Generative fan blade optimization design method based on generative adversarial network

Abstract

A wind turbine blade generative optimization design method based on a generative adversarial network comprises the steps that detailed data of existing wind turbine blades are obtained through a simulation algorithm, and a training data set is formed; and generating a new fan blade three-dimensional structure through the encoder and the generative adversarial network. And evaluating the power generation efficiency of the newly designed fan blade by using the generative adversarial network. And optimizing the fan blade design through adversarial training of the generative neural network and the predictive neural network, and finally outputting the optimized fan blade design. According to the method, the problems that multiple design targets such as aerodynamic performance, structural strength and cost are difficult to consider at the same time during fan blade optimization, and local optimal solutions are likely to be caught in are solved; through multi-target training, optimization design meeting multiple targets at the same time is generated, through feedback of a discriminator, the generated design is improved step by step, and the blade shape with the optimal comprehensive performance is found.

Description

Technical Field

[0001] The present invention belongs to the field of wind power generation, and in particular relates to a generative optimization design method for wind turbine blades based on a generative adversarial network. Background Art

[0002] The capacity of modern wind turbines is getting bigger and bigger, from hundreds of kilowatts in the early days to several megawatts (MW) or even more than ten MW now. The development of large-scale units has greatly improved the power generation capacity of wind farms. Most wind turbine blades use new materials and optimized design technologies, making modern wind turbine blades lighter and more efficient. Wind turbine blade optimization is a key link in wind power generation technology and directly affects the efficiency and reliability of wind turbines. With the advancement of computing power and algorithm technology, wind turbine blade optimization methods are also constantly developing.

[0003] At present, wind turbine blade optimization mostly uses CFD to simulate aerodynamic performance with high precision, combined with large-scale climate models and small-scale local flow field models, to perform multi-scale coupling simulation, accurately predict wind resources and blade performance, evaluate the performance of blade design under different wind conditions, and optimize the shape, angle and structure of blades through simulation to improve wind capture efficiency. However, CFD simulation and optimization algorithms require a lot of computing resources, especially when it comes to multi-objective optimization and high-resolution simulation, the computing time and cost are very high. Traditional optimization algorithms such as genetic algorithms (GA) and particle swarm optimization algorithms (PSO) can simulate natural selection and genetic mechanisms, and find the optimal solution for blade design through iterative search, but their computing cost and time are long, and each fitness evaluation may require complex simulation or calculation, especially in wind turbine blade optimization, fitness evaluation usually involves CFD simulation, which further increases the computing overhead. With the development of artificial intelligence technology, wind turbine blade optimization methods will continue to develop in the direction of intelligence, automation and integration in the future. Artificial intelligence and machine learning technologies will be more deeply integrated into the optimization process to improve design efficiency and quality. At the same time, multidisciplinary optimization methods will be further developed to achieve the global optimal design of wind turbine blades. In terms of computing resources and experimental verification, more advanced computing technologies and experimental methods will be used to ensure the practical feasibility and reliability of the optimization results.

[0004] Based on the above analysis, the technical problem to be solved by the present invention is: it is difficult to take into account multiple design objectives at the same time in the optimization of fan blades, such as aerodynamic performance, structural strength and cost, and it is easy to fall into a local optimal solution; it is difficult to generate an optimized design that meets multiple objectives at the same time through multi-objective training, and to gradually improve the generated design through the feedback of the discriminator to find the blade shape with the best comprehensive performance. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a generative optimization design method for fan blades based on a generative adversarial network, which solves the problem that it is difficult to take into account multiple design objectives at the same time in the optimization of fan blades, such as aerodynamic performance, structural strength and cost, and it is easy to fall into a local optimal solution; through multi-objective training, an optimized design that meets multiple objectives at the same time is generated, and through the feedback of the discriminator, the generated design is gradually improved to find the blade shape with the best comprehensive performance.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: A generative optimization design method for wind turbine blades based on a generative adversarial network, the steps are as follows: Step 1: Use simulation algorithms to perform large-scale numerical simulations on existing wind turbine blades to obtain blade geometry data, grid unit division results, lift and other characteristic evaluations, as well as stress, strain and deformation data; analyze the vibration characteristics of different wind turbine blades in different environments and the influence mechanism of various structural parameters on the dynamic response of wind turbine blades to form a training data set; obtain a data set that can be used in subsequent steps; Step 2: Obtain the front view, left view and top view of the existing wind turbine blade structure to form an original data set; obtain the original data sets of three different views; use the encoder to reduce the dimension of the original data sets of each view to eliminate the influence of the connection structure between the buoys on the subsequent generative optimization design, and only retain the key information of the buoys in the image; obtain the key information data set after dimensionality reduction; input the encoded original data set into the generative adversarial network to generate a new three-view image of the wind turbine blade structure; The generated coded image is decoded and restored to the same dimension as the original data set through a dimensionality-up operation, and the three-dimensional structure of the wind turbine blade is reconstructed according to the three-view image; a new three-dimensional structure of the wind turbine blade is obtained, that is, a newly designed wind turbine blade structure; Step 3: Use the generative adversarial network to evaluate the power generation of the new wind turbine blades and obtain the power generation efficiency of the new blades; Step 4: Based on the original data set, the generative neural network is used to analyze the original data set and identify data features and attributes, and the data attributes are modified by adding noise to the attributes to obtain the modified data; the modified data is passed to the prediction neural network; the prediction neural network calculates the probability that the generated output belongs to the original data set, provides a reference for the generative neural network, and reduces the randomization of noise in the next cycle; and is used to obtain the probability result; Step 5: Conduct adversarial training on the generative neural network and the predictive neural network. The generative neural network maximizes the possibility of error in the predictive neural network, while the predictive neural network minimizes the possibility of its own error. During the training iteration process, the two neural networks will continue to evolve and confront each other until a balance is reached. Finally, the optimized and trained wind turbine blade design is output.

[0007] Preferably, the sub-steps of step 1 are: Step 1.1: First, determine a suitable simulation algorithm and set relevant parameters, including simulation accuracy and calculation range; Then, the model of the existing wind turbine blade is imported into the simulation software, and the simulation process is started; through the simulation, the geometric data of the blade is obtained, and the geometric data includes information such as the shape and size of the blade; the grid unit division result is obtained, Finally, the blade model is divided into fine grids to more accurately analyze the characteristics of the blade; evaluate the lift and drag characteristics of the blade, as well as the stress, strain and deformation data under various load conditions; Step 1.2: Analyze the vibration characteristics of different wind turbine blades under different environments and the influence mechanism of various structural parameters on the dynamic response of wind turbine blades; collect the operating data of different wind turbine blades under different environmental conditions, including wind speed and wind direction; These data are analyzed to study the vibration characteristics of the blades and determine which structural parameters have a significant impact on the dynamic response of the blades. Through the analysis, a training data set is formed, which contains various characteristic data and structural parameter information of the blades, providing data support for subsequent steps.

[0008] Preferably, the sub-steps of step 2 are: Step 2.1: Obtain the front view, left side view and top view of the existing wind turbine blade structure; use image acquisition equipment or software tools to shoot or model the existing wind turbine blade from multiple angles to obtain the front view, left side view and top view of the blade structure; combine the three views to form an original data set, which contains the structural information of the blade at different angles; Step 2.2: Perform dimensionality reduction processing on the original data set of each view; select a suitable encoder and input the original data set of each view into the encoder; the encoder performs dimensionality reduction processing on the data by learning the characteristics of the data, eliminates the influence of the connection structure between the buoys on the subsequent generative optimization design, and only retains the key information of the buoys in the image; after the dimensionality reduction processing, a reduced dimensionality key information data set is obtained; Step 2.3: Input the encoded original data set into the generative adversarial network; set the parameters of the generative adversarial network, such as the structure and learning rate of the generator and discriminator; input the encoded original data set into the generative adversarial network and start the network training process; the generative adversarial network generates a new three-view diagram of the wind turbine blade structure through continuous learning and optimization; Step 2.4: Decode the generated coded image and reconstruct the three-dimensional structure; input the generated coded image into the decoder and restore it to the same dimension as the original data set through dimensionality increase operation; then, based on the information of the three-view drawing, use the three-dimensional modeling software or algorithm to reconstruct the three-dimensional structure of the wind blade; finally, obtain a new three-dimensional structure of the wind blade, that is, a newly designed wind blade structure.

[0009] Preferably, the sub-steps of step 3 are: Step 3.1, input the newly designed wind turbine blade structure into the generative adversarial network, and the network predicts the power generation of the blade under different working conditions by analyzing the structure and characteristics of the blade; Step 3.2: Determine whether the newly designed blade has higher power generation efficiency by evaluating the power generation efficiency of the new blade.

[0010] Preferably, the sub-steps of step 4 are: Step 4.1, using the generative neural network to analyze the original data set; select the appropriate generative neural network structure and parameters, and input the original data set into the generative neural network; the generative neural network analyzes the original data set and identifies the characteristics and attributes of the data by learning the characteristics and attributes of the data; Step 4.2: Modify data attributes by adding noise or random changes to the attributes. In the generative neural network, determine the attributes that need to be modified, and add noise or random changes to these attributes. By modifying the data attributes, the diversity of the data is increased, providing more possibilities for the subsequent training process. Step 4.3, obtain the modified data and pass it to the prediction neural network; the modified data is used as the output of the generating neural network to obtain the modified data; the modified data is passed to the prediction neural network, and the prediction neural network will analyze and evaluate the data; Step 4.4: The prediction neural network calculates the probability that the output generated belongs to the original data set. The prediction neural network calculates the probability that the generated output data belongs to the original data set by learning the characteristics and distribution of the original data set. This probability value will provide a reference for the generation neural network, helping the generation neural network to adjust parameters and reduce the randomization of noise points in the next cycle. Step 4.5, obtain the probability result; predict the probability result obtained by the output calculation of the neural network, and this result will be fed back to the generated neural network to guide the optimization process of the generated neural network.

[0011] Preferably, the sub-steps of step 5 are: Step 5.1, conduct adversarial training; combine the generative neural network and the predictive neural network to conduct adversarial training; during the training process, the generative neural network maximizes the possibility of error of the predictive neural network, while the predictive neural network minimizes the possibility of its own error; Step 5.2: The two neural networks continue to evolve and compete with each other. During the training iteration process, the generating neural network and the predicting neural network will continue to learn and optimize their own parameters to improve performance. The two neural networks will continue to evolve and compete with each other, and through continuous interaction and adjustment, the performance of the entire system will gradually improve. Step 5.3, until a balance state is reached and the optimized trained fan blade design is output; as the training progresses, the generative neural network and the predictive neural network will gradually reach a balance state; in the balance state, the generative neural network can generate more realistic and optimized fan blade design data, and the predictive neural network can also accurately evaluate the authenticity of these data; finally, the optimized trained fan blade design is output to provide a reference for the actual fan blade production and application.

[0012] Preferably, in step 2, the encoder solves the problem that the connection structure between the buoys affects the generation of the optimized design structure through encoding and decoding operations; the operation method is: Select 3 in turn on the original image 3-sized regions and perform convolution calculations with convolution kernels of the same size:

[0013] In the formula, Indicates encoding, represents the pixel value of the i-th row and j-th column of the image, represents the convolution operation, p Represents the original image data; h represents the convolution kernel, and h The values ​​of each parameter in the above equation should be determined through a large number of numerical simulation studies to form an existing wind turbine blade database and training; The decoding operation is achieved by restoring the high-order features of the image and upgrading the dimension. A feature pixel is selected on the encoded image and The deconvolution kernel of 3 is used for convolution calculation:

[0014] In the formula, Indicates decoding, p Represents any pixel on the image; C represents the deconvolution kernel, and C The values ​​of each parameter should be determined through a large number of numerical simulation studies to form an existing wind turbine blade database and training.

[0015] A generative optimization design system for wind turbine blades based on a generative adversarial network adopts the generative optimization design method for wind turbine blades based on a generative adversarial network.

[0016] The present invention can achieve the following beneficial effects: 1. Use generative adversarial networks to optimize the structure of wind turbine blades. By combining the advantages of the generator and the discriminator, efficient design generation, multi-objective optimization, data enhancement and simulation cost reduction, improved design innovation, automation and efficiency, and improved adaptability and generalization capabilities are achieved.

[0017] 2. By using deep learning models such as autoencoders and variational autoencoders, complex 3D structures can be reconstructed and generated. The automation of 3D modeling can be improved, and the time and cost of manual modeling can be reduced. The encoder compresses the high-dimensional 3D blade structure data into a low-dimensional latent space, effectively reducing the amount of data storage and processing. The decoder can accurately reconstruct the original 3D structure from the low-dimensional latent representation, ensuring the integrity of the design details. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 The fan blade simulation model of the present invention; Figure 2 It is a schematic diagram of a three-dimensional structure reconstruction method based on encoding-decoding of the present invention; Figure 3 Generate a neural network schematic for the present invention; Figure 4 A schematic diagram of a generative adversarial network for the present invention. DETAILED DESCRIPTION

[0019] The preferred solution is Figures 1 to 4 As shown in FIG. 1 , a generative optimization design method for wind turbine blades based on a generative adversarial network is as follows: Step 1: Data Preparation: Step 1.1: Determine the simulation algorithm and parameters and perform simulation: First, when determining the appropriate simulation algorithm, factors such as the physical properties of the wind turbine blades, the working environment, and the required accuracy requirements need to be considered. For example, a finite element analysis algorithm can be selected, which can perform accurate mechanical analysis on complex structures. At the same time, set relevant parameters, such as setting the simulation accuracy to high, medium, and low levels, and select the appropriate accuracy level according to specific needs. For the calculation range, it can be determined according to the size of the wind turbine blades and the working environment to ensure that the behavior of the blades under various working conditions can be covered.

[0020] Next, the model of the existing wind turbine blade is imported into the simulation software. This process requires ensuring the accuracy and completeness of the model, including information such as the blade's geometry and material properties. After starting the simulation process, the software will analyze the blade according to the set algorithms and parameters.

[0021] Through simulation, the geometric data of the blade is obtained, including shape information such as the length, width, thickness of the blade, and the curve characteristics of the blade surface. At the same time, the grid unit division results are obtained. Dividing the blade model into fine grids can improve the accuracy of the analysis. For example, the blade can be divided into thousands or even more small grid units. This can more accurately analyze the stress, strain and deformation of the blade under different loads. In addition, the lift, drag and other characteristics of the blade can also be evaluated. These characteristics are important performance indicators of the blade during the working process. Through simulation, the lift and drag values ​​of the blade under different wind speeds, wind directions and other conditions can be obtained.

[0022] Step 1.2: Analyze vibration characteristics and influencing mechanisms and form a training data set: Collect the operating data of different wind turbine blades under different environmental conditions, such as wind speed, wind direction, etc. This data can be obtained through field measurements, sensor data collection, or simulation experiments. For wind speed, an anemometer can be used to measure and record wind speed values ​​at different locations and times; for wind direction, a wind vane can be used to determine it.

[0023] The collected data is analyzed to study the vibration characteristics of the blades. The vibration frequency, amplitude and other characteristics of the blades are determined by monitoring and analyzing the vibration of the blades at different wind speeds and wind directions. At the same time, the influence mechanism of various structural parameters on the dynamic response of the wind turbine blades is analyzed, such as the influence of parameters such as the thickness, length, and width of the blades on the performance indicators such as lift, drag, and stress of the blades.

[0024] Through analysis, a training data set is formed. This data set contains various characteristic data and structural parameter information of the blade, such as the blade's geometric data, lift and drag characteristics, stress and strain deformation data, and environmental data such as wind speed and direction. These data will provide data support for subsequent steps and be used to train the generative adversarial network and neural network.

[0025] Step 2: New design generation: Step 2.1: Get multiple views and form the original dataset: Use image acquisition equipment or software tools to photograph or model the existing wind turbine blades from multiple angles. If you use an image acquisition device, you can use a high-resolution camera to photograph the blades from different angles to ensure that you can obtain the front view, left side view, and top view of the blades. If you use software tools, you can model the blades using 3D modeling software and then capture views from different angles.

[0026] After obtaining the front view, left side view and top view of the blade structure, these three views are combined to form the original data set. This data set contains the structural information of the blade at different angles, which can provide input data for subsequent dimensionality reduction processing and generative adversarial networks.

[0027] Step 2.2: Perform dimensionality reduction: Select a suitable encoder. The selection of the encoder needs to consider the characteristics of the data and the goal of dimensionality reduction. For example, you can choose methods such as autoencoders or principal component analysis (PCA). Input the original data set of each view into the encoder, which will reduce the dimensionality of the data by learning the characteristics of the data.

[0028] During the dimensionality reduction process, the encoder eliminates the influence of the connection structure between the buoys on the subsequent generative optimization design. This is because the connection structure may interfere with the generative adversarial network's learning and generation of key information about the blade. Through dimensionality reduction, only the key information of the buoys in the image is retained, such as the shape, size, and position of the buoys.

[0029] After dimensionality reduction, the reduced-dimensional key information dataset is obtained, which will be used as the input of the generative adversarial network.

[0030] Step 2.3: Input the generative adversarial network to generate three images: Set the parameters of the generative adversarial network, such as the structure of the generator and discriminator, the learning rate, etc. The structure of the generator can be designed according to the characteristics of the three-view image of the wind turbine blade structure to be generated, for example, a convolutional neural network (CNN) structure can be used. The structure of the discriminator needs to be able to accurately determine whether the generated image is real. The setting of the learning rate needs to be adjusted according to the scale and complexity of the training data to ensure that the network can converge quickly.

[0031] The encoded original data set is input into the generative adversarial network to start the network training process. During the training process, the generative adversarial network generates new three-view images of the wind turbine blade structure through continuous learning and optimization. The generator will try to generate realistic three-view images, while the discriminator will determine whether the generated image is real and give feedback to the generator, prompting the generator to continuously improve.

[0032] Step 2.4: Decode and reconstruct the 3D structure: The generated encoded image is input into the decoder, which has a structure corresponding to the encoder and can be restored to the same dimension as the original dataset through dimensionality upscaling. During the dimensionality upscaling process, the decoder will restore the details and features of the image, making the generated image clearer and more realistic.

[0033] Then, based on the information of the three views, the three-dimensional structure of the wind turbine blade is reconstructed using three-dimensional modeling software or algorithms. The three-dimensional model of the blade can be constructed by fusing the information of the three views using geometric modeling technology. For example, the lines and contours in the three views can be converted into the surface of the three-dimensional model using the triangular patch stitching method.

[0034] Finally, a new three-dimensional structure of the wind turbine blade is obtained, that is, a newly designed wind turbine blade structure. This structure will be used as input for subsequent steps to conduct power generation evaluation and optimization training.

[0035] Step 3: Power generation evaluation: Step 3.1: Input the new blade structure for power generation prediction: The newly designed wind turbine blade structure is input into the generative adversarial network. The network will predict the power generation of the blade under different working conditions by analyzing the structure and characteristics of the blade. For example, the network can calculate the lift and drag of the blade under different working conditions based on information such as the shape, size, material, and environmental factors such as wind speed and wind direction, and then predict the power generation.

[0036] Step 3.2: Evaluate power generation efficiency to determine performance: The new blades are evaluated for their power generation efficiency. Power generation efficiency can be determined by calculating the ratio of power generated by the blades to the wind energy input. The power generation efficiency of the newly designed blades is compared with the existing blades to determine whether the newly designed blades have a higher power generation efficiency. If the new blades have a higher power generation efficiency, the design is successful; if the power generation efficiency does not improve or decreases, the design needs to be further optimized.

[0037] Step 4: Optimize training: Step 4.1: Analyze the original dataset using the generated neural network: Select the appropriate neural network structure and parameters. The structure of the neural network can be selected according to the characteristics of the original data set and the optimization goal. For example, a deep neural network (DNN) or a recurrent neural network (RNN) can be used. The parameter settings include the number of network layers, the number of neurons, the activation function, etc., which need to be adjusted according to the scale and complexity of the data.

[0038] The original data set is fed into a generative neural network. The generative neural network analyzes the original data set and identifies the features and properties of the data by learning the features and properties of the data. For example, the network can learn features such as the geometry of the blade, material properties, lift and drag characteristics, and the relationships between these features.

[0039] Step 4.2: Add noise to modify data attributes: In the generative neural network, the attributes that need to be modified are determined. These attributes can be selected based on the optimization goal and the characteristics of the data. For example, geometric parameters such as the thickness, length, and width of the blade, or performance indicators such as lift and drag can be selected.

[0040] Add noise or random changes to these attributes. You can use methods such as random number generators or normal distribution to add noise. By modifying data attributes, you can increase the diversity of data and provide more possibilities for subsequent training processes. For example, you can generate data of blades of different thicknesses by changing the thickness of the blades, so as to explore the impact of different thicknesses on blade performance.

[0041] Step 4.3: Pass the modified data to the prediction neural network: The modified data is used as the output of the generative neural network to obtain the modified data, which will contain new blade structure and performance information.

[0042] The modified data is passed to the prediction neural network. The prediction neural network will analyze and evaluate the data. The structure and parameters of the prediction neural network also need to be selected and adjusted according to the characteristics of the data and the optimization goals.

[0043] Step 4.4: Calculate the probability that the output belongs to the original dataset: The prediction neural network learns the characteristics and distribution of the original data set and calculates the probability that the generated output data belongs to the original data set. This probability value can reflect the similarity between the generated data and the original data. If the probability value is high, it means that the generated data is similar to the original data; if the probability value is low, it means that the generated data is very different from the original data.

[0044] This probability value will provide a reference for the generative neural network to help it adjust parameters and reduce the randomization of noise in the next cycle. For example, if the probability value is low, the generative neural network can adjust the amplitude of adding noise or select different attributes for modification to improve the quality of the generated data.

[0045] Step 4.5: Get probability results and guide optimization: The prediction neural network outputs the calculated probability result. This result will be fed back to the generation neural network to guide the optimization process of the generation neural network. The generation neural network can adjust its parameters and operations according to the probability results to improve the quality and authenticity of the generated data. At the same time, this process can also help the prediction neural network to continuously learn and optimize its judgment ability and improve the accuracy of the evaluation of the generated data.

[0046] Step 5: Adversarial Training: Step 5.1: Perform adversarial training: The generative neural network and the predictive neural network are combined for adversarial training. During the training process, the goal of the generative neural network is to maximize the possibility of error in the predictive neural network, that is, the generated data can deceive the predictive neural network and make it think it is real data. The goal of the predictive neural network is to minimize the possibility of its own error, that is, to accurately judge whether the generated data is real.

[0047] Step 5.2: Neural network evolution and adversarial: During the training iteration process, the generative neural network and the predictive neural network will continuously learn and optimize their own parameters to improve performance. The generative neural network will continuously adjust the strategy of adding noise and the method of modifying data attributes to generate more realistic data. The predictive neural network will continuously learn the characteristics and distribution of the original data set to improve the accuracy of the judgment of the generated data.

[0048] The two neural networks will continue to evolve and compete with each other. Through constant interaction and adjustment, the performance of the entire system will gradually improve. For example, the data generated by the generative neural network will become more and more realistic, and the judgment ability of the predictive neural network will become stronger and stronger.

[0049] Step 5.3: Reach equilibrium and output the optimized design: As the training progresses, the generative neural network and the predictive neural network will gradually reach a state of equilibrium. In this state of equilibrium, the generative neural network can generate more realistic and optimized wind blade design data, and the predictive neural network can also accurately evaluate the authenticity of this data.

[0050] Finally, the optimized wind turbine blade design is output. This design will provide a reference for the actual production and application of wind turbine blades, which can improve the power generation efficiency, reliability and performance of wind turbines.

[0051] In step 2, the encoder solves the problem that the connection structure between the buoys affects the generation of the optimized design structure through encoding-decoding operations. The specific operation method is as follows: Select 3×3 areas on the original image and perform convolution calculations with convolution kernels of the same size. This process can be understood as extracting local features from the image. The function of the convolution kernel is to perform weighted summation on the local area of ​​the image to extract the features of the image.

[0052] The formula is:

[0053] In the formula, p Represents the original image data; h represents the convolution kernel, and hThe values ​​of each parameter should be determined through a large number of numerical simulation studies to form an existing wind turbine blade database and training.

[0054] The decoding operation is achieved by restoring the high-level features of the image and upgrading the dimension. Specifically, a feature pixel is selected from the encoded image and the size is 3 The deconvolution kernel of 3 is used for convolution calculation:

[0055] In the formula, p Represents any pixel on the image; C represents the deconvolution kernel, and C The values ​​of each parameter should be determined through a large number of numerical simulation studies to form an existing wind turbine blade database and training.

[0056] Finally, a new wind turbine blade structure was proposed by using the generative optimization design method and the encoding-decoding based three-dimensional structure reconstruction method. The vibration reduction effectiveness of the new wind turbine blade was verified by combining numerical simulation and large-scale water tank tests.

[0057] The generative adversarial network of the present invention is a deep learning architecture that trains two neural networks to compete with each other to generate more realistic new data from a given training data set, such as generating new images from an existing image database. An adversarial network means that it trains two different networks and makes them compete with each other, one network generates new data by taking input data samples and modifying them as much as possible, and the other network tries to predict whether the generated data output belongs to the original data set, or the prediction network decides whether the generated data is true or false, and the generation network generates updated and improved versions of the false data values ​​until the prediction network can no longer distinguish between the false data values ​​and the original data values.

[0058] A generative optimization design system for wind turbine blades based on generative adversarial networks adopts the above design method. The system includes the following main modules: Data acquisition module: responsible for collecting various data of existing wind turbine blades, including geometric data, vibration characteristic data, operation data, etc. These data can be obtained through sensors, measuring equipment or databases. Simulation module: Use simulation algorithms to perform large-scale numerical simulation of existing wind turbine blades, obtain various characteristic data of blades, and analyze the influence mechanism of vibration characteristics and structural parameters on dynamic response under different environments.

[0059] View acquisition and processing module: obtains the front view, left view and top view of the existing wind turbine blade structure, and performs dimensionality reduction and encoding and decoding operations to provide input data for the generative adversarial network.

[0060] Generative adversarial network module: Generates a new three-view image of the wind blade structure, and obtains a new three-dimensional structure of the wind blade through decoding and three-dimensional reconstruction.

[0061] Power generation evaluation module: Evaluate the power generation of new wind turbine blades to determine their power generation efficiency.

[0062] Neural network module: includes generating neural network and predicting neural network, which optimizes training by adding noise, modifying data attributes, calculating probability and other operations.

[0063] Adversarial training module: performs adversarial training on the generative neural network and the predictive neural network to improve the performance of the system and the quality of the output results.

[0064] Output module: Output optimized and trained wind turbine blade designs to provide reference for actual production and application.

[0065] Through the collaborative work of these modules, the system can achieve generative optimization design of wind turbine blades based on generative adversarial networks, thereby improving the performance and power generation efficiency of wind turbine blades.

[0066] The above embodiments are only preferred technical solutions of the present invention and should not be regarded as limiting the present invention. The protection scope of the present invention shall be the technical solutions recorded in the claims, including equivalent replacement solutions of the technical features in the technical solutions recorded in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present invention.

Claims

1. A generative optimization design method for wind turbine blades based on generative adversarial networks, characterized in that The following steps are involved: Step 1: Use simulation algorithms to perform large-scale numerical simulations on existing wind turbine blades to obtain blade geometry data, grid unit division results, lift and other characteristic evaluations, as well as stress, strain and deformation data; Analyze the vibration characteristics of different fan blades in different environments and the influence mechanism of various structural parameters on the dynamic response of fan blades to form a training data set; Obtain a data set that can be used in subsequent steps; Step 2, obtaining a front view, a left view, and a top view of an existing wind turbine blade structure to form an original data set; The original data sets of three different views are obtained; the encoder is used to reduce the dimension of the original data sets of each view, eliminating the influence of the connection structure between the buoys on the subsequent generative optimization design, and only retaining the key information of the buoys in the image; Obtain the key information data set after dimensionality reduction; input the encoded original data set into the generative adversarial network to generate a new three-view drawing of the wind turbine blade structure; The generated coded image is decoded and restored to the same dimension as the original data set through dimensionality increase operation, and the three-dimensional structure of the wind turbine blade is reconstructed based on the three-view image. A new three-dimensional structure of the fan blade is obtained, that is, a completely new designed fan blade structure; Step 3: Use the generative adversarial network to evaluate the power generation of the new wind turbine blades; Obtain the power generation efficiency of the new blade; Step 4: Based on the original data set, the generative neural network is used to analyze the original data set and identify data features and attributes, and the data attributes are modified by adding noise to the attributes to obtain the modified data; the modified data is passed to the prediction neural network; the prediction neural network calculates the probability that the generated output belongs to the original data set, which provides a reference for the generative neural network and reduces the randomization of noise in the next cycle; Used to obtain probability results; Step 5: Conduct adversarial training on the generative neural network and the predictive neural network. The generative neural network maximizes the possibility of error in the predictive neural network, while the predictive neural network minimizes the possibility of its own error. During the training iteration process, the two neural networks will continue to evolve and confront each other until a balance is reached. Finally, the optimized and trained wind turbine blade design is output.

2. The method for generative optimization design of wind turbine blades based on generative adversarial networks according to claim 1, characterized in that: The sub-steps of step 1 are: Step 1.1: First, determine a suitable simulation algorithm and set relevant parameters, including simulation accuracy and calculation range; Then, the model of the existing wind turbine blade is imported into the simulation software and the simulation process is started; Acquire geometric data of the blade through simulation, wherein the geometric data includes information including the shape and size of the blade; Get the grid unit division result, Finally, the blade model is divided into fine grids to more accurately analyze the characteristics of the blade; evaluate the lift and drag characteristics of the blade, as well as the stress, strain and deformation data under various load conditions; Step 1.2: Analyze the vibration characteristics of different fan blades under different environments and the influence mechanism of various structural parameters on the dynamic response of fan blades; Collect operating data of different wind turbine blades under different environmental conditions, including wind speed and direction; The operating data is analyzed to study the vibration characteristics of the blades and determine which structural parameters have a significant impact on the dynamic response of the blades. Through the analysis, a training data set is formed, which includes various characteristic data and structural parameter information of the blades.

3. The method for generative optimization design of wind turbine blades based on generative adversarial networks according to claim 1, characterized in that: The sub-steps of step 2 are: Step 2.1: Obtain the front view, left side view and top view of the existing wind turbine blade structure; Use image acquisition equipment or software tools to take photos or model existing wind turbine blades from multiple angles to obtain the front view, left view, and top view of the blade structure; combine the three views to form an original data set, which contains the structural information of the blade at different angles; Step 2.2: Perform dimensionality reduction processing on the original data sets of each view; Select a suitable encoder and input the original data set of each view into the encoder; the encoder performs dimensionality reduction processing on the data by learning the characteristics of the data, eliminating the influence of the connection structure between the buoys on the subsequent generative optimization design, and only retaining the key information of the buoys in the image; after the dimensionality reduction processing, the key information data set after dimensionality reduction is obtained; Step 2.3: Input the encoded original dataset into the generative adversarial network; Set the parameters of the generative adversarial network, such as the structure and learning rate of the generator and discriminator; input the encoded original data set into the generative adversarial network to start the network training process; The generative adversarial network generates new three-view images of the wind turbine blade structure through continuous learning and optimization; Step 2.4: Decode the generated coded image and reconstruct the three-dimensional structure; input the generated coded image into the decoder and restore it to the same dimension as the original data set through dimensionality increase operation; Then, based on the information of the three-view drawing, the three-dimensional structure of the wind blade is reconstructed using three-dimensional modeling software or an algorithm; ultimately, a new three-dimensional structure of the wind blade is obtained, that is, a newly designed wind blade structure.

4. The method for generative optimization design of wind turbine blades based on generative adversarial networks according to claim 1, characterized in that: The sub-steps of step 3 are: Step 3.1, input the newly designed wind turbine blade structure into the generative adversarial network, and the network predicts the power generation of the blade under different working conditions by analyzing the structure and characteristics of the blade; Step 3.2: Determine whether the newly designed blade has higher power generation efficiency by evaluating the power generation efficiency of the new blade.

5. The method for generative optimization design of wind turbine blades based on generative adversarial networks according to claim 1, characterized in that: The sub-steps of step 4 are: Step 4.1, using the generative neural network to analyze the original data set; select the appropriate generative neural network structure and parameters, and input the original data set into the generative neural network; the generative neural network analyzes the original data set and identifies the characteristics and attributes of the data by learning the characteristics and attributes of the data; Step 4.2: Modify data attributes by adding noise or random changes to the attributes. In the generative neural network, determine the attributes that need to be modified, and add noise or random changes to these attributes. By modifying the data attributes, the diversity of the data is increased, providing more possibilities for the subsequent training process. Step 4.3, obtain the modified data and pass it to the prediction neural network; the modified data is used as the output of the generating neural network to obtain the modified data; the modified data is passed to the prediction neural network, and the prediction neural network will analyze and evaluate the data; Step 4.4: The prediction neural network calculates the probability that the output generated belongs to the original data set. The prediction neural network calculates the probability that the generated output data belongs to the original data set by learning the characteristics and distribution of the original data set. This probability value will provide a reference for the generation neural network, helping the generation neural network to adjust parameters and reduce the randomization of noise points in the next cycle. Step 4.5, obtain the probability result; predict the probability result obtained by the output calculation of the neural network, and this result will be fed back to the generation neural network to guide the optimization process of the generation neural network.

6. The method for generative optimization design of wind turbine blades based on generative adversarial networks according to claim 1, characterized in that: The sub-steps of step 5 are: Step 5.1, conduct adversarial training; combine the generative neural network and the predictive neural network to conduct adversarial training; during the training process, the generative neural network maximizes the possibility of error of the predictive neural network, while the predictive neural network minimizes the possibility of its own error; Step 5.2: The two neural networks continue to evolve and compete with each other. During the training iteration process, the generating neural network and the predicting neural network will continue to learn and optimize their own parameters to improve performance. The two neural networks will continue to evolve and compete with each other, and through continuous interaction and adjustment, the performance of the entire system will gradually improve. Step 5.3, until a balance state is reached and the optimized trained wind turbine blade design is output; as the training progresses, the generated neural network and the predicted neural network will gradually reach a balance state; In a balanced state, the generative neural network can generate more realistic and optimized wind blade design data, and the predictive neural network can also accurately evaluate the authenticity of these data; Finally, the optimized and trained wind blade design is output to provide a reference for the actual production and application of wind blades.

7. The method for generative optimization design of wind turbine blades based on generative adversarial networks according to claim 1, characterized in that: In step 2, the encoder solves the problem that the connection structure between the buoys affects the generation of the optimized design structure through encoding and decoding operations; the operation method is: Select 3 in turn on the original image 3-sized regions and perform convolution calculations with convolution kernels of the same size: In the formula, Indicates encoding, represents the pixel value of the i-th row and j-th column of the image, represents the convolution operation, p Represents the original image data; h represents the convolution kernel, and h The values ​​of each parameter in the above equation should be determined through a large number of numerical simulation studies to form an existing wind turbine blade database and training; The decoding operation is achieved by restoring the high-order features of the image and upgrading the dimension. A feature pixel is selected on the encoded image and The deconvolution kernel of 3 is used for convolution calculation: In the formula, Indicates decoding, p Represents any pixel on the image; C represents the deconvolution kernel, and C The values ​​of each parameter should be determined through a large number of numerical simulation studies to form an existing wind turbine blade database and training.

8. A generative optimization design system for wind turbine blades based on a generative adversarial network, characterized by: A generative optimization design method for wind turbine blades based on a generative adversarial network according to any one of claims 1 to 7 is adopted.

Images