A method for assessing the adequacy of a wind power generation system considering daily and seasonal characteristics

Abstract

A method for assessing the adequacy of a wind power generation system taking into account daily and seasonal characteristics is disclosed. The method collects hourly wind speed data from a wind farm, and constructs a 24-hour wind speed joint probability distribution model based on a Gaussian copula function according to the maximum likelihood estimation method. Samples are drawn from the 24-hour wind speed joint probability distribution to generate simulated I wind speed taking into account daily characteristics. A deviation function about the seasonal correction coefficient is constructed to derive the optimal seasonal correction coefficient, and the deviation between the seasonal mean and standard deviation of the simulated wind speed sample before correction and the seasonal mean and standard deviation of the measured wind speed sample is minimized to obtain simulated II wind speed. Using simulated I wind speed and simulated II wind speed, a seasonal adequacy assessment procedure for a wind power combined power generation system taking into account daily and seasonal characteristics is established, and the seasonal pattern, the number of wind turbines and the system peak load are analyzed to analyze the influence of the seasonal pattern, the number of wind turbines and the system peak load on the seasonal adequacy of the system. The present invention provides a basis for studying the influence of the seasonal pattern, the number of wind turbines and the system peak load on the seasonal adequacy of the system.

Description

Technical Field

[0001] The invention belongs to the technical field of power system simulation, and in particular relates to a method for assessing the adequacy of a wind power generation system taking into account daily characteristics and seasonal characteristics. Background Art

[0002] At high wind power penetration levels, wind power will have a significant impact on wind power generation systems. The generation of wind power mainly depends on wind speed, which is affected by factors such as terrain and season, and has an uncertain characteristic. Therefore, it is extremely important to consider its uncertainty when modeling wind speed.

[0003] It is generally believed that wind speed has daily and seasonal variations, which affects the uneven distribution of wind resources throughout the year. Therefore, it is very important to propose an appropriate wind speed model to consider the daily and seasonal characteristics of wind speed. Compared with the wind speed analytical model, the wind speed simulation model can take into account multiple types of wind speed characteristics more finely and is a better solution.

[0004] At present, many scholars have used several wind speed analysis and simulation models to study uncertain wind speed patterns. However, there are relatively few works that use wind speed simulation models to study the diurnal and seasonal variations of wind speed. An existing non-homogeneous Markov wind speed time series model takes into account the characteristics of diurnal and seasonal variations. However, there may be two obvious limitations: first, an unrealistic assumption is made that the wind speed is uniformly distributed in several intervals. In practical applications, we found that the wind speed is not uniformly distributed; second, the simulated wind speed is corrected by using seasonal indices to make the mean of the simulated wind speed close to the mean of the actual wind speed in each season, but the standard deviation of the simulated wind speed may be inconsistent with the standard deviation of the actual wind speed in each season. Another method groups the wind speed data by day and season, and uses the kernel density method to estimate the WSPD of each cluster, which is used to generate wind speed samples in the non-sequential Monte Carlo simulation of the wind integrated IEEE-RTS power generation system. However, it may be difficult to determine the optimal bandwidth of the kernel density, and the inverse cumulative density function of the kernel density model has no parameterized expression, which brings difficulties to wind speed sampling. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a method for assessing the adequacy of a wind power generation system taking into account daily characteristics and seasonal characteristics. The method takes the wind speed simulation model as the research object, takes into account daily characteristics and seasonal characteristics, and designs a two-stage wind speed simulation model taking into account daily characteristics and seasonal characteristics according to the daily cycle variation characteristics and seasonal variation characteristics of the wind speed. This provides a basis for studying the influence of seasonal patterns, the number of wind turbines and the peak load of the system on the seasonal adequacy of the system.

[0006] The technical solution adopted by the present invention is:

[0007] A method for assessing the adequacy of a wind power generation system taking into account daily characteristics and seasonal characteristics comprises the following steps:

[0008] Step 1: Collect hourly wind speed data from the wind farm, determine the parameters of multiple candidate probability distributions based on the maximum likelihood estimation method, calculate the AIC index of multiple candidate probability distributions, and select the candidate probability distribution with the smallest AIC index as the optimal fitting hourly wind speed marginal probability distribution model;

[0009] Step 2: Construct a 24-hour wind speed joint probability distribution model based on the Gaussian copula function;

[0010] Step 3: Based on the Cholesky decomposition method and standard normal distribution random numbers, sampling is performed from the 24-hour wind speed joint probability distribution to generate simulated I wind speed taking into account daily characteristics;

[0011] Step 4: Introduce the quarterly correction coefficient, correct the simulation-Ⅰ wind speed sample through the correction equation, construct the deviation function of the quarterly correction coefficient, and derive the optimal quarterly correction coefficient The deviations between the seasonal mean and standard deviation of the simulated wind speed samples before correction and the seasonal mean and standard deviation of the measured wind speed samples are minimized to obtain the simulated II wind speed samples.

[0012] Step 5: Using the simulated wind speed samples I and II, establish a seasonal adequacy assessment procedure for the wind power combined power generation system that considers daily and seasonal characteristics, and analyze the impact of seasonal patterns, the number of wind turbines, and system peak load on the seasonal adequacy of the system.

[0013] The step 1 comprises the following steps:

[0014] S1.1: Access the open meteorological database and collect the 24-hour wind speed data set of the wind farm.

[0015] S1.2: Using wind speed data, perform maximum likelihood estimation to calculate parameter values ​​of nine candidate probability distributions, including Weibull, Rayleigh, Lognormal, Gamma, Inverse Gaussian, Birnbaum-Saunders, Nakagami, Logistic, and GeneralizedExtremeValue;

[0016] S1.3: Calculate the AIC index of nine candidate probability distributions. Taking Weibull candidate probability distribution and wind speed V1 as an example, the AIC index calculation formula is:

[0017] S1.4: Select the candidate probability distribution with the smallest AIC index as the best fitting hourly wind speed marginal probability distribution model. In step 2, since Gaussian Copula can describe high-dimensional correlation with a relatively clear expression, Gaussian Copula is used. The formula of the 24-hour wind speed joint probability distribution model based on the Gaussian copula function is as follows:

[0018]

[0019] In the above function: f(·) is the 24-hour wind speed joint probability density function, v(i,j) represents the wind speed at the jth hour on the i-th day, F j (·) and f j (·) are the cumulative probability density function and probability density function of the j-th hour marginal wind speed determined by the AIC criterion from the 9 sets of candidate parameter probability distributions;

[0020] is the Gaussian copula probability density function, which is expressed as:

[0021]

[0022] In the above formula, the symbol vector x i Expressed as I is the identity matrix, Φ -1 (·) is the inverse function of the standard normal cumulative probability density function. The symbol |·| indicates that the determinant of the matrix is ​​obtained. R is the correlation coefficient matrix of the Gaussian copula function, and it is also the only parameter of the Gaussian copula function to be determined, which can be obtained by Estimated. i T , R -1 , N are x i The transposed matrix of , the inverse matrix of R and the number of days on which wind speed samples were collected.

[0023] In the step 3, according to the Cholesky decomposition method and the standard normal distribution random number, The simulated I wind speed is obtained by sampling from the 24-hour wind speed joint probability density function; To simulate the simulated I wind speed at the jth (j=1,...,24) hour on the kth (k=1,...,M) day, M is not necessarily equal to N, and M can be any positive integer.

[0024] In the step 4, the correction equation is:

[0025]

[0026] Where:(s,t) and (s,t) is the seasonal correction coefficient of season s in period t, is the simulated I wind speed before correction, is the corrected simulation II wind speed.

[0027] According to the sample mean and standard deviation formula, the mean and standard deviation of simulated wind speed I and simulated wind speed II satisfy the following two relationships:

[0028]

[0029] Where: and are the mean and standard deviation of the simulated I wind speed in the tth period and sth season before correction;

[0030] and are the mean and standard deviation of the simulated II wind speed in the tth period and sth season, respectively;

[0031] The deviation function of the quarterly correction coefficient is:

[0032]

[0033] μ (s,t) , σ (s,t) are the average and standard deviation of the actual wind speed in period t and season s respectively. When the deviation function takes the minimum value, the optimal correction coefficient expression is derived:

[0034]

[0035] The optimal correction is achieved by applying the optimal quarterly correction coefficient to the correction equation. This optimal correction minimizes the deviation between the mean and standard deviation of the corrected simulated II wind speed and the mean and standard deviation of the measured wind speed, thereby obtaining the optimal simulated II wind speed sample with daily and seasonal characteristics.

[0036] In the step 5, actual wind speed samples, unit parameters, and system load demand data of the wind power cogeneration system are collected to calculate three seasonal adequacy indexes: seasonal power shortage time expected value SLOLE, seasonal power shortage expected value SLOEE, and seasonal power shortage frequency SLOLF; the steps include:

[0037] S5.1: First, sample the failure time and repair time of all generator sets to determine the status of the wind power generation system.

[0038] The failure time formula is: TTF = -ln(U) / λ; where: U is a uniformly distributed random number in the interval [0,1], and λ is the failure rate of the generating unit;

[0039] The repair time formula is: TTR = -ln(U) / γ, where: γ is the failure rate of the generation unit.

[0040] S5.2: Calculate the available capacity of conventional units, Where: Y up is the number of legacy units remaining in operation, G y is the rated power of the conventional unit left in operation;

[0041] S5.3: The formula for calculating the output power using the simulated II wind speed is: Where: Z up is the number of wind turbines in operation, is the wind turbine output power at simulated wind speed II in the jth period k season.

[0042] The power output formula of a wind turbine is:

[0043] Where: v ci 、v r and v out are cut-in wind speed, rated wind speed and cut-out wind speed respectively, P r is the rated power of the wind turbine. Coefficients A, B and C are expressed in terms of v ci and v r to be determined

[40] , as shown below:

[0044]

[0045]

[0046]

[0047] S5.4: Using P WF and P G Seasonal system load demand, calculate the LLD for the qth sample season q Loss of load duration, ENS q Low power (energy not supplied), LLO qThe number of power outages (loss of load occurrence). Then the seasonal power outage time expectation (seasonal loss of load expectation, SLOLE), seasonal energy shortage expectation (seasonal loss of energy expectation, SLOEE) and seasonal power outage frequency (seasonal loss of load frequency, SLOLF) are obtained; the calculation formulas of the three indicators are:

[0048]

[0049]

[0050]

[0051] Where: LLDq is the loss of load duration in the qth sample season, ENSq is the energy not supplied in the qth sample season, LLOq is the loss of load occurrence in the qth sample season. Q is the total number of seasons.

[0052] S5.5: Repeat steps S5.1 to S5.4 above until the coefficient of variation of the SLOEE indicator is less than 0.05 or the sample season Q is greater than 1,000,000.

[0053] S5.6: Output three seasonal adequacy indexes: SLOLE, SLOEE and SLOLF. Based on these three indexes, analyze the impact of seasonal patterns, number of wind turbines and system peak load on the seasonal adequacy of the system.

[0054] The present invention provides a method for assessing the adequacy of a wind power generation system taking into account daily characteristics and seasonal characteristics, and the technical effects are as follows:

[0055] 1) Aiming at the daily cycle variation characteristics of wind speed, the present invention designs a 24-hour wind speed joint probability distribution model taking into account daily characteristics. The parameters of nine candidate probability distributions are determined by the maximum likelihood estimation method, and the AIC criterion is used to make the selected hourly wind speed marginal probability distribution have the best fitting performance. Considering the characteristics of the large number of random variables in the 24-hour wind speed probability distribution established in this paper, the Gaussian copula function is selected, and then the simulated I wind speed is generated by the cholesky decomposition method and the standard normal distribution random number sampling, so that the sampling is reliable and simple.

[0056] 2) Aiming at the seasonal variation characteristics of wind speed, the present invention designs a simulation wind speed sample correction model that takes into account seasonal characteristics. The two quarterly correction coefficients proposed by the model make the mean and standard deviation of the simulation wind speed samples present seasonal characteristics. The simulated I wind speed is corrected by the proposed quarterly correction coefficient and correction equation, taking into account the seasonal characteristics of wind speed. Then, based on the derived optimal quarterly correction coefficient, the difference and standard deviation between the mean and standard deviation of the simulated I wind speed in each season and the actual wind speed are minimized, and the optimal correction coefficient is calculated. The optimal quarterly correction coefficient is applied to the correction equation to correct the simulated I wind speed generated by sampling, and the simulated II wind speed showing daily and seasonal characteristics is obtained.

[0057] 5) The present invention establishes a seasonal adequacy assessment procedure for a wind power generation system that takes into account daily characteristics and seasonal patterns, and studies the impact of seasonal patterns, the number of wind turbines, and system peak load on the seasonal adequacy of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1(a) shows the average values ​​of actual, simulated I, and simulated II wind speeds in the spring of 10 periods in the Kando area of ​​North Dakota, USA;

[0059] Figure 1(b) shows the average values ​​of actual, simulated I, and simulated II wind speeds in the summer of 10 periods in the Kando area of ​​North Dakota, USA.

[0060] Figure 1(c) shows the average values ​​of actual, simulated I, and simulated II wind speeds in the fall over 10 periods in the Kando area of ​​North Dakota, USA.

[0061] Figure 1(d) shows the average values ​​of actual, simulated I, and simulated II wind speeds in the Kando area of ​​North Dakota, USA, during 10 periods of winter.

[0062] Figure 2(a) shows the standard deviations of actual, simulated I, and simulated II wind speeds in the spring of 10 periods in the Kando region of North Dakota, USA;

[0063] Figure 2(b) shows the standard deviations of actual, simulated I, and simulated II wind speeds in the spring of 10 periods in the Kando region of North Dakota, USA;

[0064] Figure 2(c) shows the standard deviations of actual, simulated I, and simulated II wind speeds in the spring of 10 periods in the Kando region of North Dakota, USA;

[0065] Figure 2(d) shows the standard deviations of actual, simulated I, and simulated II wind speeds in the spring over 10 periods in the Kando area of ​​North Dakota, USA.

[0066] Figure 3 The figure is a flow chart of the evaluation method of the present invention. DETAILED DESCRIPTION

[0067] A method for assessing the adequacy of a wind power generation system taking into account daily and seasonal characteristics is proposed. On the one hand, a 24-hour wind speed joint probability distribution model taking into account daily characteristics is designed for the daily periodic variation characteristics of wind speed. The model first determines the parameters of nine candidate probability distributions based on the maximum likelihood estimation method, and constructs the best fitting hourly wind speed marginal probability distribution model with the help of the AIC index. Then, the Gaussian copula function is selected to construct a 24-hour wind speed joint probability distribution model, and the simulated-Ⅰ sample wind speed taking into account daily characteristics is sampled from it based on the cholesky decomposition method and the standard normal distribution random number.

[0068] On the other hand, in view of the characteristics of seasonal changes in wind speed, a simulation wind speed sample correction model taking into account seasonal characteristics is designed. The correction model introduces a correction equation and a quarterly correction coefficient, constructs a deviation function about the quarterly correction coefficient, and derives the optimal quarterly correction coefficient. The optimal quarterly correction coefficient is substituted into the correction equation to correct the simulation I sample wind speed generated by sampling to achieve the optimal correction, and determine the simulation II optimal corrected wind speed sample showing seasonal characteristics.

[0069] Finally, using the simulated wind speed I and simulated wind speed II, the three indicators of seasonal loss of load expectation (SLOLE), seasonal loss of energy expectation (SLOEE) and seasonal loss of load frequency (SLOLF) are calculated, and a seasonal adequacy assessment procedure for the wind power combined power generation system considering daily and seasonal characteristics is established, so as to discuss in detail the impact of seasonal patterns, the number of wind turbines and the system peak load on the seasonal adequacy of the system.

[0070] A method for assessing the adequacy of a wind power generation system taking into account daily characteristics and seasonal characteristics comprises the following steps:

[0071] Step 1: Collect hourly wind speed data from the wind farm, determine the parameters of multiple candidate probability distributions based on the maximum likelihood estimation method, calculate the AIC index of multiple candidate probability distributions, and select the candidate probability distribution with the smallest AIC index as the optimal fitting hourly wind speed marginal probability distribution model;

[0072] The step 1 comprises the following steps:

[0073] S1.1: Access the open meteorological database and collect the 24-hour wind speed data set of the wind farm.

[0074] S1.2: Using wind speed data, perform maximum likelihood estimation to calculate parameter values ​​of nine candidate probability distributions, including Weibull, Rayleigh, Lognormal, Gamma, Inverse Gaussian, Birnbaum-Saunders, Nakagami, Logistic, and GeneralizedExtremeValue;

[0075] S1.3: Calculate the AIC index of nine candidate probability distributions. Taking Weibull candidate probability distribution and wind speed V1 as an example, the AIC index calculation formula is: S1.4: Select the candidate probability distribution with the smallest AIC index as the best fitting hourly wind speed marginal probability distribution model.

[0076] Step 2: Construct a 24-hour wind speed joint probability distribution model based on the Gaussian copula function;

[0077] In step 2, Gaussian Copula is used because it can describe high-dimensional correlation with a relatively clear expression. The formula of the 24-hour wind speed joint probability distribution model based on the Gaussian copula function is as follows:

[0078]

[0079] In the above function: f(·) is the 24-hour wind speed joint probability density function, v(i,j) represents the wind speed at the jth hour on the i-th day, F j (·) and f j (·) are the cumulative probability density function and probability density function of the j-th hour marginal wind speed determined by the AIC criterion from the 9 sets of candidate parameter probability distributions;

[0080] is the Gaussian copula probability density function, which is expressed as:

[0081]

[0082] In the above formula, the symbol vector x i Expressed as I is the identity matrix, Φ -1 (·) is the inverse function of the standard normal cumulative probability density function. The symbol |·| indicates that the determinant of the matrix is ​​obtained. R is the correlation coefficient matrix of the Gaussian copula function, and it is also the only parameter of the Gaussian copula function to be determined, which can be obtained by Estimated. R-1 , N are x i The transposed matrix of , the inverse matrix of R and the number of days on which wind speed samples were collected.

[0083] Step 4: Introduce the quarterly correction coefficient, correct the simulation-Ⅰ wind speed sample through the correction equation, construct the deviation function of the quarterly correction coefficient, and derive the optimal quarterly correction coefficient The deviations between the seasonal mean and standard deviation of the simulated wind speed samples before correction and the seasonal mean and standard deviation of the measured wind speed samples are minimized to obtain the simulated II wind speed samples.

[0084] In the step 4, the correction equation is:

[0085]

[0086] Where: (s,t) and (s,t) is the seasonal correction coefficient of season s in period t, is the simulated I wind speed before correction, is the corrected simulation II wind speed.

[0087] According to the sample mean and standard deviation formula, the mean and standard deviation of simulated wind speed I and simulated wind speed II satisfy the following two relationships:

[0088]

[0089] Where: and are the mean and standard deviation of the simulated I wind speed in the tth period and sth season before correction;

[0090] and are the mean and standard deviation of the simulated II wind speed in the tth period and sth season, respectively;

[0091] The deviation function of the quarterly correction coefficient is:

[0092]

[0093] μ (s,t) , σ (s,t) are the mean and standard deviation of the actual wind speed in period t and season s respectively

[0094] When the deviation function takes the minimum value, the optimal correction coefficient expression is derived:

[0095]

[0096] The optimal correction is achieved by applying the optimal quarterly correction coefficient to the correction equation. This optimal correction minimizes the deviation between the mean and standard deviation of the corrected simulated II wind speed and the mean and standard deviation of the measured wind speed, thereby obtaining the optimal simulated II wind speed sample with daily and seasonal characteristics.

[0097] Step 5: Using the simulated wind speed samples I and II, establish a seasonal adequacy assessment procedure for the wind power combined power generation system that considers daily and seasonal characteristics, and analyze the impact of seasonal patterns, the number of wind turbines, and system peak load on the seasonal adequacy of the system.

[0098] In the step 5, actual wind speed samples, unit parameters, and system load demand data of the wind power cogeneration system are collected to calculate three seasonal adequacy indexes: seasonal power shortage time expected value SLOLE, seasonal power shortage expected value SLOEE, and seasonal power shortage frequency SLOLF; the steps include:

[0099] S5.1: First, the failure time and repair time of all generator sets are sampled to determine the status of the wind power generation system. The failure time formula is: TTF = -ln(U) / λ; where: U is a uniformly distributed random number in the interval [0,1], and λ is the failure rate of the generating unit; the repair time formula is: TTR = -ln(U) / γ, where: γ is the failure rate of the generating unit.

[0100] S5.2: Calculate the available capacity of conventional units, Where: Y up is the number of legacy units remaining in operation, G y is the rated power of the conventional unit left in operation;

[0101] S5.3: The formula for calculating the output power using the simulated II wind speed is: Where: Z up is the number of wind turbines in operation, is the wind turbine output power at simulated wind speed II in the jth period k season.

[0102] The power output formula of a wind turbine is:

[0103] Where: v ci 、v r and v out are cut-in wind speed, rated wind speed and cut-out wind speed respectively, P r is the rated power of the wind turbine. Coefficients A, B and C are expressed in terms of v ci and v r to be determined

[40] , as shown below:

[0104]

[0105]

[0106]

[0107] S5.4: Using P WF and P G Seasonal system load demand, calculate the LLD for the qth sample season q Loss of load duration, ENS q Low power (energy not supplied), LLO q The number of power outages (loss of load occurrence). Then the seasonal loss of load expectation (SLOLE), seasonal loss of energy expectation (SLOEE) and seasonal loss of load frequency (SLOLF) are obtained;

[0108] The calculation formulas for the three indicators are:

[0109]

[0110]

[0111]

[0112] Where: LLDq is the loss of load duration in the qth sample season, ENSq is the energy not supplied in the qth sample season, LLOq is the loss of load occurrence in the qth sample season. Q is the total number of seasons.

[0113] S5.5: Repeat steps S5.1 to S5.4 above until the coefficient of variation of the SLOEE indicator is less than 0.05 or the sample season Q is greater than 1,000,000.

[0114] S5.6: Output three seasonal adequacy indexes: SLOLE, SLOEE and SLOLF. Based on these three indexes, analyze the impact of seasonal patterns, number of wind turbines and system peak load on the seasonal adequacy of the system.

[0115] Example:

[0116] The proposed model is validated using a 10-year (March 2010-March 2020) sample of actual wind speeds in the Cando area of ​​North Dakota, USA. The seasonal adequacy of the IEEE-RTS wind power generation system is evaluated. The effects of seasonal patterns, the number of wind turbines, and the system peak load on the seasonal adequacy of the system are discussed in detail.

[0117] Figure 1(a) to Figure 1(d) and Figure 2(a) to Figure 2(d) The mean and standard deviation of actual, simulated I and simulated II wind speeds for 10 periods in the Licando area of ​​North Dakota, USA, are shown in spring (Figure 1(a), summer (Figure 1(b), autumn (Figure 1(c)) and winter (Figure 1(d). It can be seen that despite the different periods, the mean and standard deviation of simulated II wind speed are very close to the mean and standard deviation of actual wind speed in each season, indicating that simulated II wind speed presents a similar seasonal pattern to actual wind speed. However, there is a significant gap between the mean and standard deviation of simulated I wind speed and the mean and standard deviation of actual wind speed in each season, indicating that the seasonal pattern of simulated I wind speed is inconsistent with the seasonal pattern of actual wind speed. In most seasons of summer and autumn, the mean of simulated I wind speed is higher than the mean of actual and simulated II wind speeds, indicating that simulated I wind speed presents more abundant wind resources in summer and autumn. On the contrary, in most seasons of spring and winter, the mean of simulated I wind speed is lower than the mean of actual and simulated II wind speeds, indicating that simulated I wind speed presents more insufficient wind resources in spring and winter.

[0118] From the above observations, it can be concluded that the optimal seasonal coefficient appropriately adjusts the simulated I wind speed so that the simulated II wind speed shows a similar seasonal pattern to the actual wind speed. In addition, the simulated II wind speed has a similar 24-hour wind speed mean and standard deviation as the actual wind speed, which indicates that the simulated II wind speed also shows a similar daily variation pattern as the actual wind speed. It can be demonstrated and verified that the proposed two-phase wind speed simulation model can accurately consider daily and seasonal variations. The wind resources of the simulated I wind speed do not match the wind resources of the actual wind speed, which may bring large errors to the seasonal adequacy assessment of the wind power generation system.

[0119] To evaluate the impact of seasonal patterns on system adequacy, a hypothetical wind farm is assumed to be installed in Cando. The wind farm consists of 200 identical wind turbines. The cut-in wind speed, rated wind speed, cut-out wind speed, rated capacity, and hub height of the wind turbines are 3 m / s, 11.5 m / s, 25 m / s, 1.5 MW, and 80 m, respectively. The failure rate and maintenance rate of the wind turbines are 2.4% / 150 hours / year, respectively. The wind farm is assumed to be integrated into the IEEE-RTS power generation system. The system contains 32 conventional units with a total capacity of 3405 MW. The system peak load is 2850 MW. Before the wind farm integration, the SLOLE, SLOEE, and SLOLF of the original IEEE-RTS power generation system are evaluated and given in Table 1. Table 2 lists the SLOLE, SLOEE, and SLOLF of the wind-integrated IEEE-RTS power generation system using Simulation I and Simulation II wind speeds.

[0120] Table 1 Calculation of SLOLE, SLOEE and SLOLF of IEEE-RTS system using actual wind speed

[0121]

[0122] Table 2 SLOLE, SLOEE and SLOLF of IEEE-RTS wind power generation system calculated by using wind speed of simulation I and simulation II

[0123]

[0124] Table 3 SLOLE, SLOEE and SLOLF of IEEE-RTS wind power generation system calculated by using simulation I and simulation II wind speed under different numbers of wind turbines

[0125]

[0126] Comparing Table 1 with Table 2, the following observations are made.

[0127] After the integration of wind farms, the SLOLE, SLOEE and SLOLF of the four seasons are reduced; the SOLE, SLOEE and SLOLF of the summer and autumn of the simulation II wind speed treatment are higher than those of the simulation I wind speed, and the SOLE, SLOEE and SLOLF of the spring and winter of the simulation II wind speed treatment are lower than those of the simulation I wind speed treatment.

[0128] In order to evaluate the impact of the number of wind turbines on the seasonal system adequacy, the number of wind turbines in the wind farm was increased to 300, 400 and 500 respectively. Using the proposed seasonal adequacy evaluation method, the SLOLE, SLOEE and SLOLF of the IEEE-RTS wind power system with different numbers of wind turbines were evaluated. The system peak load was 2850MW, and the simulation I wind speed and simulation II wind speed were used for evaluation. The results are shown in Table 3. It can be seen that with the increase in the number of wind turbines, the SLOLE, SLOEE and SLOLF of the system are further reduced, indicating that the seasonal adequacy of the system will benefit more from the increase in the number of wind turbines. In addition, despite the increasing number of wind turbines, ignoring the seasonal pattern of wind speed will lead to underestimated suitability indicators in summer and autumn, while overestimation of suitability indicators in spring and winter.

[0129] In addition, the mismatch in the seasonal distribution of wind speed leads to an underestimation of the suitability index in summer and autumn, and an overestimation of the suitability index in spring and winter. Figure 1(a) to Figure 1(d) and Figure 2(a) to Figure 2(d) As shown in Figure 2, the mean and standard deviation of the simulated II wind speed are consistent with the mean and standard deviation of the actual wind speed, while the mean and standard deviation of the simulated I wind speed are inconsistent. This results in the seasonal sufficiency index being underestimated or overestimated in some seasons.

Claims

1. A method for assessing the adequacy of a wind power generation system taking into account daily and seasonal characteristics, characterized in that The following steps are involved: Step 1: Collect hourly wind speed data from the wind farm, determine the parameters of multiple candidate probability distributions based on the maximum likelihood estimation method, calculate the AIC index of multiple candidate probability distributions, and select the candidate probability distribution with the smallest AIC index as the optimal fitting hourly wind speed marginal probability distribution model; Step 2: Construct a 24-hour wind speed joint probability distribution model based on the Gaussian copula function; Step 3: Based on the Cholesky decomposition method and standard normal distribution random numbers, sampling is performed from the 24-hour wind speed joint probability distribution to generate simulated I wind speed taking into account daily characteristics; Step 4: Introduce the quarterly correction coefficient, correct the simulation-Ⅰ wind speed sample through the correction equation, construct the deviation function of the quarterly correction coefficient, and derive the optimal quarterly correction coefficient Minimize the deviation between the seasonal mean and standard deviation of the simulated wind speed samples before correction and the seasonal mean and standard deviation of the measured wind speed samples to obtain the simulated II wind speed samples; Step 5: Using the simulated wind speed samples of I and II, establish a seasonal adequacy assessment procedure for the wind power combined power generation system that considers daily and seasonal characteristics, and analyze the impact of seasonal patterns, the number of wind turbines, and the system peak load on the seasonal adequacy of the system; In the step 4, the correction equation is: Where: (s,t) and (s,t) is the seasonal correction coefficient of season s in period t, is the simulated I wind speed before correction, is the corrected simulation II wind speed; According to the sample mean and standard deviation formula, the mean and standard deviation of simulated wind speed I and simulated wind speed II satisfy the following two relationships: Where: and are the mean and standard deviation of the simulated I wind speed in the tth period and sth season before correction; and are the mean and standard deviation of the simulated II wind speed in the tth period and sth season, respectively; The deviation function of the quarterly correction coefficient is: μ (s,t) , σ (s,t) are the mean and standard deviation of the actual wind speed in period t and season s, respectively; When the deviation function takes the minimum value, the optimal correction coefficient expression is derived: By applying the optimal quarterly correction coefficient to the correction equation, the optimal correction is achieved, which minimizes the deviation between the mean and standard deviation of the simulated II wind speed after correction and the mean and standard deviation of the measured wind speed, thereby obtaining the optimal simulated II wind speed sample with daily and seasonal characteristics; The step five comprises the following steps: S5.1: First, the failure time and repair time of all generator sets are sampled to determine the state of the wind power generation system. The failure time formula is: TTF = -ln(U) / λ; where: U is a uniformly distributed random number in the interval [0,1], and λ is the failure rate of the generating unit; the repair time formula is: TTR = -ln(U) / γ, where: γ is the failure rate of the generating unit; S5.2: Calculate the available capacity of conventional units, Where: Y up is the number of legacy units remaining in operation, G y is the rated power of the conventional unit left in operation; S5.3: The formula for calculating the output power using the simulated II wind speed is: Where: Z up is the number of wind turbines in operation, is the wind turbine output power at simulated II wind speed in the jth period and kth season; The power output formula of a wind turbine is: Where: v ci 、v r and v out are cut-in wind speed, rated wind speed and cut-out wind speed respectively, P r is the rated power of the wind turbine; coefficients A, B and C are expressed in terms of v ci and v r To determine, as shown below: S5.4: Using P WF and P G Seasonal system load demand, calculate the LLD for the qth sample season q Power failure duration, ENS q Low battery, LLO q The number of power outages; then, the seasonal power outage time expected value SLOLE, seasonal power shortage expected value SLOEE and seasonal power outage frequency SLOLF are obtained; The calculation formulas for the three indicators are: Where: LLDq is the duration of power shortage in the qth sample season, ENSq is the power shortage in the qth sample season, LLOq is the number of power shortages in the qth sample season; Q is the total number of seasons; S5.5: Repeat steps S5.1 to S5.4 above until the coefficient of variation of the SLOEE index is less than 0.05 or the sample season Q is greater than 1,000,000; S5.6: Output three seasonal adequacy indexes: SLOLE, SLOEE and SLOLF. Based on these three indexes, analyze the impact of seasonal patterns, number of wind turbines and system peak load on the seasonal adequacy of the system.

2. A method for assessing the adequacy of a wind power generation system taking into account daily and seasonal characteristics according to claim 1, characterized in that: The step 1 comprises the following steps: S1.1: Access the open meteorological database and collect the 24-hour wind speed data set of the wind farm. S1.2: Using wind speed data, perform maximum likelihood estimation to calculate parameter values ​​of nine candidate probability distributions, including Weibull, Rayleigh, Lognormal, Gamma, Inverse Gaussian, Birnbaum-Saunders, Nakagami, Logistic, and GeneralizedExtremeValue; S1.3: Calculate the AIC index of nine candidate probability distributions. Taking Weibull candidate probability distribution and wind speed V1 as an example, the AIC index calculation formula is: S1.4: Select the candidate probability distribution with the smallest AIC index as the best fitting hourly wind speed marginal probability distribution model.

3. The method for assessing the adequacy of a wind power generation system taking into account daily and seasonal characteristics according to claim 1, characterized in that: In step 2, the formula of the 24-hour wind speed joint probability distribution model based on the Gaussian copula function is as follows: In the above function: f(·) is the 24-hour wind speed joint probability density function, v(i,j) represents the wind speed at the jth hour on the i-th day, F j (·) and f j (·) are the cumulative probability density function and probability density function of the j-th hour marginal wind speed determined by the AIC criterion from the 9 sets of candidate parameter probability distributions; is the Gaussian copula probability density function, which is expressed as: In the above formula, the symbol vector x i Expressed as I is the identity matrix, Φ -1 (·) is the inverse function of the standard normal cumulative probability density function. The symbol |·| indicates that the determinant of the matrix is ​​obtained. R is the correlation coefficient matrix of the Gaussian copula function, and it is also the only parameter of the Gaussian copula function to be determined, which can be obtained by It is estimated that R -1 , N are x i The transposed matrix of , the inverse matrix of R and the number of days on which wind speed samples were collected.

Images