Wake flow model parameterization optimization method and system based on historical operation data of wind power plant

Abstract

The invention relates to the technical field of wind power generation, and discloses a wake flow model parameterization optimization method and system based on historical operation data of a wind power plant, and the method comprises the steps: collecting the historical operation data of the wind power plant, carrying out the preprocessing of the collected historical operation data of the wind power plant, and obtaining a training set and a test set; based on an Ishihara single machine wake flow model, constructing an engineering experience wake flow model; establishing an optimization problem about a wake flow growth rate proportionality coefficient in an engineering experience wake flow model, solving by adopting a particle swarm optimization algorithm, and optimizing the wake flow growth rate proportionality coefficient by taking a mean absolute error MAE between a power predicted value and a power measured value of the whole wind power plant in a training set as a fitness function; and applying the optimized wake flow model applied to the engineering experience to predict the power of the wind power plant group. The power prediction error of the wake flow model can be obviously reduced, and the power prediction MAE of the offshore wind power plant group is reduced and the annual energy output is improved.

Description

Technical Field

[0001] This invention belongs to the field of wind power generation technology and relates to a method and system for parameterized optimization of wake models based on historical operating data of wind farms. It is particularly suitable for power prediction and intelligent control systems of offshore wind farm clusters. Background Technology

[0002] As wind farm construction develops towards clustering, base development, and large-scale operations, large-scale, grid-connected wind farm clusters are gradually becoming one of the important forms of wind power generation in China. Due to the large number of wind turbines in a wind farm cluster, the wake effect is widespread, typically reaching up to 10 km. For offshore wind farm clusters, the turbulence intensity is lower than onshore, and wake recovery is slower. Offshore wind farm wakes not only significantly impact economic returns but also exhibit more complex power characteristics due to the variable weather conditions, significant seasonal variations, and marked spatiotemporal differences at sea. Therefore, studying the wake of offshore wind farm clusters is a crucial prerequisite for assessing their power generation and guiding their precise operation.

[0003] Research on wind turbine wakes mainly focuses on two aspects: the study of wake development principles and the development and application of engineering wake models. In the study of wake development principles, the main characteristics of the wake are verified and reconstructed through field experiments, wind tunnel simulations, and CFD (Computational Fluid Dynamics) methods, and engineering wake models can be developed based on this. The application of wind farm engineering wake models refers to proposing methods to optimize wind farm layout or operational indicators based on existing wake models, or conducting research on wind turbine wakes.

[0004] For wake interference at the wind farm cluster level, current research methods mainly fall into two categories: top-down and bottom-up. The top-down approach treats a single wind farm as an object, equating it to additional surface roughness or momentum sinking, and considering the influence of atmospheric thermal stability to alter the average wind speed of the incoming flow from downstream wind farms. The bottom-up approach, based on a single-unit wake model, calculates the wind speed at the hub center of each wind turbine and applies a wind speed superposition method to obtain the wake distribution of the entire wind farm cluster.

[0005] Currently, in wake control research, the engineering empirical wake models, which serve as the basic tool, typically use fixed empirical parameters given in literature or wind tunnel tests. However, different wind farms vary greatly in geographical location, climate, and turbulence characteristics. Using a fixed set of parameters will result in insufficient calculation accuracy of the model for that specific wind farm group, thereby affecting the reliability of subsequent power prediction and control optimization. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for parameterized optimization of wake models based on historical operating data of wind farms. This method can significantly reduce the power prediction error of wake models, reduce the MAE of power prediction for offshore wind farm clusters, and increase annual power generation. It is applicable to intelligent control systems for megawatt-class offshore wind power bases.

[0007] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution.

[0008] In a first aspect, this invention proposes a parameterized optimization method for wake model based on historical operating data of wind farms, comprising the following steps:

[0009] Historical operation data of wind farms are collected, and the collected historical operation data of wind farms are preprocessed to obtain training set and test set;

[0010] Based on the Ishihara single-unit wake model, and combining the linear wind speed superposition method based on the wind turbine definition and the turbulence intensity superposition method considering wake interaction, an engineering empirical wake model is constructed.

[0011] Establish the wake growth rate ratio coefficient in the engineering empirical wake model The optimization problem is solved using the particle swarm optimization algorithm, with the mean absolute error (MAE) between the predicted and measured power values ​​of the entire wind farm in the training set as the fitness function. Optimization is performed to obtain the optimized result. ;

[0012] The optimized result It is applied to engineering empirical wake models for power prediction of wind farm clusters.

[0013] In conjunction with the first aspect, further, after optimizing the obtained After being applied to the engineering experience wake model, the following steps are also included: inputting the data from the test set into the engineering experience wake model to obtain the predicted power of the wind farm.

[0014] In conjunction with the first aspect, further, the engineering empirical wake model is constructed based on the Ishihara single-unit wake model, combined with the linear wind speed superposition method based on the wind turbine definition and the turbulence intensity superposition method considering wake interaction, including:

[0015] The expression for the linear wind speed superposition method based on the wind turbine definition is:

[0016] ;

[0017] In the formula, It is the wind speed of the target wind turbine. It is the inflow wind speed of the wind farm. It refers to the number of upstream wind turbine units; It is the serial number of the upstream wind turbine unit. It is an upstream wind turbine Wind speed; the superposition order is calculated sequentially from upstream to downstream in the windward direction;

[0018] The expression for the superposition method of turbulence intensity of the wake interaction is:

[0019] ;

[0020] ;

[0021] In the formula, It is the turbulence intensity; It is the intensity of environmental turbulence; It is the wake-added turbulence.

[0022] Building upon the first aspect, we further introduce the wake growth rate. Select proportionality coefficient To optimize the object, then The calculation method is as follows:

[0023] ;

[0024] In the formula, This is the thrust coefficient of the yaw wind turbine, which is a corrected parameter. It is the intensity of environmental turbulence;

[0025] because The calculation method causes the shape of the Gaussian distribution to change, while the maximum value remains unchanged. Therefore, it is necessary to correct the velocity loss and additional turbulence in the wake formula.

[0026] The corrected formula for calculating velocity loss is as follows:

[0027] ;

[0028] In the formula, It is after the wind turbine Reduced velocity standard deviation at position; It is the wind speed at the center of the wind turbine; It is the offset of the wake centerline; It refers to the wheel hub height; Indicates the diameter of the wind turbine; , , It is a parameter; It is the radial distance;

[0029] The corrected formula for the additional turbulence calculation method is as follows:

[0030] ;

[0031] in, It is after the wind turbine The standard deviation of the velocity at position; It is the intensity of environmental turbulence; It is a turbulence intensity correction term; , , , , It is a parameter.

[0032] In conjunction with the first aspect, the expression for the fitness function is further as follows:

[0033] ;

[0034] In the formula, These are measured values, corresponding to the measured power of the wind farm; It is a predicted value, corresponding to the predicted power of the wind farm, which is output after being calculated by the engineering experience wake model; This refers to the number of samples, corresponding to the number of wind turbine units. , .

[0035] In conjunction with the first aspect, further, the aforementioned Optimize to obtain the optimized version ,include:

[0036] With the objective of minimizing the mean absolute error between predicted and measured power values, Optimize to obtain the optimized version Specifically:

[0037] Optimize parameters When the time is right, the optimization problem is as follows:

[0038] ;

[0039] in, Represents the fitness function; Indicates constraints; These are measured values ​​of wind farm power. The predicted power of the wind farm is calculated using a wind farm group power calculation method based on an engineering experience wake model.

[0040] Secondly, this invention proposes a wake model parameterization optimization system based on historical wind farm operation data, used to implement the aforementioned wake model parameterization optimization method based on historical wind farm operation data, including:

[0041] The data acquisition module is configured to collect historical operating data of wind farms, preprocess the collected historical operating data of wind farms, and obtain training sets and test sets.

[0042] The engineering experience wake model construction module is configured to construct an engineering experience wake model based on the Ishihara single-machine wake model, combining the linear wind speed superposition method based on the wind turbine definition and the turbulence intensity superposition method considering wake interaction.

[0043] The optimization solution module is configured to establish the proportion coefficient of the wake growth rate in the engineering empirical wake model. The optimization problem is solved using the particle swarm optimization algorithm, with the mean absolute error (MAE) between the predicted and measured power values ​​of the entire wind farm in the training set as the fitness function. Optimization is performed to obtain the optimized result. ;

[0044] The prediction module is configured to use the optimized result It is applied to engineering empirical wake models for power prediction of wind farm clusters.

[0045] Thirdly, the present invention proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-mentioned wake model parameterization optimization method based on historical operating data of wind farms.

[0046] Fourthly, the present invention provides a computer device comprising:

[0047] Memory, used to store computer programs;

[0048] A processor is used to execute the computer program to implement the steps of the above-described wake model parameterization optimization method based on historical wind farm operating data.

[0049] Fifthly, the present invention proposes a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described wake model parameterization optimization method based on historical operating data of wind farms.

[0050] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0051] (1) This invention can significantly reduce the power prediction error of the engineering experience wake model, reduce the MAE of offshore wind farm group power prediction by 14.78%, and increase the annual power generation by about 0.65%. It is suitable for intelligent control system of megawatt-level offshore wind power base.

[0052] (2) This invention addresses the wake control of offshore wind farm clusters. Based on the Ishihara single-unit engineering experience wake model, wind speed superposition method, and turbulence superposition method, a wind farm cluster engineering experience wake model is constructed. For the constructed wind farm cluster engineering experience wake model, a wake model parameter optimization method is proposed. Based on historical data, the parameters of the engineering experience wake model are optimized to make the engineering experience wake model adaptable to offshore wind farm clusters.

[0053] (3) The present invention uses the wake growth rate ratio coefficient in the wake model. As optimization variables, the Particle Swarm Optimization (PSO) algorithm is used to automatically find the most suitable algorithm for a given wind farm, with the objective of minimizing the mean absolute error between the predicted and measured historical power values. Optimal value. The parameters are used to optimize the engineering experience wake model, thereby obtaining a wake model suitable for a given wind farm. Attached Figure Description

[0054] Figure 1 This is a flowchart illustrating the parameter optimization method in Embodiment 1 of the present invention;

[0055] Figure 2 This is a wind farm layout diagram in Embodiment 2 of the present invention;

[0056] Figure 3 This is a graph showing the wind speed and wind turbine operating power over 21 days in Embodiment 2 of the present invention;

[0057] Figure 4 This is a 21-day wind direction and ambient temperature graph from Embodiment 2 of the present invention;

[0058] Figure 5 This is the wind speed histogram in Embodiment 2 of the present invention;

[0059] Figure 6 This is the wind rose diagram in Embodiment 2 of the present invention;

[0060] Figure 7 This is a power comparison diagram from Embodiment 2 of the present invention;

[0061] Figure 8 This is a diagram illustrating the PSO optimization iteration process for conditions 2 and 3 in Embodiment 2 of the present invention. Detailed Implementation

[0062] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0063] The term "and / or" simply describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0064] Example 1

[0065] like Figure 1 As shown in this embodiment, the wake model parameterization optimization method based on historical wind farm operation data proposed includes the following steps:

[0066] Step S1: Acquire at least 14 consecutive days of 15-minute resolution data from the wind farm data acquisition and monitoring system, i.e., collect historical operating data of the wind farm. Preprocess the historical operating data, interpolating or removing missing, abnormal, or 0° or 360° wind direction jumps to obtain training and testing sets. The historical operating data of the wind farm includes time series data on wind speed, wind direction, wind turbine operating power, and ambient temperature.

[0067] Step S2: Based on the Ishihara single-unit wake model, and combining the linear wind speed superposition method based on the wind turbine definition and the turbulence intensity superposition method considering wake interaction, an engineering empirical wake model is constructed to calculate the wind speed and total power at the hub center of all wind turbines in the wind farm group.

[0068] Step S3: Establish the proportional coefficient of the wake growth rate in the engineering empirical wake model. The optimization problem is solved using the particle swarm optimization algorithm, with the mean absolute error (MAE) between the predicted and measured power values ​​of the entire wind farm in the training set as the fitness function. Optimization is performed to obtain the optimized result. .

[0069] Among them, minimizing the average absolute error between predicted and measured power values ​​is the objective. Optimize to obtain the optimized version .

[0070] Step S4, the optimized result obtained in step S3 It is applied to engineering empirical wake models for power prediction or optimization control of wind farm groups, thereby optimizing the parameters of the engineering empirical wake model and reducing the error in power calculation by the engineering empirical wake model.

[0071] In one specific implementation of this embodiment, since the downstream wind turbine is affected by the wake of multiple upstream wind turbines, in order to evaluate the wake effect of multiple wind turbines, the wind speed superposition method of the wake adopts the linear wind speed superposition method based on the definition of the wind turbine (i.e., the linear momentum conservation superposition method based on the definition of the wind turbine), and its formula is shown in equation (1):

[0072] (1);

[0073] In the formula, It is the wind speed of the target wind turbine. It is the inflow wind speed of the wind farm. It refers to the number of upstream wind turbine units; It is the serial number of the upstream wind turbine unit. It is an upstream wind turbine The wind speed; the superposition order is calculated sequentially from upstream to downstream in the windward direction.

[0074] The expression for the superposition method of turbulence intensity in wake interaction is:

[0075] ;

[0076] ;

[0077] In the formula, It is the turbulence intensity; It is the intensity of environmental turbulence; It is wake-added turbulence. Simultaneously, the wake growth rate is introduced. This invention automatically finds the optimal [solution / mechanism] using the particle swarm optimization algorithm. This allows the empirical wake model to more accurately reflect the diffusion characteristics of the actual wake, thus improving the accuracy of the empirical wake model; selecting proportionality coefficient To optimize the object, then The calculation method is changed to equation (2):

[0078] (2);

[0079] In the formula, It is the proportional coefficient of the tail current growth rate; This is the thrust coefficient of the yaw wind turbine, which is a corrected parameter. It refers to the intensity of environmental turbulence.

[0080] because Changing the calculation method to Equation (2) will cause the shape of the Gaussian distribution to change, while the maximum value remains unchanged. Therefore, it is necessary to correct the velocity loss and additional turbulence in the wake formula. The corrected formula for calculating velocity loss is as follows:

[0081] (3);

[0082] In the formula, It is after the wind turbine Reduced velocity standard deviation at position; It is the wind speed at the center of the wind turbine; It is the offset of the wake centerline; It refers to the wheel hub height; Indicates the diameter of the wind turbine; , , It is a parameter; It is the radial distance, calculated using the following formula;

[0083] (4);

[0084] (5);

[0085] (6);

[0086] (7);

[0087] in, The calculation formula is: ; It is the wind turbine thrust coefficient, which is a baseline parameter; ; Yaw angle; For wind turbine units In position Axis coordinates.

[0088] The corrected formula for the additional turbulence calculation method is as follows:

[0089] (8);

[0090] in, It is after the wind turbine The standard deviation of the velocity at position; It is the intensity of environmental turbulence; It is a turbulence intensity correction term; , , , , The formula for calculating the parameters is as follows:

[0091] (9);

[0092] (10);

[0093] (11);

[0094] (12);

[0095] (13);

[0096] (14).

[0097] Furthermore, the mean absolute error is selected as the loss function to determine the calculation accuracy of the engineering experience wake model of the present invention. The calculation method is shown in the following formula:

[0098] (15);

[0099] In the formula, These are measured values, corresponding to the measured power of the wind farm; These are predicted values, corresponding to the predicted power of the wind farm, which are calculated and output by the engineering experience wake model of this invention. This refers to the number of samples, corresponding to the number of wind turbine units. , .

[0100] In one specific implementation of this embodiment, the wake model parameterization optimization method of the present invention further includes a verification step considering ambient temperature, namely step S5, which specifically includes:

[0101] Step S5, In step S3, introduce a first-order polynomial of ambient temperature. As an optional extension, it degenerates into a constant if the correlation with ambient temperature is insufficient. ,Right now Either take a fixed value or use an empirical function; where, and All of these represent temperature optimization parameters; Indicates time; Indicates ambient temperature; This represents the proportional coefficient of the optimized wake growth rate, taking into account ambient temperature.

[0102] Step S5 demonstrates that the ambient temperature has little correlation with the output of the engineering experience wake model of this invention, indicating that the parameterization optimization method for the wake model of this invention does not need to consider the ambient temperature. The specific verification process is as follows:

[0103] Based on Table 1, set the particle dimension, number of particles, number of iterations, inertia weight, and learning factor for the PSO algorithm.

[0104] Table 1. Model training parameter settings in the engineering experience wake model parameter optimization process.

[0105]

[0106] The training set obtained in step S1 is used to calculate the predicted power of the wind farm using the wind farm group power calculation method based on the engineering experience wake model constructed in this invention. The time series of wind speed, wind direction, ambient temperature, and wind turbine operating power in the training set are then input into the training set. .

[0107] Different optimization methods are selected and represented by different operating conditions, as follows:

[0108] Condition 1: No optimization performed. For fixed values, the initial values ​​of the wake model based on engineering experience are used for operating condition 1. .

[0109] Operating Condition 2: Optimization of Parameters The optimization problem is as follows:

[0110] (16);

[0111] In the formula, Represents the fitness function; Indicates constraints; These are measured values ​​of wind farm power. The predicted power of the wind farm is calculated using a wind farm group power calculation method based on an engineering experience wake model, which is the power in the above formula (15). .

[0112] Operating Condition 3: Considering ambient temperature, an ambient temperature correction term is introduced to correlate the relationship between ambient temperature and the output accuracy of the engineering experience wake model of this invention. The effect of ambient temperature on the accuracy of the wake model is explored using a first-order polynomial. The influence of setting initial values ​​for optimization parameters. , ,in, For time The function is used to reflect the different environmental conditions at different times. The temperature optimization parameters are: and The optimization problem is shown in the following equation:

[0113] (17);

[0114] In the formula, It represents the ambient temperature, which is a function of time. This means that the ambient temperature is taken into account. The optimized wake growth rate ratio coefficient will be obtained after optimization. It is applied to engineering experience wake models.

[0115] The optimization parameters obtained from formulas (16) and (17) for two different optimization problems Substituting these values ​​into formulas (2), (3), and (4), the accuracy of the engineering experience wake model parameter optimization method of the present invention is verified through MAE error analysis.

[0116] In summary, this invention collects at least 14 consecutive days of historical wind farm operation data at a 15-minute resolution from the wind farm data acquisition and monitoring system. After integrity and rationality checks, training and testing sets are constructed through preprocessing. Based on the Ishihara Gaussian wake model, the wake growth rate is... The proportionality coefficient in Set as the variable to be optimized, and modify the engineering experience wake model; use particle swarm optimization algorithm to automatically find the optimal value with the goal of minimizing the average absolute error of the total power; then optimize the result. By substituting the empirical wake model into the data, the wind speed and power at the hub of each turbine are recalculated. This invention can significantly reduce the power prediction error of the empirical wake model, thereby reducing the MAE (Maximum Estimated Power) of offshore wind farm clusters and increasing annual power generation.

[0117] Example 2

[0118] This embodiment further illustrates the wake model parameter optimization method based on historical wind farm operation data of the present invention using more specific data.

[0119] This embodiment uses measured data from an actual offshore wind farm cluster in the southeastern coastal region of China for verification. The location map of the wind turbines in this offshore wind farm cluster is shown below. Figure 2 As shown, Figure 2 This is a wind turbine coordinate orientation map with the Y-axis pointing to true north and the X-axis pointing to true south. All dimensions are in kilometers. Figure 2 Each colored dot represents a wind turbine, and different colors represent wind turbines of different capacities. Figure 2 There are 3 gray dots, representing wind measurement towers; among them, the wind measurement tower used to measure wind speed and direction is 110m high, and the wind measurement tower used to measure ambient temperature is 10m high.

[0120] Figure 2 The offshore wind farm cluster comprises three offshore wind farms, designated as Wind Farm #1 (four rows from the left), Wind Farm #2 (two rows from the upper right), and Wind Farm #3 (two rows from the lower right). Each wind farm has one wind measurement tower. Wind Farm #1 has 42 wind turbines with an installed capacity of 250MW; Wind Farm #2 has 42 wind turbines with an installed capacity of 250MW; and Wind Farm #3 has 50 wind turbines with an installed capacity of 300MW. Therefore, the total installed capacity of this offshore wind farm cluster is 800MW. The data in this embodiment is from Wind Farm #3.

[0121] The specific layout of the wind turbine units is as follows: Figure 2 , Figure 2 The data for Wind Farm #3 (bottom two rows) shows that it has 50 wind turbines with an installed capacity of 300MW. The total installed capacity of this wind farm complex is 800MW.

[0122] In this embodiment, the dataset is historical operational data, including measured wind speed and direction, temperature, and wind turbine operating power (active power of wind turbines is directly exported from the wind farm data acquisition and monitoring system). The time resolution of the historical operational data is 15 minutes. The dataset in this embodiment covers data measured over 21 consecutive days in a certain winter, totaling 21*24*4=2016 data points. After preprocessing in step S1 of embodiment 1, the wind speed, wind direction, ambient temperature, and wind turbine operating power of wind farm #3 are as follows: Figure 3 and Figure 4 As shown.

[0123] A two-parameter Weibull distribution was used to fit the collected wind speeds, with the shape parameter k set to 2.078 and the scale parameter c to 8.915. The wind speed histogram is shown below. As shown, wind direction data that has only undergone completeness and reasonableness checks and unreasonable data processing, but has not undergone elimination of abrupt changes in wind direction values, is plotted as a wind rose diagram, as follows. Figure 6 As shown; where, Figure 6 N indicates northerly wind, 0° / 360°; NNE indicates north-northeast wind, 22.5°; NE indicates northeasterly wind, 45°; ENE indicates east-northeast wind, 67.5°; E indicates easterly wind, 90°; ESE indicates east-southeast wind, 112.5°; SE indicates southeasterly wind, 135°; SSE indicates south-southeast wind, 157.5°; S indicates southerly wind, 180°; SSW indicates south-southwest wind, 202.5°; SW indicates southwesterly wind, 225°; WSW indicates west-southwest wind, 247.5°; W indicates westerly wind, 270°; WNW indicates west-northwest wind, 292.5°; NW indicates northwesterly wind, 315°; NNW indicates north-northwest wind, 337.5°.

[0124] Depend on Figure 6 It can be seen that the wind speed in the collected historical operational data conforms to the Weibull distribution. Among them, the frequency of low to medium wind speed (4-8 m / s) is relatively high, while the frequency of medium and medium to high wind speed (8-14 m / s) is relatively low, reflecting that the overall wind speed in the region is low in winter. The prevailing wind direction is N, with a frequency of 41.7%, which is similar to the winter monsoon wind direction in the region. Therefore, this dataset can be used to verify the effectiveness of the wake model parameter optimization method proposed in this invention. Based on this historical operational data, the engineering experience wake model represented by formulas (3) and (8) is constructed using Python.

[0125] We used data from day 1 to day 14 of the historical operation data, and divided this data into training and test sets. Days 1 to 7 were used as the training set, and days 8 to 14 were used as the test set.

[0126] Using the PSO algorithm to solve formula (16), the optimization problem converged after 50 iterations, and the optimization result is: Using the PSO algorithm to solve formula (17), the optimization problem converged after 50 iterations, and the optimization result is: The optimization iterative process is as follows: As shown.

[0127] Table 2 shows the errors between the calculated power and the collected operating power of the engineering experience wake model for operating conditions 2 and 3 on the test set. The comparison values ​​of the actual power (i.e., the measured power value of the wind farm) are as follows: Figure 8 As shown; where, Figure 8 The horizontal axis represents the number of iterations, and the vertical axis represents the... The value.

[0128] Table 2 shows the error between the calculated power and the collected operating power from the engineering experience wake model on the test set.

[0129]

[0130] Compared to Condition 1, the optimization effect of Condition 3 is not significant, indicating that the ambient temperature does not fit well with the accuracy of the engineering empirical wake model of this invention under this wind farm condition, proving that the correlation between the ambient temperature and the output of the engineering empirical wake model of this invention is poor. Compared to Condition 1, Condition 2 reduces the mean square error (MSE) by 8.10%, the mean absolute error (MAE) by 14.78%, and the root mean square error (RMSE) by 7.68%. In conclusion, a first-order polynomial cannot be used for fitting. The relationship with ambient temperature, The parameter optimization can improve the accuracy of the wake model's power calculation. In other words, the wake model parameter optimization method of this invention does not need to consider the ambient temperature, which further proves the effectiveness of the wake model parameter optimization method of this invention.

[0131] Example 3

[0132] Based on the same inventive concept as Embodiment 1, this embodiment introduces a wake model parameterization optimization system based on historical wind farm operation data, used to implement the wake model parameterization optimization method based on historical wind farm operation data of Embodiment 1, including:

[0133] The data acquisition module is configured to collect historical operating data of wind farms, preprocess the collected historical operating data of wind farms, and obtain training sets and test sets.

[0134] The engineering experience wake model construction module is configured to construct an engineering experience wake model based on the Ishihara single-machine wake model, combining the linear wind speed superposition method based on the wind turbine definition and the turbulence intensity superposition method considering wake interaction.

[0135] The optimization solution module is configured to establish the proportion coefficient of the wake growth rate in the engineering empirical wake model. The optimization problem is solved using the particle swarm optimization algorithm, with the mean absolute error (MAE) between the predicted and measured power values ​​of the entire wind farm in the training set as the fitness function. Optimization is performed to obtain the optimized result. ;

[0136] The prediction module is configured to use the optimized result It is applied to engineering empirical wake models for power prediction of wind farm clusters.

[0137] Example 4

[0138] Based on the same inventive concept as other embodiments, this embodiment introduces a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described wake model parameterization optimization method based on historical operating data of a wind farm.

[0139] Example 5

[0140] Based on the same inventive concept as other embodiments, this embodiment introduces a computer device, including: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the above-described wake model parameterization optimization method based on historical wind farm operating data.

[0141] Example 6

[0142] Based on the same inventive concept as other embodiments, this embodiment introduces a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described wake model parameterization optimization method based on historical operating data of wind farms.

[0143] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0144] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0145] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0146] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0147] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other modifications under the guidance of the present invention, and these modifications are all within the protection scope of the present invention.

Claims

1. A parameterized optimization method for wake model based on historical operating data of wind farms, characterized in that, Includes the following steps: Historical operation data of wind farms are collected, and the collected historical operation data of wind farms are preprocessed to obtain training set and test set; Based on the Ishihara single-unit wake model, and combining the linear wind speed superposition method based on the wind turbine definition and the turbulence intensity superposition method considering wake interaction, an engineering empirical wake model is constructed. Establish the wake growth rate ratio coefficient in the engineering empirical wake model The optimization problem is solved using the particle swarm optimization algorithm, with the mean absolute error (MAE) between the predicted and measured power values ​​of the entire wind farm in the training set as the fitness function. By performing optimization, the optimized result is obtained. ; The optimized result It is applied to engineering empirical wake models for power prediction of wind farm clusters.

2. The method for parameterizing and optimizing the wake model based on historical wind farm operating data according to claim 1, characterized in that, After optimization After being applied to the engineering experience wake model, the following steps are also included: inputting the data from the test set into the engineering experience wake model to obtain the predicted power of the wind farm.

3. The method for parameterizing and optimizing the wake model based on historical wind farm operating data according to claim 1, characterized in that, The aforementioned wake model based on the Ishihara single-unit wake model, combined with a linear wind speed superposition method based on the wind turbine definition and a turbulence intensity superposition method considering wake interactions, constructs an engineering empirical wake model, including: The expression for the linear wind speed superposition method based on the wind turbine definition is: ； In the formula, It is the wind speed of the target wind turbine. It is the inflow wind speed of the wind farm. It refers to the number of upstream wind turbine units; It is the serial number of the upstream wind turbine unit. It is an upstream wind turbine Wind speed; the superposition order is calculated sequentially from upstream to downstream in the windward direction; The expression for the superposition method of turbulence intensity of the wake interaction is: ； ； In the formula, It is the turbulence intensity; It is the intensity of environmental turbulence; It is the wake-added turbulence.

4. The wake model parameterization optimization method based on historical wind farm operation data according to claim 3, characterized in that, Introducing the wake growth rate Select proportionality coefficient To optimize the object, then The calculation method is as follows: ； In the formula, This is the thrust coefficient of the yaw wind turbine, which is a corrected parameter. It is the intensity of environmental turbulence; because The calculation method causes the shape of the Gaussian distribution to change, while the maximum value remains unchanged. Therefore, it is necessary to correct the velocity loss and additional turbulence in the wake formula. The corrected formula for calculating velocity loss is as follows: ； In the formula, It is after the wind turbine Reduced velocity standard deviation at position; It is the wind speed at the center of the wind turbine; It is the offset of the wake centerline; It refers to the wheel hub height; Indicates the diameter of the wind turbine; , , It is a parameter; It is the radial distance; The corrected formula for the additional turbulence calculation method is as follows: ； in, It is after the wind turbine The standard deviation of the velocity at position; It is the intensity of environmental turbulence; It is a turbulence intensity correction term; , , , , It is a parameter.

5. The method for parameterizing and optimizing the wake model based on historical wind farm operating data according to claim 1, characterized in that, The expression for the fitness function is: ； In the formula, These are measured values, corresponding to the measured power of the wind farm; It is a predicted value, corresponding to the predicted power of the wind farm, which is output after being calculated by the engineering experience wake model; This refers to the number of samples, corresponding to the number of wind turbine units. , .

6. The method for parameterizing and optimizing the wake model based on historical wind farm operating data according to claim 1, characterized in that, The pair Optimize to obtain the optimized version ,include: With the objective of minimizing the mean absolute error between predicted and measured power values, Optimize to obtain the optimized version Specifically: Optimize parameters When the time is right, the optimization problem is as follows: ； in, Represents the fitness function; Indicates constraints; These are measured values ​​of wind farm power. The predicted power of the wind farm is calculated using a wind farm group power calculation method based on an engineering experience wake model.

7. A parameterized optimization system for wake model based on historical wind farm operating data, characterized in that, The method for parametric optimization of wake model based on historical wind farm operation data as described in any one of claims 1 to 6 includes: The data acquisition module is configured to collect historical operating data of wind farms, preprocess the collected historical operating data of wind farms, and obtain training sets and test sets. The engineering experience wake model construction module is configured to construct an engineering experience wake model based on the Ishihara single-machine wake model, combining the linear wind speed superposition method based on the wind turbine definition and the turbulence intensity superposition method considering wake interaction. The optimization solution module is configured to establish the proportion coefficient of the wake growth rate in the engineering empirical wake model. The optimization problem is solved using the particle swarm optimization algorithm, with the mean absolute error (MAE) between the predicted and measured power values ​​of the entire wind farm in the training set as the fitness function. By performing optimization, the optimized result is obtained. ; The prediction module is configured to use the optimized result It is applied to engineering empirical wake models for power prediction of wind farm clusters.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the wake model parameterization optimization method based on historical operating data of wind farms as described in any one of claims 1 to 6.

9. A computer device, characterized in that, include: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the steps of the wake model parameterization optimization method based on historical wind farm operating data as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that: When the computer program is executed by the processor, it implements the steps of the wake model parameterization optimization method based on historical operating data of wind farms as described in any one of claims 1 to 6.

Images