Post-processing methods, apparatus, equipment, and media for CFD simulation of hydrogen fuel flow

Abstract

The embodiment of the application provides a kind of hydrogen fuel flow simulation CFD's post-processing method, device, equipment and medium, it is related to numerical simulation technical field, wherein, the method is realized by the following steps: the hydrogen fuel flow CFD simulation based on aero-engine is executed, and unstructured grid data is generated;Unstructured grid data is converted into a plurality of simulation result files, and the grid geometry information and numerical distribution data in simulation result file are sorted, time snapshot matrix is generated, and the mode corresponding to each eigenvector of hydrogen fuel flow field is obtained by calculating time snapshot matrix accompanying matrix S;The magnification g i And modal coefficient amplitude b a Of each eigenvector of hydrogen fuel flow field is calculated by the mode corresponding to the mode superposition method, dynamic reconstruction and prediction are carried out on hydrogen fuel flow field.The scheme can reconstruct and predict the flow field state at any time by the method of modal superposition, improve the efficiency of post-processing.

Description

Technical Field

[0001] This invention relates to the field of numerical simulation technology, and in particular to a post-processing method, apparatus, equipment and medium for CFD simulation of hydrogen fuel flow. Background Technology

[0002] CFD simulation of hydrogen fuel flow is a key technology in the aerospace hydrogen energy utilization field, as its flow characteristics directly affect combustion efficiency and system safety. Traditional CFD-based time-averaged flow field analysis methods have significant shortcomings: First, they rely on manually set monitoring parameters, making them insensitive to microscale flow anomalies in hydrogen fuel, leading to difficulties in timely detection of safety hazards; second, conventional frequency domain analysis methods cannot effectively correlate flow modes with mixing characteristics. Existing commercial CFD post-processing software (such as Fluent and CFX) lacks dedicated analytical modules for hydrogen fuel flow, resulting in problems such as delayed mixing state assessment and insufficient early warning accuracy. Research shows that the modal distribution changes of specific frequency bands (such as the characteristic frequencies of hydrogen molecule diffusion) in the dynamic flow field are significantly correlated with mixing uniformity. This provides an important basis for developing intelligent post-processing methods for hydrogen fuel flow with multi-physics coupling analysis capabilities and is also a core technical approach to improve the operational efficiency and safety monitoring level of hydrogen energy systems. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a post-processing method for hydrogen fuel flow simulation CFD to solve the technical problems of low sensitivity in fault feature identification, poor spatiotemporal modal correlation, and insufficient diagnostic efficiency in existing traditional CFD post-processing methods. The method includes:

[0004] Based on the set input variables, perform CFD simulation of hydrogen fuel flow based on aero-engines to generate unstructured mesh data;

[0005] The unstructured mesh data is converted into multiple simulation result files. The mesh geometry and numerical distribution data in these files are then sorted according to physical time to generate a time snapshot matrix. Through the time snapshot matrix The adjoint matrix S is calculated, and the mode corresponding to each eigenvector of the hydrogen fuel flow field is calculated using the adjoint matrix S. ;

[0006] The mode corresponding to each feature vector of the hydrogen fuel flow field The magnification was calculated. g i and modal coefficient amplitude b a Based on the magnification g i and the modal coefficient amplitudeb a The hydrogen fuel flow field is dynamically reconstructed and predicted using the modal superposition method.

[0007] This invention also provides a post-processing device for hydrogen fuel flow simulation CFD, to solve the technical problems of low sensitivity in fault feature identification, poor spatiotemporal modal correlation, and insufficient diagnostic efficiency in existing CFD post-processing methods. The device includes:

[0008] The grid data generation module is used to perform CFD simulation of hydrogen fuel flow based on aero-engines according to the set input variables and generate unstructured grid data.

[0009] The modal calculation module is used to convert the unstructured mesh data into multiple simulation result files, sort the mesh geometric information and numerical distribution data in the simulation result files according to physical time, and generate a time snapshot matrix. Through the time snapshot matrix The adjoint matrix S is calculated, and the mode corresponding to each eigenvector of the hydrogen fuel flow field is calculated using the adjoint matrix S. ;

[0010] The flow field reconstruction and prediction module is used to reconstruct and predict the modes corresponding to each feature vector of the hydrogen fuel flow field. The magnification was calculated. g i and modal coefficient amplitude b a Based on the magnification g i and the modal coefficient amplitude b a The hydrogen fuel flow field is dynamically reconstructed and predicted using the modal superposition method.

[0011] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-mentioned post-processing method for any of the hydrogen fuel flow simulation CFD methods, thereby solving the technical problems of low sensitivity in fault feature identification, poor spatiotemporal modal correlation, and insufficient diagnostic efficiency in the traditional CFD post-processing methods of the prior art.

[0012] This invention also provides a computer-readable storage medium storing a computer program that executes any of the above-described post-processing methods for hydrogen fuel flow simulation CFD, in order to solve the technical problems of low sensitivity in fault feature identification, poor spatiotemporal modal correlation, and insufficient diagnostic efficiency in traditional CFD post-processing methods in the prior art.

[0013] Compared with the prior art, the beneficial effects that at least one technical solution adopted in the embodiments of this specification can achieve include at least:

[0014] By calculating the flow field modes, modal coefficient amplitudes, and modal amplification rates, and combining them with the initial values ​​of the flow field, the flow field state at any given time can be reconstructed and predicted using the modal superposition method. This avoids a large amount of repetitive simulation work and improves post-processing efficiency. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart of a post-processing method for hydrogen fuel flow simulation CFD provided in an embodiment of the present invention;

[0017] Figure 2 This is a flowchart of a post-processing method for implementing the above-described hydrogen fuel flow simulation CFD, provided by an embodiment of the present invention;

[0018] Figure 3 This is a structural block diagram of a computer device provided in an embodiment of the present invention;

[0019] Figure 4 This is a structural block diagram of a post-processing device for hydrogen fuel flow simulation CFD provided in an embodiment of the present invention. Detailed Implementation

[0020] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0021] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. This application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] In this embodiment of the invention, a post-processing method for hydrogen fuel flow simulation CFD is provided, such as... Figure 1 and Figure 2 As shown, the method includes:

[0023] Step S101: Perform a CFD simulation of hydrogen fuel flow based on an aero-engine according to the set input variables to generate unstructured mesh data;

[0024] Step S102: Convert the unstructured mesh data into multiple simulation result files, sort the mesh geometry information and numerical distribution data in the simulation result files according to physical time, and generate a time snapshot matrix. Through the time snapshot matrix The adjoint matrix S is calculated, and the mode corresponding to each eigenvector of the hydrogen fuel flow field is calculated using the adjoint matrix S. ;

[0025] Step S103: Through the modes corresponding to each eigenvector of the hydrogen fuel flow field The magnification was calculated. g i and modal coefficient amplitude b a Based on the magnification g i and the modal coefficient amplitude b a The hydrogen fuel flow field is dynamically reconstructed and predicted using the modal superposition method.

[0026] In practice, the unstructured mesh data is converted into multiple simulation result files through the following steps:

[0027] The unstructured mesh data is analyzed to extract the mesh topology and spatial distribution data of physical quantities. The mesh topology includes node coordinates and cell connection relationships, and the physical quantities include velocity field, pressure field, and temperature field. A physical quantity mapping relationship based on cell number is established, and the physical quantities are separated by type based on the physical quantity mapping relationship. Based on the spatial distribution data, an independent simulation result file is generated for each type of physical quantity. The simulation result file includes the mesh geometry information and the numerical distribution data of the physical quantities.

[0028] In this embodiment of the invention, the following steps are used to sort the mesh geometry information and numerical distribution data in the simulation result file according to physical time, thereby generating a time snapshot matrix. Through the time snapshot matrix The adjoint matrix S is calculated as follows:

[0029] The mesh geometry information and numerical distribution data in the simulation result file are sorted according to physical time to generate a time snapshot matrix. ,in, , v 1. v 2...... v N These are variables at different time steps, including velocity, pressure, temperature, and turbulence; expressed through the time snapshot matrix. Before obtaining N Snapshot matrix at time -1 and after N Snapshot matrix at time -1 According to the previous N Snapshot matrix at time -1 and after N Snapshot matrix at time -1 The mapping relationship is used to obtain the mapping matrix A of the whole sequence. The order of the mapping matrix A of the whole sequence is reduced, and the adjoint matrix S is calculated.

[0030] In this embodiment of the invention, the following steps are used to achieve the following: N Snapshot matrix at time -1 and after N Snapshot matrix at time -1 The mapping relationship is used to obtain the mapping matrix A of the whole sequence. The order of the mapping matrix A of the whole sequence is reduced to obtain the adjoint matrix S:

[0031] Using the singular value decomposition method to analyze the previous N Snapshot matrix at time -1 Dimensionality reduction is performed to generate a diagonal matrix Σ composed of singular values. This diagonal matrix Σ is then truncated to generate a truncated diagonal matrix Σ. The result is obtained by using a unitary matrix U containing left singular vectors, the truncated diagonal matrix Σ, and the subsequent... N Snapshot matrix at time -1 We obtain the mapping matrix A, where, J is the unitary matrix of the right singular vector, and U is the unitary matrix containing the left singular vector; through the mapping matrix A and the unitary matrix U, the reduced-order adjoint matrix S is obtained, where, U * Let be the conjugate transpose of U.

[0032] In this embodiment of the invention, the following steps are used to calculate the mode corresponding to each eigenvector of the hydrogen fuel flow field using the adjoint matrix S. :

[0033] By delving into the dynamic characteristics of the hydrogen fuel flow field through eigenvalue decomposition, the eigenvalues ​​of the adjoint matrix S are solved. λ i and eigenvector K i After passing N Snapshot matrix at time -1 The feature vector K i By combining the truncated diagonal matrix Σ, the modes corresponding to each eigenvector of the hydrogen fuel flow field are calculated. ,in, J is the unitary matrix of the right singular vector.

[0034] In this embodiment of the invention, the modes corresponding to each feature vector of the hydrogen fuel flow field are realized through the following steps. The magnification was calculated. g i and modal coefficient amplitude b a Based on the magnification g i and the modal coefficient amplitude b a The hydrogen fuel flow field is dynamically reconstructed and predicted using the modal superposition method:

[0035] The mode corresponding to each feature vector of the hydrogen fuel flow field and the variables of the first time step v 1. Calculate the modal coefficient amplitude. b a ,in, ; through the first i Characteristic frequencies of the first-order modes w i and the adjoint matrix S of the first i eigenvalues ​​of order λ i Calculate the first i Magnification of the order g i ,in, , t The change in time step; based on the first i Characteristic frequencies of the first-order modes w i The first i Magnification of the order g i and the modal coefficient amplitude b a The hydrogen fuel flow field is dynamically reconstructed and predicted using the modal superposition method.

[0036] In this embodiment of the invention, the following steps are used to achieve the goal based on the first...i Characteristic frequencies of the first-order modes w i The first i Magnification of the order g i and the modal coefficient amplitude b a The hydrogen fuel flow field is dynamically reconstructed and predicted using the modal superposition method:

[0037] ,in, It is the reconstructed time snapshot matrix. e It is a natural constant. r The modal order used for reconstruction, For the first i The dynamic modes of the first order, b i For the first i The modal coefficient amplitude of the first order, t For time step, w i For the first i The characteristic frequencies of the order mode.

[0038] Specifically, this can be achieved through the following steps:

[0039] Step (1): Construct a monitoring environment for variables such as speed, pressure, and temperature.

[0040] In CFD simulation, the full-variable output of velocity, pressure, temperature and turbulence parameters of all grid points is configured. An adaptive time step strategy is adopted to ensure that the sampling frequency meets the modal frequency sampling requirements. The transient flow field data is stored in blocks according to a specified format through parallel I / O technology. At the same time, a data quality control process including variable range checking and physical rationality verification is established to provide a complete spatiotemporal flow field data foundation for subsequent modal analysis.

[0041] Specifically, in the CFD simulation of hydrogen fuel flow in aero-engines, the solver is first configured with full-variable outputs of velocity (axial / radial / tangential components), pressure (static / total pressure), temperature, and turbulence parameters (such as turbulent kinetic energy and dissipation rate) for all flow field grid points. Key flow regions such as the aero-engine blade surface, tip clearance, and wake region are monitored. An adaptive time step strategy based on impeller speed is adopted to ensure that the sampling frequency meets the requirement of being more than twice the frequency of the highest mode of interest. Transient flow field data is stored in blocks according to time steps using parallel post-processing I / O technology, and a control process including pressure-temperature correlation checks and vorticity rationality verification is established. Finally, a complete spatiotemporal flow field database is formed, providing a data foundation for subsequent impeller mechanical fault modal analysis.

[0042] Step (2) converts the variable storage format in the output file (the storage format of the spatiotemporal flow field database), and the specific implementation is as follows:

[0043] For unstructured mesh data files output by CFD software, a dedicated format conversion module was developed. This module first parses the mesh topology (including node coordinates, element connectivity, and other geometric information) and the distribution data of each mesh's geometry and numerical distribution data (velocity, pressure, temperature, etc.) from the original result file. Then, it establishes a mapping table between mesh numbers and physical variables, employing sparse matrix storage technology to efficiently organize large-scale mesh data. Finally, different physical variables are output as independent format files according to the field separation principle. Each file contains complete mesh geometric information and corresponding single physical field data, while retaining metadata such as time step numbers. This processing method ensures rapid location and retrieval of specific variable data during subsequent analysis and reduces the memory overhead of single-file processing through a data divide-and-conquer strategy.

[0044] Specifically, this step developed a professional format conversion and decomposed variable storage scheme for the unstructured mesh data output from CFD simulations of hydrogen fuel flow in aero-engines. First, the original data file is completely parsed to accurately extract the mesh topology (including node coordinates and cell connectivity) and the spatial distribution data of physical quantities such as velocity field (three components), pressure field, and temperature field. Then, a physical quantity mapping relationship based on cell numbering is established, and a sparse matrix format is used to efficiently organize the large-scale mesh data. Finally, each physical quantity is separated by type and output as an independent text file (simulation result file). Each file contains complete mesh geometry information and the numerical distribution of the corresponding physical quantities, while retaining key metadata such as time step number and rotational speed. This divide-and-conquer storage strategy ensures rapid location and retrieval of target variables during subsequent dynamic modal analysis and significantly reduces memory overhead per processing run through data separation, making it particularly suitable for processing large-scale transient flow field data generated by high-precision simulations of turbomachinery.

[0045] Step (3): Form a time snapshot matrix from the simulation results and perform linear mapping. The specific method is as follows:

[0046] The simulation result files (independent text format files) of the turbomachinery, after storage format conversion in the above steps, are sorted according to physical time to form a time snapshot matrix as shown in the following formula, where v Represents variables, subscripts N Indicates the moment of a snapshot.

[0047] The extracted and processed variables are arranged into a time snapshot matrix according to the output time step, and the matrix form is as follows:

[0048] ;

[0049] In the formula v 1. v 2...... v N These are variables at different time steps, such as the pressure field distribution, velocity distribution, or vorticity distribution of turbomachinery. After forming a time snapshot matrix, it is split and the previous values ​​are taken from each time step. N -1 moment and after N The snapshot matrix at time -1 is shown in the following formula.

[0050] ;

[0051] ;

[0052] In aero-engines, we assume that data from a certain snapshot can be linearly mapped from a previous snapshot. In this case, we can... See as By performing a linear mapping, it can be described by the following formula:

[0053] .

[0054] A is a mapping matrix of the entire sequence, with a rank of [value missing]. q The square formation, q Let be the number of meshes in the turbomachinery model. Since the number of meshes in a turbomachinery model is very large, directly calculating the mapping matrix would consume a significant amount of computational resources; therefore, the matrix needs to be reduced in order.

[0055] Step (4): Reduce the order of the mapping matrix. The specific implementation method is as follows:

[0056] For each column vector in the time snapshot sequence, when the physical time interval between two consecutive snapshots is sufficiently small and there are enough time snapshots in the sequence, the data of any time snapshot can be regarded as a linear combination of the remaining time snapshots, as shown in the following formula. Where a i The coefficients representing snapshot data from different times. r It is the residual vector.

[0057] ;

[0058] The current mapping matrix is ​​A, which is a rank of q The square formation, q The number of grid cells is the mapping matrix. For large grid models, the data volume of the mapping matrix will be enormous. To address this problem, we combine the above equation with the mapping formula, transforming the mapping matrix A, which should be solved, into solving the adjoint matrix S, as shown in the following equation:

[0059] ;

[0060] In the formula, the S matrix is ​​a (N-1)×(N-1) square matrix. For aero-engine models, the number of grids is often much larger than the number of output time snapshots. Therefore, this formula can reduce the computational resource requirements. The expression for the S matrix is ​​as follows:

[0061] ;

[0062] At this point, matrix A is transformed into a rank of... N The adjoint matrix S of -1, due to N This is the number of time snapshots, which is usually much smaller than the number of grids. This step can reduce the amount of computation.

[0063] Step (5), for The specific steps for performing singular value decomposition are as follows:

[0064] The singular value decomposition (SVD) method is used to reduce the dimensionality of the preprocessed flow field snapshot matrix. First, the spatiotemporal matrix is ​​processed. Decomposition yields the following formula, where Σ is a diagonal matrix composed of singular values, with its non-zero elements arranged in descending order; U and V are unitary matrices containing the left and right singular vectors, satisfying UU*=J*J=I, where I is the identity matrix. In the formula, U and J* are both unitary matrices, with rank q for U and rank q for J*. Σ is a diagonal matrix of rank P. Note that the size of P is originally equal to the original value of Σ. They have the same rank.

[0065] ;

[0066] Step (6) truncate the Σ matrix. The specific steps are as follows:

[0067] At this point, matrices U, J, and Σ still have the problem of storing a large amount of data. Therefore, the matrices are truncated, and only the first three parts of each matrix are taken. F For large mesh models, the order element typically has: F << q It is easy to see that the above truncation process effectively compresses the data.

[0068] Specifically, due to the number of grids in an aircraft engine qThe Σ, U, and J* matrices are relatively large, and the q×q matrix U still needs to be stored during the calculation, which inevitably leads to the singular value decomposition process still consuming a large amount of memory. Since the Σ, U, and J* matrices are obtained from singular value decomposition, it can be considered that the flow characteristics of the flow field in the time series can be characterized by Σ, U, and J*. The diagonal elements of the Σ matrix are arranged in descending order, and the larger the value of the diagonal element, the higher the contribution of the corresponding vector in the matrix U and J* to the flow characteristics. Therefore, the Σ matrix is ​​reduced in order, that is, only the first few elements of the Σ matrix are selected. F The order elements, along with their corresponding U and J* data, are used for approximate substitution. Due to the nonlinear characteristics of the flow field in aero-engine turbomachinery, the order is discontinuous. F It should be ensured that the selected elements account for more than 99% of the total elements of matrix Σ. The matrix Σ before and after truncation is shown below. Simultaneously, matrices U and J* are truncated to the first F order data, resulting in... q × F as well as F × N A matrix of -1. For a typical turbomachinery model, the number of meshes is... q Typically, the numbers are in the hundreds of millions, but after processing with this method, F The values ​​are generally controlled in the hundreds. Although this truncation process introduces some numerical error, based on the flow characteristics of aero-engine turbomachinery, the preceding... F The first mode has been able to fully capture key aerodynamic features, including blade passage frequency and disk vibration mode.

[0069] .

[0070] Step (7): Calculate the mapping matrix A and the adjoint matrix S. The specific steps are as follows:

[0071] After obtaining the reduced-order singular value decomposition results (reduced U, J*, and Σ) of the time snapshot sequence, the expression for the mapping matrix A can be obtained by simultaneously solving the mapping formula, as shown in the following formula:

[0072] ;

[0073] Multiplying this expression by U* on the left and U on the right, and simultaneously applying the singular value decomposition formula, we obtain the expression for the adjoint matrix S, as follows:

[0074] .

[0075] Step (8): Calculate the modal changes of the variable of interest over time. The specific steps are as follows:

[0076] Obtain the reduced adjoint matrix S (the adjoint matrix S is constructed based on the reduced data, and its dimension is determined by the number of basis vectors after the reduction). rAfter the decision is made, the dynamic characteristics of the hydrogen fuel flow field are analyzed through eigenvalue decomposition. After obtaining the elements of the adjoint matrix S, its eigenvalues ​​are further calculated. λ i With feature vectors K i The formula is as follows:

[0077] ;

[0078] Since the adjoint matrix S is a reduced order of the mapping matrix A, and K i If is an eigenvector of the adjoint matrix, then the dynamic mode corresponding to each eigenvector of the hydrogen fuel flow field can be calculated by the following formula:

[0079] ;

[0080] Step (9): Calculate the modal coefficient amplitude and amplification. The specific steps are as follows:

[0081] After obtaining the various DMD modes of hydrogen fuel flow within an aero-engine, it is necessary to quantitatively characterize their impact. Modal coefficient amplitudes are used as an example. b a As an evaluation index, its calculation expression is shown in the following formula. The physical meaning of this parameter is that, due to the transient characteristics of unsteady flow in turbomachinery, the difference between the flow field at each time step and the initial state can be decomposed into a linear superposition of different modes. Specifically, the modal coefficient amplitude... b a The value is obtained by calculating the product of the inverse of the mode and the initial flow field snapshot. Its magnitude directly reflects the contribution weight of the mode to the overall flow evolution.

[0082] As shown in the formula below:

[0083] ,in, It is a snapshot of the flow field variables at the first time step.

[0084] The frequencies and amplification of each mode can be obtained by mapping their eigenvalues, where the real part is... g i The imaginary part represents the magnification of the flow field mode corresponding to the eigenvalue. w i The characteristic frequency of a mode. λ i These are the eigenvalues ​​of the adjoint matrix S. It's important to note that the amplification factor describes the changing characteristics of this mode during the time series evolution. If... g i A value greater than 0 indicates that the mode is unstable and will gradually worsen over time; conversely, a value greater than 0 indicates that the mode is unstable and will gradually worsen over time. gi <0 indicates that the mode is a stable mode, which gradually weakens and eventually dissipates over time; g i =0 indicates that the mode is a periodic mode. See the following formula:

[0085] , t This represents the change in the time step.

[0086] Step (10): Restoring the snapshot results from different times:

[0087] Based on the above processing flow, the entire process of compressed storage and dynamic mode decomposition of unsteady flow field data for hydrogen fuel flow in aero-engines was completed. For flow phenomena with significant periodicity, such as rotor-stator interference, the various modes and their amplification rates were obtained. g i and characteristic frequencies w i Subsequently, the flow field can be dynamically reconstructed and predicted using the modal superposition method.

[0088] Given the modality, modal magnification, and initial time snapshot data of a time snapshot, the variable data at different times can be reconstructed using the following formula.

[0089] ,in, It is the reconstructed time snapshot matrix. It is the first i The dynamic modes of the first order, g i For the first i Magnification of the order, w i For the first i The characteristic frequency of the order, b i For the first i The modal coefficient amplitude of the first order, t For time step.

[0090] The reconstructed hydrogen fuel time snapshot matrix can then be written as follows:

[0091] .

[0092] In this embodiment, a computer device is provided, such as... Figure 3 As shown, it includes a memory 301, a processor 302, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the post-processing method for the above-mentioned arbitrary hydrogen fuel flow simulation CFD.

[0093] Specifically, the computer device can be a computer terminal, a server, or a similar computing device.

[0094] In this embodiment, a computer-readable storage medium is provided, which stores a computer program that performs any of the above-described post-processing methods for hydrogen fuel flow simulation CFD.

[0095] Specifically, computer-readable storage media include both permanent and non-permanent, removable and non-removable media, which can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer-readable storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable storage media do not include transient media, such as modulated data signals and carrier waves.

[0096] Based on the same inventive concept, this invention also provides a post-processing device for hydrogen fuel flow simulation CFD, as described in the following embodiments. Since the principle of the post-processing device for hydrogen fuel flow simulation CFD is similar to that of the post-processing method for hydrogen fuel flow simulation CFD, the implementation of the post-processing device for hydrogen fuel flow simulation CFD can refer to the implementation of the post-processing method for hydrogen fuel flow simulation CFD, and repeated details will not be elaborated further. As used below, the terms "unit" or "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0097] Figure 4 This is a structural block diagram of a post-processing device for hydrogen fuel flow simulation CFD according to an embodiment of the present invention, such as... Figure 4 As shown, it includes: a grid data generation module 401, a modal calculation module 402, and a flow field reconstruction and prediction module 403. The structure is described below.

[0098] The grid data generation module 401 is used to perform CFD simulation of hydrogen fuel flow based on aero-engine according to the set input variables and generate unstructured grid data.

[0099] Modal calculation module 402 is used to convert the unstructured mesh data into multiple simulation result files, sort the mesh geometric information and numerical distribution data in the simulation result files according to physical time, and generate a time snapshot matrix. Through the time snapshot matrix The adjoint matrix S is calculated, and the mode corresponding to each eigenvector of the hydrogen fuel flow field is calculated using the adjoint matrix S. ;

[0100] The flow field reconstruction and prediction module 403 is used to reconstruct and predict the modes corresponding to each feature vector of the hydrogen fuel flow field. The magnification was calculated. g i and modal coefficient amplitude b a Based on the magnification g i and the modal coefficient amplitude b a The hydrogen fuel flow field is dynamically reconstructed and predicted using the modal superposition method.

[0101] In one embodiment, the modal calculation module includes:

[0102] The distributed data extraction unit is used to parse the unstructured grid data and extract the spatial distribution data of the grid topology and physical quantities. The grid topology includes node coordinates and cell connection relationships, and the physical quantities include velocity field, pressure field and temperature field.

[0103] A type separation unit is used to establish a physical quantity mapping relationship based on the unit number, and to separate the physical quantities according to their types based on the physical quantity mapping relationship;

[0104] The simulation result file generation unit is used to generate independent simulation result files based on the spatial distribution data, with each type of physical quantity as the unit. The simulation result files include the mesh geometry information and the numerical distribution data of the physical quantities.

[0105] In one embodiment, the modal calculation module further includes:

[0106] The time snapshot matrix construction unit is used to sort the mesh geometry information and numerical distribution data in the simulation result file according to physical time, and generate a time snapshot matrix. ,in, , v 1. v 2...... v NThese are variables at different time steps, including velocity, pressure, temperature, and turbulence;

[0107] The snapshot matrix construction unit is used to construct the snapshot matrix before and after the snapshot. Before obtaining N Snapshot matrix at time -1 and after N Snapshot matrix at time -1 ;

[0108] The adjoint matrix construction unit is used to construct the adjoint matrix based on the previous N Snapshot matrix at time -1 and after N Snapshot matrix at time -1 The mapping relationship is used to obtain the mapping matrix A of the whole sequence. The order of the mapping matrix A of the whole sequence is reduced, and the adjoint matrix S is calculated.

[0109] In one embodiment, the adjoint matrix construction unit is used to perform singular value decomposition on the preceding matrix. N Snapshot matrix at time -1 Dimensionality reduction is performed to generate a diagonal matrix Σ composed of singular values. This diagonal matrix Σ is then truncated to generate a truncated diagonal matrix Σ. The result is obtained by using a unitary matrix U containing left singular vectors, the truncated diagonal matrix Σ, and the subsequent... N Snapshot matrix at time -1 We obtain the mapping matrix A, where, J is the unitary matrix of the right singular vector, and U is the unitary matrix containing the left singular vector; through the mapping matrix A and the unitary matrix U, the reduced-order adjoint matrix S is obtained, where, U * Let be the conjugate transpose of U.

[0110] In one embodiment, the flow field reconstruction and prediction module includes:

[0111] The eigenvalue vector unit is used to delve into the dynamic characteristics of the hydrogen fuel flow field through eigenvalue decomposition, and to solve for the eigenvalues ​​of the adjoint matrix S. λ i and eigenvector K i ;

[0112] Modal calculation unit, used for post-processing N Snapshot matrix at time -1 The feature vector K i By combining the truncated diagonal matrix Σ, the modes corresponding to each eigenvector of the hydrogen fuel flow field are calculated. ,in, J is the unitary matrix of the right singular vector.

[0113] In one embodiment, the modal calculation module further includes:

[0114] The coefficient amplitude calculation unit is used to calculate the mode corresponding to each eigenvector of the hydrogen fuel flow field. and the variables of the first time step v 1. Calculate the modal coefficient amplitude. b a ,in, ;

[0115] Magnification calculation unit, used to calculate the magnification by the first i Characteristic frequencies of the first-order modes w i and the adjoint matrix S of the first i eigenvalues ​​of order λ i Calculate the first i Magnification of the order g i ,in, , t This represents the change in the time step.

[0116] Reconstruction prediction unit, used for based on the first i Characteristic frequencies of the first-order modes w i The first i Magnification of the order g i and the modal coefficient amplitude b a The hydrogen fuel flow field is dynamically reconstructed and predicted using the modal superposition method.

[0117] In one embodiment, the reconstructed prediction unit is used for ,in, It is the reconstructed time snapshot matrix. e It is a natural constant. r The modal order used for reconstruction, For the first i The dynamic modes of the first order, b i For the first i The modal coefficient amplitude of the first order, t For time step, w i For the first i The characteristic frequencies of the order mode.

[0118] The embodiments of this invention achieve the following technical effects: First, this invention proposes a modal decomposition method, a data compression method for hydrogen fuel flow simulation results in aero-engines. Considering the large-scale computational demands of aero-engine turbomachinery, this method reduces the order of output results at different times and stores them in modal form, effectively reducing the data storage space required for post-processing. Second, this invention provides an effective method for extracting hydrogen fuel flow characteristics in aero-engines. Addressing the periodic evolution of the flow field near the turbomachinery under the rotation of the blades, it accurately extracts flow field characteristic parameters corresponding to different frequencies, including key indicators such as pressure distribution and turbulent kinetic energy distribution. Based on the precise analysis of these characteristic parameters, accurate location of fault sources can be achieved, thus providing reliable technical support for fault diagnosis and maintenance of turbomachinery. Third, this invention also proposes an efficient post-processing method for hydrogen fuel flow simulation in aero-engines. By calculating the flow field modes, modal coefficient amplitudes, and modal amplification rates, and combining them with the initial flow field values, the flow field state at any time can be reconstructed and predicted using modal superposition, thereby avoiding a large amount of repetitive simulation work and improving post-processing efficiency.

[0119] Obviously, those skilled in the art should understand that the modules or steps of the above-described embodiments of the present invention can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented here, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the embodiments of the present invention are not limited to any particular hardware and software combination.

[0120] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the embodiments of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A post-processing method of hydrogen fuel flow simulation CFD, characterized by, The method comprises: performing a hydrogen fuel flow CFD simulation based on an aero-engine according to set input variables to generate unstructured grid data; Converting the unstructured grid data into a plurality of simulation result files, sorting the grid geometry information and numerical distribution data in the simulation result files according to physical time to generate a time snapshot matrix , calculating an adjoint matrix S through the time snapshot matrix , and calculating a mode corresponding to each eigenvector of the hydrogen fuel flow field through the adjoint matrix S ; a mode corresponding to each feature vector of the hydrogen fuel flow field , the magnification is calculated g i and the modal coefficient amplitude b a , based on the magnification g i and the modal coefficient amplitude b a , the hydrogen fuel flow field is dynamically reconstructed and predicted by a modal superposition method, comprising: a mode corresponding to each eigenvector of the hydrogen fuel flow field and the variable of the first time step v 1, the mode coefficient amplitude is calculated b a wherein ; By the first i order modal characteristic frequency w i and the first i order eigenvalue of the companion matrix S λ i , the first i order amplification is calculated g i wherein , t is the change in the time step based on the first i characteristic frequency of the modal of the order w i , the first i amplification of the order g i and the modal coefficient amplitude b a , the hydrogen fuel flow field is dynamically reconstructed and predicted by the modal superposition method, comprising: ,in, It is the reconstructed time snapshot matrix. e It is a natural constant. r The modal order used for reconstruction, For the first i The dynamic modes of the first order, b i For the first i The modal coefficient amplitude of the first order, t For time step, w i For the first i The characteristic frequencies of the order mode.

2. The post-processing method of hydrogen fuel flow simulation CFD of claim 1, wherein, converting the unstructured grid data into a plurality of simulation result files, comprising: parsing the unstructured grid data to extract grid topology and spatial distribution data of physical quantities, wherein the grid topology comprises node coordinates and element connection relationships, and the physical quantities include velocity field, pressure field and temperature field; establishing a physical quantity mapping relationship based on element numbers, and separating the physical quantities by type based on the physical quantity mapping relationship; based on the spatial distribution data, generating independent simulation result files for each type of physical quantity, wherein the simulation result files include the grid geometry information and the numerical distribution data of the physical quantities.

3. The post-processing method of hydrogen fuel flow simulation CFD of claim 1, wherein, According to physical time, the grid geometry information and the numerical distribution data in the simulation result file are sorted to generate a time snapshot matrix The accompanying matrix S is calculated through the time snapshot matrix including: According to physical time, the grid geometry information and numerical distribution data in the simulation result file are sorted to generate a time snapshot matrix wherein, , v 1、 v 2...... v N are variables of different time steps, the variables including velocity, pressure, temperature and turbulence; through the time snapshot matrix before N -1 snapshot matrix of the time and after N -1 snapshot matrix of the time ; According to the mapping relationship between the snapshot matrixes of the two time points N -1 time point and the mapping relationship between the snapshot matrixes of the two time points N -1 time point , a mapping matrix A of the whole sequence is obtained, and the mapping matrix A of the whole sequence is reduced in order to obtain a companion matrix S.

4. The post-processing method of hydrogen fuel flow simulation CFD of claim 3, wherein, According to the foregoing N -1 snapshot matrix at a time and the latter N -1 snapshot matrix at a time mapping relationship, obtain the mapping matrix A of the overall sequence, and reduce the order of the mapping matrix A of the overall sequence to obtain the accompanying matrix S. The singular value decomposition method is used to process the former N -1 snapshot matrix at a time The dimensionality reduction processing is performed to generate a diagonal matrix Σ composed of singular values, and the diagonal matrix Σ is truncated to generate a truncated diagonal matrix Σ; by a unitary matrix U comprising left singular vectors, the truncated diagonal matrix Σ and a post N - a snapshot matrix at time instant , obtaining a mapping matrix A, wherein J is a unitary matrix of right singular vectors and U is a unitary matrix comprising left singular vectors. By means of the mapping matrix A and the unitary matrix U, a reduced order adjoint matrix S is obtained, wherein , U * is the conjugate transpose matrix of U.

5. The post-processing method of hydrogen fuel flow simulation CFD of claim 1, wherein, By means of said companion matrix S, the mode corresponding to each eigenvector of the hydrogen fuel flow field is calculated comprising: solving the eigenvalues of the companion matrix S by eigenvalue decomposition of the dynamic characteristics of the hydrogen fuel flow field λ i and eigenvectors K i ; By post N -1 snapshot matrix at time , the feature vector K i and the truncated diagonal matrix Σ, the mode corresponding to each feature vector of the hydrogen fuel flow field is calculated where, , J is the unitary matrix of the right singular vector.

6. A post-processing device for hydrogen fuel flow simulation CFD, characterized by, The method comprises: a grid data generation module for performing a hydrogen fuel flow CFD simulation based on an aero-engine according to set input variables to generate unstructured grid data; A modal calculation module is configured to convert the unstructured grid data into a plurality of simulation result files, sort grid geometry information and numerical distribution data in the simulation result files according to physical time, and generate a time snapshot matrix An adjoint matrix S is calculated through the time snapshot matrix Each eigenvector of the hydrogen fuel flow field corresponds to a mode calculated through the adjoint matrix S ​ The flow field reconstruction and prediction module is used to reconstruct and predict the modes corresponding to each feature vector of the hydrogen fuel flow field. The magnification was calculated. g i and modal coefficient amplitude b a Based on the magnification g i and the modal coefficient amplitude b a The hydrogen fuel flow field is dynamically reconstructed and predicted using the modal superposition method; a flow field reconstruction and prediction module, comprising: a coefficient amplitude calculation unit configured to calculate a modal coefficient amplitude of each of the feature vectors of the hydrogen fuel flow field and the variable of the first time step v 1, the modal coefficient amplitude is calculated b a wherein ; a magnification calculation unit for calculating a magnification of the i order mode from the characteristic frequency of the w i order mode of the i λ i characteristic value of the i g i wherein , t is a change in the time step.​​ a reconstruction prediction unit configured to reconstruct a flow field based on the first i characteristic frequency of the modal of the order w i , the first i amplification of the modal of the order g i and the modal coefficient amplitude b a , and predict the flow field of hydrogen fuel by modal superposition method The reconstruction prediction unit is further configured to wherein, is the reconstructed temporal snapshot matrix, e is a natural constant, r is a modal order used for reconstruction, is a dynamic modal of order i is a modal coefficient amplitude of order b i is a modal coefficient amplitude of order i is a modal coefficient amplitude of order t is a time step, w i is a modal characteristic frequency of order i is a modal characteristic frequency of order 7. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the post-processing method of the hydrogen fuel flow simulation CFD according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program for executing the post-processing method of the hydrogen fuel flow simulation CFD according to any one of claims 1 to 5.

Images