Modeling method and turbulence prediction method for wind turbine airfoil turbulence model

Abstract

The invention discloses a modeling method and a turbulence prediction method of a wind turbine airfoil turbulence model. The modeling method comprises the following steps: acquiring an incoming flow working condition and a geometric shape of a wind turbine airfoil, performing grid division, and calculating a wall surface distance; constructing an RANS turbulence model, and calculating a flow field data prediction value of each grid point; correcting the flow field data predicted value by adopting a data assimilation algorithm to obtain a flow field data high confidence value; calculating a vortex viscosity correction factor and turbulent flow characteristic data of each grid point, and forming a characteristic set of the grid points by using the wall surface distances of the grid points and the turbulent flow characteristic data; a symbolic regression algorithm is adopted to extract a vortex viscosity correction factor function expression, and a data assimilation algorithm is adopted to correct the vortex viscosity correction factor function expression; and correcting the RANS turbulence model by adopting the corrected vortex viscosity correction factor function expression to obtain the wind turbine airfoil turbulence model. The wind turbine airfoil turbulence model constructed by the method can describe the wind turbine airfoil turbulence more accurately, and the prediction precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fluid dynamics, and in particular to a wind turbine airfoil turbulence modeling method and a turbulence prediction method. Background Art

[0002] Turbulence is one of the most complex phenomena in fluid mechanics. Its randomness, multi-scale and nonlinear characteristics make its simulation and prediction important in the fields of aviation, aerospace, energy, etc. The numerical calculation methods for turbulence research can be divided into: the direct numerical simulation (DNS) method that simulates the flow of all scales in the flow field by numerical solution, the large eddy simulation (LES) method that only directly simulates large structures in the flow, and the Reynolds-averaged NS equations (RANS) method that simulates the time-averaged flow of the flow field. Among them, the DNS method and the LES method are the most accurate and can obtain accurate information about the flow field, but the calculation is very time-consuming and the required computing resources are often large. Although the RANS method is not as accurate as the first two methods, it is short in calculation time and requires fewer computing resources. It is a common practice in the industry to predict complex flow field structures.

[0003] At present, the prediction of wind turbine airfoil turbulence can be done by DNS, LES and RANS methods. However, DNS and LES methods are time-consuming and require large computing resources. The existing computing resources cannot meet the needs of numerical simulation of wind turbine airfoil turbulence under high Reynolds numbers. Therefore, RANS method is a common method for predicting wind turbine airfoil turbulence under high Reynolds numbers. However, RANS method works better under low and medium Reynolds numbers. However, under high Reynolds numbers, due to the complexity and nonlinearity of the flow, RANS method often cannot make accurate predictions, resulting in deviations between the prediction results and the actual situation.

[0004] The prediction results of the RANS method are highly dependent on the accuracy of the RANS turbulence model. However, as the computational conditions of complex geometric configurations and high Reynolds number flows become increasingly demanding, the traditional RANS turbulence model can no longer meet the requirements of computational accuracy for complex flow fields, and the simulation accuracy of turbulence under high Reynolds numbers is low, resulting in inaccurate prediction results of the RANS method based on the RANS turbulence model for predicting wind turbine airfoil turbulence under high Reynolds numbers. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a wind turbine airfoil turbulence modeling method and a turbulence prediction method. The wind turbine airfoil turbulence model constructed by the present invention can more accurately describe the wind turbine airfoil turbulence and improve the prediction accuracy of the wind turbine airfoil turbulence.

[0006] In order to solve the above problems, the present invention adopts the following technical solutions:

[0007] A method for modeling a wind turbine airfoil turbulence model of the present invention comprises the following steps:

[0008] S1: Obtain the incoming flow condition and the geometric shape of the wind turbine airfoil, mesh the wind turbine airfoil, and calculate the wall distance of each grid point;

[0009] S2: constructing a RANS turbulence model, and calculating the predicted value of the flow field data of each grid point of the wind turbine airfoil under the incoming flow condition according to the RANS turbulence model, wherein the flow field data includes the eddy viscosity coefficient and multiple other flow field characteristics;

[0010] S3: Use the data assimilation algorithm to correct the predicted value of the flow field data at each grid point to obtain a high confidence value of the flow field data at each grid point;

[0011] S4: Calculate the eddy viscosity correction factor of each grid point according to the predicted value of the eddy viscosity coefficient and the high confidence value of the eddy viscosity coefficient of each grid point;

[0012] S5: Calculate turbulence characteristic data of each grid point according to the high confidence value of the flow field data of each grid point, wherein the turbulence characteristic data includes a number of characteristics that can represent the law of turbulent flow, and form a feature set of the grid point by combining the wall distance and the turbulence characteristic data of the grid point;

[0013] S6: According to the eddy viscosity correction factor of each grid point and the feature set, a symbolic regression algorithm is used to extract the mapping relationship between the eddy viscosity correction factor and the features in the feature set, and the function expression of the eddy viscosity correction factor is obtained;

[0014] S7: using a data assimilation algorithm to correct the eddy viscosity correction factor function expression, and obtaining a corrected eddy viscosity correction factor function expression;

[0015] S8: The modified eddy viscosity correction factor function expression is used to correct the Reynolds stress term function expression in the RANS turbulence model to obtain a new RANS turbulence model, and the new RANS turbulence model is used as the wind turbine airfoil turbulence model.

[0016] This method can effectively extract the spatial structure information of the turbulent flow field, optimize the eddy viscosity correction factor function expression obtained by symbolic regression through a bidirectionally coupled data assimilation algorithm, thereby correcting the Reynolds stress term in the traditional RANS turbulence model and generating a wind turbine airfoil turbulence model with high accuracy and stability. Compared with the traditional RANS turbulence model, the wind turbine airfoil turbulence model of this method can more accurately describe the wind turbine airfoil turbulence, improve the prediction accuracy of wind turbine airfoil turbulence, and show higher generalization ability when dealing with complex flows and different geometric shapes. The data assimilation algorithm in this method adopts the ensemble Kalman filter algorithm.

[0017] Preferably, the incoming flow conditions include incoming flow Mach number, Reynolds number, incoming flow temperature, and angle of attack.

[0018] Preferably, the multiple other flow field characteristics are fluid density, fluid velocity, and fluid pressure.

[0019] Preferably, step S3 comprises the following steps:

[0020] S31: obtaining high-precision observation values ​​of flow field data of each grid point of the wind turbine airfoil under the incoming flow condition;

[0021] S32: According to the high-precision observation value of the flow field data of each grid point, a data assimilation algorithm is used to correct the predicted value of the flow field data of each grid point to obtain a high-confidence value of the flow field data of each grid point.

[0022] The high-precision observation values ​​of flow field data can be physical quantities directly measured in experiments, or they can be values ​​calculated using the DNS method or the LES method.

[0023] Preferably, the calculation formula of the eddy viscosity correction factor of the pth grid point in step S4 is as follows, l≤p≤m, and m is the number of grid points:

[0024]

[0025] Among them, β p is the eddy viscosity correction factor of the pth grid point, μ t _DA_p is the high confidence value of the eddy viscosity coefficient at the pth grid point, μ t _SA_p is the predicted value of the eddy viscosity coefficient at the pth grid point.

[0026] Preferably, the turbulence characteristic data include one or more of vorticity, strain rate tensor, rotation rate tensor, pressure gradient, RANS equation invariant, and RANS equation working variable. The RANS equation invariant includes the trace of the square of the strain rate tensor and / or the trace of the square of the rotation rate tensor.

[0027] Preferably, the eddy viscosity correction factor function expression in step S6 is:

[0028] β=f(A1,A2…A n ),

[0029] Among them, β is the eddy viscosity correction factor, n is the number of features in the feature set that have a mapping relationship with the eddy viscosity correction factor, and A q is the i-th feature that has a mapping relationship with the eddy viscosity correction factor, l≤q≤n.

[0030] Preferably, step S7 comprises the following steps:

[0031] S71: Decompose the eddy viscosity correction factor function expression into a linear part and a nonlinear part. The formula is as follows:

[0032] β=β x +β y ,

[0033]

[0034] Among them, β x is the linear part, β y is the nonlinear part, a q For feature A q The corresponding linear coefficient, b q For feature A q The corresponding nonlinear coefficient, Ψ(A q ) is the description feature A q Function of nonlinear relationship with eddy viscosity correction factor;

[0035] S72: According to the high-precision observation value of the flow field data at each grid point, the linear coefficient of the linear part and the nonlinear coefficient of the nonlinear part are corrected by using a data assimilation algorithm to obtain a corrected eddy viscosity correction factor function expression.

[0036] Preferably, the method for correcting the Reynolds stress term function expression in the RANS turbulence model by using the corrected eddy viscosity correction factor function expression in step S8 is as follows:

[0037] Multiply the eddy viscosity coefficient in the Reynolds stress term function expression by the corrected eddy viscosity correction factor function expression to obtain the new Reynolds stress term function expression:

[0038]

[0039] Among them, τ ij is the Reynolds stress term, i and j represent the coordinate components in the Cartesian coordinate system, μ t is the eddy viscosity coefficient, is the average velocity component of the fluid in the i direction, is the average velocity component of the fluid in the j direction, x i is the coordinate in the i direction, x j is the coordinate in the j direction, k is the turbulent kinetic energy, δ ij is the Kronecker symbol.

[0040] A method for predicting turbulence on a wind turbine airfoil of the present invention comprises the following steps:

[0041] F1: Obtain the incoming flow conditions and the geometric shape of the wind turbine airfoil, and perform mesh division on the wind turbine airfoil;

[0042] F2: Calculate the predicted value of the flow field data of each grid point of the wind turbine airfoil under the incoming flow condition according to the wind turbine airfoil turbulence model, wherein the wind turbine airfoil turbulence model is modeled using the above-mentioned wind turbine airfoil turbulence model modeling method.

[0043] The beneficial effects of the present invention are: it can effectively extract the spatial structure information of the turbulent flow field, optimize the eddy viscosity correction factor function expression obtained by symbolic regression through a bidirectionally coupled data assimilation algorithm, thereby correcting the Reynolds stress term in the traditional RANS turbulence model, and generating a wind turbine airfoil turbulence model with high accuracy and stability. Compared with the traditional RANS turbulence model, the wind turbine airfoil turbulence model of this method can more accurately describe the wind turbine airfoil turbulence, improve the prediction accuracy of the wind turbine airfoil turbulence, and show higher generalization ability when dealing with complex flows and different geometric shapes. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a flow chart of an embodiment;

[0045] Figure 2 is a streamline diagram of the S809 airfoil flow field calculated by the SA turbulence model in the example of the embodiment;

[0046] Figure 3 is a flow field streamline diagram of an S809 airfoil obtained by calculating the turbulence model of a wind turbine airfoil in an example of an embodiment;

[0047] Figure 4 It is a comparison chart of the prediction results of the lift coefficient of the S809 airfoil as a function of the angle of attack in the example of the embodiment. DETAILED DESCRIPTION

[0048] The technical solution of the present invention is further specifically described below through embodiments and in conjunction with the accompanying drawings.

[0049] Embodiment: A wind turbine airfoil turbulence modeling method of this embodiment is as follows: Figure 1 As shown, the following steps are included:

[0050] S 1: Obtain the incoming flow conditions and the geometric shape of the wind turbine airfoil, mesh the wind turbine airfoil, and calculate the wall distance of each grid point. The incoming flow conditions include the incoming flow Mach number, Reynolds number, incoming flow temperature, and angle of attack.

[0051] S2: Construct a RANS turbulence model, and use a CFD solver based on the RANS turbulence model to calculate the predicted value of the flow field data at each grid point of the wind turbine airfoil under the incoming flow condition. The flow field data includes eddy viscosity coefficient, fluid density, fluid velocity, and fluid pressure.

[0052] S3: The data assimilation algorithm is used to correct the predicted value of the flow field data at each grid point to obtain a high confidence value of the flow field data at each grid point.

[0053] Step S3 includes the following steps:

[0054] S31: obtaining high-precision observation values ​​of flow field data of each grid point of the wind turbine airfoil under the incoming flow condition;

[0055] S32: According to the high-precision observation value of the flow field data of each grid point, a data assimilation algorithm is used to correct the predicted value of the flow field data of each grid point to obtain a high-confidence value of the flow field data of each grid point.

[0056] The high-precision observation value of the flow field data can be a physical quantity directly measured by the experiment, or a value calculated by the DNS method or the LES method. The data assimilation algorithm adopts the ensemble Kalman filter algorithm. In step S32, the Kalman gain is calculated according to the predicted value of the flow field data of each grid point and the high-precision observation value of the flow field data. In the process of coupling the calculation of the Kalman gain, the model parameters of the RANS turbulence model are corrected, and the high-confidence value of the flow field data is calculated according to the corrected RANS turbulence model.

[0057] S4: Calculate the eddy viscosity correction factor of each grid point based on the predicted value of the eddy viscosity coefficient and the high confidence value of the eddy viscosity coefficient at each grid point.

[0058] The calculation formula of the eddy viscosity correction factor of the pth grid point is as follows, l≤p≤m, m is the number of grid points:

[0059]

[0060] Among them, β p is the eddy viscosity correction factor of the pth grid point, μ t _DA_p is the high confidence value of the eddy viscosity coefficient at the pth grid point, μ t _SA_p is the predicted value of the eddy viscosity coefficient at the pth grid point.

[0061] S5: Calculate the turbulence characteristic data of each grid point according to the high confidence value of the flow field data of each grid point. The turbulence characteristic data include vorticity, strain rate tensor, rotation rate tensor, pressure gradient, trace of the square of strain rate tensor, trace of the square of rotation rate tensor, and one or more of the working variables of the RANS equation. The wall distance of the grid point and the turbulence characteristic data are combined into a feature set of the grid point.

[0062] For example: the turbulence characteristic data include vorticity, strain rate tensor, rotation rate tensor, pressure gradient, trace of the square of the strain rate tensor, and trace of the square of the rotation rate tensor. Then the feature set includes wall distance, vorticity, strain rate tensor, rotation rate tensor, pressure gradient, trace of the square of the strain rate tensor, and trace of the square of the rotation rate tensor, that is, there are seven features in the feature set.

[0063] S6: According to the eddy viscosity correction factor of each grid point and the feature set, the symbolic regression algorithm is used to extract the mapping relationship between the eddy viscosity correction factor and the features in the feature set, and the function expression of the eddy viscosity correction factor is obtained.

[0064] The function expression of eddy viscosity correction factor is:

[0065] β=f(A1,A2…A n ),

[0066] Among them, β is the eddy viscosity correction factor, n is the number of features in the feature set that have a mapping relationship with the eddy viscosity correction factor, and A q is the i-th feature that has a mapping relationship with the eddy viscosity correction factor, 1≤q≤n.

[0067] S7: The data assimilation algorithm is used to correct the eddy viscosity correction factor function expression to obtain a corrected eddy viscosity correction factor function expression.

[0068] Step S7 includes the following steps:

[0069] S71: Decompose the eddy viscosity correction factor function expression into a linear part and a nonlinear part. The formula is as follows:

[0070] β=β x +β y ,

[0071]

[0072] Among them, β x is the linear part, β y is the nonlinear part, a q For feature A q The corresponding linear coefficient, b q For feature A q The corresponding nonlinear coefficient, Ψ(A q ) is the description feature A q Function of nonlinear relationship with eddy viscosity correction factor;

[0073] S72: According to the high-precision observation value of the flow field data at each grid point, the linear coefficient of the linear part and the nonlinear coefficient of the nonlinear part are corrected by using a data assimilation algorithm to obtain a corrected eddy viscosity correction factor function expression.

[0074] The data assimilation algorithm adopts the ensemble Kalman filter algorithm. The eddy viscosity correction factor function expression is decomposed into a linear part and a nonlinear part, which reduces the system dimension. The uncorrected eddy viscosity correction factor function expression is introduced into the RANS equation for numerical calculation (that is, the eddy viscosity coefficient in the Reynolds stress term function expression is multiplied by the uncorrected eddy viscosity correction factor function expression). The ensemble Kalman filter algorithm consistent with step S3 is adopted, and the linear coefficient and nonlinear coefficient of the eddy viscosity correction factor function expression in the RANS equation are iteratively corrected using the high-precision observation value of the flow field data to obtain the corrected eddy viscosity correction factor function expression.

[0075] S8: The modified eddy viscosity correction factor function expression is used to correct the Reynolds stress term function expression in the RANS turbulence model to obtain a new RANS turbulence model, and the new RANS turbulence model is used as the wind turbine airfoil turbulence model.

[0076] The Reynolds stress term function expression in the original RANS turbulence model is as follows:

[0077]

[0078] The method of using the modified eddy viscosity correction factor function expression to correct the Reynolds stress term function expression in the RANS turbulence model is as follows:

[0079] Multiply the eddy viscosity coefficient in the Reynolds stress term function expression by the corrected eddy viscosity correction factor function expression to obtain the new Reynolds stress term function expression:

[0080]

[0081] Among them, τ ij is the Reynolds stress term, i and j represent the coordinate components in the Cartesian coordinate system (which can represent the components in the X, Y, and Z directions respectively), μ t is the eddy viscosity coefficient, is the average velocity component of the fluid in the i direction, is the average velocity component of the fluid in the j direction, x i is the coordinate in the i direction, x j is the coordinate in the j direction, k is the turbulent kinetic energy, δ ij is the Kronecker symbol.

[0082] In this solution, the RANS turbulence model constructed in step S2 can be an SA turbulence model, a BL turbulence model, a k-∈ turbulence model, a k-ω turbulence model, etc. The features in the turbulence feature data can fully capture the main physical characteristics of the turbulent flow to ensure the accuracy and effectiveness of the subsequent symbolic regression.

[0083] This method first uses high-precision observations of flow field data and RANS turbulence model to obtain high confidence values ​​of flow field data through data assimilation, then obtains eddy viscosity correction factor based on eddy viscosity coefficient prediction value and high confidence value of eddy viscosity coefficient, constructs feature set, and uses symbolic regression algorithm to extract the mapping relationship between eddy viscosity correction factor and features in feature set, i.e. eddy viscosity correction factor function expression. The eddy viscosity correction factor function expression is decomposed into linear part and nonlinear part, and bidirectional coupling optimization is performed through data assimilation algorithm to obtain the corrected eddy viscosity correction factor function expression. Finally, the corrected eddy viscosity correction factor function expression is integrated back into RANS turbulence model, so that the generated wind turbine airfoil turbulence model can be stably and convergently applied to different situations.

[0084] This method can effectively extract the spatial structure information of the turbulent flow field, optimize the eddy viscosity correction factor function expression obtained by symbolic regression through a bidirectional coupled data assimilation algorithm, thereby correcting the Reynolds stress term in the traditional RANS turbulence model and generating a wind turbine airfoil turbulence model with high accuracy and stability. Compared with the traditional RANS turbulence model, the wind turbine airfoil turbulence model of this method can more accurately describe the wind turbine airfoil turbulence, improve the prediction accuracy of wind turbine airfoil turbulence, and show higher generalization ability when dealing with complex flows and different geometric shapes.

[0085] For example:

[0086] The wind turbine airfoil is S809, and the incoming flow condition is: the incoming flow Mach number is 0.2, and the Reynolds number is 2×10 6 , the incoming flow temperature is 288K and the angle of attack is 14.24°.

[0087] S1: Obtain the incoming flow conditions and the geometric shape of the S809 airfoil, use Pointwise software to mesh the S809 airfoil, and calculate the wall distance of each grid point;

[0088] S2: Construct a RANS turbulence model (the SA turbulence model is selected as the RANS turbulence model), and use the CFD solver to calculate the predicted value of the flow field data of each grid point of the S809 airfoil under the incoming flow condition according to the RANS turbulence model. The flow field data includes eddy viscosity coefficient, fluid density, fluid velocity, and fluid pressure;

[0089] S3: Using the ensemble Kalman filter algorithm to correct the predicted value of the flow field data at each grid point, and obtain a high confidence value of the flow field data at each grid point;

[0090] S4: Calculate the eddy viscosity correction factor of each grid point according to the predicted value of the eddy viscosity coefficient and the high confidence value of the eddy viscosity coefficient of each grid point;

[0091] S5: Calculate the turbulence characteristic data of each grid point according to the high confidence value of the flow field data of each grid point. The turbulence characteristic data includes the strain rate tensor S, the rotation rate tensor R, the trace Tr(S 2 ), the trace of the square of the rotation rate tensor Tr(R 2 ), RANS equation working variable r, the wall distance of the grid point and the turbulence characteristic data are combined into a feature set of the grid point, that is, the feature set includes the wall distance, the strain rate tensor S, the rotation rate tensor R, the trace of the square of the strain rate tensor Tr(S 2 ), the trace of the square of the rotation rate tensor Tr(R 2 ), RANS equation working variable r;

[0092] The wall distance is used to reflect the viscosity effect in wall turbulence, the strain rate tensor S is used to measure the strain intensity in the turbulent flow field, the rotation rate tensor R is used to measure the rotation intensity in the turbulent flow field, and the trace of the square of the strain rate tensor Tr(S 2 ) is used to quantify the total intensity of strain effects in turbulent flow fields. The trace of the square of the rotation rate tensor Tr(R 2 ) is used to quantify the total intensity of the rotation effect in the turbulent flow field, and the RANS equation working variable r is used to reflect the turbulent viscosity or kinetic energy;

[0093] S6: According to the eddy viscosity correction factor of each grid point and the feature set, the symbolic regression algorithm is used to extract the mapping relationship between the eddy viscosity correction factor and the features in the feature set, and the function expression of the eddy viscosity correction factor is obtained:

[0094] β=0.51017+d 2 (1-tanh(d))×S,

[0095] Among them, β is the eddy viscosity correction factor, d is the wall distance, and S is the strain rate tensor;

[0096] S7: The data assimilation algorithm is used to correct the eddy viscosity correction factor function expression, and the corrected eddy viscosity correction factor function expression is obtained:

[0097] β=0.4062+1.0027×d 2 (1-tanh(d))×S;

[0098] S8: The modified eddy viscosity correction factor function expression is used to correct the Reynolds stress term function expression in the RANS turbulence model to obtain a new RANS turbulence model, and the new RANS turbulence model is used as the wind turbine airfoil turbulence model;

[0099] The modified Reynolds stress term function expression is as follows:

[0100]

[0101] The SA turbulence model and the above-mentioned wind turbine airfoil turbulence model are used to calculate the S809 airfoil under the incoming flow condition (the incoming flow Mach number is 0.2 and the Reynolds number is 2×10 6 , the flow field streamline diagram under the condition of the incoming flow temperature of 288K and the angle of attack of 14.24°, the flow field streamline diagram of S809 airfoil calculated by SA turbulence model is as follows Figure 2 As shown, the S809 airfoil flow field streamline diagram calculated by the above wind turbine airfoil turbulence model is as follows Figure 3 As shown, Figure 2 , Figure 3 The horizontal axis is the X-axis of the Cartesian coordinate system, the vertical axis is the Y-axis of the Cartesian coordinate system, and u is the fluid velocity. Figure 2 , Figure 3 It can be seen that the eddy viscosity field distribution area of ​​the S809 airfoil flow field streamline diagram calculated by the wind turbine airfoil turbulence model in the separation zone is significantly increased, indicating that the wind turbine airfoil turbulence model enhances the prediction of turbulence intensity, and the wind turbine airfoil turbulence model can predict the vortex that the SA turbulence model fails to identify, indicating that the wind turbine airfoil turbulence model can identify more complex flow phenomena more accurately.

[0102] The calculation state is selected as the incoming flow Mach number is 0.2 and the Reynolds number is 2×10 6 , the incoming flow temperature is 288K, the turbulence prediction method based on the SA turbulence model and the turbulence prediction method based on the above wind turbine airfoil turbulence model are used to predict the flow field data of the S809 airfoil at different angles of attack respectively, and the lift coefficient of the S809 airfoil is calculated according to the predicted flow field data. Figure 4 shown. Figure 4 The horizontal axis is the angle of attack α, and the vertical axis is the lift coefficient C L , curve Exp is the curve of lift coefficient varying with angle of attack obtained by experimental measurement, curve SA is the curve of lift coefficient varying with angle of attack corresponding to the SA turbulence model, and curve DASR is the curve of lift coefficient varying with angle of attack corresponding to the above-mentioned wind turbine airfoil turbulence model.

[0103] from Figure 4 It can be seen that the lift coefficient of curve SA is higher at a larger angle of attack, and curve DASR is closer to curve Exp obtained by experimental measurement, especially at a larger angle of attack, showing a trend similar to curve Exp. When the angle of attack reaches about 15 degrees, curve DASR and curve Exp almost coincide, which shows that the turbulence model based on the above wind turbine airfoil is more accurate in predicting the lift coefficient before and after stall. Figure 4 The accuracy and stability of the wind turbine airfoil turbulence model modeled by the wind turbine airfoil turbulence model modeling method of this embodiment can be verified.

[0104] A wind turbine airfoil turbulence prediction method according to this embodiment includes the following steps:

[0105] F1: Obtain the incoming flow conditions and the geometric shape of the wind turbine airfoil, and perform mesh division on the wind turbine airfoil;

[0106] F2: Based on the wind turbine airfoil turbulence model, a CFD solver is used to calculate the predicted value of the flow field data of each grid point of the wind turbine airfoil under the incoming flow condition. The wind turbine airfoil turbulence model is modeled using the above-mentioned wind turbine airfoil turbulence model modeling method.

Claims

1. A method for modeling a wind turbine airfoil turbulence model, characterized in that: The following steps are involved: S1: Obtain the incoming flow condition and the geometric shape of the wind turbine airfoil, mesh the wind turbine airfoil, and calculate the wall distance of each grid point; S2: constructing a RANS turbulence model, and calculating the predicted value of the flow field data of each grid point of the wind turbine airfoil under the incoming flow condition according to the RANS turbulence model, wherein the flow field data includes the eddy viscosity coefficient and multiple other flow field characteristics; S3: Use the data assimilation algorithm to correct the predicted value of the flow field data at each grid point to obtain a high confidence value of the flow field data at each grid point; S4: Calculate the eddy viscosity correction factor of each grid point according to the predicted value of the eddy viscosity coefficient and the high confidence value of the eddy viscosity coefficient of each grid point; S5: Calculate turbulence characteristic data of each grid point according to the high confidence value of the flow field data of each grid point, wherein the turbulence characteristic data includes a number of characteristics that can represent the law of turbulent flow, and form a feature set of the grid point by combining the wall distance and the turbulence characteristic data of the grid point; S6: According to the eddy viscosity correction factor of each grid point and the feature set, a symbolic regression algorithm is used to extract the mapping relationship between the eddy viscosity correction factor and the features in the feature set, and the function expression of the eddy viscosity correction factor is obtained; S7: using a data assimilation algorithm to correct the eddy viscosity correction factor function expression, and obtaining a corrected eddy viscosity correction factor function expression; S8: The modified eddy viscosity correction factor function expression is used to correct the Reynolds stress term function expression in the RANS turbulence model to obtain a new RANS turbulence model, and the new RANS turbulence model is used as the wind turbine airfoil turbulence model.

2. A method for modeling a wind turbine airfoil turbulence model according to claim 1, characterized in that: The incoming flow conditions include the incoming flow Mach number, Reynolds number, incoming flow temperature, and angle of attack.

3. A method for modeling a wind turbine airfoil turbulence model according to claim 1, characterized in that: The other flow field characteristics are fluid density, fluid velocity, and fluid pressure.

4. A method for modeling a wind turbine airfoil turbulence model according to claim 1, 2 or 3, characterized in that: The step S3 comprises the following steps: S31: obtaining high-precision observation values ​​of flow field data of each grid point of the wind turbine airfoil under the incoming flow condition; S32: According to the high-precision observation value of the flow field data of each grid point, a data assimilation algorithm is used to correct the predicted value of the flow field data of each grid point to obtain a high-confidence value of the flow field data of each grid point.

5. A method for modeling a wind turbine airfoil turbulence model according to claim 1, 2 or 3, characterized in that: The calculation formula of the eddy viscosity correction factor of the pth grid point in step S4 is as follows, 1≤p≤m, m is the number of grid points: Among them, β p is the eddy viscosity correction factor of the pth grid point, μ t_DA_p is the high confidence value of the eddy viscosity coefficient at the pth grid point, μ t_SA_p is the predicted value of the eddy viscosity coefficient at the pth grid point.

6. According to a method for modeling a wind turbine airfoil turbulence model according to claim 1, 2 or 3, the turbulence characteristic data includes one or more of vorticity, strain rate tensor, rotation rate tensor, pressure gradient, RANS equation invariant, and RANS equation working variable.

7. According to the method for modeling a wind turbine airfoil turbulence model according to claim 4, the function expression of the eddy viscosity correction factor in step S6 is: β=f(A1,A2…A n ), in, β is the eddy viscosity correction factor, n is the number of features in the feature set that have a mapping relationship with the eddy viscosity correction factor, A q is the i-th feature that has a mapping relationship with the eddy viscosity correction factor, 1≤q≤n.

8. According to the method for modeling a wind turbine airfoil turbulence model as claimed in claim 7, the step S7 comprises the following steps: S71: Decompose the eddy viscosity correction factor function expression into a linear part and a nonlinear part. The formula is as follows: β＝β x +β y ， Among them, β x is the linear part, β y is the nonlinear part, a q Feature A q The corresponding linear coefficient, b q Feature A q The corresponding nonlinear coefficient, Ψ(A q ) is the description feature A q Function of nonlinear relationship with eddy viscosity correction factor; S72: According to the high-precision observation value of the flow field data at each grid point, the linear coefficient of the linear part and the nonlinear coefficient of the nonlinear part are corrected by using a data assimilation algorithm to obtain a corrected eddy viscosity correction factor function expression.

9. According to the method for modeling a wind turbine airfoil turbulence model in claim 8, the method for correcting the Reynolds stress term function expression in the RANS turbulence model using the corrected eddy viscosity correction factor function expression in step S8 is as follows: Multiply the eddy viscosity coefficient in the Reynolds stress term function expression by the corrected eddy viscosity correction factor function expression to obtain the new Reynolds stress term function expression: in, τ ij is the Reynolds stress term, i and j represent the coordinate components in the Cartesian coordinate system, μ t is the eddy viscosity coefficient, is the average velocity component of the fluid in the i direction, is the average velocity component of the fluid in the j direction, x i is the coordinate in the i direction, x j is the coordinate in the j direction, k is the turbulent kinetic energy, δ ij is the Kronecker symbol.

10. A method for predicting turbulence on a wind turbine airfoil, characterized in that: The following steps are involved: F1: Obtain the incoming flow conditions and the geometric shape of the wind turbine airfoil, and perform mesh division on the wind turbine airfoil; F2: Calculate the predicted value of the flow field data of each grid point of the wind turbine airfoil under the incoming flow condition based on the wind turbine airfoil turbulence model, wherein the wind turbine airfoil turbulence model is modeled using a wind turbine airfoil turbulence model modeling method described in any one of claims 1 to 9.

Images