End region design method and device and optimization design platform

Abstract

The invention discloses an end area design method and device and an optimization design platform, the method and device are applied to optimization design of a wide and long vaneless diffuser of a centrifugal compressor, and specifically, the modeling area of the wide and long vaneless diffuser is determined; establishing an initial database based on an experimental design method, wherein the initial database comprises multiple groups of optimization target variables; based on the initial database, establishing an initial prediction model by adopting a radial basis function proxy model; performing hyper-parameter optimization processing on the initial prediction model to obtain an optimal end region prediction agent model; and carrying out multi-objective optimization design by adopting a genetic algorithm based on the optimal end region prediction agent model to obtain an optimal modeling design scheme of the end region. The technical scheme does not depend on artificial experience, so that the stability margin of the centrifugal compressor can be effectively improved.

Description

Technical Field

[0001] The present application relates to the technical field of industrial design, and more specifically, to an end region design method, device, and optimization design platform. Background Art

[0002] The main function of the vaneless diffuser of a centrifugal compressor is to decelerate and pressurize the high-speed air flow at the impeller outlet. As the air flow radially flows in the vaneless diffuser, the flow passage area gradually increases, causing the air flow velocity to decrease while the pressure increases, thus forming an adverse pressure gradient. When the adverse pressure gradient exceeds a certain limit, the resistance on the low-speed air flow near the wall increases and the speed decreases. When the kinetic energy is insufficient to overcome the adverse pressure, flow separation will occur. Flow separation is an important precursor to stall. The separated air flow forms vortices, disturbing the normal flow of the main stream, thus causing stall in the compressor diffuser. Therefore, to improve the stability margin of the centrifugal compressor, flow control needs to be carried out on the vaneless diffuser of the centrifugal compressor to delay or weaken flow separation.

[0003] The inventors of the present application found in practice that the non-axisymmetric end wall shaping technology can weaken the transverse pressure gradient in the end region, regulate the load distribution, and has remarkable effects in improving the performance and stability of the compressor. However, the traditional empirical non-axisymmetric end wall shaping technology overly relies on manual experience, has limited stall expansion effects, is prone to form local optima, and cannot effectively improve the stability margin of the centrifugal compressor. Summary of the Invention

[0004] In view of this, the present application provides an end region design method, device, and optimization design platform for realizing the end region optimization design of the wide and long vaneless diffuser of a centrifugal compressor to improve the stability margin of the centrifugal compressor.

[0005] To achieve the above object, the following solutions are proposed:

[0006] An end region design method applied to the end region design method of a wide and long vaneless diffuser of a centrifugal compressor, the end region design method comprising the steps of:

[0007] Determine the shaping region of the wide and long vaneless diffuser;

[0008] Based on the shaping region, establish an initial database, the initial database including multiple groups of optimization target variables;

[0009] Based on the initial database, establish an initial prediction model using a radial basis function surrogate model;

[0010] Perform hyperparameter optimization on the initial prediction model to obtain an optimal end region prediction surrogate model;

[0011] Based on the optimal end region prediction proxy model, a genetic algorithm is used for multi-objective optimization design to obtain the optimal shape design scheme of the end region.

[0012] Optionally, determining the final shape region of the width-length vaneless diffuser includes the steps of:

[0013] Parametrize the wall surface of the width-length vaneless diffuser;

[0014] Based on the wall surface parameters, a shape design is carried out, and the shape region is determined through sensitivity analysis.

[0015] Optionally, establishing the initial database based on the experimental design method includes the steps of:

[0016] Use the optimal Latin hypercube method in the experimental design method to obtain the geometric sample data of the wall surface parametrization;

[0017] According to the geometric sample data, a shape design is carried out on the wall surface of the width-length vaneless diffuser to obtain a variety of different shape design schemes;

[0018] Based on the CFD simulation chain established by Isight, different shape design schemes are simulated and processed;

[0019] According to the simulation results and the stability margin formula, the corresponding optimization target variables are obtained;

[0020] Based on the optimization target variables, the initial database is established.

[0021] Optionally, the stability margin formula is as follows:

[0022]

[0023] Among them, m is the mass flow rate, π is the pressure ratio, the subscript NS represents the near-stall point, PEW represents the end wall shape, and SW represents the prototype.

[0024] Optionally, when performing hyperparameter optimization on the initial prediction model, the performance evaluation formula used is:

[0025]

[0026] Among them, NMSE is the normalized root mean square error, x is the prediction result of the radial basis function proxy model, and x0 is the simulation result.

[0027] An end region design device is applied to the width-length vaneless diffuser of a centrifugal compressor. The end region design device includes:

[0028] A shape determination module configured to determine the shape region of the width-length vaneless diffuser;

[0029] A database establishment module, configured to establish an initial database based on the profiling region, the initial database including multiple groups of optimization target variables;

[0030] A model construction module, configured to establish an initial prediction model by using a radial basis function surrogate model based on the initial database;

[0031] A model optimization module, configured to perform hyperparameter optimization processing on the initial prediction model to obtain an optimal end region prediction surrogate model;

[0032] A design execution module, configured to perform multi-objective optimization design based on the optimal end region prediction surrogate model to obtain an optimal profiling design scheme for the end region.

[0033] Optionally, the profiling determination module includes:

[0034] A parameterization processing unit, configured to parameterize the wall surface of the wide-length vaneless diffuser to obtain near-wall surface parameters;

[0035] A determination execution unit, configured to perform profiling design based on the wall surface parameters and determine the profiling region through sensitivity analysis.

[0036] Optionally, the database establishment module includes:

[0037] A sample acquisition unit, configured to obtain geometric sample data of the wall surface parameters by using the optimal Latin hypercube method in the experimental design method;

[0038] A profiling processing unit, configured to perform profiling processing on the wall surface of the wide-length vaneless diffuser according to the geometric sample data to obtain multiple different profiling design schemes;

[0039] A simulation processing unit, configured to perform simulation processing on different profiling design schemes;

[0040] A variable calculation unit, configured to obtain corresponding optimization target variables according to the simulation results and the stability margin formula;

[0041] A construction execution unit, configured to establish the initial database for the optimization target variables based on the experimental design method.

[0042] An optimization design platform, the optimization design platform including a CFD simulation chain and an optimization process, wherein:

[0043] The CFD simulation chain is used to store computer programs or instructions and can automatically perform parameterized geometric modeling, mesh generation, numerical simulation solution, and post-processing performance parameter extraction;

[0044] The optimization process is used to perform the optimization design of the end-wall profile so that the wide and long vaneless diffuser of the centrifugal compressor realizes the end-region design method as described above.

[0045] As can be seen from the above technical solutions, the present application discloses an end-region design method, device, and optimization design platform. The method and device are applied to the optimization design of the wide and long vaneless diffuser of a centrifugal compressor, specifically to determine the profiling area of the wide and long vaneless diffuser; establish an initial database based on the experimental design method, where the initial database includes multiple groups of optimization target variables; establish an initial prediction model using a radial basis function surrogate model based on the initial database; perform hyperparameter optimization on the initial prediction model to obtain an optimal end-region prediction surrogate model; perform multi-objective optimization design using a genetic algorithm based on the optimal end-region prediction surrogate model to obtain an optimal profiling design solution for the end region. This technical solution does not rely on manual experience, thereby effectively improving the stability margin of the centrifugal compressor. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0047] Figure 1 It is a partial schematic diagram of the wide and long vaneless diffuser of the embodiment of the present application;

[0048] Figure 2 It is a flowchart of an end-region optimization design method of the embodiment of the present application;

[0049] Figure 3 It is a parametric schematic diagram of the profiling area determined in the embodiment of the present application;

[0050] Figure 4 It is a schematic diagram of the optimal profiling design solution of the embodiment of the present application;

[0051] Figure 5 It is a schematic diagram of the CFD simulation chain built by Isight software in the embodiment of the present application;

[0052] Figure 6 It is a block diagram of an end-region design device of the embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0054] The technical solution provided by the present application is used for optimizing the design of the end region of the wide-length vaneless diffuser of a centrifugal compressor to improve the stability margin of the centrifugal compressor. The wide-length vaneless diffuser is as Figure 1 shown, with a width ratio b / r1 > 0.1 and a radial ratio r2 / r1 > 1.8. The specific solution of the present application is described as follows.

[0055] Figure 2 It is a flowchart of an end-region optimization design method for an embodiment of the present application.

[0056] As Figure 2 shown, the end-region optimization design method provided in this embodiment is applied to the end-region optimization design of the wide-length vaneless diffuser of a centrifugal compressor to improve the stability margin of the centrifugal compressor. This optimization design platform can be understood as a computer, server, or cloud platform with data calculation capabilities and information processing capabilities. The end-region optimization design method of the present application includes the following steps:

[0057] S1. Determine the shaping region of the wide-length vaneless diffuser.

[0058] Specifically, the shaping region of the wide-length vaneless diffuser is determined through parametric operations. The specific process is as follows:

[0059] First, parameterize the single-channel disk side of the wide-length vaneless diffuser of the centrifugal compressor to obtain wall parameters;

[0060] Then, use the single-variable method for shaping design and determine the final shaping region through sensitivity analysis.

[0061] In this example, the shaping region is the first 80% region along the flow direction of the disk side of the wide-length vaneless diffuser (starting from the diffuser inlet), as Figure 3 shown. Among them, 6 control lines are evenly arranged circumferentially, and 6 control points are evenly distributed on each control line. To ensure smooth connection between control points and between control points and the flat wall, the boundary control point offset is 0, and there are a total of 20 free control points.

[0062] S2. Establish an initial database based on the experimental design method.

[0063] Specifically, by determining the shaping region of the wide-length vaneless diffuser, an initial database is established. The specific process is as follows:

[0064] First, to ensure that the randomly generated geometric sample data is evenly distributed within the set space, the method of the optimal Latin hypercube in DOE (Design of Experiments) is used to obtain the geometric sample data of the disk side parameters. DOE is a statistical method that explores the influence of multiple factors on the results through systematic arrangement of experiments, aiming to identify key factors and their interactions with the least number of experiments to achieve the optimal result.

[0065] Then, based on the geometric sample data, the disk side of the wide - length vane - less diffuser is modeled. B - spline curves are used to connect the offsets of each control point, and multiple different modeling design schemes are obtained.

[0066] After that, based on the CFD simulation chain established by Isight, the different modeling design schemes are simulated, and the corresponding simulation results are obtained.

[0067] After that, according to the simulation results and the stability margin formula, the corresponding optimization target variables are obtained.

[0068] Finally, the initial database is established based on the free variables and their corresponding optimization target variables.

[0069] Among them, the stability margin formula is as follows:

[0070]

[0071] Among them, m is the mass flow rate, π is the pressure ratio, the subscript NS represents the near - stall point, PEW represents the end - wall shape, and SW represents the prototype.

[0072] S3. An initial prediction model is established using a radial basis function surrogate model.

[0073] According to the initial database obtained above, an initial prediction model is established using a radial basis function surrogate model.

[0074] S4. Hyperparameter optimization is performed on the initial prediction model.

[0075] Multiple optimization processes can be implemented. Each optimization uses k - fold cross - validation to obtain different training sets and test sets for hyperparameter optimization. Set the numerical range of the optimization parameter variables, and use the genetic algorithm for global optimization in the process of establishing the initial prediction model to obtain the optimal values of the parameter variables in the radial basis function surrogate model, so that the prediction accuracy of the radial basis function surrogate model is the highest, and thus an optimal end - region prediction surrogate model with satisfactory accuracy is obtained.

[0076] During the optimization process, it is judged whether the parameter variables converge. If they do not converge, continue with the optimization process. If they have converged, execute the subsequent steps.

[0077] S5. Perform multi-objective optimization design based on the optimal end-region prediction surrogate model.

[0078] Use the radial basis function surrogate model after hyperparameter optimization in the above steps to obtain the non-linear relationship between the design variables with the highest prediction accuracy and the optimization objectives. And taking the improvement of the stability margin and the maximization of the design condition efficiency as the optimization objectives, use the non-dominated sorting genetic algorithm II (NSGA-II) optimization algorithm to perform multi-objective optimization to obtain the optimal styling design variables. Calculate the true values through CFX numerical simulation, add the true results to the database for continuous iterative optimization, and obtain a new radial basis function surrogate model and a new optimal styling design scheme. And so on, continuously perform iterative optimization calculations until the optimization stop condition (optimization convergence) is met to obtain the final optimal styling design scheme. As Figure 4 shown, where the parameter h represents the relative height of the diffuser end wall, a positive value represents a protrusion, and a negative value represents a depression.

[0079] The final optimal styling design scheme obtained in this example improves the stability margin of the centrifugal compressor by 6.89% and the efficiency at the design condition by 0.63%.

[0080] As can be seen from the above technical solutions, this embodiment provides an end-region design method, which is applied to the optimization design of the wide-length vaneless diffuser of a centrifugal compressor. Specifically, it is to determine the styling area of the wide-length vaneless diffuser; establish an initial database based on the experimental design method, and the initial database includes multiple groups of optimization target variables; based on the initial database, establish an initial prediction model using the radial basis function surrogate model; perform hyperparameter optimization on the initial prediction model to obtain the optimal end-region prediction surrogate model; perform multi-objective optimization design using the genetic algorithm based on the optimal end-region prediction surrogate model to obtain the optimal styling design scheme of the end region. This technical solution does not rely on artificial experience, thereby effectively improving the stability margin and efficiency of the centrifugal compressor.

[0081] Since the calculation of the styling design scheme for establishing the initial sample database and subsequent optimization iterative calculations in this application is large in amount and the process is repetitive and cumbersome, in order to make full use of computing resources and save computing time, a CFD simulation chain is built with the help of Isight software, including automatically modeling, meshing, simulation, post-processing result acquisition, and saving according to the obtained geometric sample data.

[0082] The performance evaluation formula in the optimization system of this application is:

[0083]

[0084] Among them, NMSE is the normalized root mean square error, which is used to measure the gap between the prediction result x of the radial basis function surrogate model and the simulation result x0.

[0085] This application is implemented through a high-level programming language, based on the geometric data of the shape design scheme optimized by the genetic algorithm of the radial basis function surrogate model and the predicted stability margin; the high-level programming language uses Python or MATLAB;

[0086] In this application, the model training and optimized shape design scheme are realized through a radial basis function surrogate model, and the CFD simulation chain built by Isight software is used for simulation calculation to obtain the true value x0. This CFD simulation chain is as Figure 5 shown.

[0087] This application adopts an automatic performance evaluation and optimization algorithm module, builds an optimization design platform through a high-level programming language and Isight software, and uses an optimization search method to continuously iterate and update the design variables for optimization until the optimization stop condition is met, obtaining the optimal shape design scheme of the centrifugal compressor wide-length vaneless diffuser.

[0088] Aiming at the near-wall boundary layer separation of the vaneless diffuser of the centrifugal compressor, this application proposes an end-region design method applicable to the wide-length vaneless diffuser of the centrifugal compressor, which can solve or delay the compressor stall caused by the flow separation of the vaneless diffuser of the centrifugal compressor, filling the research gap in the non-axisymmetric end-wall shape design of the vaneless diffuser of the centrifugal compressor.

[0089] To solve the problems that the traditional empirical non-axisymmetric end-wall shaping technology relies too much on manual experience, the stability augmentation effect is limited, it is easy to form local optima, and the direct use of a single optimization algorithm for global optimization has high computational cost and low efficiency, the present invention adopts a dynamic radial basis function surrogate model, which is suitable for small-sample point optimization, and while ensuring the computational accuracy, improves the optimization calculation efficiency.

[0090] This application uses a global optimization method to optimize the hyperparameters of the radial basis function surrogate model, so that the obtained prediction model has the highest accuracy and is closest to its actual simulation result, reducing the number of optimization iterations and improving the optimization efficiency. The true simulation results of the optimal design shape of each iteration optimization are added to the database, the sample points are continuously updated, and the surrogate model is reconstructed. Both the surrogate model and the sample points change during the entire optimization process, effectively accelerating the optimization convergence speed.

[0091] During the optimization process, with the maximization of improving the stability margin and efficiency as the optimization goal, the non-dominated sorting genetic algorithm II (NSGA-II) optimization algorithm is used for multi-objective optimization, while expanding the stable operating range of the compressor, taking into account the compressor efficiency and flow capacity.

[0092] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a portion of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that, in some alternative implementations, the functions noted in the blocks may occur in a different order than that noted in the drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or by a combination of dedicated hardware and computer instructions.

[0093] Although the operations are depicted in a particular order, this should not be construed as requiring that the operations be performed in the particular order shown or in sequential order. In certain circumstances, multitasking and parallel processing may be advantageous.

[0094] It should be understood that the various steps recited in the method embodiments of the present disclosure may be executed in a different order or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this regard.

[0095] Computer program code for carrying out operations of the present disclosure may be written in one or more programming languages or combinations thereof, including but not limited to object-oriented programming languages such as Java, Smalltalk, C++, and also including conventional procedural programming languages such as the C language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server.

[0096] Figure 6 It is a block diagram of an end region design device according to an embodiment of the present application.

[0097] As Figure 6As shown, the end region design device provided in this embodiment is applied to the optimized design of the end region of a wide and long vaneless diffuser for a centrifugal compressor to improve the stability margin and efficiency of the centrifugal compressor. The optimization design platform can be understood as a computer, server, or cloud platform with data computing and information processing capabilities. The end region design device of this application includes a shaping region confirmation module 10, a database establishment module 20, a prediction model construction module 30, a prediction model optimization module 40, and a global optimization module 50.

[0098] The shaping area confirmation module is used to determine the shaping area of the wide and long vaneless diffuser.

[0099] Specifically, the modeling area of the wide and long vaneless diffuser is determined through parameterization operation. This module specifically includes a parameterization processing unit and a determination execution unit.

[0100] The parameterization processing unit parameterizes the single flow channel wall surface of the wide and long bladeless diffuser of the centrifugal compressor;

[0101] The execution unit is determined to be used for modeling design using the single variable method, and the final modeling area is determined through sensitivity analysis.

[0102] In this example, the shaping area is the 80% area along the flow direction of the disc side of the wide and long vaneless diffuser (starting from the diffuser inlet), such as Figure 3 Among them, 6 control lines are evenly set along the circumference, and 6 control points are evenly distributed on each control line. In order to ensure smooth connection between control points and between control points and flat walls, the offset of boundary control points is 0, and there are 20 free control points in total.

[0103] The database establishment module is used to establish an initial database based on the experimental design method.

[0104] Specifically, the initial database is established by determining the styling area of the wide and long vaneless diffuser and building a CFD simulation chain based on Isight. This module includes a sample acquisition unit, a styling processing unit, a simulation processing unit, a variable calculation unit, and a build execution unit.

[0105] To ensure that the randomly generated geometric sample data is evenly distributed within the specified space, the sample acquisition unit uses the optimal Latin hypercube method within DOE (Design of Experiments) to obtain geometric sample data for wall parameters. DOE is a statistical method that systematically arranges experiments to explore the influence of multiple factors on results. It aims to identify key factors and their interactions to achieve the best results with the minimum number of experiments.

[0106] The shaping processing unit is used to perform shaping design on the wall surface of the wide-length bladeless diffuser according to the geometric sample data, and B-spline curves are used to connect the offsets of each control point to obtain a variety of different shaping design schemes.

[0107] The simulation processing unit is used to perform simulation processing on different shaping design schemes to obtain simulation results.

[0108] The variable calculation unit is used to obtain the corresponding optimization target variables according to the simulation results and the stability margin formula.

[0109] The construction execution unit is used to establish the initial database based on the optimization target variables.

[0110] Among them, the stability margin formula is as follows:

[0111]

[0112] Among them, m is the mass flow rate, π is the pressure ratio, the subscript NS represents the near-stall point, PEW represents the end-wall shaping, and SW represents the prototype.

[0113] The prediction model construction module is used to establish an initial prediction model by using a radial basis function surrogate model.

[0114] According to the initial database obtained above, an initial prediction model is established by using a radial basis function surrogate model.

[0115] The prediction model optimization module is used to perform hyperparameter optimization processing on the initial prediction model.

[0116] Multiple optimization processes can be implemented. Each optimization uses k-fold cross-validation to obtain different training sets and test sets for hyperparameter optimization. Set the numerical range of the optimization parameter variables, and use the genetic algorithm to perform global optimization during the process of establishing the initial prediction model to obtain the optimal values of the parameter variables in the radial basis function surrogate model, so that the prediction accuracy of the radial basis function surrogate model is the highest, and thus an optimal end-region prediction surrogate model with satisfactory accuracy is obtained.

[0117] The global optimization module is used to perform multi-objective optimization design based on the optimal end-region prediction surrogate model.

[0118] The nonlinear relationship with the highest prediction accuracy between the design variables and the optimization objective is obtained using the radial basis function surrogate model after hyperparameter optimization in the utilization solution. With the maximization of improving the stability margin and efficiency as the optimization objective, the multi-objective optimization is carried out using the non-dominated sorting genetic algorithm II (NSGA-II) optimization algorithm to obtain the optimal profile design variables. The true values are obtained through CFX numerical simulation calculations, and the true results are added to the database for continuous iterative optimization to obtain a new radial basis function surrogate model and a new optimal profile design scheme. And so on, continuously performing iterative optimization calculations until the optimization stop condition (optimization convergence) is met to obtain the final optimal profile design scheme, as Figure 4 shown, where the parameter h represents the relative height of the diffuser end wall, a positive value represents a protrusion, and a negative value represents a depression.

[0119] The final optimal profile design scheme obtained in this example increases the stability margin of the centrifugal compressor by 6.89% and improves the efficiency at the design condition by 0.63%.

[0120] It can be seen from the above technical solution that this embodiment provides an end region design device, which is applied to the optimization design of the wide-length vaneless diffuser of a centrifugal compressor. Specifically, it determines the profiling region of the wide-length vaneless diffuser; establishes an initial database based on the experimental design method, and the initial database includes multiple groups of optimization objective variables; based on the initial database, establishes an initial prediction model using the radial basis function surrogate model; performs hyperparameter optimization processing on the initial prediction model to obtain the optimal end region prediction surrogate model; based on the optimal end region prediction surrogate model, uses the genetic algorithm for multi-objective optimization design to obtain the optimal profile design scheme of the end region. This technical solution does not rely on manual experience, thereby effectively improving the stability margin of the centrifugal compressor.

[0121] This embodiment also provides an embodiment of an optimization design platform.

[0122] The above computer-readable storage medium is applied to the optimization design platform and carries one or more computer programs. When the one or more computer programs are executed by the optimization design platform, the optimization design platform determines the profiling region of the wide-length vaneless diffuser; establishes an initial database based on the experimental design method, and the initial database includes multiple groups of optimization objective variables; based on the initial database, establishes an initial prediction model using the radial basis function surrogate model; performs hyperparameter optimization processing on the initial prediction model to obtain the optimal end region prediction surrogate model; based on the optimal end region prediction surrogate model, uses the genetic algorithm for multi-objective optimization design to obtain the optimal profile design scheme of the end region. This technical solution does not rely on manual experience, thereby effectively improving the stability margin of the centrifugal compressor.

[0123] Each embodiment in this specification is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. For the same or similar parts among the embodiments, reference can be made to each other.

[0124] Although the preferred embodiments of the embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications to these embodiments once they know the basic creative concept. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the embodiments of the present invention.

[0125] Finally, it should also be noted that in this text, relative terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or terminal device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or terminal device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, article or terminal device comprising the element.

[0126] The technical solutions provided by the present invention have been introduced in detail above. Specific examples are used herein to illustrate the principles and implementation manners of the present invention. The description of the above embodiments is only for helping to understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for end region design, which is applied to an optimization design platform and is used for the optimization design of a centrifugal compressor wide and long vaneless diffuser, characterized in that The end region design method includes the steps of: Determine the final shaping region of the wide-length vaneless diffuser; Establish an initial database based on the experimental design method, where the initial database includes multiple groups of optimization objective variables; Based on the initial database, establish an initial prediction model using a radial basis function surrogate model; Perform hyperparameter optimization on the initial prediction model to obtain an optimal end region prediction surrogate model; Based on the optimal end region prediction surrogate model, perform multi-objective optimization design using a genetic algorithm to obtain an optimal shaping design scheme for the end region.

2. The end region design method according to claim 1, wherein, The determination of the final shaping region of the wide-length vaneless diffuser includes the steps of: Parameterize the wall surface of the wide-length vaneless diffuser to obtain wall surface parameters; Based on the wall surface parameters, perform shaping design and determine the final shaping region through sensitivity analysis.

3. The end region design method according to claim 2, wherein The establishment of the initial database based on the experimental design method includes the steps of: Use the optimal Latin hypercube method in the experimental design method to obtain geometric sample data of the wall surface parameters; According to the geometric sample data, perform shaping design on the wall surface of the wide-length vaneless diffuser to obtain multiple different shaping design schemes; Perform simulation processing on multiple shaping design schemes based on the CFD simulation chain constructed by Isight; Obtain corresponding optimization objective variables according to the simulation results and the stability margin formula; Establish the initial database based on the free variables and their corresponding optimization objective variables.

4. The end region design method according to claim 3, wherein, The stability margin formula is as follows: Where m is the mass flow rate, π is the pressure ratio, the subscript NS represents the near-stall point, PEW represents the end wall shaping, and SW represents the prototype.

5. The end region design method according to claim 1, wherein When performing hyperparameter optimization on the initial prediction model, the performance evaluation formula used is: Where NMSE is the normalized root mean square error, x is the prediction result of the radial basis function surrogate model, and x0 is the simulation result.

6. A tip region design device is applied to an optimization design platform and is used for the optimization design of a centrifugal compressor wide and long vaneless diffuser, and is characterized in that The end region design device includes: A shaping determination module configured to determine the final shaping region of the wide-length vaneless diffuser; A database establishment module configured to establish an initial database based on the experimental design method, where the initial database includes multiple groups of optimization objective variables; A model construction module configured to establish an initial prediction model using a radial basis function surrogate model based on the initial database; A model optimization module configured to perform hyperparameter optimization on the initial prediction model to obtain an optimal end region prediction surrogate model; A design execution module configured to perform multi-objective optimization design using a genetic algorithm based on the optimal end region prediction surrogate model to obtain an optimal shaping design scheme for the end region.

7. The end region design device according to claim 6, characterized in that, The shaping determination module includes: A parameterization processing unit configured to parameterize the wall surface of the wide-length vaneless diffuser to obtain wall surface parameters; A determination execution unit configured to perform shaping design based on the wall surface parameters and determine the final shaping region through sensitivity analysis.

8. The end region design device according to claim 7, wherein ​ ​ A shaping processing unit, configured to perform shaping design on the wall surface of the wide-length vane-less diffuser according to the geometric sample data to obtain a variety of different shaping design schemes; A simulation processing unit, configured to perform simulation processing on different ones of the shaping design schemes based on a CFD simulation chain constructed by Isight; A variable calculation unit, configured to obtain corresponding optimization target variables according to the simulation results and the stability margin formula; A construction execution unit, configured to establish the initial database based on the free variables and their corresponding optimization target variables.

9. The end region design device according to claim 8, wherein, The stability margin formula is as follows: where m is the mass flow rate, π is the pressure ratio, the subscript NS represents the near-stall point, PEW represents the end-wall shaping, and SW represents the prototype.

10. An optimized design platform, characterized in that, The optimization design platform includes a CFD simulation chain and an optimization process, where: The CFD simulation chain is used to store computer programs or instructions, and can automatically perform parametric geometric modeling, mesh generation, numerical simulation solution, and post-processing performance parameter extraction; The optimization process is used to execute end-wall shaping optimization design so that the wide-length vane-less diffuser of the centrifugal compressor realizes the end-region design method according to any one of claims 1 to 5.

Images