Generating the wind speed state transition matrix

[0076] A first-order Markov chain X = [X1, X2, X3, ..., X...] t In the state space S = [ω1, ω2, ω3, ..., ω...], the wind speed value range is divided into s states, and the state space is S = [ω1, ω2, ω3, ..., ω...]. s The state ω is represented by [ ]. i Corresponding wind speed range v i As shown in equation (1).

[0077]

[0078] Among them, v max This represents the maximum historical wind speed over a period of twenty years.

[0079] Generally, the climate of a region can be considered relatively stable over a period of several decades, and therefore the wind speed model within this time frame remains statistically unchanged. A first-order Markov chain X satisfies equation (2), and for all j, ω0, ω1, ..., ω... t The property of equation (2) is called the Markov property, which holds true for both ∈S and t∈{0,1,…}.

[0080]

[0081] Among them, X t ω represents the element in the time series corresponding to the first-order Markov chain; j represents a certain state of wind speed; ω t Let P be the element in the Markov time series composed of wind speed states; P(·|·) is the conditional transition probability. Based on the S-statistics obtained from the wind speed states, the wind speed state transition matrix P for the i-th month of the year is shown in Equation (3) below. i,tra .

[0082]

[0083] Among them, P 11 This represents the probability that the wind speed state is ω1 at time t and changes to ω2 at time t+1, and so on.

[0084] 2. Time-series power output simulation of wind farm clusters based on the Johnson distribution system

[0085] 2.1 Generating the correlation matrix of multiple wind farms

[0086] Let the historical time-series wind speed data of N wind farms in a certain region be denoted as V = [V1, V2, V3, ..., V]. N ] T The formula for calculating the Pearson correlation coefficient (PCC) of the wind speed of the i-th and j-th wind farms is shown in equation (4).

[0087]

[0088] Among them, Cov(V i V j ) represents the covariance, σ i and σ j Let $P_i$ and $P_j$ be the variances of the wind speed values ​​for the $i$-th and $j$-th wind farms, respectively. When $P_i$ is greater than 0, the wind speeds of the two wind farms are positively correlated; when $P_j$ is less than 0, the wind speeds of the two wind farms are negatively correlated; the larger the absolute value, the stronger the correlation. The correlation between any two wind farms is statistically analyzed, and a correlation matrix $ρ$ is established. V .

[0089] 2.2 Generation of Time-Series Wind Speed ​​Data Based on First-Order Markov Chain

[0090] The annual time-series wind speed curve is generated based on the 12-month wind speed state transition matrix obtained in section 1.2. The specific operation steps are as follows:

[0091] (1) Randomly sample the wind speed status in the first hour of the first month;

[0092] (2) Sampling is performed based on the wind speed state transition matrix of the first month to obtain the wind speed state of all hours in the current month;

[0093] (3) Based on the wind speed status in the last hour of the i-th month (1<=i<=11) and P i+1,tra The wind speed status for the (i+1)th month is obtained by sampling.

[0094] (4) Convert the wind speed status of all hours into specific wind speed values, state ω i Corresponding wind speed range v i As shown in equation (1), sampling within the corresponding range can be obtained.

[0095] 2.3 Normalization of Time-Series Wind Speed ​​in Wind Farms

[0096] Based on the known time-series wind speed data from Section 1, the relationship between historical time-series wind speed data and a standard normal variable can be obtained using the Johnson distribution system, i.e., the distribution of the simulated variable. The arbitrary unimodal distribution of time-series wind speed data x can be converted to a standard normal distribution using the following formula.

[0097]

[0098] S L :z=γ+ηln(x-ζ)x>ζ (5)

[0099] S U :z=γ+ηarcsinh((x-ζ) / λ)

[0100] S N z = (x - μ) / σ

[0101] Where γ, η, ζ, and λ are the parameters of the normal transformation, and μ and σ represent the mean and standard deviation of the variable x, respectively; S B S represents a family of bounded distribution functions. L S represents a family of log-normal distribution functions. U S represents a family of unbounded distribution functions. N The above four curves can cover any unimodal probability distribution function, that is, any unimodal variable can be expressed as a function of the standard normal variable as shown in equation (5).

[0102] To perform the transformation shown in equation (5), it is first necessary to determine which family of distribution functions the variable belongs to, i.e., to select a suitable Johnson distribution curve. The following describes how to select a suitable Johnson curve.

[0103] (1) Test whether the sample data of the variable conforms to a normal distribution. If it does, calculate the mean and variance and express them as a linear function of the standard normal distribution; otherwise, proceed to the following steps.

[0104] (2) Choose a real number y that is greater than 0, and find the probability values ​​p corresponding to -3y, -y, y, and 3y according to the cumulative distribution function of the standard normal distribution. -3y p -y p y and p 3y ;

[0105] (3) Find the corresponding quantile x in the input sample of the time-series wind speed data x. -3y x -y x y and x 3y ;

[0106] (4) Let m = x 3y -x y n = x -y -x -3y p = x y -x -y And calculate k = mn / p 2 The value of .

[0107] If k < 1, choose S. B Distribution; if k=1, choose S L Distribution; if k>1, choose S U distributed.

[0108] The following describes the calculation methods for parameters of different curve families:

[0109] (1)S U distributed

[0110]

[0111] (2)S B distributed

[0112]

[0113] (3)S L distributed

[0114]

[0115] 2.3 Generating Time-Series Active Power Output Data for Wind Farms

[0116] (1) For ρ V Decomposition yields the lower triangular matrix L

[0117] Due to ρ V If Q is a nonnegative definite matrix, then there exist an orthogonal matrix Q and a nonnegative diagonal matrix Λ such that Q T ρ V Q = Λ, and by transforming this equation, we get equation (9).

[0118]

[0119] (2) Generate relevant wind speed data for wind farms.

[0120] The time-series wind speed data of N wind farms in a certain region, based on a first-order Markov chain, are all transformed into normally distributed data using the method in section 2.1, denoted as V. n =[V n1 V n2 V n3 ,…,V nN ] T V n It is an independent standard normal random column vector, while preserving the fluctuating characteristics of wind speed. It has a correlation coefficient matrix ρ. V The standard normal vector Z can be calculated using equation (10).

[0121] Z = LV n (10)

[0122] The above transformation can be used to obtain the variable Z of the standard normal distribution. Substituting the obtained Z into formula (11) will yield the wind speed data of the wind farm with correlation.

[0123]

[0124]

[0125]

[0126] S N :x=μ+σz.

[0127] (3) Convert wind speed data into wind turbine output data

[0128] The wind speed value is converted into wind turbine output according to the following formula (12), where the active power output of the i-th wind turbine in a single wind farm is denoted as P. W,i .

[0129]

[0130] Where v is the wind speed, v in vout and v r These are the cut-in wind speed, cut-out wind speed, and rated wind speed of the wind turbine unit, respectively. Let be the rated output power of the i-th fan.

[0131] 3. Reliability assessment of power systems including wind farm clusters

[0132] The basic idea of ​​Monte Carlo simulation is to generate random numbers through a large number of random samples, use these random numbers to simulate the behavior of the system, and finally obtain the required statistical results. Monte Carlo simulation is further divided into sequential Monte Carlo simulation and non-sequential Monte Carlo simulation. In order to take into account the situation that the power output of the wind farm changes continuously over time, this paper adopts the sequential Monte Carlo method.

[0133] 3.1 Generate the timing operation state vectors of all units

[0134] (1) Assuming all components are initially in a fault-free operating state, generate the fault-free operating time t. TTF and fault repair duration t TTR Assuming that components in the system have only two states—fault and normal—and that both the failure rate and repair rate follow an exponential distribution, the fault-free operating time of the k-th component can be generated using random number sampling.

[0135]

[0136] Where r is a random number uniformly distributed in [0,1], and λ k Let λ be the failure rate of the k-th component. k Turn into repair rate μ k The fault repair duration of the kth component can be obtained.

[0137] (2) Generate the timing state vector within the evaluation period T. Taking the k-th element as an example, the t obtained from equation (13) is... TTF and t TTR Sequentially insert the time-series state vector ST k ,but Repeat step (1) until the following equation is satisfied:

[0138]

[0139] (3) Repeat step (2) until the timing state vectors of all components in the system are generated. The components to be simulated include various types of units in the system. When the power supply equipment is in repair state, the output is zero.

[0140] 3.2 Failure Consequence Analysis and Reliability Index Statistics.

[0141] When a component fails, a failure consequence analysis of the system is required. If the remaining generating capacity on the generator bus can compensate for the unavailable capacity on the same bus caused by the loss of the generator, then the generator can still supply power to the load normally. Otherwise, it indicates that the bus has failed. The difference between the power demand of the load and the generating capacity after the generator failure is the expected power shortage (EENS).

[0142] By continuously accumulating reliability metrics over a given evaluation period, the overall system reliability metric is obtained. The following metrics can be selected: Loss Duration (LOLD), Expected Loss Duration (LOLE), and Expected Loss Power (EENS). Let LLO be the number of power outages occurring in the i-th sampling year. i The duration of power outage is LLD i The power shortage in the sampled year was ENS. i There are a total of N sampling years, and the formula for calculating the reliability index is as follows:

[0143]

[0144] 3.3 Sequential Monte Carlo Simulation Based on Variance Coefficients

[0145] In sequential Monte Carlo sampling, the variance coefficient is used as the convergence criterion for Monte Carlo sampling.

[0146] (1) Construct a standard normal distribution of the variance coefficient based on the definition of the variance coefficient.

[0147] The variance coefficient is calculated as follows:

[0148]

[0149] Where χ represents the variance coefficient; The variance of the sampled data; σ is the sample mean; when n is sufficiently large, σ is the sample standard deviation. When the variance coefficient is sufficiently small, Monte Carlo sampling is considered convergent, and sampling stops. Based on mathematical statistics, a standard normal distribution is constructed, as shown in equation (x):

[0150]

[0151] (2) Based on the properties of the standard normal distribution, the expression for the accuracy of the Monte Carlo convergence criterion is obtained.

[0152] For a population that follows a standard normal distribution, the following equation holds:

[0153]

[0154] Where, μ α / 2φ(μ) represents the upper quantile of the standard normal distribution with respect to α / 2. α / 2 1-α is the probability at the upper quantile in a normal distribution; 1-α is the confidence probability. Transforming equation (x) yields the expression for the precision of the Monte Carlo convergence criterion:

[0155]

[0156] (3) Set appropriate variance coefficients and significance levels, and calculate the error of Monte Carlo sampling.

[0157] Given α = 0.05, its corresponding upper quantile α / 2 = 1.96. Setting the variance coefficient χ ≤ 0.05, the error accuracy at α = 0.05 can be calculated using the above formula:

[0158]

[0159] As can be seen from the above formula, under a given reasonable confidence level α = 0.05, the error between the true value y of the ideal reliability level and the estimated value y obtained by sequential Monte Carlo sampling is small, within ±10%.

[0160] 4. Reliable capacity search of wind farm clusters based on equal power supply reliability criteria, such as... Figure 2 As shown:

[0161] Assume the capacity of conventional generating units in the power system is Cap. con The load level is L0, and its reliability level can be expressed as Re{Cap con The smaller the value of L0, the higher the system's reliability level. When the capacity is Cap... wind The wind turbines are connected to the power grid, improving the power supply reliability level to Re{Cap. con +Cap wind The capacity value of newly connected wind turbines can be represented by the increased load capacity (L0). When the load level increases to L0+ΔL, if the equal reliability criterion holds, then the increase in load capacity ΔL is called the connected capacity Cap. wind The reliable capacity of distributed wind and solar power units. Calculating reliable capacity is a one-dimensional search process that requires repeatedly probing load increases. The specific steps for reliable capacity search based on the binary search method are as follows:

[0162] (1) Do not connect wind turbines, and keep the system load level L0 unchanged. Based on the solution results in 3.2, calculate the power shortage caused by system faults and assess the current reliability level Re{Cap. con ,L0}.

[0163] (2) The access capacity is Cap windFor wind power, maintaining a constant system load level L0 while improving system reliability, denoted as Re{Cap wind +Cap con ,L0}, set the initial value l1=1.

[0164] (3) Increase the load level to L0+l1*L' and assess the reliability level Re{Cap. wind +Cap con ,L0+l1*L'}. Determine if Re{Cap} is satisfied. wind +Cap con ,L0+l1*L'}>Re{Cap con ,L0};If no, proceed to step (4);If yes, proceed to step (5).

[0165] (4) Let l1 = l1 + 1, and return to step (3).

[0166] (5) The load increase corresponding to the reliable capacity is considered to be between [L0+(l1-1)*L',L0+l1*L']. Based on the binary search method, the reliable capacity of the wind turbine is searched until the load level L0+ΔL (ΔL∈[L0+(l1-1)*L',L0+l1*L']) corresponds to Re{Cap. wind +Cap con If L0+ΔL} satisfies the equal reliability criterion shown in equation (x), then ΔL obtained is the reliable capacity of the wind turbine.

[0167] Re{Cap con ,L0}=Re{Cap con +Cap wind ,L0+ΔL} (21)

[0168] Although the present invention has been described above, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many modifications under the guidance of the present invention without departing from the spirit of the present invention, and these modifications are all within the protection scope of the present invention.

Claims

1. A method for assessing the capacity value of wind farm clusters based on the Johnson distribution system, characterized in that, Includes the following steps: Step 1: Establish a wind farm cluster fluctuation model by analyzing the power output of the wind farm cluster using the first-order Markov chain algorithm; Step 2: Establish relevant models for different wind farm clusters through time-series power output analysis using the Johnson distribution system; including:

201. Calculate the wind speed state transition matrix for a specific month in the wind farm cluster fluctuation model according to the year requirements; 202. Obtain the correlation matrix of different wind farms by processing the wind speed state transition matrix of a certain month using the Pearson correlation coefficient; wherein: the Pearson correlation coefficient is used to calculate the Pearson correlation coefficient of the wind speed of the i-th and j-th wind farms by denoting the historical time-series wind speed data of N wind farms in a certain region as V=[V1,V2,V3,…,VN]T. in, For covariance, and denoted as the variances of the wind speed values ​​for the i-th and j-th wind farms, respectively; 203. Establish a time-series wind speed normal distribution model for each wind farm based on random number sampling; in: , , and The parameters for the normal transformation are... and Representing variables respectively x The mean and standard deviation; S B Describes a family of bounded distribution functions. S L Describes a family of log-normal distribution functions. S U Describes a family of unbounded distribution functions. S N Represents the normal distribution function; 204. Generate correlated time-series wind speed and power output vectors for each wind farm based on the Johnson distribution system: Convert the wind speed value into wind turbine output using the following formula, where the first wind speed value is the first wind turbine output value in a single wind farm. i The active power output of the typhoon generator is recorded as P W,i Where v is the wind speed. , and These are the cut-in wind speed, cut-out wind speed, and rated wind speed of the wind turbine unit, respectively. Let be the rated output power of the i-th fan; Step 3: Based on the given load level and wind farm capacity, conduct a reliability assessment of the power system according to the power adequacy index. Step 4: Use the binary search method to adjust the load of the wind farm system to search for the reliable capacity of the wind farm group.

2. The method for assessing the capacity value of wind farm clusters based on the Johnson distribution system according to claim 1, characterized in that, The process of generating the time-series wind speed vector for each wind farm in step 204 includes: Test whether the sample data of the variable conforms to a normal distribution. If it does, calculate the mean and variance and express them as a linear function of the standard normal distribution; otherwise, proceed to the following steps. Choose real numbers greater than 0. y Based on the cumulative distribution function of the standard normal distribution, find the value corresponding to -3. y - y , y , and 3 y probability value p -3y , p -y , p y and p 3y ; Time-series wind speed data x Find the corresponding quantile in the input sample x -3y , x -y , x y and x 3y ; make , , and calculate Value: like k <1, Select S B Distribution; if k =1, select S L Distribution; if k >1, Select S U Distribution, in which: S U distributed: S B distributed: S L distributed 。 3. The method for assessing the capacity value of wind farm clusters based on the Johnson distribution system according to claim 1, characterized in that, The process of generating the wind farm output vector in step 204 includes: The correlation coefficient matrix is ​​processed according to the following formula. Decomposition to obtain the lower triangular matrix L in: Indicates an orthogonal array; Represent a non-negative diagonal matrix; such that ; The correlation coefficient matrix is ​​processed according to the following formula. Standard normal vector Z： Variables based on the standard normal distribution Z-combination Wind speed data models for each wind farm with correlation to the time-series wind speed normal distribution model: 。 4. The method for assessing the capacity value of a wind farm cluster based on the Johnson distribution system according to claim 1, characterized in that, Step 4, which employs a bisection method to search for the reliable capacity of the wind farm cluster by adjusting the load of the wind farm system, includes: The capacity of conventional generating units in the power system is set as follows: Cap con The load level is L 0, its reliability level can be expressed as Re { Cap con , L 0}, when the capacity is Cap wind The wind turbines are connected to the power transmission system, improving the power supply reliability level to Re { Cap con + Cap wind , L 0}, the capacity value of newly connected wind turbine units can be represented by the amount of additional load they carry; When the load level rises to L 0 + L If the reliability criterion holds, then the load-bearing capacity will increase. L For access capacity Cap wind The reliable capacity of distributed wind and solar turbine units; where: The steps for implementing trusted capacity search based on the binary search method are as follows:

401. Do not connect to wind turbine clusters; maintain system load level. L Keeping 0 constant, assess the current reliability level. Re { Cap con , L 0}; 402, access capacities are respectively Cap wind Wind power supply to maintain system load level L With 0 unchanged, the system reliability is improved, denoted as Re { Cap wind + Cap con , L 0}, set the initial value l 1 = 1; 403. Increase the load level to L 0+ l 1 L’ Assess reliability level Re { Cap wind + Cap con , L 0+ l 1 L’ }; Determine if it satisfies Re { Cap wind + Cap con , L 0+ l 1 L’ } > Re { Cap con , L 0}; If no, proceed to step 404; If yes, proceed to step 405; 404, Order l 1= l 1+1, return to step 403; 405. The load increase corresponding to the reliable capacity is considered to be between [ L 0+( l 1-1) L’ , L 0+ l 1 L’ Between [a certain point], a binary search method is used to find the reliable capacity of wind turbine units until the load level [is reached]. L 0+ L ( L ∈ [ L 0+( l 1-1) L’ , L 0+ l 1 L’ ]) corresponding Re { Cap wind + Cap con , L 0+ L } Satisfies the equal reliability criterion shown in equation (x), and the obtained result is L This refers to the reliable capacity of the wind turbine; that is: 。

Images