Wind power plant frequency modulation capability prediction method and system considering wind speed uncertainty

Abstract

The invention discloses a wind power plant frequency modulation capability prediction method considering wind speed uncertainty. The method comprises the steps that a dimension raising correlation model among the wind speed of a wind power plant, the rotating speed of a draught fan and a wind power plant droop coefficient is built based on the Koopman theory; acquiring historical operation data information of the target wind power plant; according to the obtained data information, a Koopman linear matrix of the target wind power plant is obtained through calculation based on the constructed model; calculating a probability density expression of the maximum droop coefficient of the target wind power plant based on the relationship between the system frequency of the wind power plant and the droop coefficient of the wind power plant; and calculating to obtain frequency modulation capability prediction data of the target wind power plant so as to complete prediction of the frequency modulation capability of the target wind power plant considering the wind speed uncertainty. The invention also discloses a system for realizing the wind power plant frequency modulation capability prediction method considering the wind speed uncertainty. The wind power plant frequency modulation capability prediction considering the wind speed uncertainty can be realized, the reliability is higher, the accuracy is better, and the efficiency is higher.

Description

Technical Field

[0001] This invention belongs to the field of electrical automation, and specifically relates to a method and system for predicting the frequency regulation capability of wind farms considering the uncertainty of wind speed. Background Technology

[0002] With economic and technological development and the improvement of people's living standards, electricity has become an indispensable secondary energy source in people's production and daily life, bringing endless convenience. Therefore, ensuring a stable and reliable supply of electricity has become one of the most important tasks of the power system.

[0003] Currently, environmental problems are becoming increasingly serious, leading to a growing number of wind power systems being integrated into the power grid. However, the fluctuations and randomness in wind power output pose a significant threat to the safe and stable operation of the power system. Therefore, to reduce the impact of high-penetration wind power systems on grid frequency security, it is necessary for wind power systems to participate in power system frequency regulation.

[0004] When wind farms participate in system frequency regulation, they can simulate the droop characteristics of traditional synchronous turbines through power electronic adjustment of the wind turbines. By changing their active power output according to the system frequency deviation, they can improve the frequency stability of the power system. However, the exchange of kinetic energy between the wind turbine blades and the system's electromagnetic power causes changes in the turbine blade speed. Excessively low or high turbine blade speeds can lead to grid disconnection or even damage to the turbine, resulting in a secondary drop in system frequency. Therefore, it is crucial to ensure that the turbine speed remains within a reasonable range during frequency regulation control. Thus, predicting the frequency regulation capability of wind farms is particularly important.

[0005] Currently, some researchers use electromechanical transient models of wind farms to perform analytical calculations of their frequency regulation capabilities. However, these methods are based on traditional large-scale time-domain models, which involve large computational loads and long calculation times, and cannot track rapid changes in wind speed. Meanwhile, other researchers use currently popular neural network models to predict the frequency regulation capabilities of wind farms; however, these methods calculate based on real-time measured wind speeds and cannot predict the ultimate frequency regulation capabilities of wind farms. Summary of the Invention

[0006] One of the objectives of this invention is to provide a highly reliable, accurate, and efficient method for predicting the frequency regulation capability of wind farms while taking into account wind speed uncertainties.

[0007] The second objective of this invention is to provide a system for implementing the wind farm frequency regulation capability prediction method that takes into account wind speed uncertainty.

[0008] The wind farm frequency regulation capability prediction method considering wind speed uncertainty provided by this invention includes the following steps:

[0009] S1. Based on Koopman theory, construct an upgraded correlation model between wind speed, wind turbine speed and wind farm droop coefficient;

[0010] S2. Obtain historical operational data information of the target wind farm;

[0011] S3. Based on the data obtained in step S2 and the model constructed in step S1, calculate the Koopman linear matrix of the target wind farm.

[0012] S4. Based on the data obtained in step S3, and the relationship between the system frequency of the wind farm and the droop coefficient of the wind farm, calculate the probability density expression of the maximum droop coefficient of the target wind farm.

[0013] S5. Based on the data obtained in step S4, calculate the predicted frequency regulation capability of the target wind farm to complete the prediction of the frequency regulation capability of the target wind farm considering the uncertainty of wind speed.

[0014] Step S1, which involves constructing a higher-dimensional correlation model based on Koopman theory to link wind speed, turbine speed, and droop coefficient in a wind farm, specifically includes the following steps:

[0015] With the wind speed of the wind farm As output, the final speed of the fan and wind farm droop coefficient As input to build the model, it is represented as:

[0016] In the formula The model to be constructed;

[0017] According to Koopman's theory, by increasing the dimensionality of the input variables, the nonlinear relationship between the input and output can be transformed into a higher-dimensional linear relationship, expressed as:

[0018] In the formula It is a Koopman linear matrix; The input variables after dimensionality increase; The corresponding Koopman linear matrix The submatrix; The corresponding Koopman linear matrix The submatrix; For input variables, include and ; For input variables The dimension-upgrading function;

[0019] If the number of input variables is set to n, then the j-th dimension-upgrading function is expressed as:

[0020] In the formula Let j be the dimension-upgrading function; Let j be the basis vectors of the up-dimensional function. The value is determined by the input variable. Randomly select a value from the range of possible values; The dimension-upgrading function used is, and , It is an intermediate dimension-upgrading function and m is the input variable Dimensions Input variables The i-th variable, basis vectors The i-th element in.

[0021] Step S2, which involves obtaining historical operational data of the target wind farm, specifically includes the following steps:

[0022] Obtain historical operational data of the target wind farm;

[0023] The operational data information includes the wind speed of the wind farm. Final fan speed and wind farm droop coefficient ;

[0024] The historical final rotational speed of the wind turbine will be obtained and historical wind farm droop coefficient Construct a training set for the input variables. , represented as ,in Let S be the input variable formed by the i-th historical data, and S be the number of data points.

[0025] Historical wind speeds Construct the training set for the output variables , represented as ,in The output variable is formed by the i-th historical data.

[0026] Step S3, which involves calculating the Koopman linear matrix of the target wind farm based on the data obtained in step S2 and the model constructed in step S1, specifically includes the following steps:

[0027] The training set of input variables obtained in step S2 Based on the model constructed in step S1, the corresponding upgraded variable dataset is calculated. , represented as ;in, for The corresponding upgraded data;

[0028] The Koopman linear matrix of the target wind farm was calculated using the least squares method. :

[0029] In the formula is the Moore-Penrose inverse of the matrix.

[0030] Step S4, based on the data obtained in step S3, calculates the probability density expression for the maximum droop coefficient of the target wind farm based on the relationship between the system frequency and the droop coefficient of the wind farm. This specifically includes the following steps:

[0031] To calculate the maximum droop coefficient of the target wind farm The probability density expression must be calculated. The cumulative probability density function (CDF) is expressed as:

[0032] In the formula It is the cumulative probability density function; The probability that the condition in parentheses is satisfied; For the cumulative probability density function CDF;

[0033] At the final speed of the wind turbine Under certain conditions: when the system frequency is higher than the rated frequency, the wind farm droop coefficient is negatively correlated with wind speed; when the system frequency is lower than the rated frequency, the wind farm droop coefficient is positively correlated with wind speed. Therefore, the relationship between these conditions needs to be considered. The cumulative probability density function (CDF) is used for calculation:

[0034] When the system frequency is higher than the rated frequency:

[0035] Will The cumulative probability density function CDF is expressed as Thus, the solution The cumulative probability density function (CDF) is transformed into solving for the probability density function CDF. The probability value; where, Indicates when Values At that time, the corresponding ;

[0036] because The distribution of follows a Weibull distribution and its probability density is known; therefore, we obtain:

[0037] In the formula For wind speed less than or equal to The cumulative probability density function;

[0038] Due to the function for use and The function expressed, therefore Convert to The expression is represented as

[0039] In the formula The wind speed is a function of the final rotational speed and the frequency modulation coefficient when the frequency is higher than the rated frequency;

[0040] Combined expression and ,get The cumulative probability density function CDF is expressed as: ;

[0041] right Differentiating the cumulative probability density function CDF, we obtain the probability density expression PDF as follows: ;

[0042] When the system frequency is lower than the rated frequency:

[0043] Will The cumulative probability density function (CDF) is expressed as:

[0044]

[0045] In the formula The wind speed is a function of the final rotational speed and the frequency modulation coefficient when the frequency is lower than the rated frequency;

[0046] The final probability density expression PDF is as follows: ;

[0047] Considering both cases where the system frequency is higher than the rated frequency and cases where the system frequency is lower than the rated frequency, the probability density function expression for the droop coefficient of the target wind farm is obtained:

[0048] In the formula This is the actual frequency of the system; The system's rated frequency;

[0049] set up The maximum droop coefficient of the target wind farm is obtained. probability density expression :

[0050] In the formula This refers to the boundary value of the fan speed, and when the system frequency is higher than the rated frequency. When the system frequency is lower than the rated frequency ; This is the maximum speed of the fan; This is the minimum speed of the fan.

[0051] Step S5 involves calculating the predicted frequency regulation capability of the target wind farm based on the data obtained in step S4, in order to complete the prediction of the frequency regulation capability of the target wind farm considering wind speed uncertainty. Specifically, this includes the following steps:

[0052] Based on the maximum droop coefficient of the target wind farm obtained in step S4 probability density expression The calculation yields:

[0053] The physical meaning is: the maximum droop coefficient provided by the target wind farm under the corresponding wind speed probability distribution. quantiles and the maximum droop coefficient provided are The probability is no less than ;

[0054] When predicting the frequency regulation capability of a target wind farm, the Weibull probability density distribution parameters of wind speed for the target wind farm over a future set time period are obtained, based on the final turbine rotational speed. Based on the given settings, the probability density expression for different wind farm droop coefficients is calculated, and according to... The maximum droop coefficient of the corresponding wind farm is calculated under the following conditions. probability density expression Then calculate to get confidence level This is to complete the prediction of the frequency regulation capability of the target wind farm considering the uncertainty of wind speed.

[0055] This invention also provides a system for implementing the wind farm frequency regulation capability prediction method considering wind speed uncertainty, comprising a model building module, a data acquisition module, a matrix calculation module, a density calculation module, and a capability prediction module; the model building module, data acquisition module, matrix calculation module, density calculation module, and capability prediction module are connected in series; the model building module is used to construct a higher-dimensional correlation model between wind speed, wind turbine speed, and wind farm droop coefficient based on Koopman theory, and upload the data information to the data acquisition module; the data acquisition module is used to acquire historical operating data information of the target wind farm based on the received data information, and upload the data information to the matrix calculation module; the matrix calculation module... The first module calculates the Koopman linear matrix of the target wind farm based on the received and acquired data and the constructed model, and uploads the data to the density calculation module. The second module calculates the probability density expression of the maximum droop coefficient of the target wind farm based on the received and acquired data and the relationship between the system frequency and the droop coefficient of the wind farm, and uploads the data to the capacity prediction module. The third module calculates the frequency regulation capacity prediction data of the target wind farm based on the received and acquired data, thus completing the prediction of the frequency regulation capacity of the target wind farm considering wind speed uncertainty.

[0056] The wind farm frequency regulation capability prediction method and system considering wind speed uncertainty provided by this invention constructs an upgraded correlation model based on Koopman theory and calculates the probability density expression of the maximum droop coefficient of the target wind farm based on a data-driven scheme. Therefore, this invention can not only predict the wind farm frequency regulation capability considering wind speed uncertainty, but also has higher reliability, better accuracy and higher efficiency. Attached Figure Description

[0057] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0058] Figure 2 This is a schematic diagram illustrating the accuracy analysis of a data-driven model for scenarios where the frequency is higher than the rated frequency, as described in an embodiment of the method of the present invention.

[0059] Figure 3 This is a schematic diagram illustrating the accuracy analysis of a data-driven model for scenarios where the frequency is below the rated frequency, as described in an embodiment of the method of the present invention.

[0060] Figure 4 This is a schematic diagram of a primary frequency modulation simulation process in a scenario where the frequency is lower than the rated frequency, as described in an embodiment of the method of the present invention.

[0061] Figure 5 This is a schematic diagram of the wind speed distribution prediction curve in an embodiment of the method of the present invention.

[0062] Figure 6 This is a schematic diagram of the probability density distribution curves of the droop coefficient of a wind farm under different expected final rotational speeds in scenarios where the frequency is higher than the rated frequency, as an embodiment of the method of the present invention.

[0063] Figure 7 This is a schematic diagram of the probability density distribution curves of the droop coefficient of a wind farm under different expected final rotational speeds in a scenario where the frequency is lower than the rated frequency, as an embodiment of the method of the present invention.

[0064] Figure 8 This is a schematic diagram of the probability density distribution of the maximum droop coefficient in a scenario where the frequency is higher than the rated frequency, as shown in the embodiment of the method of the present invention.

[0065] Figure 9 This is a schematic diagram of the probability density distribution of the maximum droop coefficient in a scenario where the frequency is lower than the rated frequency, as shown in the embodiment of the method of the present invention.

[0066] Figure 10 This is a schematic diagram of extreme wind speed distribution used for comparing point prediction and probability prediction in an embodiment of the method of the present invention.

[0067] Figure 11 This is a schematic diagram of the droop coefficient distribution used for comparing point prediction and probability prediction in an embodiment of the method of the present invention.

[0068] Figure 12 This is a schematic diagram of the functional modules of the system of the present invention. Detailed Implementation

[0069] like Figure 1 The diagram shown is a flowchart of the method of the present invention: The wind farm frequency regulation capability prediction method considering wind speed uncertainty disclosed in this invention includes the following steps:

[0070] S1. Based on Koopman theory, construct a higher-dimensional correlation model between wind speed, wind turbine speed, and wind farm droop coefficient; specifically including the following steps:

[0071] With the wind speed of the wind farm As output, the final speed of the fan and wind farm droop coefficient As input to build the model, it is represented as:

[0072] In the formula The model is constructed to represent the complex nonlinear relationship between the output and the input;

[0073] According to Koopman's theory, by increasing the dimensionality of the input variables, the nonlinear relationship between the input and output can be transformed into a higher-dimensional linear relationship, expressed as:

[0074] In the formula It is a Koopman linear matrix; The input variables after dimensionality increase; The corresponding Koopman linear matrix The submatrix; The corresponding Koopman linear matrix The submatrix; For input variables, include and ; For input variables The dimension-upgrading function;

[0075] In theory, the above equation holds strictly only when the dimension is increased to infinite dimensions. However, in practical applications, considering the computational time cost, it is sufficient to increase it to a certain number of finite dimensions to obtain adequate accuracy.

[0076] If the number of input variables is set to n, then the j-th dimension-upgrading function is expressed as:

[0077] In the formula Let j be the dimension-upgrading function; Let j be the basis vectors of the up-dimensional function. The value is determined by the input variable. Randomly select a value from the range of possible values; The dimension-upgrading function used is polyharmonic, which can take various forms. Considering the dynamic complexity of the wind farm frequency regulation capability assessment in this invention, a polyharmonic dimension-upgrading function is adopted, expressed as follows: , It is an intermediate dimension-upgrading function and m is the input variable Dimensions Input variables The i-th variable, basis vectors The i-th element in.

[0078] S2. Obtain historical operational data of the target wind farm; specifically including the following steps:

[0079] Obtain historical operational data of the target wind farm;

[0080] The operational data information includes the wind speed of the wind farm. Final fan speed and wind farm droop coefficient ;

[0081] The historical final rotational speed of the wind turbine will be obtained and historical wind farm droop coefficient Construct a training set for the input variables. , represented as ,in Let S be the input variable formed by the i-th historical data, and S be the number of data points.

[0082] Historical wind speeds Construct the training set for the output variables , represented as ,in The output variable is composed of the i-th historical data.

[0083] S3. Based on the data obtained in step S2 and the model constructed in step S1, calculate the Koopman linear matrix of the target wind farm; specifically, this includes the following steps:

[0084] The training set of input variables obtained in step S2 Based on the model constructed in step S1, the corresponding upgraded variable dataset is calculated. , represented as ;in, for The corresponding upgraded data;

[0085] The Koopman linear matrix of the target wind farm was calculated using the least squares method. :

[0086] In the formula The Moore-Penrose inverse of the matrix;

[0087] S4. Based on the data obtained in step S3, and considering the relationship between the system frequency and the droop coefficient of the wind farm, calculate the probability density expression for the maximum droop coefficient of the target wind farm; specifically including the following steps:

[0088] To calculate the maximum droop coefficient of the target wind farm The probability density expression must be calculated. The cumulative probability density function (CDF) is expressed as:

[0089] In the formula It is the cumulative probability density function; The probability that the condition within the curly braces is satisfied; For the cumulative probability density function CDF;

[0090] At the final speed of the wind turbine Under certain conditions: when the system frequency is higher than the rated frequency, the wind farm droop coefficient is negatively correlated with wind speed; when the system frequency is lower than the rated frequency, the wind farm droop coefficient is positively correlated with wind speed, meaning the wind farm droop coefficient and wind speed exhibit a monotonic relationship. Therefore, the relationship should be considered on a case-by-case basis. The cumulative probability density function (CDF) is used for calculation:

[0091] When the system frequency is higher than the rated frequency:

[0092] Will The cumulative probability density function CDF is expressed as Thus, the solution The cumulative probability density function (CDF) is transformed into solving for the probability density function CDF. The probability value; where, Indicates when Values At that time, the corresponding ;

[0093] because The distribution of follows a Weibull distribution and its probability density is known, so we only need to find the probability density of each . Corresponding That's it; therefore, we get:

[0094] In the formula For wind speed less than or equal to The cumulative probability density function;

[0095] Due to the function for use and The function expressed, therefore Convert to The expression is represented as

[0096] In the formula The wind speed is a function of the final rotational speed and the frequency modulation coefficient when the frequency is higher than the rated frequency;

[0097] Combined expression and ,get The cumulative probability density function CDF is expressed as: ;

[0098] right Differentiating the cumulative probability density function CDF, we obtain the probability density expression PDF as follows: ;

[0099] When the system frequency is lower than the rated frequency:

[0100] Referring to the calculation scheme when the system frequency is higher than the rated frequency, The cumulative probability density function (CDF) is expressed as:

[0101]

[0102] In the formula The wind speed is a function of the final rotational speed and the frequency modulation coefficient when the frequency is lower than the rated frequency;

[0103] The final probability density expression PDF is as follows: ;

[0104] Considering both cases where the system frequency is higher than the rated frequency and cases where the system frequency is lower than the rated frequency, the probability density function expression for the droop coefficient of the target wind farm is obtained:

[0105] In the formula This is the actual frequency of the system; The system's rated frequency;

[0106] set up The maximum droop coefficient of the target wind farm is obtained. probability density expression :

[0107] In the formula This refers to the boundary value of the fan speed, and when the system frequency is higher than the rated frequency. When the system frequency is lower than the rated frequency ; This is the maximum speed of the fan; This is the minimum speed of the fan;

[0108] S5. Based on the data obtained in step S4, calculate the predicted frequency regulation capability of the target wind farm to complete the prediction of the frequency regulation capability of the target wind farm considering wind speed uncertainty; specifically including the following steps:

[0109] Based on the maximum droop coefficient of the target wind farm obtained in step S4 probability density expression The calculation yields:

[0110] The physical meaning is: the maximum droop coefficient provided by the target wind farm under the corresponding wind speed probability distribution. quantiles and the maximum droop coefficient provided are The probability is no less than ;

[0111] When predicting the frequency regulation capability of a target wind farm, the Weibull probability density distribution parameters of wind speed for the target wind farm over a future set time period are obtained, based on the final turbine rotational speed. Based on the given settings, the probability density expression for different wind farm droop coefficients is calculated, and according to... The maximum droop coefficient of the corresponding wind farm is calculated under the following conditions. probability density expression Then calculate to get confidence level This is to complete the prediction of the frequency regulation capability of the target wind farm considering the uncertainty of wind speed.

[0112] In practice, steps S1 to S4 can be completed offline; step S5 is completed online.

[0113] The method of this invention mainly adopts a data-driven approach rather than traditional time-domain simulation, thus having the advantage of fast calculation speed. It can quickly obtain the probability distribution of wind farm frequency regulation capability based on the distribution of predicted wind speed. The method of this invention only needs to perform offline training once in advance, and can then update the distribution of wind farm frequency regulation capability curve in real time according to the changes in the predicted wind speed probability density curve, which is highly efficient and timely. Moreover, unlike Monte Carlo simulation, the method of this invention does not require a large amount of random sampling, so it is not affected by random error of samples, and can obtain a unified form of the expression for wind farm frequency regulation capability.

[0114] The effects of the method of the present invention will be further illustrated below with reference to an embodiment:

[0115] By selecting appropriate and realistic wind farm physical model parameters, numerical examples are analyzed in scenarios with frequencies higher than and lower than the rated frequency, and compared with time-domain simulation combined with Monte Carlo simulation method, thereby highlighting the advantages of the method of this invention.

[0116] The wind farm has 16 wind turbines, each with a rated power of 4MW. The turbine blade speed range is... The training set is selected in scenarios with frequencies higher than the rated frequency. MW / Hz, selection for scenarios with frequencies below the rated frequency MW / Hz; Training wind speed m / s; Training set rotational speed range Since the Koopman linearization method is a global linearization method, the training set does not need to cover the extreme frequency modulation scenario. Therefore, the test set can exceed the coverage of the training set. This invention will conduct testing and comparative analysis in multiple test scenarios. The Koopman linear model selects an increased dimension of 1500, which can greatly reduce the computation time cost while ensuring accuracy.

[0117] Verifying the accuracy of the Koopman data-driven augmentation model: As mentioned earlier, the trained linear relationship is strictly equivalent to the original nonlinear relationship only when the augmented dimension is infinite. In practice, it is necessary to appropriately select the number of augmented dimensions to ensure performance. To obtain historical primary frequency regulation data, the primary frequency regulation process is activated by simulating load fluctuations. The cases of frequencies above and below the rated frequency are discussed separately. In the test scenarios, the expected final rotor speeds for scenarios with frequencies above the rated frequency are 1.38, 1.40, 1.42, and 1.44 pu, respectively, and the expected final rotor speeds for scenarios with frequencies below the rated frequency are 0.76, 0.74, 0.72, and 0.7 pu, respectively. Given different wind speeds, the droop coefficient can be calculated by the data-driven augmentation model or by time-domain simulation. The results of the proposed method are compared with the time-domain simulation results to verify its accuracy. The relative error is as follows: Figure 2 and Figure 3 As shown, the accuracy of the method of the present invention is sufficient to meet the needs of the application.

[0118] To verify that the method of this invention is unaffected by the accuracy of wind farm parameters, random errors, such as radius deviation of -5% and moment of inertia deviation of -10%, are introduced into the relevant parameters. The dynamic process without parameter errors under the time-domain method is used as the comparison basis. Assuming the frequency is lower than the rated frequency, the proposed augmented-dimensional model is used to calculate the wind farm droop coefficient, and the calculated value is substituted into the frequency regulation model. The dynamic process of primary frequency regulation is as follows: Figure 4 As shown, the dynamic curves of rotor speed, system frequency, and wind farm power under the method of this invention are almost completely consistent with the time-domain simulation. Conversely, in the time-domain simulation, inaccurate model parameters can lead to exceeding speed limits or conservative results, demonstrating the advantage of the method of this invention in resisting parameter errors.

[0119] Based on the desired final rotor speed, it can be determined according to... Figure 5 The wind speed prediction distribution shown yields the probability distribution of the wind farm droop coefficient under different final rotational speeds. Based on... Figure 5 of The probability distribution of wind speed for different time periods, and the probability distribution of wind farm droop coefficient under scenarios where the frequency is higher or lower than the rated frequency, are as follows: Figure 6 and Figure 7As shown, it can be seen that the closer the final rotational speed is to the limiting rotational speed, the more the peak probability density of the droop coefficient shifts to the right, indicating a larger droop coefficient.

[0120] according to Figure 5 The uncertain wind speed prediction shown can also be obtained , , The probability distribution corresponding to the time period, such as Figure 8 and Figure 9 As shown in the figure, the results indicate that when the wind speed probability distribution shifts to the right, the primary frequency regulation capability tends to shift in the opposite direction when the frequency is above the rated frequency. In contrast, when the frequency is below the rated frequency, the primary frequency regulation capability shifts to the right. To verify the accuracy of the probability prediction, the figure also shows... Within a given time period, the statistical histogram of the maximum droop coefficient obtained through time-domain simulation combined with the Monte Carlo method was used, and the corresponding probability density expression was derived based on kernel density estimation. It can be seen that the kernel density estimation results are highly consistent with the method of this invention, proving the accuracy of the physical information scheme adopted in this invention. Conversely, due to errors in the parameters of the wind farm dynamic model, time-domain simulation inevitably leads to inaccurate prediction results. Adding a -10% deviation to the rotor inertia in the time-domain simulation significantly deviates the original curve. Furthermore, when time-domain simulation is combined with the Monte Carlo method, the probability density value of the primary frequency regulation capability cannot be directly obtained analytically. Instead, it is obtained using statistical methods, which is inevitably affected by random sampling errors. Theoretically, the more sampling points, the more accurate the results obtained from the Monte Carlo simulation. However, the time cost of the calculation process also increases significantly.

[0121] This paper discusses point prediction and probabilistic prediction under extreme wind speed distributions: probabilistic prediction contains more comprehensive information compared to simple point prediction. To illustrate the importance of probabilistic prediction, an example of an increasing frequency scenario is used. Figure 10 and 11 As shown, if point prediction is used, the wind speed in scenario 1 is 9.2409 m / s, and the wind speed in scenario 2 is 8.3652 m / s. Based on point prediction, the maximum droop coefficient can be calculated, with a result of 21.2918 MW / Hz for scenario 1 and 29.0867 MW / Hz for scenario 2. Clearly, the point prediction of the next frequency regulation capability in scenario 1 is smaller than that in scenario 2, which is consistent with current intuitive understanding. However, in scenario 1, the probabilistic prediction result of the maximum droop coefficient is 17.0366 MW / Hz, and in scenario 2 it is 15.2348 MW / Hz. It can be concluded that the maximum droop coefficient in scenario 1 is greater than that in scenario 2, which is opposite to the trend of point prediction. This shows that, after considering the probability distribution, the probabilistic prediction result of the maximum droop coefficient satisfying a certain probability value may not be consistent with the point prediction result, proving the necessity of probabilistic prediction.

[0122] like Figure 12 The diagram shows the functional modules of the system of this invention: The system disclosed in this invention for predicting the frequency regulation capability of a wind farm considering wind speed uncertainty includes a model building module, a data acquisition module, a matrix calculation module, a density calculation module, and a capability prediction module; the model building module, data acquisition module, matrix calculation module, density calculation module, and capability prediction module are connected in series; the model building module is used to construct a higher-dimensional correlation model between wind speed, wind turbine speed, and wind farm droop coefficient based on Koopman theory, and upload the data information to the data acquisition module; the data acquisition module is used to obtain historical operating data information of the target wind farm based on the received data information, and upload the data information to the matrix calculation module. The module comprises several components: a matrix calculation module, a density calculation module, and a capacity prediction module. The matrix calculation module calculates the Koopman linear matrix of the target wind farm based on the received and acquired data and the constructed model, and uploads this data to the density calculation module. The density calculation module calculates the probability density expression of the maximum droop coefficient of the target wind farm based on the relationship between the system frequency and the droop coefficient, and uploads this data to the capacity prediction module. The capacity prediction module calculates the predicted frequency regulation capacity of the target wind farm based on the received and acquired data, thus completing the prediction of the frequency regulation capacity of the target wind farm considering wind speed uncertainties.

Claims

1. A method for predicting the frequency regulation capability of a wind farm considering wind speed uncertainty, comprising the following steps: S1. Based on Koopman theory, construct an upgraded correlation model between wind speed, wind turbine speed and wind farm droop coefficient; S2. Obtain historical operational data information of the target wind farm; S3. Based on the data obtained in step S2 and the model constructed in step S1, calculate the Koopman linear matrix of the target wind farm. S4. Based on the data obtained in step S3, and the relationship between the system frequency of the wind farm and the droop coefficient of the wind farm, calculate the probability density expression of the maximum droop coefficient of the target wind farm. S5. Based on the data obtained in step S4, calculate the predicted frequency regulation capability of the target wind farm to complete the prediction of the frequency regulation capability of the target wind farm considering the uncertainty of wind speed.

2. The method for predicting the frequency regulation capability of a wind farm considering wind speed uncertainty according to claim 1, characterized in that... Step S1, which involves constructing a higher-dimensional correlation model based on Koopman theory to link wind speed, turbine speed, and droop coefficient in a wind farm, specifically includes the following steps: With the wind speed of the wind farm As output, the final speed of the fan and wind farm droop coefficient As input to build the model, it is represented as: In the formula The model to be constructed; According to Koopman's theory, by increasing the dimensionality of the input variables, the nonlinear relationship between the input and output can be transformed into a higher-dimensional linear relationship, expressed as: In the formula It is a Koopman linear matrix; The input variables after dimensionality increase; The corresponding Koopman linear matrix The submatrix; The corresponding Koopman linear matrix The submatrix; For input variables, include and ; For input variables The dimension-upgrading function; If the number of input variables is set to n, then the j-th dimension-upgrading function is expressed as: In the formula Let j be the dimension-upgrading function; Let j be the basis vectors of the up-dimensional function. The value is determined by the input variable. Randomly select a value from the range of possible values; The dimension-upgrading function used is, and , It is an intermediate dimension-upgrading function and m is the input variable Dimensions Input variables The i-th variable, basis vectors The i-th element in.

3. The method for predicting the frequency regulation capability of a wind farm considering wind speed uncertainty according to claim 2, characterized in that... Step S2, which involves obtaining historical operational data of the target wind farm, specifically includes the following steps: Obtain historical operational data of the target wind farm; The operational data information includes the wind speed of the wind farm. Final fan speed and wind farm droop coefficient ; The historical final rotational speed of the wind turbine will be obtained and historical wind farm droop coefficient Construct a training set for the input variables. , represented as ,in Let S be the input variable formed by the i-th historical data, and S be the number of data points. Historical wind speeds Construct the training set for the output variables , represented as ,in The output variable is formed by the i-th historical data.

4. The method for predicting the frequency regulation capability of a wind farm considering wind speed uncertainty according to claim 3, characterized in that... Step S3, which involves calculating the Koopman linear matrix of the target wind farm based on the data obtained in step S2 and the model constructed in step S1, specifically includes the following steps: The training set of input variables obtained in step S2 Based on the model constructed in step S1, the corresponding upgraded variable dataset is calculated. , represented as ;in, for The corresponding upgraded data; The Koopman linear matrix of the target wind farm was calculated using the least squares method. : In the formula is the Moore-Penrose inverse of the matrix.

5. The method for predicting the frequency regulation capability of a wind farm considering wind speed uncertainty according to claim 4, characterized in that... Step S4, based on the data obtained in step S3, calculates the probability density expression for the maximum droop coefficient of the target wind farm based on the relationship between the system frequency and the droop coefficient of the wind farm. This specifically includes the following steps: To calculate the maximum droop coefficient of the target wind farm The probability density expression must be calculated. The cumulative probability density function (CDF) is expressed as: In the formula It is the cumulative probability density function; The probability that the condition within the curly braces is satisfied; For the cumulative probability density function CDF; At the final speed of the wind turbine Under certain conditions: when the system frequency is higher than the rated frequency, the wind farm droop coefficient is negatively correlated with wind speed; when the system frequency is lower than the rated frequency, the wind farm droop coefficient is positively correlated with wind speed. Therefore, the relationship between these conditions needs to be considered. The cumulative probability density function (CDF) is used for calculation: When the system frequency is higher than the rated frequency: Will The cumulative probability density function CDF is expressed as Thus, the solution The cumulative probability density function (CDF) is transformed into solving for the probability density function CDF. The probability value; where, Indicates when Values At that time, the corresponding ; because The distribution of follows a Weibull distribution and its probability density is known; therefore, we obtain: In the formula For wind speed less than or equal to The cumulative probability density function; Due to the function for use and The function expressed, therefore Convert to The expression is represented as In the formula The wind speed is a function of the final rotational speed and the frequency modulation coefficient when the frequency is higher than the rated frequency; Combined expression and ,get The cumulative probability density function CDF is expressed as: ; right Differentiating the cumulative probability density function CDF, we obtain the probability density expression PDF as follows: ; When the system frequency is lower than the rated frequency: Will The cumulative probability density function (CDF) is expressed as: In the formula The wind speed is a function of the final rotational speed and the frequency modulation coefficient when the frequency is lower than the rated frequency; The final probability density expression PDF is as follows: ; Considering both cases where the system frequency is higher than the rated frequency and cases where the system frequency is lower than the rated frequency, the probability density function expression for the droop coefficient of the target wind farm is obtained: In the formula This is the actual frequency of the system; The system's rated frequency; set up The maximum droop coefficient of the target wind farm is obtained. probability density expression : In the formula This refers to the boundary value of the fan speed, and when the system frequency is higher than the rated frequency. When the system frequency is lower than the rated frequency ; This is the maximum speed of the fan; This is the minimum speed of the fan.

6. The method for predicting the frequency regulation capability of a wind farm considering wind speed uncertainty according to claim 5, characterized in that... Step S5 involves calculating the predicted frequency regulation capability of the target wind farm based on the data obtained in step S4, in order to complete the prediction of the frequency regulation capability of the target wind farm considering wind speed uncertainty. Specifically, this includes the following steps: Based on the maximum droop coefficient of the target wind farm obtained in step S4 probability density expression The calculation yields: The physical meaning is: the maximum droop coefficient provided by the target wind farm under the corresponding wind speed probability distribution. quantiles and the maximum droop coefficient provided are The probability is no less than ; When predicting the frequency regulation capability of a target wind farm, the Weibull probability density distribution parameters of wind speed for the target wind farm over a future set time period are obtained, based on the final turbine rotational speed. Based on the given settings, the probability density expression for different wind farm droop coefficients is calculated, and according to... The maximum droop coefficient of the corresponding wind farm is calculated under the following conditions. probability density expression Then calculate to get confidence level This is to complete the prediction of the frequency regulation capability of the target wind farm considering the uncertainty of wind speed.

7. A system for implementing the wind farm frequency regulation capability prediction method considering wind speed uncertainty as described in any one of claims 1 to 6, characterized in that... It includes a model building module, a data acquisition module, a matrix calculation module, a density calculation module, and a capacity prediction module; these modules are connected in series. The model building module is used to construct a higher-dimensional correlation model between wind speed, turbine speed, and wind farm droop coefficient based on Koopman theory, and uploads the data to the data acquisition module. The data acquisition module is used to obtain historical operating data of the target wind farm based on the received data and uploads the data to the matrix calculation module. The matrix calculation module calculates the Koopman linear matrix of the target wind farm based on the received and acquired data and the constructed model, and uploads the data to the density calculation module. The density calculation module calculates the probability density expression of the maximum droop coefficient of the target wind farm based on the received and acquired data and the relationship between the system frequency and the droop coefficient of the wind farm, and uploads the data to the capacity prediction module. The capacity prediction module calculates the frequency regulation capacity prediction data of the target wind farm based on the received and acquired data, thus completing the prediction of the frequency regulation capacity of the target wind farm considering wind speed uncertainty.

Images