An aero-engine flow passage topology design method, device, equipment and medium

Abstract

The application discloses an aero-engine flow channel topology design method, device, equipment and medium, the method comprises the steps: S1, each component calculation analysis considering thermodynamic cycle and size weight analysis is carried out, each component calculation analysis is based on engine component level modeling, and the closed loop iteration of performance prediction and structure optimization is realized through multi-program collaborative simulation; S2, the calculation sampling of compressor calculation program and turbine calculation program is reversely trained with deep neural network model to obtain compressor DNN program and turbine DNN program; S3, the topology characterization and design variable dimension reduction of engine flow channel geometry structure are established based on material field level expansion method; S4, an adaptive design domain adjustment strategy is adopted, and global optimization solving based on sequence proxy model and design domain reduction is used to obtain the engine flow channel topology configuration.The application reduces the calculation amount of engine flow channel topology optimization design, does not need sensitivity information, reduces the calculation cost, and improves the design efficiency.

Description

Technical Field

[0001] This application relates to the field of aero-engine technology, and in particular to aero-engine flow channel topology design methods, apparatus, equipment and media. Background Technology

[0002] Aero-engine flow channel design is a core aspect affecting engine aerodynamic performance, structural efficiency, and reliability, and its optimization level directly impacts key indicators such as thrust-to-weight ratio and fuel economy. As aero-engine systems evolve towards higher efficiency, lighter weight, and greater intelligence, traditional flow channel design schemes based on empirical formulas and single-disciplinary iterations struggle to effectively couple thermodynamic cycle parameters with structural dimensional constraints. Furthermore, traditional flow channel topology optimization employs density methods with an excessively high number of design variables, lacks effective topology characterization techniques for engine flow channel structures, and struggles to obtain sensitivity analysis data for complex engine flow channel designs. These factors present unprecedented challenges to the application of topology optimization techniques in aero-engine flow channel design. Summary of the Invention

[0003] This application provides a method for designing the topology of aero-engine flow channels, which solves the technical problems of existing technologies, such as difficulty in effectively coupling thermodynamic cycle parameters with structural size constraints, excessive number of design variables and large computational load, lack of effective topology characterization technology for engine flow channel structures, and difficulty in obtaining sensitivity analysis for complex engine flow channel designs.

[0004] This application is achieved through the following solution: The method for designing the flow path topology of an aero-engine includes the following steps: S1. Perform calculation analysis on each component considering thermodynamic cycle and size and weight analysis. The calculation analysis of each component is based on the fine modeling of engine component level. The closed-loop iteration of performance prediction and structural optimization is realized through multi-program collaborative simulation. The multi-program includes an overall calculation program, a weight calculation program, a compressor calculation program and a turbine calculation program. S2. The compressor DNN program and turbine DNN program are obtained by back-training the compressor calculation program and turbine calculation program with the deep neural network model through the calculation sampling. They are used for subsequent free transformation of engine flow channel topology and engine performance calculation. S3. Establishing topological representation and design variable dimensionality reduction of engine flow channel geometry based on Material Field Series Expansion (MFSE) method; S4. Adopting an adaptive design domain adjustment strategy, based on the sequence surrogate model and global optimization of design domain reduction, the overall process of global unconstrained optimization is decomposed into multiple local sub-problems. By reducing the design domain and solving the range of values ​​of dynamic constraint design variables, the global optimal solution is obtained, resulting in the final engine flow channel topology configuration, which is the flow channel design that meets the performance target requirements.

[0005] Further, in step S1, the overall calculation program obtains key performance parameters, including thrust and fuel consumption rate, by inputting component geometry and aerodynamic parameters. Then, it combines these parameters with the weight obtained by the weight calculation program using a parametric modeling method to evaluate the engine's overall performance. The compressor calculation program and turbine calculation program use a reverse solution strategy to obtain the engine flow channel configuration design: the engine flow channel configuration is determined by boundary conditions and related geometric parameters. The boundary conditions are determined according to the overall performance requirements of the engine. After the working parameters are given to the compressor calculation program and turbine calculation program, the geometric parameters related to the flow channel configuration can be obtained. Combined with CFD analysis, more accurate flow channel performance data can be obtained. Based on these geometric parameters, the flow channel configuration can be obtained.

[0006] Furthermore, step S2 specifically includes the following steps: S21. Perform calculation sampling for the corresponding compressor calculation program and turbine calculation program; S22. A deep neural network model is introduced into the sample data of the compressor and turbine for calculation and reverse training to obtain the compressor DNN program and the turbine DNN program.

[0007] Further, step S22 specifically includes the following steps: S221. Divide the collected compressor and turbine sample data into training set and test set according to a preset ratio for data training and program testing. S222. Normalize the divided training and test sets to map each feature parameter to the [0,1] interval, reducing the requirement for the amount of data and obtaining the DNN program more efficiently. S223. Input the training set after normalization preprocessing into the improved deep neural network model in batches according to a preset size, use the Adam optimizer for multiple rounds of training, save the weight file after each round of training, and prevent overfitting by using early stopping method, and finally generate the optimal model weight file. S224. Input the test set data into the trained deep neural network model as the solution model, calculate the diagnostic accuracy and loss value of the solution model on the training set, and ensure that the performance of the solution model meets the calculation requirements of the aero-engine compressor program and turbine program. S225. The trained solution model is packaged and deployed for subsequent overall solution.

[0008] Furthermore, the deep neural network model is a ResNet-18 network, and the preset ratio is 1:9 to 2:8.

[0009] Furthermore, step S3 specifically includes the following steps: S31, Using material field functions The layout of the aero-engine flow channel topology is described by uniformly selecting 10,000 to 20,000 observation points within the engine flow channel design domain. The material field is represented using the eigenvalues ​​of the correlation matrix. and eigenvectors The field problem can be expressed as a linear superposition: ; in, It is a material field function Control parameters, correlation matrix eigenvalues Sort in descending order; S32. Introducing the distance correlation function ,in Take 15-25% of the shortest side dimension of the design domain and establish any two points inside the flow channel. and The material field correlation; S33. Calculate and assemble the observation point correlation matrix with positive definite symmetry properties. Solving the generalized eigenvalue equations of the generalized eigenvalue problem The first 50 eigenvalues ​​and their corresponding eigenvectors; S34. Based on the largest eigenvalues ​​of the first 50 orders and its corresponding eigenvectors The approximate material field expression for the engine flow channel topology is obtained as follows: ; in, and The coordinates of any observation point within the design domain and the control parameters of the material field function, respectively, are represented by the material field correlation matrix. , and Let represent the k-th eigenvalue and eigenvector of the correlation matrix, respectively.

[0010] Furthermore, step S4 specifically includes the following steps: S41. Define the initial sub-design area scope. And an initial proxy model is constructed within this sub-design region using a Latin hypercube sampling strategy; S42. By integrating the maximum expected value improvement criterion and the minimum prediction error criterion, the initial surrogate model is iteratively updated and reconstructed, thereby accurately solving each sub-optimization problem and determining the optimal solution within the sub-design region. η 1; S43. The optimal solution within each sub-design region Centered on the design domain, the design domain is shrunk to obtain the reduced sub-design area. ; S44. Repeat the above operation process to perform modeling and optimization again to obtain the optimal solution in another sub-design region. η 2. By repeating this process, the globally optimal solution can be obtained. This yields the final engine flow channel topology configuration, which is the flow channel design that meets the performance target requirements.

[0011] This application also provides an apparatus for designing the flow path topology of an aero-engine, including: The component calculation and analysis module is used to perform calculation and analysis of each component considering thermodynamic cycle and size and weight analysis. The calculation and analysis of each component is based on the fine modeling of engine component level. The closed-loop iteration of performance prediction and structural optimization is realized through multi-program collaborative simulation. The multi-program includes an overall calculation program, a weight calculation program, a compressor calculation program and a turbine calculation program. The DNN program reverse training module is used to reverse train the compressor calculation program and turbine calculation program with the deep neural network model to obtain the compressor DNN program and turbine DNN program, which are used for subsequent engine flow channel topology free transformation and engine performance calculation. The topology characterization and design variable dimensionality reduction module is used to establish the topology characterization and design variable dimensionality reduction of the engine flow channel geometry based on the Material Field Series Expansion (MFSE) method. The engine flow channel topology solution module is used to adopt an adaptive design domain adjustment strategy. Based on the sequence surrogate model and global optimization of design domain reduction, it decomposes the overall process of global unconstrained optimization into multiple local sub-problems. By reducing the design domain and solving the range of values ​​of dynamic constraint design variables, the global optimal solution is obtained, resulting in the final engine flow channel topology configuration, which is the flow channel design that meets the performance target requirements.

[0012] This application also provides an electronic device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned aero-engine flow channel topology design method.

[0013] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned aero-engine flow channel topology design method.

[0014] This application also provides a computer program product, including a computer program or computer-executable instructions, which, when executed by a processor, implement the aforementioned aero-engine flow channel topology design method.

[0015] Compared with the prior art, this application can produce the following beneficial effects: This application provides a method for designing the topology of aero-engine flow channels using Material Field Series Expansion (MFSE) and Deep Neural Networks (DNN). This method encompasses a comprehensive process from component calculation and analysis, DNN model training, MFSE material field dimensionality reduction, adaptive surrogate model optimization, to the final generation of the flow channel topology configuration. First, a material field series expansion strategy is used to achieve topological representation and dimensionality reduction of complex flow channel configurations through a small number of independent design variables (material field function control parameters). Then, a deep neural network model is introduced to sample and back-train the compressor and turbine calculation programs, constructing a DNN surrogate model to replace the traditional method of frequently calling calculation programs. Finally, a global optimization process based on a sequence Kriging surrogate model and an adaptive design space adjustment strategy is established to achieve dynamic shrinkage of the design space and the solution to the global optimum. As can be seen, this application achieves topological characterization and engine performance optimization of complex engine flow channel structures through a small number of independent design variables, greatly reducing the computational load of aero-engine optimization design. It is applicable to the flow channel topology optimization design problem of multiple models and multiple objectives of aero-engines. This method does not require sensitivity information of engine-related parameters, and realizes the dynamic evolution and performance optimization of flow channel topology through low-dimensional design variables, significantly reducing computational costs. It is applicable to the flow channel topology optimization design of multiple models of aero-engines, and can be directly extended to the flow channel topology optimization design of other complex engines. It is also easy to interface with various commercial finite element software and self-developed software.

[0016] In addition to the purposes, features, and advantages described above, this application has other purposes, features, and advantages. A further detailed description of this application will be provided below with reference to the figures. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein: Figure 1 This is a flowchart illustrating the preferred embodiment of the aero-engine flow channel topology design method of this application; Figure 2 This is a schematic diagram of the flow channel geometry of the engine compressor component; Figure 3 A schematic diagram of global optimization based on the sequence Kriging surrogate model; Figure 4 This is a flowchart illustrating another preferred embodiment of the aero-engine flow channel topology design method of this application; Figure 5 This is a schematic diagram of the flow channel topology configuration of a certain engine model before optimization. Figure 6 This is a schematic diagram of the optimized flow channel topology configuration for a certain type of engine. Figure 7 This is a schematic diagram of the module of the aero-engine flow channel topology design device according to a preferred embodiment of this application; Figure 8 This is a schematic block diagram of an electronic device according to a preferred embodiment of this application; Figure 9 This is a schematic diagram of the internal structure of a computer device according to a preferred embodiment of this application. Detailed Implementation

[0019] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0020] It should be noted that the executing entity in this embodiment can be a computing service system with data processing, network communication and program running functions, such as a tablet computer, personal computer, mobile phone, etc., or an aero-engine flow channel topology design device that can realize the above functions. The following uses the aero-engine flow channel topology design device as the executing entity to describe this embodiment and the following embodiments.

[0021] like Figure 1 As shown, in view of the above-mentioned technical problems, a preferred embodiment of this application provides a method for designing the flow channel topology of an aero-engine, including the following steps: S1. In the process of aero-engine development, it is necessary to carry out interdisciplinary comprehensive calculation analysis covering thermodynamic cycle characteristics, dimensional constraints and weight control. Therefore, the calculation analysis of each component considering thermodynamic cycle and dimensional weight analysis is carried out first. The calculation analysis of each component is based on the fine modeling of engine component level. The closed-loop iteration of performance prediction and structural optimization is realized through multi-program collaborative simulation. The multi-program includes the overall calculation program, weight calculation program, compressor calculation program and turbine calculation program. S2. The compressor DNN program and turbine DNN program are obtained by back-training the compressor calculation program and turbine calculation program with the deep neural network model through the calculation sampling. They are used for subsequent free transformation of engine flow channel topology and engine performance calculation. S3. Establishing topological representation and design variable dimensionality reduction of engine flow channel geometry based on Material Field Series Expansion (MFSE) method; S4. Adopting an adaptive design domain adjustment strategy, based on the sequence surrogate model and global optimization of design domain reduction, the overall process of global unconstrained optimization is decomposed into multiple local sub-problems. By reducing the design domain and solving the range of values ​​of dynamic constraint design variables, the global optimal solution is obtained, resulting in the final engine flow channel topology configuration, which is the flow channel design that meets the performance target requirements.

[0022] This embodiment provides a method for designing the topology of aero-engine flow channels using Material Field Series Expansion (MFSE) and Deep Neural Networks (DNN). This method encompasses a complete process from component calculation and analysis, DNN model training, MFSE material field dimensionality reduction, adaptive surrogate model optimization, to the final generation of the flow channel topology configuration. First, using a material field series expansion strategy, the topological representation and dimensionality reduction of complex flow channel configurations are achieved through a small number of independent design variables (material field function control parameters). Then, a deep neural network model is introduced to sample and back-train the compressor and turbine calculation programs, constructing a DNN surrogate model to replace the traditional method of frequently calling calculation programs. Finally, a global optimization process based on a sequence Kriging surrogate model and an adaptive design space adjustment strategy is established to achieve dynamic shrinkage of the design space and the solution to the global optimum.

[0023] This aero-engine flow channel topology design method utilizes a material field series expansion strategy to achieve arbitrary topologies for aero-engine flow channel configurations with a small number of parameters, establishing a mapping relationship between these design parameters and flow channel geometry and engine performance. By solving a computational program for a specific aero-engine model, a global optimization model is established, relating the flow channel topology parameters to the engine's aerodynamic and geometric parameters. The optimization problem is then solved using a sequential Kriging surrogate model optimization algorithm. The optimization process avoids relying on sensitivity information in solving complex aero-engine design problems and effectively overcomes local optimum traps.

[0024] As can be seen, this embodiment achieves topological characterization and engine performance optimization of complex engine flow channel structures through a small number of independent design variables, greatly reducing the computational load of aero-engine optimization design. It is applicable to the flow channel topology optimization design problem of multiple models and multiple objectives of aero-engines. This method does not require sensitivity information of engine-related parameters, and realizes the dynamic evolution and performance optimization of flow channel topology through low-dimensional design variables, significantly reducing computational costs. It is applicable to the flow channel topology optimization design of multiple models of aero-engines, and can be directly extended to the flow channel topology optimization design of other complex engines. It is also easy to interface with various commercial finite element software and self-developed software.

[0025] In the development of aero-engines, interdisciplinary comprehensive computational analysis covering thermodynamic cycle characteristics, dimensional constraints, and weight control is required. This analysis system is based on refined modeling at the engine component level and achieves closed-loop iteration of performance prediction and structural optimization through multi-program collaborative simulation. Specifically, it includes four core computational modules: an overall computational program, a weight computational program, a compressor computational program, and a turbine computational program. Specifically, in step S1, the overall computational program obtains key performance parameters, including thrust and fuel consumption rate, by inputting component geometry and aerodynamic parameters. These parameters are passed as targets or constraints to subsequent computational programs. Then, it is combined with the weight obtained from the weight computational program using parametric modeling methods to evaluate the engine's overall performance. The compressor and turbine computational programs employ a reverse solution strategy to obtain the engine flow channel configuration design: the engine flow channel configuration is determined by boundary conditions and related geometric parameters. Boundary conditions are determined based on the overall engine performance requirements. After the working parameters are given to the compressor and turbine computational programs, the geometric parameters related to the flow channel configuration are obtained. Combined with CFD analysis, more accurate flow channel performance data is obtained, and the flow channel configuration can be derived from these geometric parameters. Figure 2 The diagram illustrates the flow channel structure of the compressor component of an aero-engine, where J0 represents the zero-stage stator, J1 represents the first-stage stator, J2 represents the second-stage stator, J3 represents the third-stage stator, R1 represents the first-stage rotor, R2 represents the second-stage rotor, and R3 represents the third-stage rotor. This embodiment optimizes the design for a typical aero-engine model.

[0026] Preferably, step S2 specifically includes the following steps: S21. Perform calculation sampling for the corresponding compressor calculation program and turbine calculation program; S22. A deep neural network model is introduced into the sample data of the compressor and turbine for calculation and reverse training to obtain the compressor DNN program and the turbine DNN program.

[0027] Specifically, step S22 includes the following steps: S221. Divide the collected compressor and turbine sample data into training and test sets at a ratio of 9:1 for data training and program testing. This ratio is reasonable for data training and program testing. S222. Normalize the divided training and test sets to map each feature parameter to the [0,1] interval, reducing the requirement for the amount of data and obtaining the DNN program more efficiently. S223. Input the normalized preprocessed training set into the improved deep neural network model in batches of batch_size=64. The deep neural network model is a ResNet-18 network. The Adam optimizer (initial learning rate 0.001) is used for 200 rounds of training. After each round of training, the weight file is saved. The early stopping method (patience=15) is used to prevent overfitting. Finally, the optimal model weight file is generated. S224. Input the test set data into the trained deep neural network model as the solution model, calculate the diagnostic accuracy and loss value of the solution model on the training set, and ensure that the performance of the solution model meets the calculation requirements of the aero-engine compressor program and turbine program. S225. The trained solution model is packaged and deployed for subsequent overall solution.

[0028] Because the flow path parameters obtained from the compressor calculation program and the turbine calculation program are outputs, and the frequent calls to the two programs during each calculation lead to a bottleneck in computational efficiency, this embodiment introduces a deep neural network model to perform computational sampling and reverse training on the two programs to obtain a DNN program with flow path structure parameters as input. The DNN program is then used to perform subsequent free transformation of the engine flow path topology and engine performance calculation.

[0029] Deep neural networks are a type of feedforward neural network composed of multiple hidden layers, suitable for modeling high-dimensional nonlinear mapping relationships. For constructing surrogate models of compressor and turbine programs, we employ a fully connected deep neural network. The training process of a neural network generally includes four steps: collecting training data, designing the network architecture, data preprocessing, and network training.

[0030] Training a neural network requires a large amount of data, and the training data must correspond to specific labels. For neural networks, designing the network structure mainly involves determining the network's hyperparameters, activation functions, and loss functions. Data preprocessing involves normalizing the data upon acquisition. Finally, network training is the most crucial step, directly determining the final performance of the network model. The specific mathematical expression is as follows: (1) Backpropagation algorithm: The network parameters are optimized by gradient descent. The parameter update formula is: ; in, and These represent the network parameters required for training in the t-th and t+1-th iterations, respectively. This represents the loss function value under the current parameters, where k is the learning rate and t is the number of iterations. (2) Gradient Calculation: The gradient of the loss function with respect to the network parameters is calculated using the chain rule. Using the weights of layer l... For example: ; in, The first representing the network layer( ), Representing the Linear input of layer neurons, Representing the The layer's output or activation value, The error term for the l-th layer can be recursively calculated via backpropagation: ; in, This represents the network's predicted output. For real labels, ⊙ indicates element-wise multiplication. The derivative of the activation function; (3) Optimization algorithm: The Adam optimizer is used, combined with momentum and adaptive learning rate: ; ; ; ; in, Representing the t First-moment estimate of the gradient at the next iteration This represents the second-order moment estimate of the squared gradient. and These are the attenuation coefficients for the first and second moments, respectively.

[0031] (4) Training process: Divide the dataset into training sets. Validation set and test set Mini-batch gradient descent is used, with each batch size being B. ; in, Indicates the first in a small batch i One input sample, Indicates the first i The sample at the th c The true value in each output dimension This represents the predicted value of the artificial neural network for the input sample. This represents the average loss on the current mini-batch of samples.

[0032] Preferably, step S3 specifically includes the following steps: S31, Using material field functions The layout of the aero-engine flow channel topology is described by uniformly selecting 10,000 to 20,000 observation points within the engine flow channel design domain. The material field is represented using the eigenvalues ​​of the correlation matrix. and eigenvectors The field problem can be expressed as a linear superposition: ; in, It is a material field function Control parameters, correlation matrix eigenvalues Sort in descending order; S32. Introducing the distance correlation function ,in Take 15-25% of the shortest side dimension of the design domain and establish any two points inside the flow channel. and The material field correlation; S33. Calculate and assemble the observation point correlation matrix with positive definite symmetry properties. Solving the generalized eigenvalue equations of the generalized eigenvalue problem The first 50 eigenvalues ​​and their corresponding eigenvectors; S34. Based on the largest eigenvalues ​​of the first 50 orders and its corresponding eigenvectors The approximate material field expression for the engine flow channel topology is obtained as follows: ; in, and The coordinates of any observation point within the design domain and the control parameters of the material field function, respectively, are represented by the material field correlation matrix. , and Let represent the k-th eigenvalue and eigenvector of the correlation matrix, respectively.

[0033] This embodiment uses the Material Field Series Expansion (MFSE) method to establish the topological representation of the engine flow channel geometry and reduce the dimensionality of design variables. Compared with other topology optimization methods, the MFSE method has two major advantages: First, it can avoid intermediate densities and generate clear flow channel boundaries; second, it can significantly reduce the dimensionality of the optimization problem, transforming the original high-dimensional 0 / 1 discrete variable optimization problem based on dense grids into a lower-dimensional continuous variable optimization problem. These two advantages have good applicability and benefits in engine flow channel design.

[0034] For the MFSE method, in the design domain space of the optimization problem Introducing material field functions Material fields use correlation matrix eigenvalues. and eigenvectors The field problem can be expressed as a linear superposition: ; in, It is a material field function Control parameters, characteristic values Arranged in descending order, to improve computational efficiency, the eigenvalues ​​are truncated, retaining only the first M terms with relatively large eigenvalues ​​and ignoring terms with smaller eigenvalues.

[0035] After performing material field interpolation and solving the finite element equilibrium equations, The general mathematical expression for spatial topology optimization based on MFSE is as follows: ; Among them, the MFSE coefficient These are design variables in the optimization problem. It is the response vector in finite element analysis. and These are the objective function and the constraint function, respectively. This represents a linear or nonlinear equilibrium equation.

[0036] Preferably, step S4 specifically includes the following steps: S41. Define the initial sub-design area scope. And an initial proxy model is constructed within this sub-design region using a Latin hypercube sampling strategy; S42. By integrating the maximum expected value improvement criterion and the minimum prediction error criterion, the initial surrogate model is iteratively updated and reconstructed, thereby accurately solving each sub-optimization problem and determining the optimal solution within the sub-design region. η 1; S43. The optimal solution within each sub-design region Centered on the design domain, the design domain is shrunk to obtain the reduced sub-design area. ; S44. Repeat the above operation process to perform modeling and optimization again to obtain the optimal solution in another sub-design region. η 2. By repeating this process, the globally optimal solution can be obtained. This yields the final engine flow channel topology configuration, which is the flow channel design that meets the performance target requirements.

[0037] This embodiment involves a global optimization strategy based on a sequence Kriging surrogate model and design domain reduction. Its adaptive core mechanism is to decompose the overall process of global unconstrained optimization into multiple local sub-problems until the optimization problem converges. Figure 3This is a schematic diagram of global optimization based on the Kriging proxy model, through the design domain. The global optimization solution is obtained by reducing the number of variables and changing the design variables.

[0038] When solving the Kriging surrogate model, sufficient samples are needed to construct a high-precision surrogate model for optimization. Although the MFSE method reduces the topological representation variables to a smaller number through dimensionality reduction... While this approach yields numerous results, it remains challenging for the Kriging model. As the number of design variables increases, the design space that the surrogate model needs to explore expands exponentially, leading to a sharp increase in modeling complexity. Furthermore, performing Latin hypercube sampling across the entire design domain generates a large number of invalid samples, undoubtedly increasing the complexity of building the surrogate model. Therefore, an adaptive adjustment strategy for the design domain is introduced. By dynamically constraining the range of design variable values, the sampling burden is effectively reduced while maintaining model accuracy.

[0039] Figure 4 The overall flow of an aero-engine flow channel topology design method according to another preferred embodiment is shown. The engine program is calculated by combining and updating material field design variables and engine aerodynamic parameter design variables. The program includes an overall calculation program, a weight calculation program, a compressor DNN program, and a turbine DNN program. The optimization target is composed of output parameters such as engine thrust-to-weight ratio and fuel consumption rate. The design variables and optimization target are globally optimized through the Kriging surrogate model to obtain the final engine flow channel topology configuration, which is the flow channel design that meets the performance target requirements. Figure 5 This is the flow channel topology diagram of a certain engine model before optimization. Figure 6 This is a diagram of the optimized flow channel topology for a certain type of engine.

[0040] In summary, this application addresses the problems of traditional methods relying on high-dimensional design variables, high computational cost, and difficulty in multi-objective coupled optimization by proposing an efficient topology characterization and optimization process. Based on component-level thermodynamic cycle and dimensional weight analysis, flow channel parameters are obtained. Deep neural networks are used to perform reverse proxy modeling of compressor and turbine calculation programs for rapid performance prediction. A material field series expansion method is employed to characterize the complex flow channel topology with a small number of continuous variables, achieving dimensionality reduction of design variables and generation of clear geometric boundaries. Combining a sequential proxy model and an adaptive design domain reduction strategy, a global optimization process is constructed with engine thrust-to-weight ratio as the optimization objective and relevant aerodynamic and thermodynamic parameters as constraints. This application eliminates the need for sensitivity analysis, achieving dynamic evolution and performance optimization of the flow channel topology through low-dimensional design variables, significantly reducing computational costs and making it applicable to the flow channel topology optimization design of multiple aero-engine models.

[0041] like Figure 7As shown, this application also provides an apparatus for designing the flow path topology of an aero-engine, including: The component calculation and analysis module is used to perform calculation and analysis of each component considering thermodynamic cycle and size and weight analysis. The calculation and analysis of each component is based on the fine modeling of engine component level. The closed-loop iteration of performance prediction and structural optimization is realized through multi-program collaborative simulation. The multi-program includes an overall calculation program, a weight calculation program, a compressor calculation program and a turbine calculation program. The DNN program reverse training module is used to reverse train the compressor calculation program and turbine calculation program with the deep neural network model to obtain the compressor DNN program and turbine DNN program, which are used for subsequent engine flow channel topology free transformation and engine performance calculation. The topology characterization and design variable dimensionality reduction module is used to establish the topology characterization and design variable dimensionality reduction of the engine flow channel geometry based on the Material Field Series Expansion (MFSE) method. The engine flow channel topology solution module is used to adopt an adaptive design domain adjustment strategy. Based on the sequence surrogate model and global optimization of design domain reduction, it decomposes the overall process of global unconstrained optimization into multiple local sub-problems. By reducing the design domain and solving the range of values ​​of dynamic constraint design variables, the global optimal solution is obtained, resulting in the final engine flow channel topology configuration, which is the flow channel design that meets the performance target requirements.

[0042] The aero-engine flow channel topology design device provided in this embodiment adopts the aero-engine flow channel topology design method in the above embodiments, solving the technical problems of existing technologies, such as difficulty in effectively coupling thermodynamic cycle parameters and structural size constraints, excessive number of design variables and large computational load, lack of effective engine flow channel structural topology characterization technology, and difficulty in obtaining sensitivity analysis for complex engine flow channel designs. Compared with the prior art, the beneficial effects of the aero-engine flow channel topology design device provided in this embodiment are the same as those of the aero-engine flow channel topology design method provided in the above embodiments, and other technical features in the aero-engine flow channel topology design device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0043] like Figure 8 As shown, a preferred embodiment of this embodiment also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the aero-engine flow channel topology design method in the above embodiment.

[0044] This embodiment also provides an electronic device that employs the aero-engine flow channel topology design method described in the above embodiments. This addresses the technical problems of existing technologies, such as the difficulty in effectively coupling thermodynamic cycle parameters with structural size constraints, the excessive number of design variables and computational burden, the lack of effective engine flow channel structure topology characterization technology, and the difficulty in obtaining sensitivity analysis for complex engine flow channel designs. Compared with existing technologies, the beneficial effects of the electronic device provided in this embodiment are the same as those of the aero-engine flow channel topology design method described in the above embodiments. Furthermore, other technical features of the electronic device are the same as those disclosed in the methods of the above embodiments, and will not be elaborated upon here.

[0045] like Figure 9 As shown, a preferred embodiment of this application also provides a computer device, which may be a terminal or a liveness detection server, and its internal structure diagram may be as follows. Figure 9 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used for communication with other external computer devices via a network connection. When the computer program is executed by the processor, it implements the steps of the aforementioned aero-engine flow channel topology design method.

[0046] Those skilled in the art will understand that Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the solution of this embodiment, and does not constitute a limitation on the computer device to which the solution of this embodiment is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0047] The computer device provided in this embodiment adopts the aero-engine flow channel topology design method in the above embodiments, which solves the technical problems of existing technologies, such as difficulty in effectively coupling thermodynamic cycle parameters and structural size constraints, excessive number of design variables and large amount of calculation, lack of effective engine flow channel structure topology characterization technology, and difficulty in obtaining sensitivity analysis of complex engine flow channel design. Compared with the prior art, the beneficial effects of the computer device provided in this embodiment are the same as the beneficial effects of the aero-engine flow channel topology design method provided in the above embodiments, and other technical features in the electronic device are the same as the features disclosed in the method of the above embodiments, which will not be repeated here.

[0048] A preferred embodiment of this application also provides a storage medium, the storage medium including a stored program, which, when the program is executed, controls the device where the storage medium is located to perform the steps of the aero-engine flow channel topology design method in the above embodiments.

[0049] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0050] If the functions described in this embodiment are implemented as software functional units and sold or used as independent products, they can be stored in one or more computing device-readable storage media. Based on this understanding, the parts of this embodiment that contribute to the prior art or the technical solution can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a computing device (which may be a personal computer, server, mobile computing device, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this embodiment. The aforementioned storage media include: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0051] Those skilled in the art will understand that the embodiments of this example can be provided as methods, systems, or computer program products. Therefore, this example can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this example can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code. The solutions in this example can be implemented using various computer languages, such as the object-oriented programming language C++ and the embedded programming language C.

[0052] This embodiment is described with reference to flowchart illustrations and / or block diagrams of the method, apparatus (system), and computer program product according to this embodiment. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1One or more processes and / or boxes Figure 1 A system that specifies functions in one or more boxes.

[0053] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction set implemented in a process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0054] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0055] This embodiment also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the aero-engine flow channel topology design method described above.

[0056] The computer program product provided in this embodiment solves the technical problems of existing technologies, such as the difficulty in effectively coupling thermodynamic cycle parameters and structural size constraints, the excessive number of design variables and computational burden, the lack of effective engine flow channel structure topology characterization technology, and the difficulty in obtaining sensitivity analysis for complex engine flow channel designs. Compared with the prior art, the beneficial effects of the computer program product provided in this embodiment are the same as those of the aero-engine flow channel topology design method provided in the above embodiments, and will not be repeated here.

[0057] This application can effectively overcome the computational efficiency bottleneck and the difficulty in characterizing the flow channel topology in the design of aero-engine flow channel topology optimization, and can efficiently obtain the optimal flow channel topology configuration that meets the performance goals of aero-engines. It can be applied to the flow channel topology optimization design of various engine models, so as to further improve the design capability of large aerospace equipment.

[0058] Obviously, those skilled in the art can make various modifications and variations to this embodiment without departing from the spirit and scope of this embodiment. Therefore, if these modifications and variations of this embodiment fall within the scope of the claims of this embodiment and their equivalents, this embodiment is also intended to include these modifications and variations.

Claims

1. A method for designing the flow channel topology of an aero-engine, characterized in that, Including the following steps: S1. Perform calculation analysis on each component considering thermodynamic cycle and size and weight analysis. The calculation analysis of each component is based on the fine modeling of engine component level. The closed-loop iteration of performance prediction and structural optimization is realized through multi-program collaborative simulation. The multi-program includes an overall calculation program, a weight calculation program, a compressor calculation program and a turbine calculation program. S2. The compressor DNN program and turbine DNN program are obtained by back-training the compressor calculation program and turbine calculation program with the deep neural network model through the calculation sampling. They are used for subsequent free transformation of engine flow channel topology and engine performance calculation. S3. Establishing topological representation and design variable dimensionality reduction of engine flow channel geometry based on material field series expansion method; S4. Adopting an adaptive design domain adjustment strategy, based on the sequence surrogate model and global optimization of design domain reduction, the overall process of global unconstrained optimization is decomposed into multiple local sub-problems. By reducing the design domain and solving the range of values ​​of dynamic constraint design variables, the global optimal solution is obtained, resulting in the final engine flow channel topology configuration, which is the flow channel design that meets the performance target requirements.

2. The aero-engine flow channel topology design method according to claim 1, characterized in that, In step S1, the overall calculation program obtains key performance parameters, including thrust and fuel consumption rate, by inputting component geometry and aerodynamic parameters. Then, it combines these parameters with the weight obtained by the weight calculation program using a parametric modeling method to evaluate the engine's overall performance. The compressor calculation program and turbine calculation program use a reverse solution strategy to obtain the engine flow channel configuration design: the engine flow channel configuration is determined by boundary conditions and related geometric parameters. The boundary conditions are determined according to the overall performance requirements of the engine. After the working parameters are given to the compressor calculation program and turbine calculation program, the geometric parameters related to the flow channel configuration can be obtained. Combined with CFD analysis, more accurate flow channel performance data can be obtained. Based on these geometric parameters, the flow channel configuration can be obtained.

3. The aero-engine flow channel topology design method according to claim 1, characterized in that, Step S2 specifically includes the following steps: S21. Perform calculation sampling for the corresponding compressor calculation program and turbine calculation program; S22. A deep neural network model is introduced into the sample data of the compressor and turbine for calculation and reverse training to obtain the compressor DNN program and the turbine DNN program.

4. The aero-engine flow channel topology design method according to claim 3, characterized in that, Step S22 specifically includes the following steps: S221. Divide the collected compressor and turbine sample data into training set and test set according to a preset ratio for data training and program testing. S222. Normalize the divided training and test sets to map each feature parameter to the [0,1] interval, reducing the requirement for the amount of data and obtaining the DNN program more efficiently. S223. Input the training set after normalization preprocessing into the improved deep neural network model in batches according to a preset size, use the Adam optimizer for multiple rounds of training, save the weight file after each round of training, and prevent overfitting by using early stopping method, and finally generate the optimal model weight file. S224. Input the test set data into the trained deep neural network model as the solution model, calculate the diagnostic accuracy and loss value of the solution model on the training set, and ensure that the performance of the solution model meets the calculation requirements of the aero-engine compressor program and turbine program. S225. The trained solution model is packaged and deployed for subsequent overall solution.

5. The aero-engine flow channel topology design method according to claim 4, characterized in that, The deep neural network model is a ResNet-18 network, and the preset ratio is 1:9 to 2:

8.

6. The aero-engine flow channel topology design method according to claim 1, characterized in that, Step S3 specifically includes the following steps: S31, Using material field functions The layout of the aero-engine flow channel topology is described by uniformly selecting 10,000 to 20,000 observation points within the engine flow channel design domain. The material field is represented using the eigenvalues ​​of the correlation matrix. and eigenvectors The field problem can be expressed as a linear superposition: ； in, It is a material field function Control parameters, correlation matrix eigenvalues Sort in descending order; S32. Introducing the distance correlation function ,in Take 15-25% of the shortest side dimension of the design domain and establish any two points inside the flow channel. and The material field correlation; S33. Calculate and assemble the observation point correlation matrix with positive definite symmetry properties. Solving the generalized eigenvalue equations of the generalized eigenvalue problem The first 50 eigenvalues ​​and their corresponding eigenvectors; S34. Based on the largest eigenvalues ​​of the first 50 orders and its corresponding eigenvectors The approximate material field expression for the engine flow channel topology is obtained as follows: ； in, and The coordinates of any observation point within the design domain and the control parameters of the material field function, respectively, are represented by the material field correlation matrix. , and Let represent the k-th eigenvalue and eigenvector of the correlation matrix, respectively.

7. The aero-engine flow channel topology design method according to claim 1, characterized in that, Step S4 specifically includes the following steps: S41. Define the initial sub-design area scope. And an initial proxy model is constructed within this sub-design region using a Latin hypercube sampling strategy; S42. By integrating the maximum expected value improvement criterion and the minimum prediction error criterion, the initial surrogate model is iteratively updated and reconstructed, thereby accurately solving each sub-optimization problem and determining the optimal solution within the sub-design region. η 1; S43. The optimal solution within each sub-design region Centered on the design domain, the design domain is shrunk to obtain the reduced sub-design area. ; S44. Repeat the above operation process to perform modeling and optimization again to obtain the optimal solution in another sub-design region. η 2. By repeating this process, the globally optimal solution can be obtained. This yields the final engine flow channel topology configuration, which is the flow channel design that meets the performance target requirements.

8. An aero-engine flow channel topology design device, characterized in that, include: The component calculation and analysis module is used to perform calculation and analysis of each component considering thermodynamic cycle and size and weight analysis. The calculation and analysis of each component is based on the fine modeling of engine component level. The closed-loop iteration of performance prediction and structural optimization is realized through multi-program collaborative simulation. The multi-program includes an overall calculation program, a weight calculation program, a compressor calculation program and a turbine calculation program. The DNN program reverse training module is used to reverse train the compressor calculation program and turbine calculation program with the deep neural network model to obtain the compressor DNN program and turbine DNN program, which are used for subsequent engine flow channel topology free transformation and engine performance calculation. The topology characterization and design variable dimensionality reduction module is used to establish the topology characterization and design variable dimensionality reduction of the engine flow channel geometry based on the material field series expansion method. The engine flow channel topology solution module is used to adopt an adaptive design domain adjustment strategy. Based on the sequence surrogate model and global optimization of design domain reduction, it decomposes the overall process of global unconstrained optimization into multiple local sub-problems. By reducing the design domain and solving the range of values ​​of dynamic constraint design variables, the global optimal solution is obtained, resulting in the final engine flow channel topology configuration, which is the flow channel design that meets the performance target requirements.

9. An electronic device, the electronic device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor, when executing the computer program, implements the aero-engine flow channel topology design method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the aero-engine flow channel topology design method as described in any one of claims 1 to 7.

Images