Cooling efficiency deviation prediction method for internal cooling structure of turbine blade after precision casting

Abstract

The invention discloses a cooling efficiency deviation prediction method of a turbine blade inner cooling structure after precision casting, and relates to the field of aero-engines. Casting process parameters are randomly extracted through Latin hypercube sampling and serve as input parameters of a fitting function; procast software is used for extracting point cloud data of turbine blade deformation through casting simulation; the Geomagic software is used for completing reversing and measuring the deformation amount; modifying the design model by using Catia modeling software according to the measured deformation; cFD simulation and cooling efficiency calculation of the blade are completed through Fluent, and the CFD simulation and the cooling efficiency calculation serve as output parameters of a fitting equation; and training and optimizing the Kriging proxy model, and predicting the cooling efficiency deviation by using the model. The reverse iteration period is shortened, the workload of subsequent iteration design and flow field analysis is reduced, cooling efficiency prediction is completed more quickly, and the verification period of blade cooling design is shortened.

Description

Technical Field

[0001] The invention relates to the field of aeroengines, and in particular to a method for predicting cooling efficiency deviation of an internal cooling structure of a turbine blade after precision casting. Background Art

[0002] The hot end turbine components of modern advanced aircraft engines are in high temperature, high speed and high load environment for a long time, and the working conditions are deteriorating. Therefore, in order to maintain the normal operation of turbine blades and increase the working life of turbine blades, complex and high-performance cooling structure is an indispensable technical means for modern advanced gas turbine high temperature turbines, and it is also a major technical problem recognized in China.

[0003] Whether the turbine blades can work safely and reliably in a high temperature environment mainly depends on the temperature distribution of the blades. The temperature distribution of the blades mainly depends on the internal cooling structure of the blades. In order to ensure that the engine has good reliability and durability, the premise of increasing the temperature before the turbine is to effectively cool the high-temperature components. For rotor blades, in order to improve cooling efficiency and save cooling air, blades often adopt multi-channel or multi-loop convection cooling channel design to ensure sufficient cooling of the entire blade; and the cooling channel is often provided with spoiler ribs or / and spoiler columns to enhance local heat exchange. The high performance of advanced gas engines has more stringent requirements on processing accuracy and operating performance, which urgently requires the introduction of the third level of turbine cooling design, that is, the deterioration of operating conditions, manufacturing tolerances and processing deviations in the design. In the manufacturing process of turbine blades, affected by factors such as processing accuracy, there is a certain deviation between the internal cooling structure of precision-cast turbine blades and the design model. This deviation has a certain weakening effect on the cooling effect of turbine blades. The influence of different process parameters (such as pouring temperature, mold shell temperature, and pulling speed) on the deformation of the blade also leads to a certain influence on the cooling efficiency. This will be one of the most important directions for future research on turbine cooling. Summary of the invention

[0004] The purpose of the present invention is to provide a cooling efficiency deviation prediction method for the internal cooling structure of turbine blades after precision casting based on the Kriging space interpolation method, in order to address the shortcomings of complex turbine blade internal cooling structure model design and low casting efficiency. The method replaces complex reverse modeling by modifying the design model, and replaces the iteration of the digital model by parameter iteration of the proxy model. While reducing the workload of designers in subsequent iterative design and flow field analysis, the work efficiency of designers is improved, the design and manufacturing compliance of turbine blades is improved, and the cooling effect of the internal cooling structure of turbine blades is predicted quickly, accurately and efficiently, so as to provide the best initial design scheme for the robust and reliability design of the internal cooling structure of turbine blades.

[0005] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions.

[0006] The method for predicting the cooling efficiency deviation of the internal cooling structure of a turbine blade after precision casting comprises the following steps:

[0007] 1) Determine the range of simulation parameters and randomly sample;

[0008] 2) Establish a turbine blade casting system;

[0009] 3) Simulate the directional solidification process of turbine blades to obtain point cloud data of the blades after precision casting;

[0010] 4) Preliminary model inversion and structural deformation extraction, and modification of the design model based on the deformation data;

[0011] 5) Perform CFD fluid simulation on blades deformed by casting and establish a mapping database between process parameters and cooling efficiency;

[0012] 6) Using three randomly sampled process parameters as input and the cooling efficiency parameters obtained by CFD fluid simulation as output, the Kriging proxy model is trained to predict the cooling efficiency of the internal cooling structure of turbine blades under different casting conditions.

[0013] In step 1), the simulation parameter range is determined and random sampling is performed to identify key simulation process parameters that affect the turbine blade casting process and cooling efficiency, including pouring temperature, cooling rate, mold temperature, etc.; Latin hypercube sampling is used to randomly select values ​​of the three process parameters within a certain range, so that the original samples extracted are relatively uniform and will not produce obvious aggregation phenomena, and samples are forced to be extracted from each layer to ensure the comprehensiveness of the sample results; random sampling is performed within a reasonable parameter range to simulate different casting conditions.

[0014] In step 2), the specific steps of establishing the turbine blade casting system may be: using Catia modeling software to process the turbine blade model, and establishing a casting system adapted thereto, including casting bodies such as runners, cold copper, and furnace bodies during the casting process, to save computing costs and ensure model accuracy.

[0015] In step 3), the specific steps of simulating the directional solidification process of the turbine blades may be: using Procast casting software to simulate the directional solidification process of the turbine blades, utilizing multi-scale numerical simulation of the directional solidification process on the macroscopic temperature field, utilizing a microstructure growth numerical model on the microscopic level, establishing a multi-scale numerical simulation of the directional solidification process, and obtaining point cloud data of the blades after precision casting, which data reflect the geometric shape of the blades after casting.

[0016] In step 4), the preliminary inversion of the model and the extraction of the structural deformation are to preliminarily reverse the blade point cloud data obtained by simulation in Geomagic software, extract the deformation data of the internal cooling structure of the blade (including spoiler columns and spoiler ribs), and establish a mapping relationship database between process parameters and internal structural deformation; the modification of the design model based on the deformation data is to establish a design table based on the extracted structural deformation data, so as to quickly modify the design model and ensure the accuracy of the model to reflect the deformation conditions in the actual casting process.

[0017] In step 4), the model is initially reversed and the structural deformation is extracted. The specific steps for modifying the design model based on the deformation data are as follows: after the initial reverse, the deformation data of the internal cooling structure of the blade is extracted, the deformation data includes: the radius r of the spoiler column, the height e of the spoiler rib, the spacing P of the spoiler rib, the width d of the slit, a mapping database between the process parameters and the internal structural deformation is established, a design table is established in Catia based on the extracted structural deformation data, and after obtaining new deformation data later, the design table parameters can be directly modified in Excel to update the blade model after the precision casting deformation.

[0018] In step 5), the specific steps of performing CFD fluid simulation on the blades deformed by casting may be: using Fluent numerical simulation software to perform fluid simulation analysis on the turbine blades deformed by casting, simulating the flow of cooling fluid in the cooling structure inside the blades; comparing with the original design model without casting deformation, analyzing the influence of the deformation of the internal cooling structure of the turbine blades after precision casting on the cooling efficiency of the blades, recording the cooling efficiency parameters under different process parameters, and establishing a mapping relationship database between process parameters and cooling efficiency.

[0019] In step 6), the training Kriging proxy model predicts the cooling efficiency of the internal cooling structure of the turbine blade under different casting conditions, selects a certain number of randomly sampled three process parameters as training sample inputs, and the corresponding CFD simulation cooling efficiency parameters are used as outputs, and uses these data to train the Kriging proxy model; the Kriging proxy model overcomes the limitation that the parameterized model must establish a specific mathematical model, and is an approximate technology with more statistical characteristics, and the effectiveness of the model does not change with the change of random errors; the cooling efficiency of the internal cooling structure of the turbine blade under different casting conditions is predicted by the trained Kriging proxy model, thereby realizing a fast, accurate and efficient prediction of the cooling effect of the internal cooling structure of the turbine blade.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] The present invention provides a method for predicting the cooling efficiency deviation of the inner cooling structure of a blade based on the Kriging spatial interpolation method. By using random sampling techniques such as Latin hypercube sampling, a wide range of process parameter spaces can be efficiently covered under limited computing resources, thereby quickly obtaining a large amount of simulation data. The introduction of the Kriging proxy model enables the cooling efficiency under different process parameters to be quickly predicted without the need to re-perform a complete CFD simulation, thereby greatly improving the prediction efficiency. Both the directional solidification process simulation and the CFD fluid simulation are based on advanced numerical simulation technology, and can more accurately simulate the physical phenomena and fluid flow conditions in the casting process, thereby providing reliable simulation data. The established mapping relationship database between the process parameters and the cooling efficiency, as well as the Kriging proxy model after verification and adjustment, further improve the accuracy of the prediction. The method replaces complex reverse modeling by modifying the design model, replaces the iteration of the digital model by the parameter iteration of the proxy model, shortens the reverse iteration cycle, reduces the workload of subsequent iterative design and flow field analysis, and can provide the best initial design scheme for the robust and reliable design of the internal cooling structure of the turbine blade, which is of great significance for improving the casting quality and cooling efficiency of the turbine blade. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is a flow chart of the design process of the present invention;

[0023] Figure 2 This is a schematic diagram of random value selection of three process parameters of the present invention;

[0024] Figure 3 It is a schematic diagram of the turbine blade casting system of the present invention;

[0025] Figure 4 It is a preliminary reverse schematic diagram of the internal flow field of the present invention;

[0026] Figure 5 It is a schematic diagram of the CFD calculation model of the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the following embodiments will further illustrate the present invention in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. On the contrary, the present invention encompasses any substitutions, modifications, equivalent methods and schemes made on the essence and scope of the present invention as defined by the claims. Further, in order to enable the public to have a better understanding of the present invention, some specific details are described in detail in the following detailed description of the present invention. For those skilled in the art, the present invention can be fully understood without the description of these details.

[0028] like Figure 1 As shown, the embodiment of the present invention includes the following steps:

[0029] Step 1: Random values ​​of simulation parameters

[0030] Latin hypercube sampling is used to randomly select values ​​of three process parameters within a certain range, namely, pouring temperature, mold shell temperature and pulling speed, so that the original samples extracted are relatively uniform and will not produce obvious aggregation phenomena. Samples are forced to be extracted from each layer to ensure the comprehensiveness of the sample results.

[0031] In this embodiment, a three-dimensional sampling space is set up, and random numbers are extracted within the three-dimensional space range of pouring temperature 1500-1650°C, mold shell temperature 1500-1650°C, and pulling speed 4.5-7.5mm / min. Each dimension is divided into several non-overlapping intervals so that each interval has the same probability. A point is randomly extracted from each interval in each dimension. These points constitute the population of the Latin hypercube. Then, points selected from the population are randomly extracted from each dimension and they are combined into vectors; that is, the Matlab code of Latin hypercube sampling is called to generate uniform random number samples for each dimension, and the order of each column of sample points is randomly shuffled to avoid correlation. The schematic diagram of random value selection of three process parameters is shown in the figure. Figure 2 shown.

[0032] Step 2: Establish the turbine blade casting system

[0033] Use Catia modeling software to process the turbine blade model and establish a casting system that is compatible with it, including the pouring channel, cold copper, furnace body and other casting entities during the casting process, saving calculation costs and ensuring model accuracy;

[0034] In this embodiment, Catia modeling software is used to establish a pouring system adapted to the turbine blade model, including pouring channels, cold copper, furnace bodies and other pouring bodies during the pouring process, which are specifically refined from top to bottom as follows: gate 1, pouring channel 2, blade 3, seeding section 4, spiral crystal selector 5, cold copper 6. The schematic diagram of the turbine blade pouring system is shown in FIG. Figure 3 As shown. The main research object of blade 3 is the blade body. The tenon and other parts are simplified during simulation to reduce the amount of calculation while maintaining the key features that affect the pouring system. The ceramic core is obtained after the blade is processed by Catia's Boolean operation. As a transition body to form the casting cavity, the ceramic core contains complex structures such as spoiler ribs and spoiler columns to enhance the cooling effect and internal quality of the casting. The furnace body is a heat radiation surface, which is a sheet body and is divided into high temperature zone, medium temperature zone, and low temperature zone. This is achieved by creating sheets with different temperature gradients in Catia.

[0035] Step 3: Directional solidification process simulation

[0036] Procast casting software was used to simulate the directional solidification process of the blade. Multi-scale numerical simulation of the directional solidification process was used on the macroscopic temperature field, and the microstructure growth numerical model was used on the microscopic field to establish a multi-scale numerical simulation of the directional solidification process and obtain the point cloud data of the blade after precision casting.

[0037] In this embodiment, the Procast casting software is used to simulate the directional solidification process of the blade described in step 2, and the blade model is meshed. The number of body meshes in the simulation pre-processing is between 3.5 million and 4.5 million to ensure the balance between simulation accuracy and computational efficiency. According to the n groups of process parameter simulation data randomly selected in step 1, the corresponding pouring temperature, mold shell temperature, and pulling speed are set in Procast, and the blade material can be set to nickel-based alloys, ceramic-based composite materials, etc. (such as DD6 nickel-based single crystal high-temperature alloy), and the numerical simulation of the directional solidification process is established to obtain the .stl point cloud data file of the blade after precision casting.

[0038] Step 4: Preliminary reverse engineering of the model and extraction of structural deformation, and modification of the design model based on the deformation data

[0039] Perform preliminary reverse engineering of blade point cloud data in Geomagic software, extract deformation data of blade internal cooling structure (including spoiler columns and spoiler ribs), establish a mapping database between process parameters and internal structure deformation, and establish a design table based on the extracted structural deformation data to quickly modify the design model and ensure model accuracy;

[0040] In this embodiment, the preliminary reverse engineering of the blade point cloud data is performed in Geomagic software. The schematic diagram of the preliminary reverse engineering of the internal flow field is shown in FIG. Figure 4 As shown. Import the .stl point cloud data file exported in step three into Geomagic. Perform preprocessing operations such as denoising, smoothing, and filling holes on the point cloud data to improve the accuracy of the reverse model. Based on the processed point cloud data, use Geomagic's reverse engineering tool to generate a preliminary blade model. Extract the deformation data of the internal cooling structure of the blade (including spoiler columns and spoiler ribs). The deformation data includes: spoiler column radius r, spoiler rib height e, spoiler rib spacing P, slit width d, and establish a mapping relationship database between process parameters and internal structure deformation. According to the extracted structural deformation data, a design table is established in Catia to associate process parameters with deformation data. After obtaining new deformation data later, the design table parameters can be directly modified in Excel to update the blade model after precision casting deformation, so as to achieve rapid update and correction of the model.

[0041] Step 5: Perform CFD fluid simulation on the blades deformed by casting to form a data set

[0042] Fluent numerical simulation software was used to perform fluid simulation analysis on the turbine blades after casting deformation, and compared with the original design model without casting deformation, to analyze the influence of the deformation of the internal cooling structure of the turbine blades after precision casting on the cooling efficiency of the blades, and to establish a mapping relationship database between process parameters and cooling efficiency;

[0043] In this embodiment, first open the CFD simulation software (such as ANSYS Fluent, CFX, etc.) to perform fluid simulation analysis, and import the undeformed and deformed blade design models into the CFD software respectively; set boundary conditions: velocity inlet: set velocity boundary conditions at the inlet of the blade, and set the appropriate flow rate according to the actual application scenario; pressure outlet: set pressure boundary conditions at the outlet of the blade, usually set to atmospheric pressure or set according to system requirements; wall conditions: set the blade surface to a no-slip wall, and consider thermal boundary conditions, such as the temperature distribution or heat flux density on a given blade surface; set the Realizable k-ε model as the turbulence model used in the present invention, and according to the geometric characteristics of the blade and the simulation requirements, you can choose to use regular tetrahedron or regular hexahedron mesh in the mesh division. The schematic diagram of the CFD calculation model is shown in the figure. Figure 5 As shown in the figure, a is the cold air inlet; b is the cold air outlet; c is the mainstream inlet; d is the mainstream outlet. Since the actual flow of fluid in the blade is unsteady, the final result of the steady solution of the flow field will inevitably produce periodic fluctuations with the number of solution steps. In order to eliminate the influence of this fluctuation, the study adopts the method of taking the average value, and takes the average value within a cycle as the final calculation result.

[0044] The cooling efficiency of the blade is defined as:

[0045]

[0046] Among them, T g represents the mainstream inlet temperature, T c represents the cooling flow inlet temperature, T b Represents the average temperature of the blade outer surface.

[0047] The cooling efficiency parameters of blades under different process parameters are sorted out through the cooling efficiency formula, and a mapping database between process parameters and cooling efficiency is established. The influence of process parameters on cooling efficiency is analyzed, and the key factors and sensitive areas affecting cooling efficiency are identified.

[0048] Step 6: Train the Kriging proxy model and predict blade cooling efficiency

[0049] The Kriging proxy model overcomes the limitation that parametric models must establish specific mathematical models. It is an approximation technology with more statistical characteristics, and the effectiveness of the model does not change with the change of random errors. The three randomly sampled process parameters are used as input and the cooling efficiency parameters simulated by CFD fluid are used as output. The Kriging proxy model is trained to predict the cooling efficiency of the internal cooling structure of the turbine blade under different casting conditions, thereby realizing a fast, accurate and efficient prediction of the cooling effect of the internal cooling structure of the turbine blade.

[0050] In this embodiment, a Kriging model is constructed. The Kriging model is essentially similar to linear fitting. It estimates the information of the prediction point by linearly weighting and combining known measurement points within a certain range near the prediction point. Its regression expression is:

[0051]

[0052] β0=(F T R -1 F) -1 F T R -1 y s

[0053] Where x is the model input vector, y s is the sample output data vector, F is the n-dimensional column vector: β0 is the regression coefficient;

[0054] F=[1 1 ... 1] T

[0055] R is the correlation coefficient matrix:

[0056]

[0057] Where x (i) Sample input variables.

[0058] Let column vector V Krig =R -1 (y s -β0F), then β0 and V Krig They are only related to known sample points and can be calculated and stored once after the model is trained. After that, predicting the response at any x only requires calculating r(x) and V Krig The dot product of .

[0059]

[0060] Input variables: three randomly sampled process parameters;

[0061] Output variables: cooling efficiency parameters of blade segments;

[0062] The statistical process parameter data and the cooling efficiency data obtained by numerical calculation are imported into the Matlab proxy model code for training, and the training model is trained and verified. Subsequently, the cooling efficiency of the internal cooling structure of the turbine blade under different casting conditions can be predicted.

[0063] Through the verification example, it can be seen that the effectiveness of the method for predicting the cooling efficiency deviation of the internal cooling structure of the turbine blade after precision casting. The method of the present invention combines simulation technology, reverse engineering, CFD fluid simulation and Kriging proxy model, etc., and can effectively predict the cooling efficiency deviation under different casting conditions, providing a scientific basis for the optimization of the casting process of turbine blades.

[0064] The present invention uses Latin hypercube sampling to randomly extract casting process parameters as input parameters of the fitting function; uses Procast software to extract point cloud data of turbine blade deformation through casting simulation; uses Geomagic software to complete reverse engineering and measure deformation; uses Catia modeling software to modify the design model according to the measured deformation; uses Fluent to complete CFD simulation and cooling efficiency calculation of the blade as output parameters of the fitting equation; trains and optimizes the Kriging proxy model, and uses the model to predict the cooling efficiency deviation. For a more complex turbine blade model, the present invention replaces reverse modeling with modified design models, and uses the Kriging proxy model to predict the cooling efficiency of the blade internal cooling structure under different casting conditions, shortening the reverse iteration cycle, reducing the workload of subsequent iterative design and flow field analysis, so as to complete the cooling efficiency prediction faster and more conveniently, shorten the blade cooling design verification cycle, and ensure that the design verification is successful once.

[0065] The above description is only a preferred embodiment of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the disclosure scope involved in the present invention is not limited to the technical solution formed by a specific combination of the above technical features, but also should cover other technical solutions formed by any combination of the above technical features or their equivalent features. For example, the above features are replaced with the technical features with similar functions disclosed in the present invention.

Claims

1. A method for predicting the cooling efficiency deviation of the internal cooling structure of a turbine blade after precision casting, characterized in that The following steps are involved: 1) Determine the range of simulation parameters and randomly sample; 2) Establish a turbine blade casting system; 3) Simulate the directional solidification process of turbine blades to obtain point cloud data of the blades after precision casting; 4) Preliminary model inversion and structural deformation extraction, and modification of the design model based on the deformation data; 5) Perform CFD fluid simulation on blades deformed by casting and establish a mapping database between process parameters and cooling efficiency; 6) Using three randomly sampled process parameters as input and the cooling efficiency parameters obtained by CFD fluid simulation as output, the Kriging proxy model is trained to predict the cooling efficiency of the internal cooling structure of turbine blades under different casting conditions.

2. The method for predicting the cooling efficiency deviation of the internal cooling structure of a turbine blade after precision casting according to claim 1, characterized in that In step 1), the simulation parameter range is determined and random sampling is performed to identify key simulation process parameters that affect the turbine blade casting process and cooling efficiency, including pouring temperature, cooling rate, and mold temperature; Latin hypercube sampling is used to randomly select values ​​of the three process parameters within a certain range, so that the original samples extracted are relatively uniform and will not produce obvious aggregation phenomena, and samples are forced to be extracted from each layer to ensure the comprehensiveness of the sample results; random sampling is performed within a reasonable parameter range to simulate different casting conditions.

3. The method for predicting the cooling efficiency deviation of the internal cooling structure of a turbine blade after precision casting according to claim 1, characterized in that In step 2), the specific steps of establishing the turbine blade casting system are: using Catia modeling software to process the turbine blade model, and establishing a casting system adapted thereto, including the runner, cold copper, and furnace body during the casting process, saving computing costs and ensuring model accuracy.

4. The method for predicting the cooling efficiency deviation of the internal cooling structure of a turbine blade after precision casting according to claim 1, characterized in that In step 3), the specific steps of simulating the directional solidification process of the turbine blade are: using Procast casting software to simulate the directional solidification process of the turbine blade, using multi-scale numerical simulation of the directional solidification process on the macroscopic temperature field, using a microstructure growth numerical model on the micro level, establishing a multi-scale numerical simulation of the directional solidification process, and obtaining point cloud data of the blade after precision casting, which reflects the geometric shape of the blade after casting.

5. The method for predicting the cooling efficiency deviation of the internal cooling structure of a turbine blade after precision casting according to claim 1, characterized in that In step 4), the model is initially reversed and the structural deformation is extracted by performing preliminary reverse engineering on the blade point cloud data obtained by simulation in Geomagic software, extracting the deformation data of the internal cooling structure of the blade (including spoiler columns and spoiler ribs), and establishing a mapping relationship database between process parameters and internal structural deformation; The modification of the design model according to the deformation data is to establish a design table according to the extracted structural deformation data, so as to quickly modify the design model and ensure the accuracy of the model to reflect the deformation situation in the actual casting process.

6. The method for predicting the cooling efficiency deviation of the internal cooling structure of a turbine blade after precision casting according to claim 1, characterized in that In step 4), the model is initially reversed and the structural deformation is extracted. The specific steps of modifying the design model according to the deformation data are as follows: after the initial reverse, the deformation data of the internal cooling structure of the blade is extracted, the deformation data includes: the radius r of the spoiler column, the height e of the spoiler rib, the spacing P of the spoiler rib, the width d of the slit, a mapping database between the process parameters and the internal structural deformation is established, a design table is established in Catia according to the extracted structural deformation data, and after obtaining new deformation data later, the design table parameters are directly modified in Excel, that is, the blade model after the precision casting deformation is updated.

7. The method for predicting the cooling efficiency deviation of the internal cooling structure of a turbine blade after precision casting according to claim 1, characterized in that In step 5), the specific steps of performing CFD fluid simulation on the blades deformed by casting are: using Fluent numerical simulation software to perform fluid simulation analysis on the turbine blades deformed by casting, simulating the flow of cooling fluid in the cooling structure inside the blades; comparing with the original design model without casting deformation, analyzing the influence of the deformation of the internal cooling structure of the turbine blades after precision casting on the cooling efficiency of the blades, recording the cooling efficiency parameters under different process parameters, and establishing a mapping relationship database between process parameters and cooling efficiency.

8. The method for predicting the cooling efficiency deviation of the internal cooling structure of a turbine blade after precision casting according to claim 1, characterized in that In step 6), the training Kriging proxy model predicts the cooling efficiency of the internal cooling structure of the turbine blade under different casting conditions, selects a certain number of randomly sampled three process parameters as training sample inputs, and the corresponding CFD simulation cooling efficiency parameters are used as outputs, and uses these data to train the Kriging proxy model; the Kriging proxy model overcomes the limitation that the parameterized model must establish a specific mathematical model, and is an approximate technology with more statistical characteristics, and the effectiveness of the model does not change with the change of random errors; the cooling efficiency of the internal cooling structure of the turbine blade under different casting conditions is predicted by the trained Kriging proxy model, so as to achieve a fast, accurate and efficient prediction of the cooling effect of the internal cooling structure of the turbine blade.

Images