Compressor stability extension method based on clustering modal cluster energy transient growth rate regulation and control

Abstract

The invention discloses a gas compressor stability extension method based on clustering modal cluster energy transient growth rate regulation and control, which comprises the following steps of: acquiring flow field data through unsteady numerical simulation of a centrifugal gas compressor, performing dynamic modal decomposition to extract modal characteristics, obtaining a modal cluster by utilizing clustering analysis and calculating energy, the method comprises the following steps: defining a modal cluster energy transient growth rate, determining a design variable and a change range, generating a sample set, calculating the modal cluster energy transient growth rate, constructing an agent model, carrying out multi-objective optimization by adopting a non-dominated sorting genetic algorithm, and finally obtaining an optimal design scheme. According to the optimization method based on clustering modal cluster energy transient growth rate regulation and control, performance improvement of the centrifugal compressor under different working conditions is achieved, the isentropic efficiency and the total pressure ratio of the compressor are remarkably improved, and meanwhile the operation stability of the compressor is enhanced.

Description

Technical Field

[0001] This invention belongs to the field of turbomachinery optimization design technology, and in particular relates to a compressor stabilization method based on the control of transient growth rate of energy in clustered mode clusters. Background Technology

[0002] Centrifugal compressors, as highly efficient energy conversion machines, convert mechanical energy into pressure energy by performing work on gas through their blades. They are widely used in turbocharged internal combustion engines, small and medium-sized gas turbines in ships, aero engines, and industrial compression devices. Therefore, improving the stability and efficiency of centrifugal compressors has always been a key research focus in this field.

[0003] In the early days, due to insufficient understanding of flow mechanisms, traditional trial-and-error methods often required multiple iterative design cycles, resulting in long design cycles and difficulty in obtaining a globally optimal solution. Furthermore, this method heavily relied on the designer's experience; those lacking experience experienced even longer design cycles or failed to achieve effective results. Therefore, traditional trial-and-error methods are no longer sufficient to meet the design requirements of modern high-performance compressors. In recent years, thanks to advancements in computational fluid dynamics, mathematical optimization theory, and computer hardware, rapid automatic optimization methods based on numerical simulations have been widely applied to the aerodynamic optimization of components such as centrifugal compressor inlet guide vanes, impellers, diffusers, return valves, volutes, and casing treatments. However, this method only optimizes macroscopic aerodynamic performance indicators and cannot actively control specific flow structures based on flow mechanisms.

[0004] Recently, a new optimization method for centrifugal compressors based on modal energy transfer theory and with modal energy as the objective function has been emerging. However, after modal decomposition of unsteady flow field data of centrifugal compressors, there may be a broadband mode group with similar modal energy proportions and frequencies, making the selection of the target mode difficult.

[0005] To address the aforementioned issues, there is an urgent need to propose a compressor stabilization method based on the control of transient growth rate of energy in clustered mode clusters. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a compressor stabilization method based on the control of transient growth rate of energy in clustered mode clusters.

[0007] This invention proposes a compressor stabilization method based on the control of transient growth rate of energy in clustered mode clusters, comprising the following steps:

[0008] Unsteady flow field data were obtained based on unsteady numerical simulation of a centrifugal compressor.

[0009] Dynamic mode decomposition is performed based on flow field data, and the spatial and time-varying features of the modes are extracted.

[0010] Based on the spatial and time-varying characteristics of modes, mode clusters are obtained using cluster analysis and the energy of the mode clusters is calculated;

[0011] Based on the modal cluster energy, a functional relationship for the transient growth rate of the modal cluster energy is preset;

[0012] Based on the impact of the transient growth rate of mode cluster energy on the compressor, the compressor design variables and their range of variation are determined;

[0013] A sample set is generated based on the compressor design variables. Unsteady numerical simulations are performed on each sample, and the transient growth rate of the modal cluster energy is calculated based on the functional relationship of the transient growth rate of the modal cluster energy.

[0014] Based on the calculation results, a proxy model is constructed using a generalized neural network.

[0015] Based on the surrogate model, a non-dominated sorting genetic algorithm is used to perform multi-objective optimization of the compressor to obtain the design scheme with the optimal transient growth rate.

[0016] Optionally, the process of obtaining unsteady flow field data based on unsteady numerical simulation of a centrifugal compressor includes:

[0017] Structured meshes are used to generate meshes for the inlet section, impeller, and diffuser. The impeller blades and diffuser blades use O-type meshes, while the impeller and diffuser channels use H-type, J-type, C-type, and L-type meshes.

[0018] Numerical simulations were performed using the unsteady Reynolds-averaged method based on the grid. The SST k-ω model was selected as the turbulence model, and spatial discretization was performed using a high-order difference scheme and temporal discretization using a second-order backward Euler scheme. The inlet boundary conditions were set as total pressure and total temperature, and the outlet boundary conditions were set as static pressure or mass flow rate. The solid wall adopted an adiabatic no-slip boundary condition, and the dynamic-static interface was set as the transient-to-stator method to obtain the unsteady flow field of the centrifugal compressor.

[0019] Optionally, the process of performing dynamic mode decomposition based on flow field data and extracting the spatial and time-varying features of the modes includes:

[0020] Unsteady flow field data of the spanwise section of a centrifugal compressor impeller were selected as the analysis object. The dynamic mode decomposition method was used to perform mode decomposition on the selected flow field data, extract the spatial features of the modes, and represent the spatial structure of the modes with the first five central moments of the modes. At the same time, the time-varying features of the modes were extracted, and the temporal variation characteristics of the modes were represented by the increase and decrease of frequency and time coefficient.

[0021] Optionally, the functional relationship of the transient growth rate of the mode cluster energy is:

[0022] If the total energy of the inherent or characteristic mode cluster before optimization is set to the first energy value, and the total energy of the inherent or characteristic mode cluster after optimization is set to the second energy value, then the transient growth rate of the mode cluster energy is the difference between the second energy value and the first energy value divided by the first energy value.

[0023] Optionally, the process of determining the compressor design variables and their range of variation based on the impact of the transient growth rate of mode cluster energy on the compressor includes:

[0024] The influence of a single design variable on the transient growth rate of energy of the intrinsic mode cluster and the transient growth rate of energy of the characteristic mode cluster is calculated within the range of variation. Based on the calculation results, the key design variables in the impeller, diffuser and casing treatment that have an influence on the transient growth rate of energy of the mode cluster exceeding the preset threshold are selected, and the key design variables are determined as the design space.

[0025] Optionally, the process of generating a sample set based on compressor design variables, performing unsteady numerical simulations on each sample, and calculating the transient growth rate of the mode cluster energy for each sample based on the functional relationship of the transient growth rate of the mode cluster energy includes:

[0026] Based on the compressor design variables, an optimal Latin hypercube sampling method is used to generate a sample set, ensuring that the sample points are evenly distributed throughout the design space. For each sample in the sample set, unsteady flow field data of the compressor under the highest efficiency condition and near-stall condition are obtained using an unsteady numerical simulation method. Based on the functional relationship of the transient growth rate of mode cluster energy, combined with the mode cluster energy obtained from dynamic mode decomposition, the transient growth rate of the intrinsic mode cluster energy and the transient growth rate of the characteristic mode cluster energy under the highest efficiency condition and near-stall condition are calculated for each sample.

[0027] Optionally, the process of constructing the surrogate model using a generalized neural network based on the calculation results is as follows:

[0028] Data on the transient growth rates of the energy of the inherent mode clusters and the transient growth rates of the energy of the characteristic mode clusters of the compressor under different operating conditions, obtained from unsteady numerical simulation calculations, are collected. A surrogate model is constructed using a generalized neural network, and the mapping relationship between the transient growth rates of the energy of the compressor mode clusters and key design variables is established through the surrogate model.

[0029] Optionally, the process of using a non-dominated sorting genetic algorithm to perform multi-objective optimization of the compressor based on a surrogate model to obtain the design scheme with the optimal transient growth rate includes:

[0030] Based on the mapping relationship between the transient growth rate of modal cluster energy and design variables established by the surrogate model, a non-dominated sorting genetic algorithm is used to perform multi-objective optimization of the compressor. The objective function is to maximize the transient growth rate of intrinsic modal cluster energy and minimize the transient growth rate of characteristic modal cluster energy under the compressor's highest efficiency condition and near-stall condition. Through iterative calculation of the algorithm, a Pareto front solution set is generated, and the design variable parameters corresponding to the optimal solution are selected. These parameters are verified by numerical simulation calculation, and finally the design scheme with the optimal transient growth rate is determined.

[0031] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method.

[0032] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

[0033] Compared with the prior art, the present invention has the following advantages and technical effects:

[0034] 1. This invention employs a dynamic mode decomposition method to perform mode decomposition on the compressor and extracts the spatial and time-varying features of the modes. The K-means++ algorithm is introduced to perform cluster analysis on the mode features, obtaining mode clusters containing similar modes and calculating the mode cluster energy. This solves the difficulty that the mode that dominates the compressor stability cannot be correctly selected due to the existence of similar modes, and improves the rationality and interpretability of the target mode selection.

[0035] 2. This invention defines the transient growth rate of modal cluster energy. Compared with the modal cluster energy ratio, which is only applicable to the comparison of modal clusters under the same working condition and the same geometric model, the transient growth rate of modal cluster energy is applicable to the comparison of modal clusters under different working conditions and different geometric models. Using the transient growth rate of modal cluster energy as the objective function solves the problem that the optimization results of using the modal cluster energy ratio as the objective function are unreasonable, and realizes efficient and accurate multi-objective optimization. Attached Figure Description

[0036] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0037] Figure 1 This is a schematic diagram of the modal clusters for the highest efficiency operating condition in an embodiment of the present invention;

[0038] Figure 2 This is a schematic diagram of the modal clusters under near-stall conditions according to an embodiment of the present invention;

[0039] Figure 3 This is a geometric schematic diagram of the self-circulating casing according to an embodiment of the present invention;

[0040] Figure 4 This is a comparison chart of isentropic efficiency curves before and after optimization in an embodiment of the present invention;

[0041] Figure 5 This is a comparison chart of the total pressure ratio curves before and after optimization in an embodiment of the present invention. Detailed Implementation

[0042] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0043] 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.

[0044] Example 1

[0045] This embodiment provides a compressor stabilization method based on the control of transient energy growth rate of clustered mode clusters, including the following steps:

[0046] Unsteady flow field data were obtained based on unsteady numerical simulation of a centrifugal compressor.

[0047] Dynamic mode decomposition is performed based on flow field data, and the spatial and time-varying features of the modes are extracted.

[0048] Based on the spatial and time-varying characteristics of modes, mode clusters are obtained using cluster analysis and the energy of the mode clusters is calculated;

[0049] Based on the modal cluster energy, a functional relationship for the transient growth rate of the modal cluster energy is preset;

[0050] Based on the impact of the transient growth rate of mode cluster energy on the compressor, the compressor design variables and their range of variation are determined;

[0051] A sample set is generated based on the compressor design variables. Unsteady numerical simulations are performed on each sample, and the transient growth rate of the modal cluster energy is calculated based on the functional relationship of the transient growth rate of the modal cluster energy.

[0052] Based on the calculation results, a proxy model is constructed using a generalized neural network.

[0053] Based on the surrogate model, a non-dominated sorting genetic algorithm is used to perform multi-objective optimization of the compressor to obtain the design scheme with the optimal transient growth rate.

[0054] The feasible process of obtaining unsteady flow field data based on unsteady numerical simulation of a centrifugal compressor includes:

[0055] Structured meshes were used to generate meshes for the inlet section, impeller, and diffuser. O-type meshes were used for impeller and diffuser blades, while H-, J-, C-, and L-type meshes were used for the impeller and diffuser channels. Based on these meshes, numerical simulations were performed using the unsteady Reynolds-averaged method. The SST k-ω model was selected as the turbulence model, and spatial discretization was performed using a high-order difference scheme, while temporal discretization was performed using a second-order backward Euler scheme. The inlet boundary conditions were set to total pressure and total temperature, and the outlet boundary conditions were set to static pressure or mass flow rate. Adiabatic no-slip boundary conditions were used on the solid wall, and the dynamic-static interface was set using the transient-to-stator method to obtain the unsteady flow field of the centrifugal compressor.

[0056] As a specific implementation, this embodiment uses structured meshes to generate the meshes for the inlet section, impeller, and diffuser. O-type meshes are used near the impeller and diffuser blades, H / J / C / L-type meshes are used for the impeller and diffuser passages, and unstructured meshes are used for the volute. All meshes have a y+ value less than 5 near the wall, and the mesh density meets the mesh independence requirement. Furthermore, the compressor casing is a self-circulating casing, modeled parametrically, with a structured mesh. The y+ value and mesh settings are consistent with those of the impeller. Numerical simulation is performed using the unsteady Reynolds-averaged method, with the SST k-ω model as the turbulence model. A high-order difference scheme is used for spatial discretization, and a second-order backward Euler scheme is used for temporal discretization. The inlet boundary conditions are set with a total pressure of 1 atm and a total temperature of 298 K. The outlet boundary conditions are set as static pressure outlet under high flow rate conditions and mass flow rate outlet under low flow rate conditions. The solid wall adopts adiabatic no-slip boundary conditions. The interface between the inlet section and the impeller, and between the impeller and the diffuser, is set as transient rotor-stator. The interface between the diffuser and the volute, and between the casing and the inlet section and the impeller, is set as frozen rotor. The unsteady flow field of the compressor is obtained based on the above numerical simulation method.

[0057] The feasible process of performing dynamic mode decomposition based on flow field data and extracting the spatial and time-varying features of the modes includes:

[0058] Unsteady flow field data of the spanwise section of a centrifugal compressor impeller were selected as the analysis object. The dynamic mode decomposition method was used to perform mode decomposition on the selected flow field data, extract the spatial features of the modes, and represent the spatial structure of the modes with the first five central moments of the modes. At the same time, the time-varying features of the modes were extracted, and the temporal variation characteristics of the modes were represented by the increase and decrease of frequency and time coefficient.

[0059] As a specific implementation method, dynamic mode decomposition and mode clustering analysis of the unsteady flow field of the compressor are performed. First, unsteady flow field data of the compressor impeller spanwise section are selected, and mode decomposition is performed using the dynamic mode decomposition method. Second, the modal space and time-varying features are extracted. The modal space structure is represented by the first five central moments of the modes, and the time-varying features are represented by the frequency and time coefficient increments / decreases. Then, based on the modal features, the K-means++ algorithm is used to cluster the modes to obtain mode clusters containing similar modes. Finally, based on the impeller flow field analysis, intrinsic mode clusters and characteristic mode clusters are determined. Intrinsic mode clusters characterize the self-excited disturbances and dynamic-static interference structures in the compressor internal flow field, and their modal space structure and frequencies generally do not change with compressor flow rate. Characteristic mode clusters characterize the vortex-induced disturbance structures in the compressor internal flow field, and their modal space structure and frequencies change with compressor flow rate. The total energy of the intrinsic mode clusters and the total energy of the characteristic mode clusters are calculated.

[0060] The feasible functional relationship for the transient growth rate of the mode cluster energy is as follows:

[0061] If the total energy of the intrinsic mode cluster before optimization is the first energy value and the total energy of the intrinsic mode cluster after optimization is the second energy value, then the transient growth rate of the mode cluster energy is the difference between the second energy value and the first energy value divided by the first energy value.

[0062] As a specific implementation method, the transient growth rate of modal cluster energy is defined. Assuming the total energy of the intrinsic (feature) modal clusters before optimization is E1, and the total energy of the intrinsic (feature) modal clusters after optimization is E2, then the transient growth rate of the intrinsic (feature) modal cluster energy is defined as follows:

[0063] The feasible process of determining the compressor design variables and their range of variation based on the influence of the transient growth rate of mode cluster energy on the compressor includes:

[0064] The influence of a single design variable on the transient growth rate of energy of the intrinsic mode cluster and the transient growth rate of energy of the characteristic mode cluster is calculated within the range of variation. Based on the calculation results, the key design variables in the impeller, diffuser and casing treatment that have an influence on the transient growth rate of energy of the mode cluster exceeding the preset threshold are selected, and the key design variables are determined as the design space.

[0065] As a specific implementation method, single-parameter analysis is used to determine design variables. Under the condition that the compressor geometry does not undergo severe distortion or deformation, the impact of a single design variable on the transient growth rate of the intrinsic and characteristic mode cluster energy within its variation range is calculated. Specifically, the transient growth rates of the intrinsic and characteristic mode cluster energy are calculated for different values ​​of the single variable within its variation range. Key design variables in the impeller, diffuser, and casing treatments that have an impact exceeding 4% on the transient growth rate of the mode cluster energy are selected. The range of variation of the design variables that results in a positive transient growth rate of the intrinsic mode cluster and a negative transient growth rate of the characteristic mode cluster is defined as the design space.

[0066] The feasible process of generating a sample set based on compressor design variables, performing unsteady numerical simulations on each sample, and calculating the transient growth rate of the mode cluster energy for each sample based on the functional relationship of the transient growth rate of the mode cluster energy includes:

[0067] Based on the compressor design variables, an optimal Latin hypercube sampling method is used to generate a sample set, ensuring that the sample points are evenly distributed throughout the design space. For each sample in the sample set, unsteady flow field data of the compressor under the highest efficiency condition and near-stall condition are obtained using an unsteady numerical simulation method. Based on the functional relationship of the transient growth rate of mode cluster energy, combined with the mode cluster energy obtained from dynamic mode decomposition, the transient growth rate of the intrinsic mode cluster energy and the transient growth rate of the characteristic mode cluster energy under the highest efficiency condition and near-stall condition are calculated for each sample.

[0068] As a specific implementation method, the optimal Latin hypercube sampling method is used to generate the design variable sample set. This method is an optimization based on Latin hypercube sampling, which makes the sample points evenly distributed throughout the design space. The initial number of samples in the sample set is generally related to the number of design variables, and the minimum number requirement is (n+1)×(n+2) / 2 times the number of design variables.

[0069] The feasible process of constructing the surrogate model using a generalized neural network based on the calculation results is as follows:

[0070] Data on the transient growth rates of the energy of the inherent mode clusters and the transient growth rates of the energy of the characteristic mode clusters of the compressor under different operating conditions, obtained from unsteady numerical simulation calculations, are collected. A surrogate model is constructed using a generalized neural network, and the mapping relationship between the transient growth rates of the energy of the compressor mode clusters and key design variables is established through the surrogate model.

[0071] As a specific implementation method, a mapping relationship between compressor geometric parameters and the transient growth rate of modal cluster energy is constructed. A generalized neural network (GRNN) is used to construct the mapping relationship between compressor geometric parameters and the transient growth rate of modal cluster energy, forming a surrogate model for rapid evaluation of the transient growth rate of compressor modal cluster energy. Based on the surrogate model, the transient growth rates of intrinsic and characteristic modal cluster energy under the compressor's highest efficiency condition and near-stall condition are obtained.

[0072] The feasible process of using a non-dominated sorting genetic algorithm to perform multi-objective optimization of the compressor based on a surrogate model to obtain the design scheme with the optimal transient growth rate includes:

[0073] Based on the mapping relationship between the transient growth rate of modal cluster energy and design variables established by the surrogate model, a non-dominated sorting genetic algorithm is used to perform multi-objective optimization of the compressor. The objective function is to maximize the transient growth rate of intrinsic modal cluster energy and minimize the transient growth rate of characteristic modal cluster energy under the compressor's highest efficiency condition and near-stall condition. Through iterative calculation of the algorithm, a Pareto front solution set is generated, and the design variable parameters corresponding to the optimal solution are selected. These parameters are verified by numerical simulation calculation, and finally the design scheme with the optimal transient growth rate is determined.

[0074] As a specific implementation method, combined with the surrogate model, the non-dominated sorting genetic algorithm (NSGA-II) is used for optimization. The objective function is to maximize the transient growth rate of the energy of the intrinsic mode cluster under the compressor's highest efficiency condition and near-stall condition, and to minimize the transient growth rate of the energy of the characteristic mode cluster under the compressor's highest efficiency condition and near-stall condition. Finally, the Pareto front solution set is obtained, and the optimization schemes that have a significant improvement in all objective functions are screened for verification, thereby obtaining the optimal design scheme.

[0075] Example 2

[0076] This embodiment uses a certain type of turbocharger compressor and simulates it using the unsteady numerical method in Embodiment 1 to obtain the unsteady flow field of this type of compressor under the highest efficiency condition (mass flow rate of 0.78 kg / s) and near stall condition (mass flow rate of 0.68 kg / s).

[0077] Dynamic mode decomposition was performed on unsteady flow field data of the compressor impeller at 96% blade height under the highest efficiency and near-stall operating conditions. The main modes and their corresponding frequencies are shown in Table 1. Cluster analysis was performed on the main modes under the two operating conditions, as follows: Figure 1 and Figure 2 The diagram shows the mode clusters obtained under two operating conditions. Based on the flow field analysis, the inherent mode clusters under the highest efficiency condition are determined to be mode clusters 1 and 3, and the characteristic mode cluster is mode cluster 2. Under the near-stall condition, the inherent mode clusters are mode clusters 1 and 3, and the characteristic mode cluster is mode cluster 2.

[0078] Table 1

[0079]

[0080] like Figure 3 As shown, the design variables and their ranges are determined as follows: suction groove width g1 (1mm~4mm), suction groove axial position g2 (-47.3mm~-32.7mm), suction groove angle g3 (90°~132°), return groove axial position g4 (-80.5mm~-59.5mm), return groove angle g5 (120°~160°), cavity downward tilt angle g6 (0°~3°), cavity upward tilt angle g7 (0°~3°), return groove cavity radial width g8 (1mm~5.7mm), return groove radial width g9 (0.7mm~18.7mm), cavity radial width g... 10 (22.5mm~32.5mm), axial length of cavity g 11 (77.0mm~91.9mm), diffuser consistency g 12 (6~12);

[0081] A sample set of 120 samples was determined, and the transient growth rate of energy of the intrinsic and characteristic mode clusters for each sample under the compressor’s highest efficiency condition and near-stall condition was obtained.

[0082] A surrogate model was constructed, and multi-objective optimization was performed to obtain the optimal solution: Under the highest efficiency condition, the transient energy growth rate of the intrinsic mode family was 0.08%, and the transient energy growth rate of the characteristic mode family was -1.75%; under the near-stall condition, the transient energy growth rate of the intrinsic mode family was 12.09%, and the transient energy growth rate of the characteristic mode family was -46.39%. Figure 4 and Figure 5 As shown, compared to the original model, the optimized compressor improves the isentropic efficiency and total pressure ratio by 1.2% and 1.1% respectively under the highest efficiency condition, and by 3.5% and 4.6% respectively under the near-stall condition.

[0083] Example 3

[0084] This embodiment also discloses a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in Embodiment 1.

[0085] Example 4

[0086] This embodiment also discloses a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0087] Example 5

[0088] This embodiment also discloses a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in Embodiment 1.

[0089] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A compressor stabilization method based on the transient growth rate regulation of energy in clustered mode clusters, characterized in that, Includes the following steps: Unsteady flow field data were obtained based on unsteady numerical simulation of a centrifugal compressor. Dynamic mode decomposition is performed based on flow field data, and the spatial and time-varying features of the modes are extracted. Based on the spatial and time-varying characteristics of modes, mode clusters are obtained using cluster analysis and the energy of the mode clusters is calculated; Based on the modal cluster energy, a functional relationship for the transient growth rate of the modal cluster energy is preset; Based on the impact of the transient growth rate of mode cluster energy on the compressor, the compressor design variables and their range of variation are determined; A sample set is generated based on the compressor design variables. Unsteady numerical simulations are performed on each sample, and the transient growth rate of the modal cluster energy is calculated based on the functional relationship of the transient growth rate of the modal cluster energy. Based on the calculation results, a proxy model is constructed using a generalized neural network. Based on the surrogate model, a non-dominated sorting genetic algorithm is used to perform multi-objective optimization of the compressor to obtain the design scheme with the optimal transient growth rate.

2. The method according to claim 1, characterized in that, The process of obtaining unsteady flow field data based on unsteady numerical simulation of a centrifugal compressor includes: Structured meshes are used to generate meshes for the inlet section, impeller, and diffuser. The impeller blades and diffuser blades use O-type meshes, while the impeller and diffuser channels use H-type, J-type, C-type, and L-type meshes. Numerical simulations were performed using the unsteady Reynolds-averaged method based on the grid. The SST k-ω model was selected as the turbulence model, and spatial discretization was performed using a high-order difference scheme and temporal discretization using a second-order backward Euler scheme. The inlet boundary conditions were set as total pressure and total temperature, and the outlet boundary conditions were set as static pressure or mass flow rate. The solid wall adopted an adiabatic no-slip boundary condition, and the dynamic-static interface was set as the transient-to-stator method to obtain the unsteady flow field of the centrifugal compressor.

3. The method according to claim 1, characterized in that, The process of performing dynamic mode decomposition based on flow field data and extracting the spatial and time-varying features of the modes includes: Unsteady flow field data of the spanwise section of a centrifugal compressor impeller were selected as the analysis object. The dynamic mode decomposition method was used to perform mode decomposition on the selected flow field data, extract the spatial features of the modes, and represent the spatial structure of the modes with the first five central moments of the modes. At the same time, the time-varying features of the modes were extracted, and the temporal variation characteristics of the modes were represented by the increase and decrease of frequency and time coefficient.

4. The method according to claim 1, characterized in that, The functional relationship of the transient growth rate of the energy of the mode cluster is as follows: The total energy of the intrinsic or characteristic mode clusters before optimization is set as the first energy value, and the total energy of the intrinsic or characteristic mode clusters after optimization is set as the second energy value. Then, the transient growth rate of the mode cluster energy is the difference between the second energy value and the first energy value divided by the first energy value. The transient growth rate of the intrinsic mode cluster energy and the transient growth rate of the characteristic mode cluster energy are calculated based on the functional relationship of the transient growth rate of the mode cluster energy.

5. The method according to claim 4, characterized in that, The process of determining compressor design variables and their range of variation based on the impact of the transient growth rate of modal cluster energy on the compressor includes: The influence of a single design variable on the transient growth rate of energy of the intrinsic mode cluster and the transient growth rate of energy of the characteristic mode cluster is calculated within the range of variation. Based on the calculation results, the key design variables in the impeller, diffuser and casing treatment that have an influence on the transient growth rate of energy of the mode cluster exceeding the preset threshold are selected, and the key design variables are determined as the design space.

6. The method according to claim 1, characterized in that, The process of generating a sample set based on compressor design variables, performing unsteady numerical simulations on each sample, and calculating the transient growth rate of the modal cluster energy for each sample based on the functional relationship of the transient growth rate of the modal cluster energy includes: Based on the compressor design variables, an optimal Latin hypercube sampling method is used to generate a sample set, ensuring that the sample points are evenly distributed throughout the design space. For each sample in the sample set, unsteady flow field data of the compressor under the highest efficiency condition and near-stall condition are obtained using an unsteady numerical simulation method. Based on the functional relationship of the transient growth rate of mode cluster energy, combined with the mode cluster energy obtained from dynamic mode decomposition, the transient growth rate of the intrinsic mode cluster energy and the transient growth rate of the characteristic mode cluster energy under the highest efficiency condition and near-stall condition are calculated for each sample.

7. The method according to claim 1, characterized in that, The process of constructing a proxy model using a generalized neural network based on the calculation results is as follows: Data on the transient growth rates of the energy of the inherent mode clusters and the transient growth rates of the energy of the characteristic mode clusters of the compressor under different operating conditions, obtained from unsteady numerical simulation calculations, are collected. A surrogate model is constructed using a generalized neural network, and the mapping relationship between the transient growth rates of the energy of the compressor mode clusters and key design variables is established through the surrogate model.

8. The method according to claim 1, characterized in that, The process of using a non-dominated sorting genetic algorithm to perform multi-objective optimization of the compressor based on a surrogate model to obtain the design scheme with the optimal transient growth rate includes: Based on the mapping relationship between the transient growth rate of modal cluster energy and design variables established by the surrogate model, a non-dominated sorting genetic algorithm is used to perform multi-objective optimization of the compressor. The objective function is to maximize the transient growth rate of intrinsic modal cluster energy and minimize the transient growth rate of characteristic modal cluster energy under the compressor's highest efficiency condition and near-stall condition. Through iterative calculation of the algorithm, a Pareto front solution set is generated, and the design variable parameters corresponding to the optimal solution are selected. These parameters are verified by numerical simulation calculation, and finally the design scheme with the optimal transient growth rate is determined.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the method according to any one of claims 1-8.

Images