Wind power blade aerodynamic performance simulation analysis method

Abstract

The invention provides a wind power blade aerodynamic performance simulation analysis method, which comprises the following steps of: preprocessing original wind condition data by adopting a data cleaning algorithm according to time sequence data of physical quantities such as wind speed, wind direction and turbulence intensity, removing noise interference and abnormal points in the data through methods such as abnormal value detection and missing value filling, and calculating the aerodynamic performance of a wind power blade. Obtaining cleaned wind regime time sequence data; a data encryption and compression algorithm is adopted to encode a wind regime boundary condition file, the security and efficiency of data transmission and storage are improved, and seamless joint and quick calling of wind regime data and a computational fluid mechanics model are realized by establishing a correlation index between the wind regime data and a blade model.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to a method for simulating and analyzing the aerodynamic performance of a wind power blade. Background Art

[0002] In the process of designing wind turbine blades, it is necessary to import data such as wind speed, wind direction and turbulence intensity collected in the actual wind field environment into the computational fluid dynamics software to simulate and analyze the aerodynamic performance and flow field characteristics of the blades in the real wind environment. However, the wind condition data in the actual wind field environment is usually random, intermittent and non-steady, with huge data volume and complex and diverse formats. How to efficiently and accurately process and convert these raw wind condition data and import them into the computational fluid dynamics software for numerical simulation calculations is a technical problem that needs to be solved urgently.

[0003] The original wind data usually contains time series data of multiple physical quantities such as wind speed, wind direction and turbulence intensity. The data sampling frequency is high, the time span is large, and the data format and units between different physical quantities are also different. In addition, there may be problems such as outliers, missing values ​​and noise interference in the original data, which require data cleaning and preprocessing. At the same time, computational fluid dynamics software has specific requirements for the format and organization of imported data, and the processed wind data needs to be converted into a standard format supported by the software.

[0004] How to achieve efficient data processing and format conversion while ensuring data accuracy and completeness, and seamlessly connect with the data interface of computational fluid dynamics software is a major technical challenge faced in the data import process of wind turbine blade design. Summary of the invention

[0005] The present invention provides a method for simulating and analyzing the aerodynamic performance of a wind turbine blade, which mainly includes: According to the time series data of physical quantities such as wind speed, wind direction, and turbulence intensity, a data cleaning algorithm is used to preprocess the original wind condition data. Through methods such as outlier detection and missing value filling, noise interference and outliers in the data are removed to obtain the cleaned wind condition time series data. For the problem of inconsistent data formats and units of different physical quantities, a data standardization method is used to normalize the cleaned wind condition time series data, converting the data of different physical quantities into a unified dimensionless representation, eliminating the influence of dimension and order of magnitude differences, and obtaining the standardized wind condition time series data. According to the requirements of the wind condition data time and space resolution for the aerodynamic performance simulation analysis of wind turbine blades, downsampling processing is performed on the standardized wind condition time series data. Through methods such as time window averaging or interpolation, the sampling frequency and time span of the data are adjusted to obtain the wind condition time series data that matches the time step of the computational fluid dynamics simulation. The wavelet analysis method is used to perform multi-scale decomposition on the downsampled wind condition time series data, extracting the fluctuation characteristics of wind speed, wind direction, and turbulence intensity at different frequency scales. Through threshold screening and reconstruction of wavelet coefficients, the stabilized wind condition time series data is obtained, reducing the randomness and intermittency effects of the wind condition data. According to the data format requirements of the inlet boundary conditions of the computational fluid dynamics software, format conversion and reorganization are performed on the stabilized wind condition time series data. Through methods such as interpolation and fitting, the time series data of physical quantities such as wind speed, wind direction, and turbulence intensity are mapped to the grid nodes of the computational domain to form a wind condition boundary condition file adapted to the grid topology structure. Data encryption and compression algorithms are used to encode the wind condition boundary condition file, improving the security and efficiency of data transmission and storage. By establishing an association index between the wind condition data and the blade model, seamless docking and rapid invocation of the wind condition data and the computational fluid dynamics model are achieved. During the aerodynamic performance simulation analysis of the blade, according to the time series of the wind condition boundary condition file, the inlet wind speed, wind direction, and turbulence intensity and other parameters of the computational fluid dynamics model are dynamically updated. By numerically solving the fluid dynamics control equations, the aerodynamic forces and flow field characteristics of the blade under actual wind condition are simulated, and key aerodynamic performance indicators such as the force distribution and wake structure of the blade are obtained, providing a basis for the blade structure design and optimization.

[0006] The technical solution provided by the embodiment of the present invention includes the following beneficial effects: The present invention discloses a method for simulating and analyzing the aerodynamic performance of wind turbine blades. This method first preprocesses and standardizes the original wind condition data, solving the problem of inconsistent data formats and units of different physical quantities. Then, through downsampling and wavelet analysis, multi-scale features of the wind condition data are extracted to reduce the randomness effect. Next, the processed wind condition data is converted into the boundary condition format required by the computational fluid dynamics software, and an association index with the blade model is established.

[0007] During the simulation analysis process, the inlet wind condition parameters are dynamically updated, realizing the seamless connection between wind condition data and the calculation model. The present invention effectively solves the application problem of actual wind condition data in the simulation of blade aerodynamic performance, improves the accuracy and efficiency of simulation analysis, and provides a reliable basis for the optimal design of blade structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is a flowchart of a method for simulating and analyzing the aerodynamic performance of a wind turbine blade according to the present invention.

[0009] Figure 2 It is a schematic diagram of a method for simulating and analyzing the aerodynamic performance of a wind turbine blade according to the present invention.

[0010] Figure 3 It is another schematic diagram of a method for simulating and analyzing the aerodynamic performance of a wind turbine blade according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0011] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0012] As Figures 1-3 , a method for simulating and analyzing the aerodynamic performance of a wind turbine blade in this embodiment may specifically include: Step S101, according to the time series data of physical quantities such as wind speed, wind direction, and turbulence intensity, preprocess the original wind condition data using a data cleaning algorithm. By methods such as outlier detection and missing value filling, remove the noise interference and outliers in the data to obtain the cleaned wind condition time series data.

[0013] According to the original time series data of physical quantities such as wind speed, wind direction, and turbulence intensity, preprocess the original wind condition data using a data cleaning algorithm. Through the outlier detection algorithm, determine whether the physical quantity value at each time point exceeds the preset normal range threshold. If it exceeds, mark the time point as an outlier. For outliers, according to the normal physical quantity values of the previous and subsequent time points, use linear interpolation to estimate the physical quantity value of the outlier, and replace the value of the outlier with the estimated value.

[0014] Using the missing value filling algorithm, determine whether there are missing values in the physical quantity time series. If there are, estimate the missing values using linear interpolation based on the normal physical quantity values before and after the missing values, and fill them into the missing positions in the original time series. For the wind condition time series data after outlier correction and missing value filling, use digital signal processing algorithms such as moving average filtering to remove the high-frequency noise interference components in the physical quantity data. Based on the wind condition time series data after outlier detection, missing value filling, and denoising processing, obtain the final cleaned wind condition data. Use the cleaned time series data of wind speed, wind direction, and turbulence intensity as the input for subsequent wind condition feature analysis and wind power prediction models to predict the aerodynamic performance of wind turbine blades.

[0015] Specifically, in a wind farm, by collecting the original time series data of physical quantities such as wind speed, wind direction, and turbulence intensity, use data cleaning algorithms to preprocess the original wind condition data. First, through the outlier detection algorithm, set the normal range of wind speed to 3 - 25 m / s, the normal range of wind direction to 0 - 360°, and the normal range of turbulence intensity to 0 - 5. Determine whether the physical quantity value at each time point exceeds the preset normal range threshold. If it does, mark that time point as an outlier. For the outlier, use the linear interpolation method. Take the normal physical quantity values at 5 time points before and after the outlier, and fit a straight line through the least squares method. Replace the physical quantity value of the outlier with the estimated value of the straight line at that point. Then, through the missing value filling algorithm, referring to the physical quantity time series, if there are 3 or more consecutive missing values, take the normal physical quantity values at 5 time points before and after the missing values, fit a smooth curve through the cubic spline interpolation method, and estimate and fill the missing values into the missing positions in the original time series. For the wind condition time series data after outlier correction and missing value filling, use a moving average filtering algorithm with a window size of 10 to remove the high-frequency noise interference components in the physical quantity data. Finally, obtain the cleaned wind condition time series data.

[0016] Step S102, for the problem of inconsistent data formats and units for different physical quantities, use the data standardization method to normalize the cleaned wind condition time series data, convert the data of different physical quantities into a unified dimensionless representation, eliminate the influence of dimension and order of magnitude differences, and obtain the standardized wind condition time series data.

[0017] According to the data formats and units of different physical quantities in the wind condition time series data, determine the types of physical quantities that need to be data standardized and the corresponding order of magnitude of dimensions. Using the min-max standardization method, normalize the data of each physical quantity, map the data into the interval [0, 1], and eliminate the differences in dimensions and orders of magnitude between the data of different physical quantities. For the dimensionless data after standardization, construct a standardized wind condition time series data set according to the physical quantity type and the time stamps of the wind condition time series. Adopt the convolutional neural network algorithm, use the standardized wind condition time series data as the input, extract the local features of the time series data through the convolutional layer, and reduce the feature dimension through the pooling layer. On the basis of the convolutional neural network, introduce the long short-term memory network (LSTM) layer to capture the long-term dependence relationship of the wind condition time series data and extract the time-dependent features of the time series. Fuse the local features extracted by the convolutional neural network and the time-dependent features extracted by the LSTM to construct a comprehensive feature representation of the wind condition time series data. Based on the fused comprehensive features of the wind condition time series, use a fully connected neural network for wind condition prediction, and obtain accurate wind condition prediction results by training and optimizing the network parameters.

[0018] Specifically, according to the data formats and units of different physical quantities in the wind condition time series data, the types of physical quantities that need to be data standardized include wind speed (m / s), wind direction (degrees), air pressure (hPa), and temperature (°C), etc. Among them, the order of magnitude of wind speed and air pressure is around 10 2 or so, the wind direction is between 10 2 and 10 3 and the temperature is around 10 1Around. Using the min-max normalization method, the data of each physical quantity is normalized. For example, if the minimum value of the wind speed data is 5 m / s and the maximum value is 28 m / s, then the normalized wind speed value = (wind speed value - 5) / (28 - 5), mapping the data into the interval [0, 1] to eliminate the dimensional and order-of-magnitude differences between the data of different physical quantities. For the dimensionless data after normalization, according to the type of physical quantity and the time stamps of the wind condition time series, a normalized wind condition time series data set is constructed. For example, data is recorded every 10 minutes, including the normalized wind speed, wind direction, air pressure, and temperature values, as well as the corresponding time stamps. Using the convolutional neural network algorithm, with the normalized wind condition time series data as the input, the local features of the time series data are extracted through a convolutional layer with a convolutional kernel size of 3×3 and a stride of 1, and the feature dimension is reduced through a 2×2 max pooling layer to reduce the computational complexity. On the basis of the convolutional neural network, a long short-term memory network (LSTM) layer is introduced. By setting the number of LSTM hidden layer units to 128, the gating mechanism of LSTM is used to capture the long-term dependence relationship of the wind condition time series data and extract the time dependence features of the time series. The local features extracted by the convolutional neural network and the time dependence features extracted by LSTM are fused, for example, by concatenation, to construct a comprehensive feature representation of the wind condition time series data. Based on the fused comprehensive features of the wind condition time series, a fully connected neural network is used for wind condition prediction. By setting the number of neurons in the fully connected layer to 256 and 128, the Adam optimization algorithm is used to train and optimize the network parameters to obtain accurate wind condition prediction results. Through the above method, the wind condition time series data can be effectively processed, the key features of the data can be extracted, and an accurate wind condition prediction model can be established, providing important data support for the optimization of the aerodynamic performance of wind turbine blades.

[0019] Step S103, according to the requirements of the wind condition data time and space resolution for the aerodynamic performance simulation analysis of the wind turbine blade, downsample the normalized wind condition time series data. By methods such as time window averaging or interpolation, adjust the sampling frequency and time span of the data to obtain wind condition time series data that matches the time step of the computational fluid dynamics simulation.

[0020] Obtain the wind condition time series data required for the aerodynamic performance simulation analysis of the wind turbine blade; judge whether the spatio-temporal resolution of the wind condition data meets the simulation analysis requirements. If not, perform downsampling; determine the target sampling frequency and time span of downsampling according to the time step of the computational fluid dynamics simulation; use methods such as time window averaging or interpolation to downsample the wind condition time series data; through downsampling, obtain wind condition time series data that matches the simulation time step; input the downsampled wind condition data into the aerodynamic performance simulation analysis model; run the simulation analysis to obtain the aerodynamic performance prediction results of the wind turbine blade under the given wind condition.

[0021] Specifically, first, obtain the high spatio-temporal resolution wind condition data of the area where the wind farm is located from the meteorological department, such as the time series of wind speed and wind direction at a height of 10 meters recorded every 10 minutes. Then, according to the requirements of the computational fluid dynamics simulation for the time step, judge whether the time resolution of the wind condition data meets the requirements. For example, if the simulation time step is 1 second and the wind condition data is once every 10 minutes, downsampling is required. Determine that the target time resolution for downsampling is 1 second, and the time span is the same as the simulation duration. Using the linear interpolation method, insert 599 data points between two adjacent 10-minute wind condition data points to increase the time resolution of the wind condition data to 1 second. After downsampling, the wind condition time series data matches the simulation time step. Input the processed high-time-resolution wind condition data into the computational fluid dynamics numerical simulation software, set the three-dimensional geometric model of the wind turbine blade, the turbulence model, the boundary conditions, etc., and run the simulation calculation. The simulation results can obtain the pressure distribution, velocity field, wake characteristics, etc. on the surface of the wind turbine blade, and then analyze the force characteristics and aerodynamic performance of the blade, such as the variation laws of the lift coefficient and drag coefficient with the wind speed. By comparing with the aerodynamic design parameters of the wind turbine blade, evaluate its aerodynamic performance and provide a basis for optimizing the aerodynamic shape of the blade.

[0022] Step S104, use the wavelet analysis method to perform multi-scale decomposition on the downsampled wind condition time series data, extract the fluctuation characteristics of wind speed, wind direction, and turbulence intensity at different frequency scales, and obtain the smoothed wind condition time series data through threshold screening and reconstruction of wavelet coefficients, reducing the randomness and intermittency effects of the wind condition data.

[0023] Perform downsampling on the wind condition time series data to reduce the data volume and improve the efficiency of subsequent analysis; use wavelet transform to perform multi-scale decomposition on the downsampled wind condition time series data, mapping the time series to the time-frequency domain; for the decomposed wavelet coefficients, screen according to the preset threshold to remove high-frequency noise and outliers; according to the screened wavelet coefficients, use the inverse wavelet transform for reconstruction to obtain the smoothed wind condition time series data; perform statistical analysis on the reconstructed wind speed, wind direction, and turbulence intensity time series data to obtain their respective fluctuation characteristic parameters; according to the fluctuation characteristic parameters of wind speed, wind direction, and turbulence intensity, construct a wind condition time variation model for the optimization control of the aerodynamic performance of the wind turbine blade; apply the optimized control strategy to the current wind turbine blade model to improve the power generation efficiency and stability of the wind farm and reduce the adverse effects of wind condition randomness and intermittency.

[0024] Specifically, for wind time series data, downsampling is first used to reduce the amount of original data to 1 / 10 of the original, such as downsampling 10Hz data to 1Hz. Then, the downsampled data is subjected to a 5-layer wavelet decomposition, and the time series is mapped to the time-frequency domain using the db4 wavelet basis function. For the decomposed wavelet coefficients, the threshold is set to twice the coefficient mean, and high-frequency noise and outliers above the threshold are removed. Based on the screened wavelet coefficients, the time series is reconstructed using the inverse wavelet transform to obtain stabilized wind data. Statistical analysis is performed on the reconstructed wind speed, wind direction, and turbulence intensity data, and the fluctuation characteristic parameters such as mean, variance, skewness, and kurtosis are calculated.

[0025] Step S105, according to the data format requirements of the computational fluid dynamics software for the inlet boundary conditions, the format of the stabilized wind time series data is converted and reorganized, and the time series data of physical quantities such as wind speed, wind direction and turbulence intensity are mapped to the grid nodes of the calculation domain through interpolation and fitting methods to form a wind boundary condition file that is compatible with the grid topology structure.

[0026] Obtain the inlet boundary condition data format requirements of the computational fluid dynamics software, and determine the format conversion and reorganization methods required for the wind time series data. Preprocess the stabilized wind time series data, remove outliers and missing values, and perform necessary data cleaning and normalization operations. According to the grid topology of the computational domain, determine the location coordinate information of the wind data that needs to be generated on the grid nodes. Use the interpolation algorithm to interpolate the time series data of physical quantities such as wind speed and wind direction to the location coordinates of the grid nodes. Common interpolation algorithms include linear interpolation and spline interpolation. If the wind data contains turbulence intensity information, use a suitable turbulence model, such as the k-ε turbulence model, to map the turbulence intensity data to the grid nodes to match the wind speed data. Use a fitting algorithm, such as the least squares method, to fit the interpolated and mapped wind data to obtain a continuous and smooth wind data surface that is suitable for the grid node position. According to the inlet boundary condition data format requirements of the computational fluid dynamics software, organize the fitted wind data into a boundary condition file, ensure that the file format is consistent with the software requirements, and save the file.

[0027] Specifically, we first need to obtain the inlet boundary condition data format requirements of the computational fluid dynamics software. For example, ANSYS Fluent software requires the boundary condition file to be in ASCII format, and the data is arranged in sequence according to the velocity components in the x, y, and z directions. Then, the wind time series data is preprocessed, the 3σ principle is used to remove outliers, the missing values ​​are filled with linear interpolation, and the data is normalized to map the wind speed data to the 0-1 interval. According to the grid topology of the computational domain, the position coordinate information of the grid nodes is determined, such as grid nodes (15, 28, 32). The cubic spline interpolation algorithm is used to interpolate the wind speed and wind direction time series data to the grid node coordinates. If the wind condition data contains turbulence intensity information, the k-ε turbulence model is used to map the turbulence intensity data to the grid nodes by solving the transport equations of turbulent kinetic energy k and turbulent dissipation rate ε to match the wind speed data. The least squares method is used for data fitting. By minimizing the objective function of the sum of square errors between the fitting surface and the data points, a continuous and smooth wind data surface that is adapted to the grid node position is obtained. The fitting function can be expressed as f(x, y, z) = a0 + a1x + a2y + a3z + a4xy + a5yz + a6xz +a7x^2 + a8y^2 + a9z^2. Finally, the fitted wind data is organized according to the boundary condition file format of the Fluent software and saved as an ASCII file for subsequent numerical simulation. Through the above series of data processing operations, the measured wind time series data can be converted into inlet boundary conditions suitable for computational fluid dynamics numerical simulation, providing high-quality meteorological data support for wind environment prediction and optimization.

[0028] Step S106, using data encryption and compression algorithms to encode the wind boundary condition file, improve the security and efficiency of data transmission and storage, and achieve seamless connection and rapid call of wind condition data and computational fluid dynamics model by establishing an association index between wind condition data and blade model.

[0029] According to the characteristics of the wind condition boundary condition file, select a suitable data encryption algorithm to encrypt the file to ensure the security of data during transmission and storage. According to the data volume and access frequency of the wind condition boundary condition file, adopt a suitable data compression algorithm to compress the file, reduce the file size, and improve the transmission and storage efficiency. Encode the encrypted and compressed wind condition boundary condition file to generate a unique file identification code for subsequent management and invocation. Establish an association index table between the wind condition data and the blade model, record the blade model parameters corresponding to each wind condition data file, and realize the association between data and model. In the computational fluid dynamics model, quickly find the wind condition boundary condition file that matches the current blade model through the association index table and read the file data. Decode and decompress the read wind condition boundary condition file to restore the original data content for subsequent model calculations. Seamlessly transfer the wind condition data to the computational fluid dynamics model to achieve fast invocation and integration of the model and data, and improve the operating efficiency of the entire system.

[0030] Specifically, according to the characteristics of the wind condition boundary condition file, the AES-256 encryption algorithm can be selected to encrypt the file. This algorithm uses a 256-bit key and has a high security strength, which can effectively prevent data from being illegally stolen or tampered with during transmission and storage. At the same time, considering the large data volume and high access frequency of the wind condition boundary condition file, the LZ4 compression algorithm can be used to compress the file. The LZ4 algorithm has a fast compression speed and a high compression ratio, which can reduce the file size by 30% - 50%, significantly improving the transmission and storage efficiency. Perform Base64 encoding on the encrypted and compressed wind condition boundary condition file to generate a unique 32-bit file identification code. Establish an association index table between the wind condition data and the blade model, adopt a B+ tree index structure, and achieve fast association between data and model. In the computational fluid dynamics model, find the matching wind condition boundary condition file through the association index table with a time complexity of O(log n), and use the mmap memory mapping technology to achieve efficient reading of the file. Perform Base64 decoding and LZ4 decompression on the read wind condition boundary condition file to restore the original wind condition data. Seamlessly transfer the wind condition data to the computational fluid dynamics model through the shared memory mechanism to achieve zero-copy invocation of the model and data, reduce the data transfer latency to the nanosecond level, and significantly improve the operating efficiency of the system.

[0031] In step S107, during the simulation and analysis of the aerodynamic performance of the blade, according to the time series of the wind condition boundary condition file, parameters such as the inlet wind speed, wind direction, and turbulence intensity of the computational fluid dynamics model are dynamically updated. By numerically solving the fluid mechanics control equations, the aerodynamic forces and flow field characteristics of the blade under actual wind condition are simulated, and key aerodynamic performance indicators such as the force distribution on the blade and the wake structure are obtained, providing a basis for the blade structure design and optimization.

[0032] According to the time series data of the wind condition, boundary condition parameters such as the inlet wind speed, wind direction angle, and turbulence intensity corresponding to each time step are obtained. The obtained boundary condition parameters are dynamically updated into the CFD model as the input conditions for the simulation calculation. Numerical methods such as the finite volume method or the finite element method are used to discretize the fluid mechanics control equations and establish a mathematical model of the flow field around the blade. According to the boundary conditions and initial conditions, the discretized equations are iteratively solved until the residual converges below a preset threshold to obtain a convergent numerical solution of the flow field. The pressure distribution data on the blade surface are extracted from the numerical solution, and the aerodynamic forces on each section of the blade are calculated to obtain the aerodynamic load distribution in the spanwise and chordwise directions of the blade. The velocity field data in the wake region of the blade are extracted from the numerical solution, and flow characteristic parameters such as the turbulence intensity and wake decay rate of the wake are analyzed. The key aerodynamic performance indicators such as the blade aerodynamic force distribution and the wake velocity field are output as the constraint conditions and objective functions for the blade structure design optimization. Optimization methods such as parametric modeling and genetic algorithms are used to perform aerodynamic optimization design on the blade shape.

[0033] Specifically, according to the time series data of wind conditions, the interpolation method can be used to obtain the boundary condition parameters such as the inlet wind speed, wind direction angle and turbulence intensity corresponding to each time step. For example, if the wind speed data at a certain moment is 10 m / s, the wind direction angle is 45°, and the turbulence intensity is 1, these values ​​can be used as the inlet boundary conditions of the CFD model at that moment. In the CFD solver, the finite volume method can be used to discretize the NS equations, and the SIMPLE algorithm can be used to solve the coupling problem of the velocity field and the pressure field. By setting the convergence residual threshold to 10-5, after hundreds of iterations, a converged numerical solution of the flow field can be obtained. Using the pressure data in the numerical solution, the aerodynamic force of each section of the blade can be obtained by integral calculation. For example, for a certain blade section, the span length is 1 m, the chord length is 5 m, the average pressure on the upper surface of the section is 101325 Pa, and the average pressure on the lower surface is 100000 Pa, then the lift of the section can be calculated to be 665 N. By integrating the aerodynamic forces of all blade segments, the aerodynamic load distribution of the entire blade can be obtained. In the wake area of ​​the blade, the velocity data in the numerical solution can be extracted, and the turbulence intensity and wake attenuation rate can be calculated using the spectral analysis method. These aerodynamic performance indicators are used as constraints and objective functions for the optimization design, and the blade shape is parametrically modeled using B-spline curves. The optimal blade aerodynamic shape is searched using a genetic algorithm. After 50 generations of evolution, a blade design with better aerodynamic performance can be obtained.

[0034] It should be noted that the above examples are only some specific embodiments of the present invention. Obviously, the present invention is not limited to the above embodiments, and there are many variations. All variations that can be directly derived or associated with the content disclosed by a person skilled in the art should be considered as the protection scope of the present invention.

Claims

1. A method for simulating and analyzing the aerodynamic performance of a wind turbine blade, characterized in that: The method comprises: According to the time series data of physical quantities such as wind speed, wind direction and turbulence intensity, the original wind condition data is preprocessed by using data cleaning algorithm to obtain the cleaned wind condition time series data; The cleaned wind time series data are normalized by using data normalization method to eliminate the influence of dimension and order of magnitude differences and obtain standardized wind time series data. According to the requirements of wind turbine blade aerodynamic performance simulation analysis on wind data time and spatial resolution, the data sampling frequency and time span are adjusted to obtain wind time series data that matches the computational fluid dynamics simulation time step; The wavelet analysis method is used to perform multi-scale decomposition on the downsampled wind time series data, extract the fluctuation characteristics of wind speed, wind direction and turbulence intensity at different frequency scales, and obtain the stabilized wind time series data. According to the data format requirements of the computational fluid dynamics software for the inlet boundary conditions, the time series data of physical quantities such as wind speed, wind direction and turbulence intensity are mapped to the grid nodes of the computational domain to form a wind boundary condition file that is compatible with the grid topology structure; Adopt data encryption and compression algorithms to encode wind boundary condition files to improve the security and efficiency of data transmission and storage; According to the time series of the wind boundary condition file, the inlet wind speed, wind direction, turbulence intensity and other parameters of the computational fluid dynamics model are dynamically updated to simulate the aerodynamic force and flow field characteristics of the blade under actual wind conditions, and obtain key aerodynamic performance indicators such as the blade force distribution and wake structure.

2. The method according to claim 1, characterized in that The method of preprocessing the original wind condition data using a data cleaning algorithm based on the time series data of physical quantities such as wind speed, wind direction and turbulence intensity to obtain cleaned wind condition time series data includes: According to the original time series data of physical quantities such as wind speed, wind direction and turbulence intensity, the original wind condition data is preprocessed by using data cleaning algorithm; Through the outlier detection algorithm, it is determined whether the physical value at each time point exceeds the preset normal range threshold. If it exceeds, the time point is marked as an outlier point; For abnormal points, the physical value of the abnormal point is estimated by linear interpolation method according to the normal physical value of the time points before and after it, and the value of the abnormal point is replaced by the estimated value; Through the missing value filling algorithm, it is determined whether there are missing values ​​in the physical quantity time series. If there are missing values, the missing values ​​are estimated by linear interpolation method according to the normal physical quantity values ​​before and after the missing values, and filled into the missing position of the original time series; For the wind time series data after outlier correction and missing value filling, digital signal processing algorithms such as moving average filtering are used to remove high-frequency noise interference components in the physical quantity data; Based on the wind time series data after anomaly detection, missing filling and denoising, the final cleaned wind data is obtained; The cleaned wind speed, wind direction and turbulence intensity time series data are used as input for subsequent wind condition characteristic analysis and wind power prediction model to carry out optimized control and power prediction of wind farms.

3. The method according to claim 1, characterized in that The data normalization method is used to normalize the cleaned wind condition time series data to eliminate the influence of dimension and order of magnitude differences, and obtain the standardized wind condition time series data, including: According to the data formats and units of different physical quantities in the wind time series data, determine the types of physical quantities that need to be standardized and the corresponding dimensional magnitudes; The minimum-maximum normalization method is used to normalize the data of each physical quantity, mapping the data to the interval [0, 1] to eliminate the dimension and order of magnitude differences between the data of different physical quantities; For the dimensionless data after standardization, a standardized wind time series dataset is constructed according to the physical quantity type and the timestamp of the wind time series; The convolutional neural network algorithm is used to take the standardized wind time series data as input, extract the local features of the time series data through the convolution layer, and reduce the feature dimension through the pooling layer; On the basis of convolutional neural network, the long short-term memory network (LSTM) layer is introduced to capture the long-term dependency of wind time series data and extract the time-dependent features of the time series; The local features extracted by the convolutional neural network and the time-dependent features extracted by the LSTM are fused to construct a comprehensive feature representation of the wind time series data; Based on the comprehensive characteristics of the fused wind time series, a fully connected neural network is used to predict wind conditions. By training and optimizing network parameters, accurate wind prediction results are obtained.

4. The method according to claim 1, characterized in that: The wavelet analysis method is used to perform multi-scale decomposition on the downsampled wind time series data, extract the fluctuation characteristics of wind speed, wind direction and turbulence intensity at different frequency scales, and obtain the stabilized wind time series data, including: Obtain wind condition time series data required for aerodynamic performance simulation analysis of wind turbine blades; Determine whether the spatiotemporal resolution of wind data meets the simulation analysis requirements. If not, downsample the data. Determine the target sampling frequency and time span of downsampling according to the computational fluid dynamics simulation time step; Use time window averaging or interpolation methods to downsample wind time series data; Through downsampling processing, wind condition time series data matching the simulation time step are obtained; The downsampled wind condition data is input into the aerodynamic performance simulation analysis model; Run simulation analysis to obtain the predicted results of the aerodynamic performance of wind turbine blades under given wind conditions.

5. The method according to claim 1, characterized in that The wavelet analysis method is used to perform multi-scale decomposition on the downsampled wind time series data, extract the fluctuation characteristics of wind speed, wind direction and turbulence intensity at different frequency scales, and obtain the stabilized wind time series data through threshold screening and reconstruction of wavelet coefficients, thereby reducing the randomness and intermittent effects of wind data, including: Downsample the wind time series data to reduce the amount of data and improve the efficiency of subsequent analysis; Wavelet transform is used to perform multi-scale decomposition on the downsampled wind time series data and map the time series to the time-frequency domain. The decomposed wavelet coefficients are screened according to the preset threshold to remove high-frequency noise and outliers; According to the selected wavelet coefficients, the inverse wavelet transform is used to reconstruct and obtain the stabilized wind condition time series data; Conduct statistical analysis on the reconstructed time series data of wind speed, wind direction and turbulence intensity to obtain their respective fluctuation characteristic parameters; According to the fluctuation characteristic parameters of wind speed, wind direction and turbulence intensity, a wind condition random process model is constructed for optimal control of wind farms; The optimized control strategy is applied to wind turbines to improve the power generation efficiency and stability of wind farms and reduce the adverse effects of randomness and intermittency of wind conditions.

6. The method according to claim 1, characterized in that According to the data format requirements of the computational fluid dynamics software for the inlet boundary conditions, the time series data of physical quantities such as wind speed, wind direction and turbulence intensity are mapped to the grid nodes of the computational domain to form a wind boundary condition file that is compatible with the grid topology structure, including: Obtain the inlet boundary condition data format requirements of the computational fluid dynamics software and determine the format conversion and reorganization methods required for wind time series data; Preprocess the stabilized wind time series data, remove outliers and missing values, and perform necessary data cleaning and normalization operations; According to the grid topology of the computational domain, determine the location coordinate information on the grid nodes where wind condition data needs to be generated; The interpolation algorithm is used to interpolate the time series data of physical quantities such as wind speed and wind direction to the position coordinates of the grid nodes; Commonly used interpolation algorithms include linear interpolation and spline interpolation; If the wind condition data contains turbulence intensity information, a suitable turbulence model, such as the k-ε turbulence model, is used to map the turbulence intensity data to the grid nodes to match the wind speed data; The interpolated and mapped wind data are fitted by a fitting algorithm, such as the least square method, to obtain a continuous and smooth wind data surface adapted to the grid node positions; According to the inlet boundary condition data format requirements of the computational fluid dynamics software, organize the fitted wind condition data into a boundary condition file, ensure that the file format is consistent with the software requirements, and save the file.

7. The method according to claim 1, characterized in that The method of encoding the wind boundary condition file using a data encryption and compression algorithm to improve the security and efficiency of data transmission and storage includes: According to the characteristics of the wind boundary condition file, select a suitable data encryption algorithm to encrypt the file to ensure the security of the data during transmission and storage; According to the data volume and access frequency of the wind boundary condition file, a suitable data compression algorithm is used to compress the file to reduce the file size and improve the transmission and storage efficiency; Encode the encrypted and compressed wind boundary condition file to generate a unique file identification code for easy subsequent management and call; Establish an association index table between wind condition data and blade models, record the blade model parameters corresponding to each wind condition data file, and realize the association between data and models; In the computational fluid dynamics model, the wind boundary condition file matching the current blade model is quickly found through the associated index table, and the file data is read; Decode and decompress the read wind boundary condition file to restore the original data content for subsequent model calculation; The wind condition data is seamlessly transmitted to the computational fluid dynamics model, enabling rapid call and integration of the model and data, thus improving the operating efficiency of the entire system.

8. The method according to claim 1, characterized in that According to the time series of the wind boundary condition file, the parameters such as the inlet wind speed, wind direction and turbulence intensity of the computational fluid dynamics model are dynamically updated to simulate the aerodynamic force and flow field characteristics of the blade under actual wind conditions, and obtain key aerodynamic performance indicators such as the force distribution and wake structure of the blade, including: According to the time series data of wind conditions, the boundary condition parameters such as inlet wind speed, wind direction angle and turbulence intensity corresponding to each time step are obtained; Dynamically update the acquired boundary condition parameters to the CFD model as input conditions for simulation calculations; Using numerical methods such as finite volume method or finite element method, discretize the fluid mechanics governing equations and establish a mathematical model of the flow field around the blades; According to the boundary conditions and initial conditions, the discretized equations are solved iteratively until the residual converges to below the preset threshold, and a converged numerical solution of the flow field is obtained; Extract the pressure distribution data on the blade surface from the numerical solution, calculate the aerodynamic force of each section of the blade, and obtain the aerodynamic load distribution in the span and chord directions of the blade; Extract the velocity field data of the blade wake area from the numerical solution and analyze the flow characteristic parameters such as the turbulence intensity and wake attenuation rate of the wake; The key aerodynamic performance indicators such as blade aerodynamic force distribution and wake velocity field are output as constraint conditions and objective functions for blade structure design optimization. Optimization methods such as parametric modeling and genetic algorithms are used to perform aerodynamic optimization design of the blade shape.

Images