A hybrid simulation method for discrete frequency aerodynamic noise
Through HD-DMD stream sound separation preprocessing technology, discrete frequency noise sound source data is extracted and added in the form of sound source boundary, which solves the problem of large data coupled storage space and poor computing stability, and achieves more efficient discrete frequency aerodynamic noise simulation.
Patent Information
- Application Number
- CN202411288110.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-14
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-09-14
AI Technical Summary
In the prior art, when coupling CFD flow field solver and CAA sound field solver to simulate discrete frequency aerodynamic noise, file-based data coupling requires too much storage space, and the near-field flow field pulsation is directly assigned to the sound source boundary. Impacts calculation accuracy and stability.
The Helmholtz-dynamic modal decomposition (HD-DMD) method is used to perform stream-acoustic separation pre-processing on CFD flow field data, extract the dominant discrete frequency acoustic mode, reduce storage requirements, and transfer the processed data to the sound source boundary. The form is added to the acoustic propagation equation for calculation.
Reduces the storage space required for data coupling, enhances computing stability and accuracy, and avoids the influence of non-acoustic modes.
Smart Images

Figure CN118798099B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the field of numerical simulation, and in particular to a discrete frequency aerodynamic noise hybrid simulation method. Background Art
[0002] Driven by stricter and more sophisticated design requirements, the demand for multi-physics coupling modeling and numerical simulation capabilities is growing rapidly. To achieve multi-physics calculations, it is necessary to run two or more mature single-physics numerical simulation tools in different disciplines at the same time and effectively exchange coupling data. Aeroacoustic problems involve flow and noise and are typical multi-physics coupling problems. Depending on the calculation strategy, the solution methods of computational aeroacoustics (CAA) are divided into two categories: direct simulation and hybrid simulation.
[0003] The direct simulation method is to couple the flow field and the acoustic field by directly solving the Navier-Stokes (NS) equations or their modularized equations in a unified computational domain, and simultaneously obtain the generation and propagation process of aerodynamic noise. This method requires not only accurate solution of the unsteady flow information in the near field, but also accurate solution of the acoustic field information in the mid- and far-fields. There is a huge difference in scale between the acoustic field information and the flow field information. Compared with the traditional computational fluid dynamics (CFD) simulation, the direct simulation CAA method has a larger computational domain, a smaller grid space scale far from the object surface, a much higher total amount of grids required, and a huge amount of calculation, which is difficult to meet with the existing computer hardware conditions. Considering comprehensive computing resources and engineering applications, the hybrid simulation method can obtain relatively accurate simulation results at a relatively low computational cost.
[0004] The hybrid simulation method decomposes the computational aeroacoustic problem into two independent processes: the generation of sound waves and the propagation of sound waves in the flow field. The specific approach is to divide the entire calculation area into two parts: the sound source area and the propagation area. The flow field calculation is performed in the sound source area, and the sound field calculation is performed in the propagation area, and different control equations are solved respectively. The sound wave is generated in the sound source area, and the sound field calculation in the propagation area is based on the unsteady simulation of the flow field in the sound source area. According to the type of NS equations and physical models used, the CFD flow field solution in the sound source area can be divided into: Unsteady Reynolds Averaged NS simulation (URANS), Detached Eddy Simulation (DES), Large Eddy Simulation (LES) and Direct Numerical Simulation (DNS). There are two ways to calculate the sound field in the propagation area: the acoustic analogy equation and the acoustic propagation equation, which are respectively applied to the solution of the acoustic far field and the midfield.
[0005] The CFD coupled sound propagation equation simulation method uses the CFD flow field solver to solve the NS equation to obtain the average flow and sound source information, and then uses the CAA sound field solver to solve the sound propagation equation to obtain the sound field in the propagation area. It is suitable for simulating aerodynamic noise propagation problems in non-uniform flow fields or flow fields with complex boundary conditions. Compared with the direct solution method, the hybrid solution method based on the sound propagation equation can greatly reduce the amount of calculation while ensuring the calculation accuracy.
[0006] When this method is implemented, it is necessary to transfer the unsteady sound source data calculated by the CFD flow field solver to the CAA sound field (sound propagation) solver to complete the data coupling of the two solvers. Memory-based data coupling relies on third-party libraries (such as preCICE and CWIPI, etc.), and the solver code needs to be modified, while file-based data coupling does not rely on third-party libraries and does not require the solver code to be modified. However, the biggest disadvantage of file-based data coupling is that this coupling requires a considerable amount of hard disk storage space, sometimes even several TB. Implementing dimensionality reduction and reconstruction preprocessing based on modal decomposition for unsteady sound source data can reduce the storage space requirements for storing sound source data files.
[0007] The derivation idea of the sound propagation equation is to decompose the physical quantity of the flow field into two parts: the average quantity and the pulsating quantity, and then bring the physical quantity decomposed into the average quantity and the pulsating quantity into the Euler equation, and obtain the linear sound propagation equation after sorting and simplification. Linear Euler Equation (LEE) and Acoustic Perturbation Equations (APE) are two different types of sound propagation equations. Although the specific forms of the two are different, the right-hand side terms of the equations are all sound source terms. The essence of the solution method of the CFD coupled sound propagation equation is to extract the sound source from the CFD flow field and add it to the linear sound propagation equation calculation as a sound source term or in the form of a sound source boundary. When added in the form of a sound source term, the source term on the right side of the sound propagation equation needs to be retained, and the sound source term calculation is performed based on the average quantity and pulsating quantity of the CFD flow field. When added in the form of a sound source boundary, the source term on the right side of the sound propagation equation can be discarded, and there is no need to calculate the sound source term. Instead, the flow field pulsating quantity is directly assigned to the sound source boundary of the sound propagation area.
[0008] The CFD coupled sound propagation equation simulation method requires two sets of grids to complete the calculation: (1) CFD grid, used for flow field calculation preparation; (2) CAA grid, used for sound field calculation. The two sets of grids have very different requirements for the total number of grid points and the density of grid points. For CFD grids, it is necessary to encrypt the grids near the wall and in areas with strong flow. In order to reduce the amount of calculation, the grid scale will generally increase rapidly in areas far away from the wall and where the flow is not strong, while the calculation results in areas with large grid scales are generally less accurate. Therefore, setting the sound source boundary of the CAA grid in an area with a larger CFD grid scale will affect the accuracy of solving the sound propagation equation.
[0009] The flow field pulsation in the near field, such as near the wall and in the area of vigorous flow, contains both compressible acoustic modes and incompressible vortex modes. The sound propagation equation (such as the LEE equation) not only supports the propagation of sound waves, but also supports the propagation of vortex waves and entropy waves. The current method of directly assigning the near-field flow field pulsation quantity to the sound source boundary of the CAA grid for sound propagation calculation will be accompanied by non-acoustic mode propagation, which will not only affect the accuracy of the sound propagation calculation, but also if the sound source boundary is close to the area of vigorous flow, it may even cause calculation instability at the sound source boundary, leading to calculation crash.
[0010] The applicant has found that the prior art faces two problems when coupling the CFD flow field solver and the CAA acoustic field (acoustic propagation) solver to simulate discrete frequency aerodynamic noise: (1) the storage space required for file-based data coupling of the two solvers is too large; and (2) when the near-field flow field pulsation quantity is directly assigned to the sound source boundary for sound propagation calculation, the accompanying non-acoustic modes affect the calculation accuracy and stability. Summary of the invention
[0011] In order to solve the above problems, the present invention provides a discrete frequency aerodynamic noise hybrid simulation method, the method comprising:
[0012] Step S1: performing unsteady fluid dynamics simulation based on the CFD grid to obtain unsteady CFD time-series flow field data including velocity field, pressure field and density field;
[0013] Step S2: Based on the unsteady CFD time-series flow field data, calculate and obtain the time-averaged flow field data and the time-series pulsating flow field data on the CFD grid;
[0014] Step S3: using mesh generation software to generate a CAA mesh required by a CAA sound field solver for sound propagation calculation;
[0015] Step S4: interpolating the time-averaged flow field data on the CFD grid to the CAA grid to obtain the background flow field data required for the sound propagation calculation;
[0016] Step S5: Perform dynamic modal decomposition on the time-series pulsating velocity field data to obtain the DMD modal set of the pulsating velocity field; extract the pulsating velocity field DMD mode corresponding to the frequency of the dominant discrete frequency noise component to be solved from the obtained pulsating velocity field DMD modal set to obtain the dominant discrete frequency pulsating velocity field DMD mode; perform Helmholtz decomposition on the dominant discrete frequency pulsating velocity field DMD mode to obtain the dominant discrete frequency pulsating velocity field acoustic mode; based on the dominant discrete frequency pulsating velocity field acoustic mode, calculate and obtain the time-evolved dominant discrete frequency acoustic modal velocity field; based on the time-evolved dominant discrete frequency acoustic modal velocity field, calculate and obtain the time-evolved acoustic modal pressure field and acoustic modal density field; based on the time-evolved dominant discrete frequency acoustic modal velocity field, the time-evolved acoustic modal pressure field and acoustic modal density field, obtain the dominant discrete frequency noise sound source data preprocessed by flow-acoustic separation;
[0017] Step S6: interpolating the dominant discrete frequency noise source data pre-processed by flow-acoustic separation on the CFD grid to the CAA grid to obtain the dominant discrete frequency aerodynamic noise source data pre-processed by flow-acoustic separation required for sound propagation calculation;
[0018] Step S7: Call the CAA sound field solver to perform sound propagation simulation, read the CAA grid, input the background flow field data, input the dominant discrete frequency aerodynamic noise source data that has been preprocessed by flow-acoustic separation, add the dominant discrete frequency aerodynamic noise source data that has been preprocessed by flow-acoustic separation in the form of a sound source boundary to the calculation of solving the sound propagation equation, and complete the numerical simulation of discrete frequency aerodynamic noise.
[0019] Among them, the unsteady motion of the fluid will generate aerodynamic noise. Due to the different configurations and speeds of the flow problem, the aerodynamic noise generated also has different spectral characteristics. Unsteady flows with significant periodicity, such as airfoil flow, cavity flow, impact jet and incompletely expanded free supersonic jet, will generate and radiate noise components with discrete frequencies, namely trailing edge single tone, self-excited oscillation noise, impact noise and whistling. These noises and their harmonics have a single frequency, and the sound intensity is much higher than other noise components, playing an absolutely dominant role in the entire noise spectrum. For such special aerodynamic noise problems, people usually focus on the generation and propagation characteristics of such discrete frequency noise, while ignoring other noise components. Therefore, this discrete frequency noise component can be extracted separately for targeted hybrid numerical simulation. Helmholtz-Dynamic Mode Decomposition (HD-DMD) is a flow-acoustic mode decomposition method suitable for discrete frequency noise flows. The flow-acoustic modes decomposed by the HD-DMD are orthogonal in time, and can extract the co-ordered structures with specific frequencies and growth rates in complex flows and the sound waves they generate. It can effectively separate the acoustic modes and non-acoustic dynamic modes in flows dominated by single-frequency noise. Therefore, when the flow and sound waves of a certain dominant frequency are the main research objects, HD-DMD has great advantages.
[0020] The present invention first obtains unsteady time-series CFD sound source data by CFD flow field solution; introduces HD-DMD into flow-acoustic separation preprocessing of unsteady CFD sound source velocity field of discrete frequency aerodynamic noise according to its characteristics, realizes separation of dominant discrete frequency acoustic mode and non-acoustic dynamic mode of sound source velocity field, obtains time-evolving velocity field dominant discrete frequency acoustic mode, and then obtains time-evolving acoustic modal velocity field containing only dominant discrete frequency components; calculates dominant discrete frequency acoustic modal pressure field of time evolution according to acoustic modal pressure Poisson equation (acoustic modal pressure-velocity relationship); calculates dominant discrete frequency acoustic modal density field of time evolution according to acoustic modal pressure density-pressure relationship; takes dominant discrete frequency acoustic modal velocity field, pressure field and density field of time evolution as input, solves CAA acoustic field (acoustic propagation), and finally realizes hybrid numerical simulation of discrete frequency aerodynamic noise problem. The method proposed in the present invention can, when implementing hybrid numerical simulation of discrete frequency aerodynamic noise problems: (1) reduce the sound source data storage space required for file-based data coupling between a CFD flow field solver and a CAA sound field (sound propagation) solver; and (2) eliminate the direct addition of unsteady sound source data in the form of sound source boundaries to the accompanying propagation of non-acoustic modes in the sound propagation calculation, thereby enhancing the computational stability of the CAA sound field solution.
[0021] Preferably, the step S1 specifically includes:
[0022] Generate a CFD calculation grid and call the CFD flow field solver to perform unsteady fluid dynamics simulation. When performing unsteady fluid dynamics simulation, set the sound pressure monitoring point and output the sound pressure data. Then, based on the sound pressure data at the monitoring point, calculate the frequency and period of the discrete frequency noise component, and then use the time interval Δ t The near-field time-series flow field data file with a time scale of several cycles is output as the unsteady CFD time-series flow field data. The unsteady CFD time-series flow field data includes velocity field, pressure field and density field data.
[0023] Preferably, dynamic mode decomposition is performed on the time-series pulsating velocity field to obtain a DMD mode set of the pulsating velocity field, which specifically includes:
[0024] Construct a data snapshot matrix based on the time-series pulsating velocity field;
[0025] Construct a similarity transformation matrix based on the data snapshot matrix;
[0026] Perform eigendecomposition on the similarity transformation matrix to obtain the eigenvectors and eigenvalues of the similarity transformation matrix;
[0027] Perform singular value decomposition on the data snapshot matrix to obtain a left singular vector matrix and a right singular vector matrix;
[0028] The DMD mode set of the pulsating velocity field is constructed based on the eigenvectors of the similarity transformation matrix and the left singular vector matrix of the data snapshot matrix;
[0029] Based on the eigenvalues of the similarity transformation matrix and the time interval of the flow field data file, the frequency of each DMD mode in the DMD mode set of the pulsating velocity field is calculated.
[0030] Preferably, the pulsating velocity field is V' , V'= ( u' , v' , w' ), u' is the streamwise pulsating velocity field, v' is the normal pulsating velocity field, w' is the spanwise pulsating velocity field;
[0031] Construct a data snapshot matrix based on the time-series pulsating velocity field, including:
[0032] ;
[0033] in, yes t k The pulsating velocity field at each moment, k is the time label value in the time series, N is the number of time-series pulsating velocity fields, Mis the product of the number of grid points and the number of velocity components, is the first data snapshot matrix, is the second data snapshot matrix, is the set of real numbers;
[0034] Construct the similarity transformation matrix S based on the data snapshot matrix, specifically:
[0035] ;
[0036] in, yes The transpose of , the superscript + represents the matrix transpose;
[0037] Perform eigendecomposition on the similarity transformation matrix to obtain the eigenvectors and eigenvalues of the similarity transformation matrix, specifically:
[0038] ;
[0039] in, m i and y i are the eigenvalues and eigenvectors of the similarity transformation matrix S, respectively. i For the i the labels of the eigenvalues and eigenvectors;
[0040] Perform singular value decomposition on the data snapshot matrix to obtain the left singular vector matrix and the right singular vector matrix matrix, specifically:
[0041] ;
[0042] in, U and W are the left singular vector matrix and the right singular vector matrix, respectively. is a diagonal matrix of singular values;
[0043] Based on the eigenvectors of the similarity transformation matrix and the left singular vector matrix of the data snapshot matrix, the DMD mode set of the pulsating velocity field is constructed, including:
[0044] , i =1,…, N ;
[0045] in, , is the unnormalized DMD mode, b i For the i The initial amplitude of each DMD mode;
[0046] ;
[0047] ;
[0048] in, b is the initial amplitude of the DMD mode set of the pulsating velocity field, Ф is the unnormalized DMD mode set of the pulsating velocity field;
[0049] Based on the eigenvalues of the similarity transformation matrix and the time interval of the flow field data file, the frequency of each DMD mode in the DMD mode set of the pulsating velocity field is calculated. oh i , specifically:
[0050] , i =1,…, N ;
[0051] Among them, Δ t For the time interval.
[0052] Preferably, the dominant discrete frequency pulsating velocity field DMD mode is subjected to Helmholtz decomposition to obtain the dominant discrete frequency pulsating velocity field acoustic mode, specifically including:
[0053] The dominant discrete frequency pulsating velocity field DMD mode is characterized by scalar potential gradient and vector potential curl, and the characterized dominant discrete frequency pulsating velocity field DMD mode is obtained;
[0054] The divergence of the dominant discrete frequency pulsating velocity field DMD mode after characterization is obtained to obtain the acoustic mode scalar potential Poisson equation, and the acoustic mode scalar potential is obtained by solving the acoustic mode scalar potential Poisson equation;
[0055] The gradient of the acoustic mode scalar potential is taken to obtain the dominant discrete frequency pulsating velocity field acoustic mode.
[0056] Preferably, the dominant discrete frequency pulsating velocity field DMD mode is characterized by a scalar potential gradient and a vector potential curl, and the characterized dominant discrete frequency pulsating velocity field DMD mode is obtained, specifically:
[0057] ;
[0058] Among them, DMD is the dominant discrete frequency pulsating velocity field DMD mode, is the acoustic mode scalar potential, is the gradient operator, is the vortex mode vector potential;
[0059] The divergence of the dominant discrete frequency pulsating velocity field DMD mode after characterization is obtained to obtain the acoustic modal scalar potential Poisson equation, and the acoustic modal scalar potential Poisson equation is solved to obtain the acoustic modal scalar potential, which is specifically:
[0060] The acoustic modal scalar potential Poisson equation is:
[0061] ;
[0062] in, is the Laplace operator;
[0063] The gradient of the acoustic mode scalar potential is taken to obtain the dominant discrete frequency pulsating velocity field acoustic mode, which is:
[0064] ;
[0065] Among them, (HD-DMD) a is the dominant discrete frequency pulsating velocity field acoustic mode, subscript a Represents the acoustic mode.
[0066] Preferably, based on the dominant discrete frequency pulsating velocity field acoustic mode, the time-evolved dominant discrete frequency acoustic mode velocity field is calculated, specifically:
[0067] ;
[0068] in, is the dominant discrete frequency acoustic mode velocity field of time evolution, x is the spatial coordinate information, m is the eigenvalue of the dominant discrete frequency pulsating velocity field DMD mode, m k is the eigenvalue m of k Power, k is the label value of the moment in the time evolution, t k+1 For the time evolution k +1 moment.
[0069] Preferably, based on the time-evolved dominant discrete frequency acoustic mode velocity field, the time-evolved acoustic mode pressure field and acoustic mode density field are calculated, specifically including:
[0070] Construct the Poisson equation for acoustic modal pressure:
[0071] ;
[0072] in, is the time-evolving acoustic modal pressure field, is the time-averaged density field, is the body derivative operator;
[0073] Solve the Poisson's equation for acoustic modal pressure to obtain the time-evolving acoustic modal pressure field;
[0074] Based on the time-evolving acoustic modal pressure field and the relationship between acoustic modal pressure and density, the time-evolving acoustic modal density field is calculated:
[0075] ;
[0076] in, is the time-evolving acoustic modal density field, is the average speed of sound.
[0077] The present invention is aimed at hybrid simulation of discrete frequency aerodynamic noise problems. It adopts a flow-acoustic separation method based on Helmholtz-dynamic mode decomposition (HD-DMD) to perform flow-acoustic separation preprocessing on the unsteady sound source data obtained by CFD calculation before sound propagation calculation. The technical effects are as follows:
[0078] The HD-DMD-based flow-acoustic separation method can reduce the dimension and reconstruct the time-evolving sound source information based on the limited unsteady CFD flow field data files, avoid storing the flow field data files at all times during the entire CFD calculation time, and reduce the storage space required for file-based data coupling between the CFD flow field solver and the CAA sound field (sound propagation) solver.
[0079] The HD-DMD-based flow-acoustic separation method can separate the acoustic mode and non-acoustic mode of the sound source information, and then eliminate the non-acoustic mode and extract and retain only the dominant discrete frequency acoustic mode, thereby eliminating the direct addition of the unsteady sound source data in the form of sound source boundary to the sound propagation calculation accompanied by the non-acoustic mode, and enhancing the accuracy and computational stability of the CAA sound field solution of the discrete frequency aerodynamic noise problem implemented by this method. BRIEF DESCRIPTION OF THE DRAWINGS
[0080] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of the present invention, and do not constitute a limitation on the embodiments of the present invention;
[0081] Figure 1 A flow chart of a discrete frequency aerodynamic noise hybrid simulation method. DETAILED DESCRIPTION
[0082] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict.
[0083] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those within the scope of this description. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.
[0084] Embodiment 1;
[0085] In order to obtain accurate sound propagation calculation results and enhance calculation stability, it is necessary to perform flow-acoustic separation preprocessing on the near-field pulsation quantity, and then assign the separated compressible acoustic modal components to the sound source boundary of the CAA grid. The present invention develops a hybrid simulation method for aerodynamic noise based on the sound propagation equation and Helmholtz-Dynamic Mode Decomposition (HD-DMD) flow-acoustic separation preprocessing to solve the two problems faced when coupling the CFD flow field solver and the CAA sound field (acoustic propagation) solver to simulate discrete frequency aerodynamic noise: (1) the problem of excessive storage space required for file-based data coupling of the two solvers; (2) when the near-field flow field pulsation quantity is directly assigned to the sound source boundary for sound propagation calculation, the accompanying other non-acoustic modes affect the calculation accuracy and stability.
[0086] Please refer to Figure 1 , Figure 1 The present invention is a flow chart of a hybrid simulation method for discrete frequency aerodynamic noise. Aiming at the hybrid numerical simulation of discrete frequency aerodynamic noise problems, the present invention proposes a flow-acoustic separation method based on Helmholtz-dynamic mode decomposition (HD-DMD) to reconstruct and preprocess near-field sound source data on the basis of a file-based data coupling method. Then, the dominant discrete frequency noise source information preprocessed by flow-acoustic is added to the sound propagation calculation in the form of sound source boundary, thereby realizing the hybrid simulation of discrete frequency aerodynamic noise problems by coupling CFD flow field solver and CAA sound field (sound propagation) solver.
[0087] The specific implementation is divided into the following 7 steps:
[0088] Step 1 (S1): Computational fluid dynamics (CFD) simulation. Based on the CFD computational grid for the discrete frequency aerodynamic noise problem to be solved generated by the grid generation software, the CFD flow field solver is called, and the URANS, DES, LES or DNS methods are selected to solve the Navier-Stokes equations and perform unsteady fluid dynamics simulation. During the CFD simulation, the sound pressure monitoring point is set, the sound pressure data is output, and then the frequency and period of the discrete frequency noise component are calculated based on the sound pressure data of the monitoring point, and then the time interval Δ t Output the near-field flow field (including velocity field, pressure field and pulsation field) data file with a time scale of several cycles as the unsteady CFD sound source data.
[0089] Step 2 (S2): Based on the unsteady CFD sound source data, the time-averaged flow field data and the time-series pulsating flow field data on the CFD grid are calculated and obtained.
[0090] Step 3 (S3): Use mesh generation software to generate the CAA mesh required by the CAA acoustic field solver for sound propagation calculations.
[0091] Step 4 (S4): Background flow field interpolation for acoustic propagation calculation. The time-averaged flow field data on the CFD grid is interpolated onto the CAA grid using interpolation methods such as Lagrange polynomial interpolation or radial basis function interpolation to obtain the background flow field data required for acoustic propagation calculation.
[0092] Step 5 (S5): Extraction of dominant discrete frequency aerodynamic noise source data and preprocessing of flow-acoustic separation based on HD-DMD flow-acoustic separation method. The HD-DMD flow-acoustic separation method is used to process the time-series pulsating velocity field data, realize the separation of the dominant discrete frequency acoustic mode and non-acoustic dynamic mode of the sound source pulsating velocity field, and obtain the dominant discrete frequency acoustic mode of the pulsating velocity field; based on the dominant discrete frequency acoustic mode of the pulsating velocity field, the time-evolving acoustic modal velocity field containing only the dominant discrete frequency component is calculated; based on this, according to the acoustic modal velocity-pressure relationship, the time-evolving dominant discrete frequency acoustic modal pressure field is calculated; according to the acoustic modal pressure-density relationship, the time-evolving dominant discrete frequency acoustic modal density field is calculated.
[0093] The specific implementation of step S5 is as follows:
[0094] S5-1: Perform dynamic mode decomposition (DMD) on the time-series pulsating velocity field to obtain the DMD mode of the pulsating velocity field.
[0095] (1-1) Reorganization of unsteady pulsating velocity field V'= ( u' , v' , w' )( V' is the vector pulsating velocity field, u' is the streamwise pulsating velocity field, v' is the normal pulsating velocity field, w' is the spanwise pulsating velocity field) time series is the data snapshot matrix:
[0096] ; (1)
[0097] in, yes t k The pulsating velocity field at each moment, N is the number of time-series pulsating velocity fields, M is the product of the number of grid points and the number of velocity components, is the set of real numbers.
[0098] (1-2) Construct a similarity transformation matrix:
[0099] ; (2)
[0100] in, yes The superscript + represents the transpose of the matrix.
[0101] Perform eigendecomposition on the matrix S:
[0102] ; (3)
[0103] in, m i and y i are the eigenvalues and eigenvectors of the matrix S respectively.
[0104] (1-3) For the matrix Perform singular value decomposition:
[0105] ; (4)
[0106] in, U and W are the left singular vector matrix and the right singular vector matrix, respectively. is a diagonal matrix of singular values.
[0107] (1-4) Construct the dynamic mode set of pulsating velocity field (DMD mode set):
[0108] , i =1,…, N ; (5)
[0109] in,
[0110] ; (6)
[0111] is the unnormalized DMD mode, b i It is i The initial amplitude of each DMD mode is calculated by the following formula:
[0112] , ; (7)
[0113] in, b is the initial amplitude of the DMD mode set of the pulsating velocity field, Ф is the unnormalized DMD mode set of the pulsating velocity field;
[0114] (1-5) Calculate the frequency of each DMD mode in the DMD mode set of the pulsating velocity field oh i:
[0115] . (8)
[0116] S5-2: Extract the DMD mode of the pulsating velocity field whose frequency corresponds to the frequency of the dominant discrete frequency noise component to be solved, and obtain the DMD mode of the dominant discrete frequency pulsating velocity field.
[0117] S5-3: Perform Helmholtz decomposition (HD) on the DMD mode of the dominant discrete frequency pulsating velocity field to obtain the acoustic mode of the dominant discrete frequency pulsating velocity field.
[0118] First, the dominant discrete frequency pulsating velocity field DMD mode is characterized by scalar potential gradient and vector potential curl:
[0119] ; (9)
[0120] Among them, DMD is the dominant discrete frequency pulsating velocity field DMD mode. The former on the right side of equation (9) represents the acoustic mode velocity, and the latter represents the vortex mode velocity, that is, is the acoustic mode scalar potential, is the vortex mode vector potential.
[0121] Secondly, the dominant discrete frequency pulsating velocity field acoustic mode is solved. Taking the divergence of both sides of the dominant discrete frequency pulsating velocity field DMD mode, i.e., equation (9), the acoustic mode scalar potential Poisson equation is obtained:
[0122] ; (10)
[0123] Numerically solve the acoustic modal scalar potential Poisson equation to obtain the acoustic modal scalar potential .
[0124] Acoustic mode scalar potential By taking the gradient, we can obtain the dominant discrete frequency pulsating velocity field acoustic mode:
[0125] . (11)
[0126] S5-4: Based on the acoustic mode of the dominant discrete frequency pulsating velocity field, the time-evolving dominant discrete frequency acoustic mode velocity field is calculated to achieve flow-acoustic separation of the sound source velocity field.
[0127] The specific calculation method is:
[0128] ; (12)
[0129] Here is the dominant discrete frequency acoustic mode velocity field of time evolution, is the spatial coordinate information, m is the eigenvalue of the dominant discrete frequency pulsating velocity field DMD mode, m k is the eigenvalue m No. k Power, k is the time label value in the time evolution, indicating the first k +1 moment.
[0130] S5-5: Based on the time-evolving dominant discrete frequency acoustic mode velocity field, the time-evolving acoustic mode pressure field and acoustic mode density field are calculated to achieve flow-acoustic separation of the sound source pressure field and density field.
[0131] First, according to the acoustic modal pressure Poisson equation:
[0132] ; (13)
[0133] Compute the time-evolving acoustic modal pressure field ,in, is the time-averaged density field, is the gradient operator, is the Laplace operator, is the body derivative operator.
[0134] Secondly, according to the acoustic modal pressure-density relationship, the time-evolving acoustic modal density field is calculated:
[0135] ; (14)
[0136] in, is the average speed of sound.
[0137] Equations (12) to (14) are the dominant discrete frequency aerodynamic noise source data after flow-acoustic separation preprocessing.
[0138] Step 6 (S6): Interpolation of sound source data pre-processed by flow-acoustic separation. Using interpolation methods such as Lagrange polynomial interpolation or radial basis function interpolation, the dominant discrete frequency noise sound source data pre-processed by flow-acoustic separation on the CFD grid is interpolated to the CAA grid to obtain the dominant discrete frequency aerodynamic noise sound source data pre-processed by flow-acoustic separation required for sound propagation calculation.
[0139] Step 7 (S7): Sound propagation simulation based on sound source preprocessed by flow-acoustic separation. The CAA sound field (sound propagation) solver is called to perform sound propagation simulation. First, the CAA calculation grid is read and the background flow field data is input. As the calculation gradually moves forward, the dominant discrete frequency aerodynamic noise sound source data preprocessed by flow-acoustic separation is input and added to the calculation of solving the sound propagation equation in the form of a sound source boundary to complete the numerical simulation of the discrete frequency aerodynamic noise problem.
[0140] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.
[0141] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
Claims
1. A discrete frequency aerodynamic noise hybrid simulation method, characterized in that: The method comprises: Step S1: performing unsteady fluid dynamics simulation based on the CFD grid to obtain unsteady CFD time-series flow field data including velocity field, pressure field and density field; Step S2: Based on the unsteady CFD time-series flow field data, calculate and obtain the time-averaged flow field data and the time-series pulsating flow field data on the CFD grid; Step S3: using mesh generation software to generate a CAA mesh required by a CAA sound field solver for sound propagation calculation; Step S4: interpolating the time-averaged flow field data on the CFD grid to the CAA grid to obtain the background flow field data required for the sound propagation calculation; Step S5: Perform dynamic modal decomposition on the time-series pulsating velocity field data to obtain the DMD modal set of the pulsating velocity field; extract the pulsating velocity field DMD mode corresponding to the frequency of the dominant discrete frequency noise component to be solved from the obtained pulsating velocity field DMD modal set to obtain the dominant discrete frequency pulsating velocity field DMD mode; perform Helmholtz decomposition on the dominant discrete frequency pulsating velocity field DMD mode to obtain the dominant discrete frequency pulsating velocity field acoustic mode; based on the dominant discrete frequency pulsating velocity field acoustic mode, calculate and obtain the time-evolved dominant discrete frequency acoustic modal velocity field; based on the time-evolved dominant discrete frequency acoustic modal velocity field, calculate and obtain the time-evolved acoustic modal pressure field and acoustic modal density field; based on the time-evolved dominant discrete frequency acoustic modal velocity field, the time-evolved acoustic modal pressure field and acoustic modal density field, obtain the dominant discrete frequency noise sound source data preprocessed by flow-acoustic separation; Step S6: interpolating the dominant discrete frequency noise source data pre-processed by flow-acoustic separation on the CFD grid to the CAA grid to obtain the dominant discrete frequency aerodynamic noise source data pre-processed by flow-acoustic separation required for sound propagation calculation; Step S7: Call the CAA sound field solver to perform sound propagation simulation, read the CAA grid, input the background flow field data, input the dominant discrete frequency aerodynamic noise source data that has been preprocessed by flow-acoustic separation, add the dominant discrete frequency aerodynamic noise source data that has been preprocessed by flow-acoustic separation in the form of a sound source boundary to the calculation of solving the sound propagation equation, and complete the numerical simulation of discrete frequency aerodynamic noise.
2. A discrete frequency aerodynamic noise hybrid simulation method according to claim 1, characterized in that: The step S1 specifically includes: Generate a CFD calculation grid and call the CFD flow field solver to perform unsteady fluid dynamics simulation. When performing unsteady fluid dynamics simulation, set the sound pressure monitoring point and output the sound pressure data. Then, based on the sound pressure data at the monitoring point, calculate the frequency and period of the discrete frequency noise component, and then use the time interval Δ t The near-field time-series flow field data file with a time scale of several cycles is output as the unsteady CFD time-series flow field data. The unsteady CFD time-series flow field data includes velocity field, pressure field and density field data.
3. A discrete frequency aerodynamic noise hybrid simulation method according to claim 1, characterized in that: Perform dynamic mode decomposition on the time-series pulsating velocity field to obtain the DMD mode set of the pulsating velocity field, including: Construct a data snapshot matrix based on the time-series pulsating velocity field; Construct a similarity transformation matrix based on the data snapshot matrix; Perform eigendecomposition on the similarity transformation matrix to obtain the eigenvectors and eigenvalues of the similarity transformation matrix; Perform singular value decomposition on the data snapshot matrix to obtain a left singular vector matrix and a right singular vector matrix; The DMD mode set of the pulsating velocity field is constructed based on the eigenvectors of the similarity transformation matrix and the left singular vector matrix of the data snapshot matrix; Based on the eigenvalues of the similarity transformation matrix and the time interval of the flow field data file, the frequency of each DMD mode in the DMD mode set of the pulsating velocity field is calculated.
4. A discrete frequency aerodynamic noise hybrid simulation method according to claim 3, characterized in that: The pulsating velocity field is V' , V'= ( u' , v' , w' ), u' is the streamwise pulsating velocity field, v' is the normal pulsating velocity field, w' is the spanwise pulsating velocity field; Construct a data snapshot matrix based on the time-series pulsating velocity field, including: ; in, yes t k The pulsating velocity field at each moment, k is the time label value in the time series, N is the number of time-series pulsating velocity fields, M is the product of the number of grid points and the number of velocity components, is the first data snapshot matrix, is the second data snapshot matrix, is the set of real numbers; Construct the similarity transformation matrix S based on the data snapshot matrix, specifically: ; in, yes The transpose of , the superscript + represents the matrix transpose; Perform eigendecomposition on the similarity transformation matrix to obtain the eigenvectors and eigenvalues of the similarity transformation matrix, specifically: ; in, μ i and y i are the eigenvalues and eigenvectors of the similarity transformation matrix S, respectively. i For the i the labels of the eigenvalues and eigenvectors; Perform singular value decomposition on the data snapshot matrix to obtain the left singular vector matrix and the right singular vector matrix matrix, specifically: ; in, U and W are the left singular vector matrix and the right singular vector matrix, respectively. is a diagonal matrix of singular values; Based on the eigenvectors of the similarity transformation matrix and the left singular vector matrix of the data snapshot matrix, the DMD mode set of the pulsating velocity field is constructed, including: , i =1,…, N ; in, , is the unnormalized DMD mode, b i For the i The initial amplitude of each DMD mode; ; ; in, b is the initial amplitude of the DMD mode set of the pulsating velocity field, Ф is the unnormalized DMD mode set of the pulsating velocity field; Based on the eigenvalues of the similarity transformation matrix and the time interval of the flow field data file, the frequency of each DMD mode in the DMD mode set of the pulsating velocity field is calculated. ω i , specifically: , i =1,…, N ; Among them, Δ t is the time interval.
5. The discrete frequency aerodynamic noise hybrid simulation method according to claim 1, characterized in that: The dominant discrete frequency pulsating velocity field DMD mode is decomposed by Helmholtz to obtain the dominant discrete frequency pulsating velocity field acoustic mode, including: The dominant discrete frequency pulsating velocity field DMD mode is characterized by scalar potential gradient and vector potential curl, and the characterized dominant discrete frequency pulsating velocity field DMD mode is obtained; The divergence of the dominant discrete frequency pulsating velocity field DMD mode after characterization is obtained to obtain the acoustic mode scalar potential Poisson equation, and the acoustic mode scalar potential is obtained by solving the acoustic mode scalar potential Poisson equation; The gradient of the acoustic mode scalar potential is taken to obtain the dominant discrete frequency pulsating velocity field acoustic mode.
6. A discrete frequency aerodynamic noise hybrid simulation method according to claim 5, characterized in that: The dominant discrete frequency pulsating velocity field DMD mode is characterized by scalar potential gradient and vector potential curl, and the characterized dominant discrete frequency pulsating velocity field DMD mode is obtained, specifically: ; Among them, DMD is the dominant discrete frequency pulsating velocity field DMD mode, is the acoustic mode scalar potential, is the gradient operator, is the vortex mode vector potential; The divergence of the dominant discrete frequency pulsating velocity field DMD mode after characterization is obtained to obtain the acoustic modal scalar potential Poisson equation, and the acoustic modal scalar potential Poisson equation is solved to obtain the acoustic modal scalar potential, which is specifically: The acoustic modal scalar potential Poisson equation is: ; in, is the Laplace operator; The gradient of the acoustic mode scalar potential is taken to obtain the dominant discrete frequency pulsating velocity field acoustic mode, which is: ; Among them, (HD-DMD) a is the dominant discrete frequency pulsating velocity field acoustic mode, subscript a Represents the acoustic mode.
7. A discrete frequency aerodynamic noise hybrid simulation method according to claim 6, characterized in that: Based on the dominant discrete frequency pulsating velocity field acoustic mode, the time-evolving dominant discrete frequency acoustic mode velocity field is calculated, specifically: ; in, is the dominant discrete frequency acoustic mode velocity field of time evolution, x is the spatial coordinate information, μ is the eigenvalue of the dominant discrete frequency pulsating velocity field DMD mode, μ k is the eigenvalue μ of k Power, k is the label value of the moment in the time evolution, t k+1 For the time evolution k +1 moment.
8. A discrete frequency aerodynamic noise hybrid simulation method according to claim 7, characterized in that: Based on the time-evolving dominant discrete frequency acoustic mode velocity field, the time-evolving acoustic mode pressure field and acoustic mode density field are calculated, including: Construct the Poisson equation for acoustic modal pressure: ; in, is the time-evolving acoustic modal pressure field, is the time-averaged density field, is the body derivative operator; Solve the Poisson's equation for acoustic modal pressure to obtain the time-evolving acoustic modal pressure field; Based on the time-evolving acoustic modal pressure field and the relationship between acoustic modal pressure and density, the time-evolving acoustic modal density field is calculated: ; in, is the time-evolving acoustic modal density field, is the average speed of sound.
