Ultra-short-term wind power plant power prediction method based on physical data dual drive

Abstract

The invention relates to an ultra-short-term wind power plant power prediction method based on physical data dual drive, which belongs to the field of wind power plant power prediction, and comprises the following steps: S1, calculating the power of a downstream wind turbine generator; s2, establishing an objective function, calculating the optimal power of the downstream wind turbine generator, and recording the optimal power as P1; s3, obtaining a power discrete image of a real downstream wind turbine generator, clustering the power of the real downstream wind turbine generator to generate a plurality of clusters, then decomposing different clusters by adopting empirical mode decomposition (EMD), and obtaining a weight corresponding to each cluster; finally, obtaining a function component sum of the power of the downstream wind turbine generator closest to reality, and recording the function component sum as P2; and S4, generating a plurality of power prediction expressions of the downstream wind turbine generator, generating a feature matrix, substituting the feature matrix into each power prediction expression, finally obtaining a power prediction expression closest to the real power of the downstream wind turbine generator, predicting the power of the downstream wind turbine generator, and recording the power as P3. The method can improve the prediction accuracy.

Description

Technical Field

[0001] The present invention belongs to the field of wind farm power prediction, and in particular relates to an ultra-short-term wind farm power prediction method based on physical data dual drive. Background Art

[0002] Traditional wind farm power prediction methods mainly use physical prediction or data prediction. Physically driven prediction mainly uses tuning algorithms, while data-driven prediction mainly uses neural networks. Both methods have their own shortcomings in predicting wind power. The disadvantage of the physical driven model is that its prediction accuracy is limited by the setting of the model and parameters, and it is easy to be affected by model errors, resulting in prediction errors and inaccurate prediction results. The common disadvantage of data driven is that a large amount of historical data needs to be accumulated, otherwise the model lacks interpretability, and a single prediction result is prone to single biased prediction, which is far from the actual prediction result. Summary of the invention

[0003] Purpose: In view of at least one of the above technical problems, the present application provides an ultra-short-term wind farm power prediction method based on dual drive of physical data, which can predict wind power in the ultra-short term with better accuracy.

[0004] The technical solution adopted in this application is:

[0005] In a first aspect, the present invention provides an ultra-short-term wind farm power prediction method based on physical data dual drive, comprising the following steps:

[0006] S1: Calculate the power of downstream wind turbines based on the Gaussian wake model;

[0007] S2: Based on the power of the downstream wind turbines calculated in S1 and the actual power of the downstream wind turbines, an objective function is established, and the optimal power of the downstream wind turbines is iteratively calculated, which is recorded as P1;

[0008] S3: Obtain the power discrete image of the real downstream wind turbines, first cluster the power of the real downstream wind turbines to generate multiple clusters, then decompose the different clusters by using empirical mode decomposition (EMD), and obtain the weights corresponding to each cluster; finally, obtain the function component sum closest to the power of the real downstream wind turbines, recorded as P2;

[0009] S4: Use the PySR symbolic regression model to randomly generate power prediction expressions for several downstream wind turbines, use P1 and P2 to generate a feature matrix and substitute it into each power prediction expression, and finally obtain the power prediction expression that is closest to the actual power of the downstream wind turbine. Use this power prediction expression to predict the power of the downstream wind turbine, recorded as P3.

[0010] Furthermore, step S1 includes the following steps:

[0011] S1.1: Establish a Gaussian wake model and calculate the speed reduction of the wind turbine based on the Gaussian wake model; the formula is:

[0012]

[0013] Among them, u hub is the wind speed at the center of the wind turbine hub;

[0014] Δu(x,y,z) is the velocity reduction at the wind turbine (x,y,z); (x,y,z) is the coordinate of the wind turbine corresponding to the origin;

[0015] C T is the thrust coefficient of the wind turbine;

[0016] ε is the initial wake diameter;

[0017] k * is the wake growth rate;

[0018] D is the diameter of the wind turbine rotor;

[0019] z h is the wind turbine hub height;

[0020] β is the ratio of the initial wake area to the rotor swept area;

[0021] S1.2: Calculate the hub center wind speed of the downstream wind turbine; the formula is:

[0022]

[0023] Among them, Δu i (x ij ,y ij ,z ij ) is the speed reduction of the upstream i-th wind turbine, calculated by formula (1);

[0024] u hub,i is the wind speed at the hub of the i-th upstream wind turbine;

[0025] u hub,j is the wind speed at the hub center of the jth wind turbine in the downstream;

[0026] u0 is the inflow wind speed of the upstream wind turbine;

[0027] S1.3 Calculate the power of the downstream wind turbine using the formula:

[0028]

[0029] Among them, P hub,jis the output power of the downstream wind turbine;

[0030] u hub,j is the wind speed at the hub center of the jth wind turbine in the downstream;

[0031] N is the number of downstream wind turbines;

[0032] g(*) is the function of the output power of the downstream wind turbine with respect to wind speed, which is determined by the type of wind turbine.

[0033] Further, step S2 includes the following steps:

[0034] S2.1: Establish the objective function, set constraints, and iteratively calculate the optimal wake growth rate of the upstream wind turbine; the formula is:

[0035] k * (t) = k opt

[0036] min.Y=MAE(P r (t),P[k * (t)])

[0037] st0.001≤j opt ≤0.055 (4)

[0038] Where min.Y is the optimization target, the fitness function is used, st is the constraint condition, MAE(*) represents the mean absolute error function; P[k*(t)] represents the function of the power of the downstream wind turbine when the wake growth rate of the upstream wind turbine is k*(t) with respect to the time series, which is obtained by the combined calculation of equations (1)-(3); P r (t) is the function of the real value of the downstream wind turbine power with respect to the time series; k*(t) is the function of the wake growth rate with respect to the time series; k opt is the wake growth rate constraint value; according to the iteration of formula (4), the optimal wake growth rate sequence of the upstream wind turbine is obtained, which is denoted as k * best (t);

[0039] S2.2: k * best (t) is substituted into equations (1)-(3) to obtain the optimal power of the downstream wind turbine as a function of the time series, denoted as: P * best (t);

[0040] S2.3: P * best (t) and P r (t) is calculated using MAE, and the function of the MAE calculation result with respect to the time series is Pm (t);

[0041] If P m (t) converges, then record P at this time * best (t) is P1;

[0042] If P m If (t) does not converge, replace k*(t) and re-substitute it into equations (1)-(3) to obtain the optimal power of the downstream wind turbine as a function of the time series, denoted as P2 * best (t), P2 * best (t) and P r (t) is calculated using MAE, and the function of the calculation result with respect to the time series is denoted as P z (t), if P z (t) converges, then P2 * best (t) is P1; otherwise, replace k*(t) again and repeat S2.1-S2.3 until convergence, and so on to obtain P1.

[0043] Further, step S3 includes the following steps:

[0044] S3.1: Obtain the actual wind power data P of the downstream wind turbines r (t); cluster the wind power data using K clustering; and then determine the number of clusters in K clustering using the contour analysis method. Let the number of clusters given by the contour analysis method be K d , K clustering the real wind power data P of downstream wind turbines r (t) Cluster into K d clusters, each of which contains multiple wind power data;

[0045] S3.2: Use EMD to decompose different clusters according to different characteristic fluctuations, and divide each cluster into multiple function components according to different characteristic fluctuations;

[0046] S3.3: Calculate the error after EMD decomposition of each cluster, the formula is:

[0047]

[0048] Where, T j represents the error after EMD decomposition of the jth cluster;

[0049] Z represents the number of function components after EMD decomposition of the jth cluster;

[0050] P tirepresents the wind power value of the i-th function component;

[0051] S i represents the weight of the i-th function component;

[0052] P total Indicates the actual wind power;

[0053] S3.4: Compare the error sizes of all clusters after EMD decomposition, select the cluster with the smallest error, calculate the sum of the function components of the cluster with the smallest error, and record the sum of the function components of the cluster at this time as P2; the formula is:

[0054]

[0055] Where Zn represents the number of function components after EMD decomposition of the cluster with the smallest error.

[0056] Furthermore, step S4 includes the following steps:

[0057] S4.1: Let the feature matrix be X2. The expression of the feature matrix is:

[0058]

[0059] S4.2: The PySR symbolic regression model generates 1000 prediction result expressions, substitutes the feature matrix X2 into the 1000 prediction result expressions, and obtains 1000 prediction results;

[0060] S4.3: Compare the 1000 prediction results with the actual wind power data P r (t) After making a difference, compare and obtain the prediction result with the smallest difference, which is recorded as P min , then P min The corresponding prediction result expression is an expression for predicting wind power. The expression for predicting wind power is used to predict the wind power of the downstream wind turbines. The predicted wind power of the downstream wind turbines is recorded as P3.

[0061] In a second aspect, the present application provides an ultra-short-term wind farm power prediction device based on physical data dual drive, including a processor and a storage medium;

[0062] The storage medium is used to store instructions;

[0063] The processor is configured to operate according to the instructions to execute the method according to the first aspect.

[0064] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the method described in the first aspect when executed by a processor.

[0065] In a fourth aspect, the present application provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the method of the first aspect when executing the computer program.

[0066] Beneficial effects: This method combines physical drive and data drive to improve the accuracy of prediction. The real data generation process is often affected by characteristics such as data conversion or parameter drift. The introduction of the dual-drive fusion method can reduce the prediction error caused by parameter or model misidentification. Even when a single prediction result is biased, the fusion can produce an unbiased prediction result. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a flow chart of an ultra-short-term wind farm power prediction method based on physical data dual drive according to an embodiment of the present application;

[0068] Figure 2 The figure is a comparison chart of the results of physical-driven prediction, data-driven prediction, and physical-data dual-driven prediction according to an embodiment of the present application. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] In the description of this application, "several" means more than one, "more" means more than two, "greater than", "less than", "exceed", etc. are understood to exclude the number itself, and "above", "below", "within", etc. are understood to include the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0071] In the description of the present application, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0072] Embodiment 1: This embodiment provides an ultra-short-term wind farm power prediction method based on physical data dual drive, such as Figure 1 As shown, the specific steps include:

[0073] S1: Calculate the power of downstream wind turbines based on the Gaussian wake model.

[0074] The specific steps include:

[0075] S1.1: Establish a Gaussian wake model and calculate the speed reduction of the wind turbine based on the Gaussian wake model; the formula is:

[0076]

[0077] Among them, u hub is the wind speed at the center of the wind turbine hub;

[0078] Δu(x,y,z) is the velocity reduction at the wind turbine (x,y,z); (x,y,z) is the coordinate of the wind turbine corresponding to the origin;

[0079] C T is the thrust coefficient of the wind turbine;

[0080] ε is the initial wake diameter;

[0081] k * is the wake growth rate;

[0082] D is the diameter of the wind turbine rotor;

[0083] z h is the wind turbine hub height;

[0084] β is the ratio of the initial wake area to the rotor swept area;

[0085] S1.2: Calculate the hub center wind speed of the downstream wind turbine.

[0086] Because the wake of the upstream wind turbine will affect the downstream wind turbine, in order to evaluate the wind speed of the downstream wind turbine, this scheme adopts a linear superposition model to calculate the wind speed at the hub center of the downstream wind turbine.

[0087] Assume that the number of upstream wind turbines is n; the number of downstream wind turbine components is N;

[0088] Then the wind speed at the hub center of the jth wind turbine in the downstream is:

[0089]

[0090] Among them, Δu i (x ij ,y ij ,zij ) is the speed reduction of the upstream i-th wind turbine, calculated by formula (1);

[0091] u hub,i is the wind speed at the hub of the i-th upstream wind turbine;

[0092] u hub,j is the wind speed at the hub center of the jth wind turbine in the downstream;

[0093] u0 is the inflow wind speed of the upstream wind turbine;

[0094] S1.3: Calculate the power of the downstream wind turbine using the formula:

[0095]

[0096] Among them, P hub,j is the output power of the downstream wind turbine;

[0097] u hub,j is the wind speed at the hub center of the jth wind turbine in the downstream;

[0098] N is the number of downstream wind turbines;

[0099] g(*) is the function of the output power of the downstream wind turbine with respect to wind speed, which is determined by the type of wind turbine.

[0100] S2: Establish the objective function and find the optimal wake growth rate k of the upstream wind turbine * best (t), according to the optimal wake growth rate k of the upstream wind turbine * best (t) Obtain the optimal power of the downstream wind turbine as a function of time series P * best (t), judge P * best (t) Whether it converges. If so, set P * best (t) is recorded as P1; if it does not converge, the optimal wake growth rate of the upstream wind turbine is replaced, and the optimal power of the downstream wind turbine is recalculated as a function of the time series until convergence, and P2 is obtained at this time. * best (t), and recorded as P1.

[0101] The specific steps include:

[0102] S2.1: Establish the objective function, set constraints, and iteratively calculate the optimal wake growth rate of the upstream wind turbine;

[0103] k*(t)=k opt

[0104] min.Y=MAE(P r (t),P[k * (t)])

[0105] st0.001≤k opt ≤0.055 (4)

[0106] Where min.Y is the optimization target, the fitness function is used, st is the constraint condition, MAE(*) represents the mean absolute error function; P[k*(t)] represents the function of the power of the downstream wind turbine when the wake growth rate of the upstream wind turbine is k*(t) with respect to the time series, which is obtained by the combined calculation of equations (1)-(3); P r (t) is the function of the real value of the downstream wind turbine power with respect to the time series. k*(t) is the function of the wake growth rate with respect to the time series. opt is the wake growth rate constraint value.

[0107] According to the iterative calculation of formula (4), the optimal wake growth rate sequence of the upstream wind turbine can be obtained, that is, k * best (t).

[0108] S2.2: k * best (t) is substituted into equations (1)-(3) to obtain the optimal wind speed of the downstream wind turbine as a function of the time series, i.e., u hub,j,best (t); the optimal power of the downstream wind turbine is also obtained as a function of the time series, denoted as: P * best (t);

[0109] S2.3: P * best (t) and P r (t) is calculated using the mean absolute error function (MAE), and the function of the MAE calculation result with respect to the time series is P m (t);

[0110] If P m (t) converges, then record P at this time * best (t) is P1.

[0111] If P m If (t) does not converge, replace k*(t) and re-substitute it into equations (1)-(3) to obtain the optimal power of the downstream wind turbine as a function of the time series, denoted as P2 * best (t), P2 * best (t) and Pr (t) is calculated using the mean absolute error function (MAE), and the function of the calculation result with respect to the time series is denoted as P z (t), if P z (t) converges, then P2 * best (t) is P1, and so on, until P1 is obtained.

[0112] S3: Obtain the real wind power data P of the downstream wind turbines r (t), the K clustering algorithm is used to cluster the wind power data; the contour analysis method is used to determine the number of clusters in the K clustering; let the number of clusters be K d indivual.

[0113] For any cluster, empirical mode decomposition (EMD) is used to decompose the cluster and divide the historical wind power data into multiple functional components according to different characteristic fluctuations;

[0114] By randomly assigning weights, the sum of the function components is minimized to minimize the error between the sum of the function components and the historical wind power data of the downstream wind turbines. The sum of the function components at this time is recorded as P2.

[0115] The specific steps include:

[0116] S3.1: Obtain the actual wind power data P of the downstream wind turbines r (t); cluster the wind power data using K clustering; and then determine the number of clusters in K clustering using the contour analysis method. Let the number of clusters given by the contour analysis method be K d , then K clusters the real wind power data P of the downstream wind turbines r (t) Cluster into K d There are multiple clusters, each of which contains multiple wind power data.

[0117] S3.2: Use EMD to decompose different clusters according to different characteristic fluctuations. Each cluster is divided into multiple function components according to different characteristic fluctuations.

[0118] Assume that the sum of wind power data in a cluster is P total ; After decomposing the cluster EMD, there are Z function components, which are denoted as P t1 , P t2 , P t3 , P t4 ,……,P tZ ; After the cluster EMD decomposition, the weights corresponding to each function component generated by EMD are S1, S2, S3, S4, ..., S Z .

[0119] S3.3: Calculate the error after EMD decomposition of the cluster. The formula is:

[0120]

[0121] Where, T j represents the error after EMD decomposition of the jth cluster;

[0122] Z represents the number of function components after EMD decomposition of the jth cluster;

[0123] P ti represents the wind power value of the i-th function component;

[0124] S i represents the weight of the i-th function component;

[0125] P total Indicates the actual wind power.

[0126] S3.4: Compare the error sizes of all clusters after EMD decomposition, select the cluster with the smallest error, calculate the sum of the function components of the cluster with the smallest error, and record the sum of the function components of the cluster at this time as P2;

[0127] Assume that the cluster with the smallest error has Z n function components, denoted as P t1 , P t2 , P t3 , P t4 ,……,P tZn ; After the EMD decomposition of the cluster, the corresponding weights randomly generated by each function component are S1, S2, S3, S4, ..., S Zn .

[0128] Then the formula for P2 is:

[0129]

[0130] S4: Generate a prediction result expression by the PySR symbolic regression model according to the four arithmetic operations and the random combination of one-variable functions; establish a feature matrix, input the feature matrix into the prediction result expression of the PySR symbolic regression model, compare the prediction results of all prediction result expressions with P r (t) is compared with P r (t) The prediction result expression with the smallest error is used to predict wind power.

[0131] The specific steps include:

[0132] S4.1: Let the feature matrix be X2. The expression of the feature matrix is:

[0133]

[0134] S4.2: The PySR symbolic regression model generates a total of 1000 prediction result expressions; the first 5 expressions are:

[0135] Equation 1: ((abs(x2)^0.50392675)-((rem(x2-1.1401341,0.000869341)*x2)-tan(cos(-0.0024737646*x2)+1.5311451)))

[0136] Equation 2: ((abs(x2)^0.50392675)-((rem(x2,0.000869341)*x2)-tan(tan(cos(-0.0024737646*x2))+1.530304)))

[0137] Equation 3: ((abs(x2)^0.50392675)-((rem(x2,0.000869341)*x2)-tan(cos(-0.0024737646*x2)+1.530304)))

[0138] Equation 4: (sqrt(abs(x2))-((rem(x2,0.0008701484)*x2)-tan(cos(-0.0024737646*x2)+1.5337806)))

[0139] Equation 5: (sqrt(abs(x2))-((rem(x2,0.000869341)*x2)-tan((-0.0024735518*x2)-0.03545753)))

[0140] Substitute the feature matrix X2 in S4.1 into the expression in S4.2 to obtain 1000 prediction results;

[0141] S4.3: Compare the 1000 prediction results with the actual wind power data P r (t) After making a difference, compare and obtain the prediction result with the smallest difference, which is recorded as P min , then P min The corresponding prediction result expression is an expression for predicting wind power, which is subsequently used to predict the wind power of the downstream wind turbines. The predicted wind power of the downstream wind turbines is recorded as P3.

[0142] Figure 2 In the figure, the Original Data line indicates the actual operating power of the downstream wind turbines;

[0143] The pwr1 line represents the operating power of the downstream wind turbine under physical drive.

[0144] The pwr2 line represents the operating power of the downstream wind turbine driven by data.

[0145] The Prediction line represents the operating power of downstream wind turbines driven by both physics and data.

[0146] from Figure 2 It can be seen that, in a few cases, operating condition 2 is closer to the actual operating power than operating conditions 1 and 3, so the physical-data dual-driven prediction of the operating power of downstream wind turbines is more reasonable.

[0147] Embodiment 2: Based on Embodiment 1, this embodiment provides an ultra-short-term wind farm power prediction device based on physical data dual drive, including a processor and a storage medium;

[0148] The storage medium is used to store instructions;

[0149] The processor is used to operate according to the instruction to execute the method according to embodiment 1.

[0150] Embodiment 3: Based on Embodiment 1, this embodiment provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method described in Embodiment 1 is implemented.

[0151] Embodiment 4: Based on Embodiment 1, this embodiment provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the method described in Embodiment 1 when executing the computer program.

[0152] Embodiment 5: Based on Embodiment 1, this embodiment provides a computer program product, including a computer program, which implements the method described in Embodiment 1 when executed by a processor.

[0153] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0154] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0155] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0156] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0157] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for ultra-short-term wind farm power prediction based on dual-drive of physical data, characterized in that: The following steps are involved: S1: Calculate the power of downstream wind turbines based on the Gaussian wake model; S2: Based on the power of the downstream wind turbines calculated in S1 and the actual power of the downstream wind turbines, an objective function is established, and the optimal power of the downstream wind turbines is iteratively calculated, which is recorded as P1; S3: Obtain a discrete power image of the real downstream wind turbines, first cluster the real power of the downstream wind turbines to generate multiple clusters, then decompose the different clusters by using empirical mode decomposition (EMD), and obtain the weights corresponding to each cluster; Finally, the function component sum that is closest to the actual power of the downstream wind turbines is obtained, which is recorded as P2; S4: Use the PySR symbolic regression model to randomly generate power prediction expressions for several downstream wind turbines, use P1 and P2 to generate a feature matrix and substitute it into each power prediction expression, and finally obtain the power prediction expression that is closest to the actual power of the downstream wind turbine. Use this power prediction expression to predict the power of the downstream wind turbine, recorded as P3.

2. The ultra-short-term wind farm power prediction method based on physical data dual drive according to claim 1 is characterized in that: Step S1 includes the following steps: S1.1: Establish a Gaussian wake model and calculate the speed reduction of the wind turbine based on the Gaussian wake model; the formula is: Among them, u hub is the wind speed at the center of the wind turbine hub; Δu(x,y,z) is the velocity reduction at the wind turbine (x,y,z); (x,y,z) is the coordinate of the wind turbine corresponding to the origin; C T is the thrust coefficient of the wind turbine; ε is the initial wake diameter; k * is the wake growth rate; D is the diameter of the wind turbine rotor; z h is the wind turbine hub height; β is the ratio of the initial wake area to the rotor swept area; S1.2: Calculate the hub center wind speed of the downstream wind turbine; the formula is: Among them, Δu i (x ij ,y ij ,z ij ) is the speed reduction of the upstream i-th wind turbine, calculated by formula (1); u hub,i is the wind speed at the hub of the i-th upstream wind turbine; u hub,j is the wind speed at the hub center of the jth wind turbine in the downstream; u0 is the inflow wind speed of the upstream wind turbine; S1.3 Calculate the power of the downstream wind turbine using the formula: Among them, P hub,j is the output power of the downstream wind turbine; u hub,j is the wind speed at the hub center of the jth wind turbine in the downstream; N is the number of downstream wind turbines; g(*) is the function of the output power of the downstream wind turbine with respect to wind speed, which is determined by the type of wind turbine.

3. The ultra-short-term wind farm power prediction method based on physical data dual drive according to claim 1 is characterized in that: Step S2 includes the following steps: S2.1: Establish the objective function, set constraints, and iteratively calculate the optimal wake growth rate of the upstream wind turbine; the formula is: k * (t)=k opt min.Y=MAE(P r (t),P[k*(t)]) s.t.0.001≤k opt ≤0.055 (4) Where min.Y is the optimization target, the fitness function is used, st is the constraint condition, MAE(*) represents the mean absolute error function; P[k*(t)] represents the function of the power of the downstream wind turbine when the wake growth rate of the upstream wind turbine is k*(t) with respect to the time series, which is obtained by the combined calculation of equations (1)-(3); P r (t) is the function of the real value of the downstream wind turbine power with respect to the time series; k*(t) is the function of the wake growth rate with respect to the time series; k opt is the wake growth rate constraint value; according to the iteration of formula (4), the optimal wake growth rate sequence of the upstream wind turbine is obtained, which is denoted as k * best (t); S2.2: k * best (t) is substituted into equations (1)-(3) to obtain the optimal power of the downstream wind turbine as a function of the time series, denoted as: P * best (t); S2.3: P * best (t) and P r (t) is calculated using MAE, and the function of MAE calculation result with respect to time series is denoted as P m (t); If P m (t) converges, then record P at this time * best (t) is P1; If P m If (t) does not converge, replace k*(t) and re-substitute it into equations (1)-(3) to obtain the optimal power of the downstream wind turbine as a function of the time series, denoted as P2 * best (t), P2 * best (t) and P r (t) is calculated using MAE, and the function of the calculation result with respect to the time series is denoted as P z (t), if P z (t) converges, then P2 * best (t) is P1; otherwise, replace k*(t) again and repeat S2.1-S2.3 until convergence, and so on to obtain P1.

4. The ultra-short-term wind farm power prediction method based on physical data dual drive according to claim 1 is characterized in that: Step S3 includes the following steps: S3.1: Obtain the actual wind power data P of the downstream wind turbines r (t); cluster the wind power data using K clustering; and then determine the number of clusters in K clustering using the contour analysis method. Let the number of clusters given by the contour analysis method be K d , K clustering the real wind power data P of downstream wind turbines r (t) Cluster into K d clusters, each of which contains multiple wind power data; S3.2: Use EMD to decompose different clusters according to different characteristic fluctuations, and divide each cluster into multiple function components according to different characteristic fluctuations; S3.3: Calculate the error after EMD decomposition of each cluster; S3.4: Compare the error sizes of all clusters after EMD decomposition, select the cluster with the smallest error, calculate the sum of the function components of the cluster with the smallest error, and record the sum of the function components of the cluster at this time as P2.

5. The ultra-short-term wind farm power prediction method based on physical data dual drive according to claim 4 is characterized in that: S3.3: Calculate the error after EMD decomposition of each cluster, including: Where, T j represents the error after EMD decomposition of the jth cluster; Z represents the number of function components after EMD decomposition of the jth cluster; P ti represents the wind power value of the i-th function component; S i represents the weight of the i-th function component; P total Indicates the actual wind power.

6. The ultra-short-term wind farm power prediction method based on physical data dual drive according to claim 4 is characterized in that: In S3.4, the function components and P2 of the cluster with the smallest error are calculated, including: Where Zn represents the number of function components after EMD decomposition of the cluster with the smallest error, P ti represents the wind power value of the i-th function component; S i Represents the weight of the i-th function component.

7. The ultra-short-term wind farm power prediction method based on physical data dual drive according to claim 1 is characterized in that: Step S4 includes the following steps: S4.1: Let the feature matrix be X2. The expression of the feature matrix is: S4.2: The PySR symbolic regression model generates 1000 prediction result expressions, substitutes the feature matrix X2 into the 1000 prediction result expressions, and obtains 1000 prediction results; S4.3: Compare the 1000 prediction results with the actual wind power data P r (t) After making a difference, compare and obtain the prediction result with the smallest difference, which is recorded as P min , then P min The corresponding prediction result expression is an expression for predicting wind power. The expression for predicting wind power is used to predict the wind power of the downstream wind turbines. The predicted wind power of the downstream wind turbines is recorded as P3.

8. An ultra-short-term wind farm power prediction device based on physical data dual drive, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

Images