Turbine blade surface skin parameterization mapping grid creation and automatic interpolation method

Abstract

The invention discloses a turbine blade surface skin parameterized mapping grid creation and automatic interpolation method, which realizes efficient post-processing of blade surface simulation data through molded line partition splitting, arc length parameterized grid generation and physical quantity data interpolation, and comprises the following steps of: firstly, performing partition preprocessing on a blade molded line according to a pressure surface, a suction surface and front and rear tail edges; constructing a body-attached parameterized grid; then simulation data are imported and automatically partitioned, an optimization interpolation algorithm is adopted for different regions, and the interpolation precision is improved in combination with an invalid value correction technology; and finally, mapping the three-dimensional data to a two-dimensional parameter plane to realize feature visualization and analysis. According to the method, manual intervention is remarkably reduced through an automatic process, the post-processing efficiency is improved while the data precision is guaranteed, and reliable support is provided for pneumatic optimization and cooling design of the turbine blade.

Description

Technical Field

[0001] The present invention belongs to the technical field of engineering thermophysics, and in particular relates to a method for creating and automatically interpolating a body-fitting parameterized mapping grid on a turbine blade surface. Background Art

[0002] Turbine blades are core components in power units such as gas turbines and aircraft engines. They are mainly responsible for converting the energy of high-temperature and high-pressure gas into mechanical energy, driving the turbine to rotate and generate power. They are usually made of high-temperature resistant alloys (such as nickel-based superalloys) or ceramic composites, and integrate precise aerodynamic shapes and internal cooling channels to ensure stable operation in extreme temperatures (up to 1500°C) and high-pressure environments. Modern turbine blades also often use advanced processes such as single crystal casting technology and thermal barrier coatings (TBC) to further improve heat resistance and efficiency. Its performance directly affects the thrust, fuel efficiency and reliability of the engine, and it is one of the key technologies in the fields of aerospace, energy generation, etc.

[0003] Computational fluid dynamics (CFD) plays an irreplaceable core role in the research and development of turbine blades. Through high-precision numerical simulation, it can not only deeply analyze the complex flow field characteristics of the blade surface (including pressure distribution, velocity field, shock wave position and secondary flow), but also optimize the design parameters of the cooling system (such as internal flow channel structure, film cooling hole arrangement, etc.), significantly improving the cooling efficiency. At the same time, CFD can perform thermal-fluid-solid coupling analysis, accurately predict the temperature field distribution and thermomechanical load of the blade under extreme working conditions, and provide a key basis for structural strength and life assessment. In addition, CFD technology can also simulate turbulent characteristics and flow control schemes to guide the aerodynamic optimization design of blades. Compared with traditional test methods, CFD greatly reduces R&D costs and time, making design iterations more efficient. With the development of high-performance computing and artificial intelligence, CFD is being deeply integrated with advanced technologies such as digital twins, driving turbine blades towards higher efficiency, longer life, and greater intelligence. However, the resulting problems of data processing and presentation are becoming increasingly prominent. The large amount of simulation data obtained by CFD often requires post-processing software to extract parameters such as temperature and surface friction required in the turbine design stage, and to generate charts in a standardized and unified format. This requires a lot of manpower and time for a large number of calculation cases.

[0004] With the development of high-performance turbomachinery towards high load and high precision design, the complexity of blade surface geometry has been significantly improved, and the surface often presents highly twisted three-dimensional curved surface characteristics. However, the complex curved surface geometry of the turbine blade faces the challenges of low data mapping efficiency, difficulty in local feature analysis and insufficient multi-physical field coupling visualization in simulation post-processing. The traditional post-processing methods such as physical quantity cloud map analysis are not comprehensive and intuitive for data analysis and feature mining. Although the general simulation post-processing software is rich in functions and convenient to use, it can provide good processing effect for many problems and simulation results, but in the simulation data post-processing of turbine blade, especially in the surface data display of flow field and thermal field simulation, there are many deficiencies, such as the perspective is not intuitive when the cloud map is displayed. SUMMARY

[0005] In order to overcome the problems existing in the prior art, the purpose of the present application is to provide a method for creating a surface body parameterized mapping grid of a turbine blade and automatically partitioning and interpolating simulation data of the blade surface based on MATLAB, which significantly reduces manual intervention through an automated process, improves post-processing efficiency while ensuring data accuracy, and provides reliable support for aerodynamic optimization and cooling design of turbine blades.

[0006] The present application is realized by the following technical solutions:

[0007] A method for creating a surface body parameterized mapping grid of a turbine blade and automatically interpolating based on MATLAB, the specific steps are as follows:

[0008] Step 1: Preprocessing of turbine blade profile, dividing the blade surface profile into pressure surface, suction surface and leading and trailing edges, and obtaining a basic data set.

[0009] Step 2: Generating a body arc length parameterized grid using the preprocessed blade profile, setting the number of arc length direction nodes num_u, and setting the number of blade radial nodes num_v.

[0010] Step 3: Importing the blade surface related physical quantity data file, and similarly, completing the partition screening.

[0011] Step 4: Using corresponding interpolation methods for different regions, and performing invalid value retrieval and correction on the interpolated data.

[0012] Step 5: Visualization of the interpolation results, mapping the data interpolated onto the three-dimensional grid to a two-dimensional parameter plane for display, so as to identify and analyze the features. Due to the structural characteristics of the parameterized grid, the interpolated data can be mathematically processed and used for deep learning and other purposes.

[0013] Further, the specific steps for obtaining the basic data set in step 1 include:

[0014]

[0014] Step 1-1: Extract its surface profile, in this invention, in order to fully capture the geometric features of the surface, it is necessary to extract the three-dimensional coordinate file of the point set at the upper and lower and middle section line position, this step is realized by the point set command in UG, in the point set command, select the corresponding curve, and set the number of equal arc length points, you can evenly arrange a large number of points on the curve.

[0015] Extract the three-dimensional coordinate file of the point set at the upper and lower and middle section line position of the blade, and export it as a.txt file;

[0016] Select three groups of point sets respectively, export the xyz geometric coordinates as.txt files, named as 'up','mean' and 'down' respectively.

[0017] Step 1-2: Import the three groups of point set coordinate files in array format in MATLAB.

[0018] Step 1-3: Execute the profile splitting module, the splitting principle is as shown in Figure 2 The module first identifies the leading edge stop point and the trailing edge end point of the profile, and then calculates the angle and distance relative to the assumed arc center for the remaining points.

[0019] Step 1-4: Divide all points into n small intervals according to the arc center angle, and obtain a point on the near center arc by taking the arithmetic mean of the point coordinates in the interval. The distance of the arc center of the points in the interval relative to this point is the basis for the suction and pressure surface partition. By adjusting n, the recognition accuracy can be improved. Store the coordinates of the divided profile points for use in the next step. Visualize the split profile results.

[0020] Further, the specific steps of generating the arc length parameterized mesh of the skin in step 2 include:

[0021] Step 2-1: Calculate the arc length of each point on the profile to the leading edge, and obtain the normalized arc position of each point by adding the distance point by point in order of arc length, i.e. the leading edge end point is 0, the trailing edge end point is 1, and the intermediate point is 0-1.

[0022] Step 2-2: Uniformly divide the arc length space [0, 1] of each profile into num_v parts, and then interpolate the original coordinate points on the profile to the parameter space [0, 1] to obtain num_v evenly distributed coordinate points.

[0023] Step 2-3: For the same parameter point v on a surface, there are three coordinate points distributed along the radial direction, and then the three coordinate points are uniformly interpolated as num_u coordinate points in the radial direction of the blade in the form of spline curve.

[0024] Steps 2-4: All interpolated coordinate points are evenly distributed in the UV parameter space, and the interpolated mesh is calculated. The mesh is split along the U direction to create the leading and trailing edge arcs. The parameter mesh is visualized.

[0025] Furthermore, the specific steps of importing physical quantity data and partitioning data points in step 3 include:

[0026] Step 3-1: After obtaining the corresponding turbine blade simulation results in simulation software such as FLUENT, export the data points of a certain physical quantity on the entire blade surface as a .txt format file.

[0027] The grid coordinate points on the suction and pressure sides are summed and averaged to obtain uniform grid points approximately on the mid-arc surface.

[0028] Step 3-2: Use the knn-search algorithm to find the nearest grid point on this grid for all imported data points, and determine the position of the data point on the blade by the angle between the arc center, grid point, and data point vector.

[0029] The data point position determination is performed separately for the leading edge and the trailing edge.

[0030] Step 3-3: Store the index of all regional data points and visualize the data point partitions.

[0031] Furthermore, in step 4, the specific steps of using corresponding interpolation methods for different regions and performing invalid value retrieval and correction on the interpolation data include:

[0032] Step 4-1: For the suction and pressure surfaces except the leading edge, use the scatteredInterpolant interpolation method based on the x,z coordinates.

[0033] Step 4-2: For the leading edge, the scatteredInterpolant interpolation method based on y and z coordinates is used. After the interpolation is completed, the interpolated data is stored for visualization and data analysis.

[0034] Step 4-3: Remove null values ​​and outliers and re-interpolate. The identification of null values ​​and outliers depends on conditional indexing. It can be specified that null values ​​and data points outside the specified range are indexed and re-interpolated. The scatteredInterpolant interpolation method is also used, and the extrapolation method is nearest.

[0035] Beneficial effects of the present invention:

[0036] 1. The present invention can automatically divide the imported profile coordinate file into feature regions and create a highly fitted arc-length parameterized interpolation grid. At the same time, it automatically completes regional classification for the data points required for interpolation, thereby improving the interpolation effect. The generated arc-length parameterized grid can also be used for many purposes, such as locating air film holes on the blade surface.

[0037] 2. The present invention greatly improves the efficiency of post-processing of turbine blade simulation data and provides a novel and intuitive display method for the visualization of physical quantities on the blade surface.

[0038] 3. The present invention reduces the reliance on manual labor in the post-processing of turbine blade surface data. Only the corresponding files need to be provided to automatically execute the entire process and output the target result image and data file. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Schematic diagram of the imported blade model profile coordinate points.

[0040] Figure 2 Schematic diagram of the line after automatic splitting by area.

[0041] Figure 3 Schematic diagram of the reconstructed arc-length parameterized blade surface.

[0042] Figure 4 Schematic diagram of the disassembled leading edge of the blade.

[0043] Figure 5 Schematic diagram of the interpolated mesh and its local details obtained by projecting the parameterized surface into the two-dimensional parameter space.

[0044] Figure 6 Schematic diagram of the point set and its local details after the imported data point set is automatically separated by region.

[0045] Figure 7 Schematic diagram of the interpolation result without local interpolation correction and its leading edge details.

[0046] Figure 8 Schematic diagram of the interpolation result after local interpolation correction and its leading edge details.

[0047] Figure 9 Schematic diagram of the results of lateral averaging of interpolated data.

[0048] Figure 10 Schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0049] The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present invention, but should not be construed as limiting the present invention.

[0050] The embodiments of the present invention are described in detail below with reference to the accompanying drawings. Figure 10 Schematic diagram of the process of the present invention.

[0051] A method for creating a body-fitting parameterized mapping grid and automatically interpolating the surface of a turbine blade based on MATLAB is described. The specific steps are as follows:

[0052] Step 1: The turbine blade object used in this example is the GE-E3 turbine high-pressure first-stage guide vane. Pre-process the turbine blade profile. First, create a section in the blade to intercept the middle profile, and the upper and lower profiles are the blade tip and blade root. Then, in the point set command, select the above three curves respectively and set the number of equal arc length points. Then, a large number of points can be evenly arranged on the curve, such as Figure 1 As shown, this will help capture the curvature and structural features of the surface. Ultimately, the geometric coordinates of the three groups of points on the three curves are written out as three sets of .txt files and named according to their locations. Import the three sets of profile .txt files into the workspace, and then execute different functional modules according to the process. First, split the imported profile into the suction side and the pressure side. For the leading edge of the high-pressure guide vane, it can generally be geometrically defined as the front end of the flow axis, and the trailing edge is defined as the farthest point from the leading edge. After determining the beginning and end of the suction and pressure sides of the profile, for the partitioning of the coordinate points in the middle area, the idea of ​​partial judgment is used here. A point is selected at the far end of the blade as the arc center. The leading edge and trailing edge form an angle θ with the line connecting them. θ is evenly divided into n parts. Then, the angle between the vector formed by the remaining points on the profile and the base point and the vector between the leading edge and the arc center is calculated to determine the angle range of each point. After all points are classified by angle, the points can be sorted in order from small to large according to the angle. Then the spatial distance of the points in each interval about the arc center is averaged. Since the number of points belonging to the suction surface and the pressure surface in each interval is basically the same, the calculated average coordinate should be located near the mid-arc line. At this time, it is easy to know that the points whose spatial distance from the point to the arc center is greater than the average coordinate-arc center coordinate distance should be on the suction side, otherwise they should be on the pressure side. In this way, the classification and sorting of the points on the profile are completed. Then, the visualization module can be used to display the profile suction surface and pressure surface separately as follows Figure 2 shown.

[0053] Step 2: Calculate the arc length from each point on the profile to the leading edge by accumulating the distance point by point in the order of arc length, and normalize the arc position of each point, that is, the leading edge endpoint is 0, the trailing edge endpoint is 1, and the middle point is 0~1. Decompose the arc length space [0,1] of each profile evenly into num_u parts. In this example, num_u is 800. Then interpolate the original coordinate point on the profile to the parameter space [0,1] to obtain num_v coordinate points that are evenly distributed. In this example, num_v is 336. For the same parameter point v on a surface, there are three coordinate points distributed radially. Then, these three coordinate points are uniformly interpolated into num_u coordinate points in the radial direction of the blade according to the spline curve. All interpolated coordinate points are evenly distributed in the U×V parameter space, and the interpolation grid calculation is completed. Split the grid into leading edge and trailing edge arc surfaces in the u direction. The parameter grid is visualized as follows Figure 3 and Figure 4 As shown, in order to finally interpolate the numerical calculation data onto the visual plane, the three-dimensional parameterized surface mesh is projected onto the two-dimensional parameter plane as shown in Figure 5 shown.

[0054] Step 3: Import the .txt file of the simulation physical quantity data related to the GE-E3 blade surface, and similarly, complete the partitioning and screening for it. In the present invention, the partitioning of the data points on the high-pressure guide vane surface is realized based on the interpolation grid. The suction surface and pressure surface grids are both U×V two-dimensional parameter space grids, and are connected end to end. Each node on the grid surface corresponds to each other. If the arithmetic average of the three-dimensional space coordinates of the corresponding grid nodes is taken, a U×V two-dimensional parameter grid is finally obtained. Here we consider it to be the mid-arc surface grid in the blade, and the trajectory is approximately the mid-arc line. Then, using the knn-search command in MATLAB, we need to find a nearest mid-arc surface grid node for each imported data point. After storing the index of each nearest neighbor point, calculate the vector of each data point and its nearest neighbor's spatial coordinates, and determine the direction. Taking the Y-axis as an example, if the vector direction is negative on the Y-axis, it means that the data point is above the mid-arc surface grid, that is, the data point is on the suction surface side, otherwise the data point is on the pressure surface side. All data points on the suction surface side and the pressure surface side are stored in the structure of the corresponding surface respectively. The split data points are as follows: Figure 6 shown.

[0055] Step 4: Use the corresponding interpolation method for different areas. For the leading edge of the turbine blade, it has a large curvature and bending angle. The interpolation effect at this position is often not good, because the grid points and data point coordinates of the leading edge arc surface are relatively high in the x-direction (blade axial direction), resulting in a large interpolation error. Therefore, the leading edge grid needs to be split out for local interpolation. The same treatment can be used for the trailing edge, but the proportion of the trailing edge is generally very small and can be ignored. The leading edge can be split simply by specifying an axial distance from the starting point of the leading edge. The definition of the leading edge does not need to be very precise. It is only for local interpolation. The interpolation result before local interpolation correction is as follows: Figure 7 As shown. Then, invalid values ​​are retrieved and corrected for the interpolation data. In the previous interpolation process, due to errors in the geometric position of the grid or exceeding the data boundary, null values ​​and abnormal values ​​may be generated, which need to be eliminated. However, it is not reasonable to simply assign them a fixed value. A feasible way is to use the data points around these values ​​to re-interpolate new values ​​for these points, which can ensure the continuity and smoothness of physical information in these areas. The identification of null values ​​and abnormal values ​​depends on conditional indexing. It can be specified that null values ​​and data points outside the specified range are indexed and re-interpolated. The scatteredInterpolant interpolation method is also used, and the extrapolation method is nearest. The final interpolated image is shown as follows. Figure 8 shown.

[0056] Step 5: Visualize the interpolation results. Map the data interpolated on the three-dimensional grid to the two-dimensional parameter plane for feature recognition and analysis. Due to the structural characteristics of the parameterized grid, mathematical processing of the interpolated data can be performed very conveniently. For example, the lateral data along the x-axis can be averaged. Figure 9 shown.

[0057] The present invention achieves efficient post-processing of blade surface simulation data through line partitioning, arc length parameterized grid generation, and physical quantity data interpolation. First, the blade line is pre-processed according to the pressure surface, suction surface, and front and rear trailing edges to construct a body-fitting parameterized grid. Then, the simulation data is imported and automatically partitioned. An optimized interpolation algorithm is used for different areas, and the invalid value correction technology is combined to improve the interpolation accuracy. Finally, the three-dimensional data is mapped to a two-dimensional parameter plane to achieve feature visualization and analysis. The automated process significantly reduces manual intervention, improves post-processing efficiency while ensuring data accuracy, and provides reliable support for the aerodynamic optimization and cooling design of turbine blades.

Claims

1. A method for creating and automatically interpolating a body-fitting parameterized mapping grid for a turbine blade surface, characterized in that: The following steps are involved: 1) Preprocess the turbine blade profile and split the blade surface profile into pressure side, suction side, and front and rear trailing edges to obtain a basic data set; 2) Generate a body-fitting arc-length parameterized mesh using the preprocessed blade profile, and set the number of nodes num_u in the arc-length direction and the number of nodes num_v in the blade radial direction; 3) Import the blade surface physical quantity data file and complete the partition screening of this data; 4) Use corresponding interpolation methods for different regions and perform invalid value retrieval and correction on interpolation data; 5) Visualize the interpolation results and map the three-dimensional data onto a two-dimensional parameter plane for easy feature recognition and analysis.

2. The method for creating and automatically interpolating a body-fitting parameterized mapping grid for a turbine blade surface according to claim 1, characterized in that: The specific steps for obtaining the basic data set in step 1) include: 1-1) Extract the 3D coordinate file of the point set at the upper, lower, and middle section positions of the blade and export it as a .txt file; 1-2) Import the three sets of point coordinate files mentioned above into MATLAB in array format; 1-3) Execute the line splitting module to identify the leading edge end point and the trailing edge end point, and calculate the angle and distance of the remaining points relative to the assumed arc center; 1-4) Divide the points in each interval by the arc center angle, and calculate a point on the arc near the center, which is used as the basis for the suction and pressure surface divisions.

3. The method for creating and automatically interpolating a body-fitting parameterized mapping grid for a turbine blade surface according to claim 1, characterized in that: The specific steps of generating the body-fitting arc-length parameterized grid in step 2) include: 2-1) Calculate the arc length from each point on the profile to the leading edge by adding up the distances point by point in arc length order, and normalize the arc position of each point, i.e., the leading edge endpoint is 0, the trailing edge endpoint is 1, and the midpoint is 0-1; 2-2) Decompose the arc length space [0,1] of each profile into num_v parts uniformly, and then interpolate the original coordinate points on the profile into the parameter space [0,1] to obtain num_v uniformly distributed coordinate points; 2-3) For the same parameter point v on a surface, there are three coordinate points distributed along the radial direction. Then these three coordinate points are uniformly interpolated into num_u coordinate points in the blade radial direction using a spline curve. 2-4) All interpolated coordinate points are evenly distributed in the U×V parameter space to form an interpolation grid.

4. The method for creating and automatically interpolating a body-fitting parameterized mapping grid for a turbine blade surface according to claim 1, characterized in that: The specific steps of importing physical quantity data and partitioning it in step 3) include: 3-1) Export the data points of a physical quantity on the entire blade surface from the simulation software into a .txt file; 3-2) Use the knn-search algorithm to find the nearest grid point for all imported data points, and determine the position of the leaf where the data point is located by the vector angle; 3-3) Store the indexes of all regional data points.

5. The method for creating and automatically interpolating a body-fitting parameterized mapping grid for a turbine blade surface according to claim 1, characterized in that: In step 4), the specific steps of using corresponding interpolation methods for different regions and retrieving and correcting invalid values ​​of interpolation data include: 4-1) For the suction and pressure surfaces except the leading edge, the scatteredInterpolant interpolation method based on x and z coordinates is used; 4-2) For the leading edge, the scatteredInterpolant interpolation method based on y and z coordinates is used; 4-3) Null value removal and outlier removal and re-interpolation: data points with null values ​​and values ​​outside the specified range are indexed and re-interpolated.

Images