A direct-drive wind farm equivalent method and system for power system stability analysis

Abstract

The present invention discloses a direct-drive wind farm equivalent method and system for power system stability analysis, which relates to the technical field of power system stability analysis. The wind speed of each direct-drive wind turbine in a large-scale direct-drive wind farm at the expected moment is predicted by adopting the xLSTM and FLORIS wake model for joint solution; the typical values ​​of the equivalent parameters of the three-machine equivalent model of the direct-drive wind farm are solved by the capacity weighting method; the equivalent parameters of the three-machine equivalent model of the direct-drive wind farm are corrected by adopting the parameter identification method based on the Stuart t hybrid model-grey wolf optimization with the goal of keeping the power response characteristics of the equivalent model consistent with the detailed model. The present invention improves the accuracy of direct-drive wind farm grouping by accurately calculating the wind speed of each direct-drive wind turbine in the direct-drive wind farm at the expected moment and the closed-loop correction of the equivalent parameters, avoids the additional equivalent errors introduced by parameter changes, and improves the accuracy of the direct-drive wind farm equivalent model in describing the transient response characteristics of the detailed model.

Description

Technical Field

[0001] The present invention relates to the technical field of power system stability analysis, and in particular to a direct-drive wind farm equivalent method and system for power system stability analysis. Background Art

[0002] In recent years, the installed capacity and penetration rate of wind power in the world have continued to grow. The output of large-scale wind farms has strong random and intermittent characteristics, which brings challenges to the safe and stable operation of power systems. In order to clarify the stability characteristics of power systems containing large-scale wind farms at the expected time, it is necessary to build an equivalent model of the wind farm and conduct simulation analysis on it.

[0003] According to the number of equivalent machines in a wind farm, the existing widely used equivalent methods are mainly divided into single-machine equivalent method and multi-machine equivalent method. The single-machine equivalent method aggregates the wind farm into one equivalent machine, and obtains the equivalent wind speed of the equivalent machine by calculating the average power of the wind turbine. The remaining parameters are obtained by the capacity weighted method. The single-machine equivalent method proposed in the literature M. Mercado-Vargas, D. G´omez-Lorente, and O. Rabaza, "Aggregated models of permanent magnet synchronous generators wind farms," ​​only selects a single characteristic quantity to describe the operating characteristics of the wind farm, so it is difficult to accurately characterize the dynamic behavior differences between multiple wind turbines in a wind farm under multiple scenarios. Overall, the single-machine equivalent method is mainly suitable for application scenarios with a small number of wind turbines in a wind farm and small wind speed differences.

[0004] The multi-machine equivalent method usually selects a single or multiple characteristic quantities of the wind turbine as a clustering index, divides the wind turbines with the same characteristics into the same group and aggregates them into an equivalent machine, so that the response characteristics of the wind farm can be more accurately characterized. The literature D.Li, C. Shen, Y. Liu, Y. Chen and S. Huang, "A Dynamic Equivalent Method for PMSG-WTG Based Wind Farms Considering Wind Speeds and Fault Severities", proposed a multi-machine equivalent method for direct-driven wind farms, which is based on the initial wind speed and fault severity of the wind turbines. Among them, the wind speed of the wind turbine at the head end of the wind farm at the expected moment is randomly selected using a uniform distribution. Based on the wind speed of the head end wind turbine, the initial wind speed of the remaining wind turbines is calculated using the Jensen model based on the linear expansion assumption. However, the method of random selection using a uniform distribution is difficult to reflect the wind speed of the actual wind farm at a specific moment, and the wind speed of hundreds of wind turbines in an actual direct-driven wind farm does not expand linearly. At the same time, when calculating the wind speed of the wind turbine in the direct-driven wind farm at the expected time, the Jensen model cannot take into account the influence of the wake center offset, turbulence intensity, yaw control, etc., and it is difficult to accurately calculate the wind speed of the direct-driven wind turbine. In addition, when constructing the equivalent model of the direct-driven wind farm, the parameters of each equivalent machine are crucial to the equivalent accuracy, and the parameters of the direct-driven wind turbine may change dynamically during long-term operation. The capacity weighted calculation method commonly used in the calculation of equivalent parameters of direct-driven wind farms does not take into account the inaccuracy of given parameters and the dynamic changes of parameters during long-term operation, which reduces the accuracy of the equivalent model in characterizing the response characteristics of the wind farm. In summary, there are two shortcomings in the existing clustering equivalent. 1) There is a deviation in the wind speed of the head wind turbine at the expected time randomly selected based on the uniform distribution. The wind farm wind speed obtained by combining the head wind turbine wind speed and the Jensen wake model is difficult to accurately reflect the actual wind speed of each wind turbine in the direct-driven wind farm at the expected time. 2) When calculating the equivalent model parameters of the direct-drive wind farm, the single capacity weighted method ignores the impact of parameter changes on the equivalent accuracy of the direct-drive wind farm, thereby reducing the equivalent accuracy of the direct-drive wind farm. Summary of the invention

[0005] The purpose of the present invention is to provide a direct-driven wind farm equivalent method and system for power system stability analysis, to construct a practical wind farm grouping equivalent method for the expected time, and to greatly improve the efficiency of wind farm stability analysis by converting the complex high-dimensional detailed wind farm model into a low-order equivalent model. At the same time, through the parameter closed-loop correction process, the additional equivalent error introduced by parameter changes is avoided, the equivalent accuracy of direct-driven wind farm grouping is improved, and the accuracy of the direct-driven wind farm equivalent model in describing the transient response characteristics of the detailed model is improved.

[0006] To achieve the above object, the present invention provides the following technical solution: a direct-drive wind farm equivalent method and system for power system stability analysis, comprising the following steps:

[0007] S1. Use the xLSTM-FLORIS joint wind speed prediction model to calculate the wind speed of each wind turbine in the wind farm: Use the xLSTM model to calculate the wind speed of each feeder head wind turbine in the wind farm at the expected time, and input it into the FLORIS wake model to predict the wind speed of the remaining direct-drive wind turbines in the actual wind farm;

[0008] S2. According to the wind speed and terminal voltage of each direct-drive wind turbine in the wind farm, the direct-drive wind farm is divided into three groups to obtain a three-machine equivalent model of the direct-drive wind farm;

[0009] S3. Using the mechanism-data hybrid driven method, the typical parameters of the equivalent model are calculated based on the capacity weighted method to obtain the identification range of the parameters; based on the parameter identification method of the Studen t hybrid model-Grey Wolf optimization, the typical parameters are corrected to obtain the equivalent parameters of the corrected direct-drive wind farm;

[0010] S4. Build an equivalent model of the direct-drive wind farm in the electromagnetic transient simulation platform, input the parameter identification results in S3, and perform stability analysis of the power system at the expected time through simulation.

[0011] Preferably, S1 comprises the following steps:

[0012] S11. Collect the historical wind speed, wind direction and the temperature, air pressure, cloud coverage, turbulence intensity and altitude of the direct-drive wind turbines to form multiple sets of sample data sets;

[0013] S12, inputting multiple groups of sample data sets into the xLSTM model, and outputting the spatial structure and electrical quantity information of the sample data sets;

[0014] S13, input the spatial structure and electrical quantity information of the sample data set into the fully connected layer, and output the prediction result, that is, the wind speed of the head end fan of the feeder where the fan is located ;

[0015] S14, adjust the wind speed of the fan at the head end Input the FLORIS wake model to calculate the wind speed of the rear direct-drive wind turbine in three-dimensional space under the effect of the wake of a single wind turbine;

[0016] S15, split the wake interaction of multiple wind turbines into the superposition of the wind speeds of the rear direct-drive wind turbines in the three-dimensional space under the wake of a single wind turbine, and calculate the wind speed of a specific wind turbine in the wake superposition area. i The actual wind speed is used to obtain the direct-drive wind turbine wind speed considering the influence of multiple machine wakes at the expected time.

[0017] Preferably, in S14, the wind speed of the rear end direct-drive fan in the three-dimensional space under the effect of the wake of a single fan is:

[0018] (1);

[0019] In the formula, are the coordinates of the rear-end fan in the horizontal direction, vertical direction and height respectively; is the thrust coefficient of the fan; are the shape parameters of Gaussian distribution; is the offset of the wake center under yaw control; is the wind direction of the direct-drive fan; They are the temperature, air pressure, cloud coverage, turbulence intensity, and altitude at the location of the direct-drive wind turbines.

[0020] Preferably, in S15, the specific wind turbine in the wake superposition area i Actual wind speed It is obtained from the following formula:

[0021] (2);

[0022] In the formula, For the upstream j Under the effect of typhoon wake alone, downstream wind turbines i The wind speed; is the initial free wind speed of the direct-drive wind farm.

[0023] Preferably, S3 comprises the following steps:

[0024] S31, taking the controller and electrical parameters of each direct-drive wind turbine given at the factory as inputs of the capacity weighting method, outputting the parameters of the direct-drive wind farm equivalent model, and considering them as typical values ​​for equivalent parameter correction;

[0025] S32, using the matrix composed of typical values ​​in S31 as a reference, setting the value interval of the equivalent parameter identification range to obtain the parameter range to be identified;

[0026] S33. Within the range of parameters to be identified, a plurality of key parameter combination sets are randomly generated based on the Monte Carlo method, and the parameters are normalized to obtain a normalized key parameter set. x ;

[0027] Through the electromagnetic transient simulation platform, a detailed model and an equivalent model of a direct-drive wind farm are built and batch simulated to generate a detailed model of a direct-drive wind farm and an equivalent model under a variety of key parameter combinations, and the binary norm of the difference between active and reactive power is output as a feature vector set. y ;

[0028] S34, the normalized key parameter set x and feature vector set y The probability density function between is characterized in the form of the probability density function of the Stuart t mixture model SMM;

[0029] S35. Normalize the key parameter set x and feature vector set y The gray wolf position output from the gray wolf optimization algorithm is input into the Studun t mixture model SMM, and the final position of the gray wolf is output. D As the final output feature of the gray wolf algorithm; the first gray wolf position is in the feature vector set y Randomly generated within the parameter range consisting of the minimum and maximum elements in;

[0030] S36, in the specific feature quantity Under the condition of , the conditional probability of the Studun t mixed model SMM and the parameter identification results of the wind farm equivalent model are calculated. ;

[0031] S37, inputting the parameter identification result into the three-machine equivalent model of the direct-drive wind farm, and finally forming the three-machine equivalent model of the direct-drive wind farm.

[0032] Preferably, in S33, the feature vector set is:

[0033] ;

[0034] in,

[0035] (3);

[0036] In the formula, are the characteristic quantities of active power equivalent error and reactive power equivalent error respectively; , are the active power and reactive power of the detailed model of the wind farm, respectively; , are the active power and reactive power output by the average model in the simulation, respectively; is the sampling times.

[0037] Preferably, in S34, the normalized key parameter set x and feature vector set y The probability density function between is:

[0038] (4);

[0039] In the formula, is the component number; For the The probability density function of the Studen t distribution with components; is the degree of freedom parameter of the Stuart t mixed model; For the The mean of the components; For the The standard deviation of the components; is the gamma function.

[0040] The probability density function of the Stuart mixture model SMM is in the form of:

[0041] (5);

[0042] In the formula, is the number of components of the Studen t distribution; The weight of each component.

[0043] Preferably, in S36, the conditional probability of the Stuart mixed model SMM is:

[0044] (6);

[0045] The parameter identification result of the wind farm equivalent model is the expected value of the conditional probability of the Studen t mixed model SMM:

[0046] (7);

[0047] In the formula, Compute the function for the expected value.

[0048] A direct-driven wind farm equivalent system for power system stability analysis, the system is used to implement a direct-driven wind farm equivalent method for power system stability analysis, including:

[0049] A memory storing executable program code;

[0050] a processor coupled to the memory;

[0051] The processor calls the executable program code stored in the memory to execute the steps of the direct-drive wind farm equivalent method for power system stability analysis.

[0052] The present invention also provides an electronic device, comprising a memory and a processor, wherein the processor is used to implement the steps of a direct-drive wind farm equivalent method for power system stability analysis when executing a computer management program stored in the memory.

[0053] The present invention also provides a computer-readable storage medium on which a computer management program is stored. When the computer management program is executed by a processor, the steps of a direct-drive wind farm equivalent method for power system stability analysis are implemented.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] The present invention adopts the xLSTM model to accurately predict the wind speed of the wind turbines at the head end of each feeder in the actual wind farm based on historical wind speed information, rather than randomly selecting based on a specific distribution, thereby improving the accuracy of wind speed prediction at the expected time.

[0056] The present invention takes into account the wake influencing factors such as the distance between each direct-drive wind turbine in the actual direct-drive wind farm in three-dimensional space, the degree of wake diffusion, the yaw effect of the wind turbine, the turbulence intensity, etc., so that when calculating the wind speed of the direct-drive wind farm at the expected time, it can more accurately characterize the wake effect of the actual specific direct-drive wind farm.

[0057] The present invention takes into account the nonlinear superposition interaction between multiple direct-drive wind turbine units, and obtains the wind speed in the wake interaction scenario of multiple wind turbines by calculating the sum of squares of wind speed losses in the wake area, so that the calculated wind speed information at the expected time is closer to the actual wind speed information of the wind farm.

[0058] The present invention adopts a mechanism-data fusion driven solution approach to improve the equivalent accuracy of direct-drive wind farms. It overcomes the existing capacity weighted method, which ignores the impact of inaccurate parameters of direct-drive wind turbines given by manufacturers or parameter changes during long-term operation on the equivalent accuracy, and reduces the additional errors introduced by the above situation when equivalent. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0060] Figure 2 This is a schematic diagram of the topological structure of a direct-drive wind farm according to an embodiment of the present invention;

[0061] Figure 3 It is a schematic diagram of a wind speed prediction model of a feeder head-end wind turbine based on an xLSTM model according to an embodiment of the present invention;

[0062] Figure 4 It is a schematic diagram of the FLORIS wake model under the influence of a single wind turbine in a 2-dimensional plane according to an embodiment of the present invention;

[0063] Figure 5 This is a schematic diagram of a three-machine equivalent model of a direct-drive wind farm according to an embodiment of the present invention;

[0064] Figure 6 The figure is a schematic diagram of a closed-loop parameter calculation process based on Grey Wolf Optimization (GWO) according to an embodiment of the present invention. DETAILED DESCRIPTION

[0065] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0066] Embodiment 1:

[0067] See also Figure 1 , a direct-drive wind farm equivalent method for power system stability analysis, comprising the following steps:

[0068] S1. Use the xLSTM-FLORIS joint wind speed prediction model to calculate the wind speed of each wind turbine in the wind farm: Use the xLSTM model to calculate the wind speed of each feeder head wind turbine in the wind farm at the expected time, and input it into the FLORIS wake model to predict the wind speed of the remaining direct-drive wind turbines in the actual wind farm;

[0069] like Figure 2 As shown in the figure, it is the topology of a direct-drive wind farm, where 100 wind turbines are connected to the point of common coupling (PCC) through 16 feeders. For such a direct-drive wind farm composed of multiple feeders, in order to analyze the operating characteristics of the wind farm at the expected time, it is necessary to first predict the wind speed of the head wind turbine of each feeder, that is, to use the xLSTM model to predict the wind speed of the 1st, 7th, 13th, 19th, 26th, 32nd, 38th, 44th, 50th, 56th, 62nd, 68th, 74th, 80th, 87th, and 94th wind turbines in the figure.

[0070] Figure 3 The figure is a schematic diagram of the wind speed prediction model of the head-end wind turbine of the feeder based on xLSTM, which consists of a data input layer, a learning layer, a fully connected layer, and an output layer. On the other hand, the predicted wind speed of the head-end wind turbine of each feeder is input into the FLORIS wake model to obtain the wind speed of the remaining wind turbines in the wind farm except the head-end wind turbine of each feeder.

[0071] The specific steps are as follows:

[0072] S11. Collect the historical wind speed, wind direction and temperature, air pressure, cloud coverage, turbulence intensity and altitude of the direct-drive wind turbines, and form multiple groups of sample data sets in the data input layer.

[0073] S12. In the learning layer, multiple groups of sample data sets are input into the xLSTM model, and the spatial structure and electrical quantity information of the sample data sets are output.

[0074] S13, input the spatial structure and electrical quantity information of the sample data set into the fully connected layer, and output the prediction result, that is, the wind speed of the head end fan of the feeder where the fan is located .

[0075] S14, adjust the wind speed of the fan at the head end Input the FLORIS wake model to calculate the wind speed of the rear direct-drive wind turbine in three-dimensional space under the influence of the wake of a single wind turbine.

[0076] like Figure 4 As shown in the figure is the wind speed of the rear wind turbine in two-dimensional space. The FLORIS wake model regards the wind speed distribution in the far wake area as a Gaussian distribution, and at the same time draws the offset of the wake center under yaw control at the time of modeling. For three-dimensional space, the wind speed of the rear direct-drive wind turbine in three-dimensional space under the action of the wake of a single wind turbine is:

[0077] (1);

[0078] In the formula, are the coordinates of the rear-end fan in the horizontal direction, vertical direction and height respectively; is the thrust coefficient of the fan; are the shape parameters of Gaussian distribution; is the offset of the wake center under yaw control; is the wind direction of the direct-drive fan; They are the temperature, air pressure, cloud coverage, turbulence intensity, and altitude at the location of the direct-drive wind turbines.

[0079] S15, split the wake interaction of multiple wind turbines into the superposition of the wind speeds of the rear direct-drive wind turbines in the three-dimensional space under the wake of a single wind turbine, and calculate the wind speed of a specific wind turbine in the wake superposition area. i The actual wind speed is used to obtain the direct-drive wind turbine wind speed considering the influence of multiple machine wakes at the expected time.

[0080] In the actual wind farm, a single downstream wind turbine i The wind speed is usually affected by the superposition of the wake effects of multiple wind turbines upstream. Figure 2 In feeder 1, the wind speed of wind turbine No. 6 is affected by wind turbines No. 1, 2, 3, 4, and 5. In order to calculate the wind speed of each wind turbine, it is necessary to calculate the wind speed of each wind turbine under the influence of multiple wind turbines. When calculating the wind speed of a specific direct-drive wind turbine, the wake interaction of multiple wind turbines can be split into the superposition of wind speeds under the action of a single wind turbine. By calculating the sum of the squares of the wind speed losses in each wake, the wind speed loss of a specific wind turbine in the wake superposition area is obtained by formula (2). i Actual wind speed size.

[0081] (2);

[0082] In the formula, For the upstream j Under the effect of typhoon wake alone, downstream wind turbines i The wind speed; is the initial free wind speed of the direct-drive wind farm.

[0083] Formula (2) can be used to obtain the wind speed of the direct-drive wind turbines in each feeder, except for the wind speed of the head-end wind turbine, at the expected time, taking into account the influence of the wake of multiple wind turbines. At this point, the joint wind speed prediction based on xLSTM-FLORIS is completed, and the wind speed of each direct-drive wind turbine in the wind farm is obtained. The result is closer to the wind speed result of the actual direct-drive wind farm at the expected time.

[0084] S2. According to the wind speed and terminal voltage of each direct-drive wind turbine in the wind farm, the direct-drive wind farm is divided into three groups, and a three-machine equivalent model of the direct-drive wind farm is obtained.

[0085] During normal operation, in order to maintain the voltage stability on the DC side, direct-drive wind turbines often adopt a fixed DC side voltage control strategy. During faults, in order to suppress voltage drops, direct-drive wind turbines need to output reactive power for active voltage support. According to the terminal voltage in the fault steady state, the upper limit of the active current, and the active current value before the fault, the fault response characteristics of the direct-drive wind turbines can be classified. When the initial wind speed and the terminal voltage in the fault steady state of each direct-drive wind turbine in the direct-drive wind farm at the expected moment are known, the direct-drive wind turbines can be grouped according to their fault response characteristics. (Existing technology, no further description)

[0086] S3. Using the mechanism-data hybrid driven method, the typical parameters of the equivalent model are calculated based on the capacity weighted method to obtain the identification range of the parameters; based on the parameter identification method of the Studen t hybrid model-Grey Wolf optimization, the typical parameters are corrected to obtain the equivalent parameters of the corrected direct-drive wind farm;

[0087] After obtaining the grouping results of the direct-drive wind turbines in the direct-drive wind farm, each group of wind turbines is equivalent to a unit, and the following is obtained: Figure 5 The three-machine equivalent model of the direct-drive wind farm is shown in the figure. Z 1 , Z 2 , Z 3 They are the equivalent impedances of the collector lines connected to the Group I, Group II, and Group III value-added machines, respectively.

[0088] In order to finally obtain the equivalent model, it is necessary to calculate the parameters of each equivalent machine in the equivalent model, where the parameters of the equivalent machine include the controller parameters and electrical parameters of the direct-drive wind turbine.

[0089] The specific steps are as follows:

[0090] S31. In terms of mechanism drive, the controller parameters and electrical parameters of each direct-drive wind turbine given at the factory are used as the input of the capacity weighted method, and the parameters of the direct-drive wind farm equivalent model are output, which are regarded as the typical values ​​of the equivalent parameter correction; the matrix composed of the typical values ​​of the key equivalent parameter correction is .

[0091] S32. In terms of data driving, the matrix composed of the typical values ​​obtained in S31 As a benchmark, set the value interval of the equivalent parameter identification range to obtain the parameter range to be identified; the calculation method is as follows:

[0092] (3);

[0093] In the formula, is a matrix composed of the maximum values ​​of the identification parameter group; is a matrix composed of the minimum values ​​of the identification parameter group; Set the value for the parameter identification range.

[0094] S33, within the value range of the parameter range to be identified set in S32, randomly generate multiple key parameter combination sets based on the Monte Carlo method, and normalize the parameters to obtain a normalized key parameter set x ;

[0095] Then, by building a detailed model and an equivalent model of the direct-drive wind farm in the electromagnetic transient simulation platform and conducting batch simulations on them, the detailed model and the equivalent model of the direct-drive wind farm under various key parameter combinations are generated, and the binary norm of the difference between active power and reactive power is output as the characteristic quantity;

[0096] (4);

[0097] In the formula, are the characteristic quantities of active power equivalent error and reactive power equivalent error respectively; , are the active power and reactive power of the detailed model of the wind farm, respectively; , are the active power and reactive power output by the average model in the simulation, respectively; is the sampling number. For the convenience of explanation, the feature vector is recorded as: ;

[0098] S34, using the maximum expectation algorithm, calculate the normalized key parameter set obtained in S33 x and feature vector set yStudent t Mixture Model (SMM) parameters, including the degree of freedom parameter of the Student t Mixture Model , No. The mean of the components , No. The standard deviation of the component , the weight of each component .

[0099] Normalized key parameter set x and feature vector set y The probability density function between can be expressed as:

[0100] (5);

[0101] In the formula, is the component number; For the The probability density function of the Studen t distribution with components; is the gamma function.

[0102] Then, the normalized key parameter set can be x and feature vector set y The probability density function of the Studon t mixture model SMM is represented as:

[0103] (6);

[0104] In the formula, is the number of components of the Studen's t distribution.

[0105] S35, such as Figure 6 As shown, the parameters are calculated based on the Gray Wolf Optimization Algorithm. The normalized key parameter set x and feature vector set y The gray wolf positions output by the gray wolf optimization algorithm are input into the Studun t mixture model SMM, and the parameter set to be identified is output. Among them, the first gray wolf position is in the feature vector set y The parameters are randomly generated within the parameter range composed of the minimum and maximum elements in the . Subsequently, the parameter set to be identified is input into the equivalent model in the electromagnetic transient simulation platform, and the equivalent model and the detailed model are simulated to calculate the fitness function corresponding to the parameter set to be identified and input into the Grey Wolf Optimization Algorithm GWO. For the calculation of equivalent parameters of wind farms, the parameter solution goal is to ensure that the power response characteristics of the equivalent model are consistent with those of the detailed model. The fitness function corresponding to the parameter set to be identified is:

[0106] (7);

[0107] By calculating the value of the fitness function, the quality of the gray wolf's position can be judged, that is, the accuracy of the equivalent parameters in the wind farm equivalent model. By repeatedly performing the correction iteration process, when the fitness function is lower than the preset specific threshold, the final position of the gray wolf is output. D .

[0108] S36. Output the final position of the gray wolf D It can be regarded as the characteristic quantity of the final output of the Grey Wolf Optimization Algorithm GWO. At this time, the parameter identification problem becomes the calculation of the Studun t mixture model SMM at a specific characteristic quantity The corresponding key parameter set in the case.

[0109] The conditional probability of the input into the Studen t mixture model SMM is obtained:

[0110] (8);

[0111] Then, the expected value of the conditional probability of the Studun t mixed model SMM is calculated, and the expected value is the parameter identification result of the wind farm equivalent model. :

[0112] (9);

[0113] In the formula, Computes a function for expected value.

[0114] S37, inputting the parameter identification result into the three-machine equivalent model of the direct-drive wind farm, and finally forming the three-machine equivalent model of the direct-drive wind farm.

[0115] S4. Build an equivalent model of the direct-drive wind farm in the electromagnetic transient simulation platform, input the parameter identification results of the wind farm equivalent model into the equivalent model, and obtain the equivalent model of the direct-drive wind farm. Connect the equivalent model of the direct-drive wind farm to the power grid simulation model, and perform the stability analysis of the power system at the expected time by measuring the power, voltage, current and other response characteristics of the wind farm according to the power system stability analysis requirements.

[0116] Embodiment 2

[0117] A direct-driven wind farm equivalent system for power system stability analysis, the system is used to implement a direct-driven wind farm equivalent method for power system stability analysis, including:

[0118] A memory storing executable program code;

[0119] a processor coupled to the memory;

[0120] The processor calls the executable program code stored in the memory to execute the steps of a direct-drive wind farm equivalent method for power system stability analysis as described in the first embodiment.

[0121] Embodiment 3

[0122] The present invention also provides an electronic device, comprising a memory and a processor, wherein the processor is used to implement the steps of a direct-drive wind farm equivalent method for power system stability analysis when executing a computer management program stored in the memory.

[0123] Embodiment 4

[0124] The present invention also provides a computer-readable storage medium on which a computer management program is stored. When the computer management program is executed by a processor, the steps of a direct-drive wind farm equivalent method for power system stability analysis are implemented.

Claims

1. A direct-drive wind farm equivalent method for power system stability analysis, characterized in that: The following steps are involved: S1. Calculate the wind speed of each wind turbine in the wind farm using the xLSTM-FLORIS joint wind speed prediction model: Calculate the wind speed of each feeder head wind turbine in the wind farm at the expected time using the xLSTM model, and input it into the FLORIS wake model to predict the wind speed of the remaining direct-drive wind turbines in the actual wind farm; including the following steps: S11. Collect the historical wind speed, wind direction and the temperature, air pressure, cloud coverage, turbulence intensity and altitude of the direct-drive wind turbines to form multiple sets of sample data sets; S12, inputting multiple groups of sample data sets into the xLSTM model, and outputting the spatial structure and electrical quantity information of the sample data sets; S13, input the spatial structure and electrical quantity information of the sample data set into the fully connected layer, and output the prediction result, that is, the wind speed of the head end fan of the feeder where the fan is located ; S14, adjust the wind speed of the fan at the head end Input the FLORIS wake model to calculate the wind speed of the rear direct-drive wind turbine in three-dimensional space under the effect of the wake of a single wind turbine: （1）； In the formula, are the coordinates of the rear-end fan in the horizontal direction, vertical direction and height respectively; is the thrust coefficient of the fan; are the shape parameters of Gaussian distribution; is the offset of the wake center under yaw control; is the wind direction of the direct-drive fan; They are the temperature, air pressure, cloud coverage, turbulence intensity, and altitude at the location of the direct-drive wind turbines; S15, split the wake interaction of multiple wind turbines into the superposition of the wind speeds of the rear direct-drive wind turbines in the three-dimensional space under the wake of a single wind turbine, and calculate the wind speed of a specific wind turbine in the wake superposition area. i Actual wind speed : （2）； In the formula, For the upstream j Under the effect of typhoon wake alone, downstream wind turbines i The wind speed; is the initial free wind speed of the direct-drive wind farm; Get the wind speed of the direct-drive wind turbine considering the influence of multiple turbine wakes at the expected time; S2. According to the wind speed and terminal voltage of each direct-drive wind turbine in the wind farm, the direct-drive wind farm is divided into three groups to obtain a three-machine equivalent model of the direct-drive wind farm; S3. Using the mechanism-data hybrid driven method, the typical parameters of the equivalent model are calculated based on the capacity weighted method to obtain the identification range of the parameters; based on the parameter identification method of the Studen t hybrid model-Grey Wolf optimization, the typical parameters are corrected to obtain the equivalent parameters of the corrected direct-drive wind farm; S4. Build an equivalent model of the direct-drive wind farm in the electromagnetic transient simulation platform, input the parameter identification results, and perform stability analysis of the power system at the expected time through simulation.

2. A direct-drive wind farm equivalent method for power system stability analysis according to claim 1, characterized in that: S3 includes the following steps: S31, taking the controller and electrical parameters of each direct-drive wind turbine given at the factory as inputs of the capacity weighting method, outputting the parameters of the direct-drive wind farm equivalent model, and considering them as typical values ​​for equivalent parameter correction; S32, using the matrix composed of typical values ​​in S31 as a reference, setting the value interval of the equivalent parameter identification range to obtain the parameter range to be identified; S33. Within the range of parameters to be identified, a plurality of key parameter combination sets are randomly generated based on the Monte Carlo method, and the parameters are normalized to obtain a normalized key parameter set. x ; The electromagnetic transient simulation platform generates a detailed model of the direct-drive wind farm and equivalent models under various key parameter combinations, and outputs the binary norm of the difference between active and reactive power as a characteristic vector set. y ; S34, the normalized key parameter set x and feature vector set y The probability density function between is characterized in the form of the probability density function of the Stuart t mixture model SMM; S35. Normalize the key parameter set x and feature vector set y The gray wolf position output from the gray wolf optimization algorithm is input into the Studun t mixture model SMM, and the final position of the gray wolf is output. D As the final output feature of the gray wolf optimization algorithm; the first gray wolf position is in the feature vector set y Randomly generated within the parameter range consisting of the minimum and maximum elements in; S36, in the specific feature quantity Under the condition of , the conditional probability of the Studun t mixed model SMM and the parameter identification results of the wind farm equivalent model are calculated. ; S37, inputting the parameter identification result into the three-machine equivalent model of the direct-drive wind farm, and finally forming the three-machine equivalent model of the direct-drive wind farm.

3. A direct-drive wind farm equivalent method for power system stability analysis according to claim 2, characterized in that: In S33, the feature vector set is: ； in, （3）； In the formula, are the characteristic quantities of active power equivalent error and reactive power equivalent error respectively; , are the active power and reactive power of the detailed model of the wind farm, respectively; , are the active power and reactive power output by the average model in the simulation, respectively; is the sampling times.

4. A direct-drive wind farm equivalent method for power system stability analysis according to claim 2, characterized in that: In S34, the normalized key parameter set x and feature vector set y The probability density function between is: （4）； In the formula, is the component number; For the The probability density function of the Studen t distribution with components; is the degree of freedom parameter of the Stuart t mixed model; For the The mean of the components; For the The standard deviation of the components; is the gamma function; The probability density function of the Stuart mixture model SMM is in the form of: （5）； In the formula, is the number of components of the Studen t distribution; The weight of each component.

5. A direct-drive wind farm equivalent method for power system stability analysis according to claim 2, characterized in that: In S36, The conditional probability of the Stuart t mixed model SMM is: （6）； The parameter identification result of the wind farm equivalent model is the expected value of the conditional probability of the Studen t mixed model SMM: （7）； In the formula, Computes a function for expected value.

6. A direct-driven wind farm equivalent system for power system stability analysis, used to implement a direct-driven wind farm equivalent method for power system stability analysis as described in any one of claims 1 to 5, comprising: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the steps of the direct-drive wind farm equivalent method for power system stability analysis.

Images