Lift-drag combined vertical axis wind turbine blade airfoil design method
Through parametric blade airfoil design, combined with deep learning agent model and particle swarm optimization algorithm, the combined vertical axis wind turbine blade airfoil is optimized, which solves the problems of multi-working condition adaptability and collaborative aerodynamic characteristics, and improves the energy capture efficiency of the wind turbine.
Patent Information
- Application Number
- CN202511107638.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-08
- Publication Date
- 2025-10-17
AI Technical Summary
In the prior art, the design of combined vertical axis wind turbine blade airfoils ignores the adaptability to multiple working conditions and the synergistic aerodynamic characteristics of lift-drag blades when pursuing the aerodynamic lift-drag ratio, resulting in design limitations.
A parametric blade airfoil design is adopted. Through NURBS curve parametric representation, deep learning agent model and particle swarm optimization algorithm, combined with multi-condition aerodynamic simulation model, the blade airfoil is optimized to improve energy capture efficiency.
It significantly improves the energy capture efficiency of the blade airfoil under different working conditions, enhances the overall performance of the wind turbine, and meets the energy supply needs of high-energy-consuming enterprises on land.
Smart Images
Figure CN120805722A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application belongs to the technical field of wind power generation, and particularly relates to a blade airfoil design method suitable for a lift-drag combined vertical axis wind turbine, and especially suitable for a distributed wind power system of a land-based high energy consumption enterprise. BACKGROUND
[0002] Compared with horizontal axis wind turbines, vertical axis wind turbines have the advantages of high operational stability, low power generation cost and no need for complex yawing devices, and are an important trend in the future development of the wind power industry. Considering the objective facts that lift-type vertical axis wind turbines have high efficiency but poor self-starting performance, and drag-type vertical axis wind turbines have good self-starting performance but low efficiency, a combined vertical axis wind turbine is created by combining the advantages of the two and overcoming their respective shortcomings, which has important significance for the energy supply of land-based high energy consumption enterprises and the sustainable development of the wind power industry. The optimization of the blade airfoil improves the energy capture efficiency of the lift-drag combined vertical axis wind turbine in the rated power wind speed range, and approaches or even reaches the efficiency level of pure lift-type vertical axis wind turbines, which plays an important role in the adjustment of national energy structure and the realization of energy saving and emission reduction targets. The existing inventions in the same field only focus on optimizing the aerodynamic lift-drag ratio of a single blade airfoil, and ignore the multi-working condition adaptability and the collaborative aerodynamic characteristics of lift blades and drag blades, which has limitations in actual application. SUMMARY
[0003] The present application provides a lift-drag combined vertical axis wind turbine blade airfoil design method to solve the technical problem that the blade airfoil of the combined vertical axis wind turbine in the prior art only pursues the aerodynamic lift-drag ratio in design, but ignores the multi-working condition adaptability and the collaborative aerodynamic characteristics of lift blades and drag blades.
[0004] To solve the above technical problems, the present application provides the following technical solutions:
[0005] The present application provides a lift-drag combined vertical axis wind turbine blade airfoil design method, comprising the following steps:
[0006] Parameterize the blade airfoil profile, generate an airfoil decision space through a sampling method, and obtain airfoil aerodynamic performance parameters based on a multi-working condition aerodynamic simulation model;
[0007] Based on the airfoil structural feature parameters and the aerodynamic performance parameters, an aerodynamic performance deep learning agent model of the airfoil decision space is constructed;
[0008] The aerodynamic performance deep learning agent model is used in combination with constraint conditions to establish a blade airfoil optimization model, and an optimization algorithm is used for global optimization to complete the airfoil profile design.
[0009] Preferably, the generating the airfoil decision space comprises the steps of: parameterizing the blade airfoil to obtain structural characteristic parameters; and sampling sample points in the structural characteristic parameters to generate the airfoil decision space.
[0010] Preferably, the obtaining the airfoil aerodynamic performance parameters comprises the steps of: dividing different operating conditions based on the unique aerodynamic characteristics of the combined vertical axis wind turbine; establishing a three-dimensional geometric model and a simulation model of the wind turbine, and performing grid division and fluid mechanics calculation on the simulation model to obtain the aerodynamic performance parameters.
[0011] Preferably, the airfoil aerodynamic performance parameters include aerodynamic torque, rated aerodynamic power and power coefficient.
[0012] Preferably, the constructing the airfoil decision space aerodynamic performance deep learning agent model comprises the steps of: obtaining a structural input vector of the airfoil decision space according to the structural characteristic parameters of the sample points; obtaining a response vector corresponding to the airfoil decision space according to the aerodynamic performance parameters of the sample points; and establishing the airfoil decision space aerodynamic performance deep learning agent model according to the structural input vector and the response vector of the airfoil decision space.
[0013] Preferably, the using the aerodynamic performance deep learning agent model to establish a blade airfoil optimization model in combination with constraint conditions, and using an optimization algorithm for global optimization to complete airfoil profile design comprises the steps of: establishing the blade airfoil optimization model with the maximum response value of the agent model representing the power coefficient as an objective function in combination with thickness and curvature constraint conditions, wherein the objective function is:
[0014] F(x)=Cp=DNN(x i ,tsr)
[0015] In the formula, x i is an airfoil structural characteristic parameter, and tsr is a tip speed ratio under different operating conditions.
[0016] Preferably, the establishing the blade airfoil optimization model with the maximum response value of the agent model representing the power coefficient as an objective function in combination with thickness and curvature constraint conditions further comprises the steps of: establishing the blade airfoil optimization model with the maximum response value of the agent model representing the power coefficient as the objective function in combination with thickness and curvature constraint conditions, and a mathematical model expression of the blade airfoil optimization model is:
[0017] Objiect:Maxmize F(x)=Cp=DNN(x i ,tsr)
[0018]
[0019] In the formula, t max is the relative maximum thickness of the airfoil, c is the maximum camber of the airfoil, L max is the position of the maximum camber of the airfoil on the chord length of the airfoil; if the optimization target value converges or reaches the set maximum number of iterations, the optimized airfoil is obtained.
[0020] Preferably, the blade airfoil optimization model further comprises the steps of: judging whether the optimized airfoil meets the airfoil accuracy requirement, if yes, outputting the optimized airfoil and ending the optimization process, and if not, increasing the sample points of the airfoil decision space, updating the surrogate model, until the airfoil accuracy requirement is met.
[0021] Preferably, the lift-drag combined vertical axis wind turbine blade airfoil design method further comprises the steps of: realizing parameterization characterization of the blade airfoil based on a NURBS curve parameterization method; obtaining the airfoil decision space based on a hyper-Latin cube sampling method; establishing the multi-working-condition aerodynamic simulation model in STARCCM+ software, generating structure grid and boundary layer grid, and setting boundary conditions and solving parameters, and performing aerodynamic flow field simulation and the aerodynamic performance parameter calculation.
[0022] Compared with the prior art, the present application has at least the following beneficial effects:
[0023] In the present application, the aerodynamic efficiency, multi-working-condition adaptability and lift-drag blade collaborative aerodynamic characteristics of the blade airfoil in one rotation of the wind turbine are comprehensively considered when designing the lift-drag combined vertical axis wind turbine blade airfoil, so that the energy capture efficiency of the blade airfoil under different working conditions is significantly improved. BRIEF DESCRIPTION OF DRAWINGS
[0024] In order to facilitate understanding of the technical solutions of the embodiments of the present application, the following briefly describes the drawings.
[0025] Figure 1 The flowchart of the lift-drag combined vertical axis wind turbine blade airfoil design method of the present application.
[0026] Figure 2 The contrast schematic diagram of the NACA0015 airfoil profile of the NURBS parameterization fitting of the present application.
[0027] Figure 3 The profile schematic diagram of the airfoil decision space of the present application.
[0028] Figure 4 The grid schematic diagram of the simulation model of the lift-drag combined vertical axis wind turbine of the present application.
[0029] Figure 5 The contrast schematic diagram of the optimized airfoil OPT-1 and the original standard airfoil of the present application.
[0030] Figure 6 Fig. 1 is a schematic diagram of the comparison between the optimized airfoil OPT-2 and the original standard airfoil.
[0031] Figure 7 Fig. 2 is a schematic diagram of the comparison between the optimized airfoil OPT-3 and the original standard airfoil. DETAILED DESCRIPTION
[0032] The application will be further described in detail below in conjunction with the accompanying drawings and examples; the examples are only used to exemplarily illustrate the implementation manners of the application, and do not constitute a limitation on the protection scope of the claims.
[0033] Due to the dramatic change of the attack angle of the blade airfoil within one rotation, the symmetric airfoil in the NACA airfoil database is usually selected according to the operating conditions and design indexes of the wind turbine. In this example, the standard airfoil NACA0015 is taken as the research object, and the operating conditions with the rated tip speed ratio of 2, 2.5 and 3 are selected for the aerodynamic design of the blade airfoil. The flow chart of the combined vertical axis wind turbine blade airfoil design method is shown in Fig. 1. Figure 1
[0034] Step (1): Parameterization of the standard airfoil. The commonly used blade airfoil characterization methods mainly include the Hicks-Henne parameterization method, the CST parameter method, the B-spline curve fitting method and the NURBS curve fitting method. NURBS curve is a widely used curve representation method in computer graphics and geometric modeling. It can accurately represent free curves and surfaces, and has high computational efficiency and stability, providing a powerful tool for modeling and optimization of complex geometric shapes. Here, NURBS curve is selected for parameterization of the blade airfoil. The mathematical expressions of the upper and lower surfaces of the lift-drag combined vertical axis wind turbine blade airfoil are as follows:
[0035]
[0036]
[0037] where P i is the control point, u is the upper surface curve parameter, and l is the lower surface curve parameter.
[0038] In the parameterization of this embodiment, in order to accurately control the shape of the upper and lower surfaces, 10 control points are selected for each of the upper and lower airfoils except the first and last two end points. The control point coordinates are shown in Table 1. The parameterization results of NACA0015 by NURBS curve are shown in Fig. 2, which can accurately represent the profile of the airfoil. Figure 2
[0039] Table 1 Control point coordinates of the upper and lower airfoils of the combined vertical axis wind turbine
[0040]
[0041] Step (2): Generating airfoil decision space. Hyper-Latin sampling is a special multidimensional stratified sampling method, which is widely used in real engineering. The distribution of sampling points is uniform, which can quickly achieve convergence. In this embodiment, the blade airfoil design variables are control point coordinates, and the generation of sample point set in the airfoil decision space is realized by Python programming.
[0042] Specifically, 50 groups of airfoil decision space sample points are generated for the design variable longitudinal coordinate with an upper and lower change of 10%. In this embodiment, the hyper-Latin sampling algorithm generates a blade airfoil decision space profile diagram as shown in Figure 3 .
[0043] Step (3): Establishment of aerodynamic performance simulation model of lift-drag combined vertical axis wind turbine. According to the structural characteristics of the combined vertical axis wind turbine, a suitable three-dimensional geometric model is established, which specifically includes the shape, size, installation angle of the lift-type and drag-type blades and their combination mode. The model should as far as possible reflect the actual structure of the wind turbine to improve the accuracy of subsequent CFD simulation.
[0044] STARCCM+ is a CAE software with extremely powerful functions, high integration and focus on multi-physical field engineering simulation. Based on STARCCM+, the boundary condition setting, mesh division and solver setting of the simulation model of the lift-drag combined vertical axis wind turbine are completed, and further, the aerodynamic torque, rated power and power coefficient are calculated. The mesh division result of the simulation model is shown in Figure 4 , and the boundary layer parameter setting of the lift-drag combined vertical axis wind turbine blade is shown in Table 2.
[0045] Table 2 Setting of boundary layer parameters of lift-drag combined vertical axis wind turbine blade
[0046]
[0047] Step (4): Establishment of deep learning proxy model. In aerodynamic optimization design, in order to reduce the calculation period of the optimization system and improve the search efficiency, a proxy model with simple structure, small calculation amount and high prediction accuracy is often introduced. Deep neural network (DNN) is a kind of feedforward neural network with multiple hidden layers in the field of machine learning, which can perform complex feature extraction and transformation on data to realize the nonlinear mapping between input data and output target. In this embodiment, DNN model is selected to perform nonlinear fitting on the structure input vector and response vector of the airfoil decision space.
[0048] Specifically, this embodiment builds a deep neural network model of the aerodynamic performance of a lift-drag combined vertical axis wind turbine based on the TensorFlow framework. In order to improve the prediction accuracy of the model, this embodiment trains three deep neural network models for three operating conditions. The model structure includes an input layer, three hidden layers, three Dropout layers, and an output layer. The input layer receives 20 structural characteristic parameters of the airfoil; the first hidden layer contains 128 neurons and uses the ReLU activation function; the second hidden layer contains 64 neurons and also uses the ReLU activation function; the third hidden layer contains 32 neurons and the activation function is ReLU; the coefficient of the three Dropout layers is 0.3; the output layer is the predicted power coefficient value, and the activation function is a linear function.
[0049] Step (5): Setting the objective function and constraints for blade airfoil optimization. Under normal operating conditions at low Reynolds numbers, increasing the blade's angle of attack within a suitable range can increase the blade's lift coefficient; increasing the blade's thickness can enable the wind turbine to start at lower wind speeds. In order to reduce the number of algorithm iterations and save time, the airfoil parameters must be constrained during the airfoil optimization process.
[0050] The mathematical expressions of the optimization model objective function and constraints are:
[0051] Object:Maxmize f(x)=Cp=DNN(x i ,tsr)
[0052]
[0053] Where, t max is the relative maximum thickness of the airfoil, c is the maximum camber of the airfoil, L max The chord length of the airfoil where the maximum camber of the airfoil occurs.
[0054] Step (6): Set up the blade airfoil optimization process of the particle swarm algorithm. As a global optimization algorithm based on swarm intelligence, the particle swarm optimization algorithm has the advantages of being easy to implement, having strong global search capabilities, fast convergence speed, and strong adaptability, and has an outstanding ability to handle complex problems. In this embodiment, the blade airfoil optimization process based on the particle swarm algorithm is as follows:
[0055] 1. Initialize the particle swarm: In airfoil optimization, each particle represents a potential solution for the airfoil. When initializing the particle swarm, a set of random airfoil parameters is generated as the particle's initial position. Also, a random initial velocity is generated for the particle, typically within a range that is determined by the airfoil parameter's range of variation.
[0056] Furthermore, it is necessary to initialize the individual extreme value P of the particle bestand the group extreme value G best .
[0057] 2. Fitness calculation: For each airfoil corresponding to a particle, a combined vertical axis wind turbine proxy model based on a deep neural network algorithm is used to predict aerodynamic performance and calculate fitness.
[0058] 3. Extreme value update: Update the current fitness value of each particle with its individual extreme value P best If the current fitness value is better, update P best is the current position. best In the equation, find the particle with the best fitness value and use its fitness value as the group extreme value G best This process ensures that the particles move towards a more optimal direction in subsequent iterations.
[0059] 4. Speed and Position Update: Based on the particle swarm algorithm's speed update formula, combined with the influence of inertia factors, individual extreme values, and group extreme values, the new speed of the particle in each airfoil parameter dimension is calculated. The new speed determines the particle's movement direction and step size in the solution space.
[0060] Furthermore, according to the position update formula, the position of the particle in each airfoil parameter dimension is adjusted according to the new speed to obtain a new airfoil parameter combination, that is, a new particle position.
[0061] 5. Loop termination condition judgment: Check whether the loop termination conditions are met, such as reaching the preset maximum number of iterations, the fitness value meeting the set accuracy requirements, or the fitness value of several consecutive generations has no significant improvement. If any termination condition is met, the iteration is stopped and the current population extreme value G is output. best The corresponding airfoil parameters are taken as the optimal solution; otherwise, return.
[0062] In this embodiment, a particle swarm optimization algorithm is used to perform parameter optimization, wherein: the particle swarm size N is set to 100; the individual learning factor c1 is set to 2; the social learning factor c2 is set to 2.1; and the maximum number of iterations U is set to 200.
[0063] Step (7): Verify the accuracy of the optimized airfoil. Perform a high-precision CFD simulation on the optimized airfoil to determine whether it meets the airfoil accuracy requirements. If so, output the optimized airfoil; if not, increase the sample points in the airfoil decision space and update the proxy model until the airfoil accuracy requirements are met.
[0064] Step (8): Comparison of the optimized airfoil with the standard airfoil. Due to the particularity of the structure of the combined vertical axis wind turbine, the shape of the blade airfoil directly affects the wind energy utilization rate of the wind turbine. Therefore, in the process of optimizing the structure of the blade of the combined vertical axis wind turbine, the main work is to complete the shape optimization design of the blade airfoil.
[0065] Specifically, the comparison of the profile of the optimized airfoil OPT-1 with the standard airfoil NACA0015 is shown in FIG. 1. Figure 5 The leading edge angle of the OPT-1 airfoil is reduced, the average thickness of the upper and lower airfoils is reduced, the maximum thickness of the airfoil is increased and the position is moved forward.
[0066] Specifically, the comparison of the profile of the optimized airfoil OPT-2 with the standard airfoil NACA0015 is shown in FIG. 2. Figure 6 The leading edge angle of the OPT-2 airfoil is almost unchanged, the maximum thickness of the upper airfoil is increased, the thickness of the lower airfoil is reduced, and the position of the maximum thickness of the airfoil is moved forward.
[0067] Specifically, the comparison of the profile of the optimized airfoil OPT-3 with the standard airfoil NACA0015 is shown in FIG. 3. Figure 7 The leading edge angle of the OPT-3 airfoil is reduced, the thickness of the upper airfoil is increased, the thickness of the lower airfoil is reduced, and the maximum thickness of the airfoil is increased but the position is almost unchanged.
[0068] Overall, the upper and lower airfoil profiles of the optimized airfoil are more complex than the standard airfoil, the maximum thickness of the upper airfoil is increased, and the average thickness of the lower airfoil is reduced.
[0069] The rated operating data of the combined vertical axis wind turbine with the three optimized airfoils are shown in Tables 3, 4 and 5. In this embodiment, the power coefficient of the OPT-1 airfoil is 0.449839, which is increased by 12.725% compared with the standard airfoil NACA0015; the power coefficient of the OPT-2 airfoil is 0.433982, which is increased by 7.395% compared with the standard airfoil NACA0015; the power coefficient of the OPT-3 airfoil is 0.375446, which is increased by 20.21% compared with the standard airfoil NACA0015.
[0070] Table 3 Rated operating data of the OPT-1 airfoil and the NACA0015 airfoil
[0071]
[0072] Table 4 Rated operating data of the OPT-2 airfoil and the NACA0015 airfoil
[0073]
[0074] Table 5 Rated operating data of the OPT-3 airfoil and the NACA0015 airfoil
[0075]
[0076] The present application comprehensively considers the aerodynamic efficiency, multi-working condition adaptability and lift-drag blade collaborative aerodynamic characteristics of the blade airfoil during one rotation of the wind turbine, and significantly improves the energy capture efficiency of the blade airfoil under different working conditions.
[0077] Compared with the prior art, the airfoil optimization design process based on the deep learning agent model has at least the following beneficial effects:
[0078] (1) The NURBS curve parameterization represents the airfoil, which is a significant advantage of the process. Compared with traditional methods such as Hicks-Henne and CST, NURBS curve can accurately fit complex airfoil profiles, accurately control airfoil shape through control point coordinates, and quantitatively adjust key parameters such as airfoil curvature and thickness. This accuracy ensures that airfoil optimization is within a reasonable range, avoiding deviations in the optimization direction caused by parameterization errors.
[0079] (2) The collaborative application of deep neural network agent model and particle swarm optimization algorithm realizes the balance between efficiency and accuracy. Deep neural network can accurately map the complex relationship between airfoil parameters and power coefficient due to its multi-layer nonlinear transformation capability, and can replace time-consuming CFD simulation to complete performance prediction. At the same time, the group optimization characteristics of particle swarm optimization algorithm are suitable for solving multi-dimensional airfoil parameter optimization problems. Through dynamic updating of individual extreme value and group extreme value, the optimal solution is quickly converged. This "deep learning agent model + intelligent optimization" architecture not only retains the accuracy advantage of CFD simulation, but also breaks through the computational efficiency bottleneck, making multi-working condition optimization of complex airfoils possible.
[0080] (3) The blade airfoil optimization model fully considers the lift-drag blade collaborative aerodynamic characteristics and engineering application requirements of the lift-drag combined vertical axis wind turbine under different working conditions. The objective function maximizes the power coefficient of the whole machine during one rotation of the wind turbine under different working conditions, considers the collaborative aerodynamic characteristics of the lift-drag blade under different working conditions, directly aims at improving the wind energy utilization rate, and meets the requirements of land-based high-energy-consuming enterprises for energy supply efficiency. The constraint conditions limit the maximum thickness, maximum curvature and its position of the airfoil, ensuring that the optimization result has high performance and manufacturability, avoiding the problem of disconnection between theoretical optimization and practical application, and providing a feasible technical solution for commercialization of the lift-drag combined vertical axis wind turbine.
[0081] The content described in the embodiments of the present application is only a list of implementation forms of the inventive concept, the protection scope of the present application should not be regarded as limited to the specific forms stated in the embodiments, and the present application can also be made into more kinds of variations and modifications, which are all within the scope of the present application.
Claims
1. A lift-drag combined vertical axis wind turbine blade airfoil design method, characterized in that: The following steps are involved: Parameterize the blade airfoil profile, generate the airfoil decision space through sampling method, and obtain the airfoil aerodynamic performance parameters based on multi-condition aerodynamic simulation model; Based on the airfoil structural characteristic parameters and the aerodynamic performance parameters, constructing an aerodynamic performance deep learning agent model of the airfoil decision space; The aerodynamic performance deep learning agent model is used in combination with constraint conditions to establish a blade airfoil optimization model, and the optimization algorithm is used to perform global optimization to complete the airfoil profile design.
2. The lift-drag combined vertical axis wind turbine blade airfoil design method according to claim 1, characterized in that: Generating the airfoil decision space comprises the following steps: Performing parameter characterization on the blade airfoil to obtain structural characteristic parameters; Sample points are sampled within the structural characteristic parameters to generate the airfoil decision space.
3. The lift-drag combined vertical axis wind turbine blade airfoil design method according to claim 1, characterized in that: The method of obtaining the aerodynamic performance parameters of the airfoil includes the following steps: Based on the unique aerodynamic characteristics of the combined vertical axis wind turbine, different operating conditions are divided; A three-dimensional geometric model and a simulation model of the wind turbine are established, and meshing and fluid dynamics calculations are performed on the simulation model to obtain the aerodynamic performance parameters.
4. The lift-drag combined vertical axis wind turbine blade airfoil design method according to claim 1, characterized in that: The aerodynamic performance parameters of the airfoil include aerodynamic torque, rated aerodynamic power and power coefficient.
5. The lift-drag combined vertical axis wind turbine blade airfoil design method according to claim 1, characterized in that: The method of constructing an aerodynamic performance deep learning agent model for an airfoil decision space includes the following steps: Obtaining a structural input vector of the airfoil decision space according to the structural characteristic parameters of each sample point; Obtaining a response vector corresponding to the airfoil decision space according to the aerodynamic performance parameters of each sample point; According to the structural input vector and response vector of the airfoil decision space, an aerodynamic performance deep learning proxy model of the airfoil decision space is established.
6. The lift-drag combined vertical axis wind turbine blade airfoil design method according to claim 1, characterized in that: The method utilizes the aerodynamic performance deep learning agent model, combines the constraints, establishes a blade airfoil optimization model, uses the optimization algorithm to perform global optimization, and completes the airfoil profile design, including the following steps: The maximum response value of the proxy model representing the power coefficient is used as the objective function, and the blade airfoil optimization model is established in combination with the thickness and camber constraints, wherein the objective function is: F(x)=Cp=DNN(x i ,tsr) Where x i is the characteristic parameter of the airfoil structure, and tsr is the tip speed ratio under different operating conditions.
7. The lift-drag combined vertical axis wind turbine blade airfoil design method according to claim 6, characterized in that: The method of establishing the blade airfoil optimization model by taking the maximum response value of the proxy model representing the power coefficient as the objective function and combining the thickness and camber constraints further includes the following steps: Taking the maximum response value of the proxy model representing the power coefficient as the objective function, combined with the thickness and camber constraints, a blade airfoil optimization model is established. The mathematical model expression of the blade airfoil optimization model is: Objiect:Maxmize F(x)=Cp=DNN(x i ,tsr) Where, t max is the relative maximum thickness of the airfoil, c is the maximum camber of the airfoil, L max is the chord length position of the airfoil where the maximum camber of the airfoil is located; If the optimization target value converges or reaches the set maximum number of iterations, the optimized airfoil is obtained.
8. The lift-drag combined vertical axis wind turbine blade airfoil design method according to claim 7, characterized in that: The blade airfoil optimization model further includes the following steps: Determine whether the optimized airfoil meets the airfoil accuracy requirement. If so, output the optimized airfoil and end the optimization process. If not, increase the sample points in the airfoil decision space and update the proxy model until the airfoil accuracy requirement is met.
9. The lift-drag combined vertical axis wind turbine blade airfoil design method according to claim 1, characterized in that: The lift-drag combined vertical axis wind turbine blade airfoil design method further includes the following steps: A parametric characterization of the blade airfoil is achieved based on a NURBS curve parameterization method; Obtaining the airfoil decision space based on a super Latin cube sampling method; The multi-condition aerodynamic simulation model is established in STARCCM+ software, a structural grid and a boundary layer grid are generated, and boundary conditions and solution parameters are set to perform aerodynamic flow field simulation and aerodynamic performance parameter calculation.