An emulation data generation method, apparatus, device, medium and product

Abstract

The application discloses a simulation data generation method and device, equipment, medium and product, relates to the technical field of engineering simulation, and the method comprises the following steps: obtaining target simulation parameters and a target file name input by a user through a user graphical interface; analyzing the target file name to obtain an extraction path of a simulation data file, a simulation data set name and a simulation data structure associated with the target file name; based on the extraction path of the simulation data file and the simulation data set name, using a block reading, lazy loading or concurrent reading strategy to extract field variable data associated with the target simulation parameters in the simulation data file; calling a pre-trained reduced-order model to predict high-dimensional data of the field variable data, and obtaining high-dimensional field variable prediction data; and reconstructing the high-dimensional field variable prediction data according to the format of the simulation data, and generating high-dimensional field variable data which is the same as the simulation data structure. The application can improve data processing efficiency and user friendliness.

Description

Technical Field

[0001] This application relates to the field of engineering simulation technology, and in particular to a method, apparatus, equipment, medium and product for generating simulation data. Background Technology

[0002] With the development of scientific computing and engineering simulation technologies, XMF / H5 format simulation data generated in fields such as fluid mechanics and structural mechanics are becoming increasingly large and complex. Generating XMF / H5 format simulation data is time-consuming, resulting in low efficiency. To improve this efficiency, reduced-order models are generally used to predict simulation data, thereby reducing its complexity and time consumption.

[0003] However, in existing technologies, if open-source libraries are used to reduce the model order, operators need to have professional programming knowledge and relatively proficient operating skills, which is not user-friendly for ordinary practitioners. Therefore, there is an urgent need to provide a simulation data generation method that can improve data processing efficiency and is user-friendly. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the related technologies, the purpose of this application is to provide a simulation data generation method, apparatus, equipment, medium and product that can improve data processing efficiency and be more user-friendly.

[0005] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a simulation data generation method, which involves obtaining target simulation parameters and target file names input by the user through a user graphical interface; parsing the target file name to obtain the extraction path, simulation dataset name, and simulation data structure of the simulation data file associated with the target file name; extracting field variable data associated with the target simulation parameters from the simulation data file using block reading, lazy loading, or concurrent reading strategies based on the extraction path and simulation dataset name; calling a pre-trained reduced-order model to predict the high-dimensional data of the field variable data to obtain high-dimensional field variable prediction data; and reconstructing the high-dimensional field variable prediction data according to the simulation data format to generate high-dimensional field variable data with the same structure as the simulation data.

[0006] Optionally, the target file is an XMF file; parsing the target file name to obtain the extraction path, simulation dataset name, and simulation data structure of the simulation data file associated with the target file name includes: performing deep parsing of the XML structure of the XMF file, using block reading and lazy loading to identify the path, dataset name, and simulation data structure of a single H5 file of the XMF file; and using multi-threading or multi-processing to read multiple H5 files in parallel to obtain parameter and field variable snapshot data.

[0007] Optionally, the step of calling a pre-trained reduced-order model to predict the high-dimensional data of the field variable data to obtain high-dimensional field variable prediction data includes: inputting the obtained target simulation parameters into the pre-trained reduced-order model, wherein the pre-trained reduced-order model predicts the coefficients of the target simulation parameters in the low-dimensional space. The coefficients of the target simulation parameters in the low-dimensional space are combined with the POD mode set and reconstructed by linear combination to obtain the high-dimensional field variable prediction data corresponding to the target simulation parameters.

[0008] Optionally, the step of reconstructing the high-dimensional field variable prediction data according to the simulation data structure to generate high-dimensional field variable data with the same format as the simulation data includes: reconstructing the high-dimensional field variable prediction data according to the simulation data structure to generate reconstructed field variable data; creating a new XMF file according to the structure of the simulation data, wherein the new XMF file contains a reference to a new H5 data file; writing the reconstructed field variable data into the new H5 file, and ensuring that the data structure, dataset name, and attributes of the reconstructed field variable data are consistent with the target file, so as to generate the high-dimensional field variable data.

[0009] Optionally, the training process of the pre-trained reduced-order model includes: acquiring simulation parameters, XMF files, and H5 files input by the user through a graphical user interface; parsing the XMF files to establish a mapping relationship between field variable names and the H5 dataset; extracting numerical data from the H5 files to obtain a snapshot data set associated with the simulation parameters; performing POD decomposition on the snapshot data set to obtain a set of POD modes and low-dimensional coefficients corresponding to the snapshot data; using the simulation parameters as input and the corresponding low-dimensional coefficients as output, training a surrogate model using the user-selected RBF, GPR, or ANN algorithm to learn the mapping relationship from parameters to low-dimensional coefficients; the trained reduced-order model includes POD modes, a machine learning model, and corresponding metadata.

[0010] Optionally, the proxy model is a reduced-order model built based on the EZyRB library, and the proxy model includes any combination of POD and RBF, GPR and ANN.

[0011] Secondly, this application provides a simulation data generation device, the simulation data generation device comprising: The acquisition module is used to acquire the target simulation parameters and target file name input by the user through the user's graphical interface; The parsing module is used to parse the target file name to obtain the extraction path, simulation dataset name, and simulation data structure of the simulation data file associated with the target file name; The extraction module is used to extract field variable data associated with the target simulation parameters from the simulation data file based on the extraction path and simulation dataset name of the simulation data file, using block reading, lazy loading, or concurrent reading strategies. The prediction module is used to call a pre-trained reduced-order model to predict the high-dimensional data of the field variable data, and obtain high-dimensional field variable prediction data. The generation module is used to reconstruct the high-dimensional field variable prediction data according to the format of the simulation data, and generate high-dimensional field variable data with the same structure as the simulation data.

[0012] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the simulation data generation method described in any one of the above.

[0013] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the simulation data generation method described in any one of the above.

[0014] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the simulation data generation method described in any one of the above.

[0015] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a simulation data generation method, apparatus, device, medium, and product. It obtains the target simulation parameters and target file name input by the user through a graphical user interface; by parsing the target file name, it obtains the extraction path, simulation dataset name, and simulation data structure of the simulation data file associated with the target file name; using the extraction path and simulation dataset name, it extracts the field variable data associated with the target simulation parameters from the simulation data file using block reading, lazy loading, or concurrent reading strategies; it predicts the high-dimensional data of the field variable data by calling a pre-trained reduced-order model to obtain high-dimensional field variable prediction data; and it reconstructs the high-dimensional field variable prediction data according to the simulation data format to generate high-dimensional field variable data with the same structure as the simulation data. This reduces the modeling complexity of high-dimensional simulation data, significantly shortens the data generation cycle, solves the problem of long processing times in traditional methods, and effectively improves data processing efficiency. The visual user interface effectively reduces the dependence on users' professional programming skills, improves tool usability, and achieves a user-friendly purpose. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, 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.

[0017] Figure 1 A schematic flowchart illustrating a simulation data generation method provided in an embodiment of this application; Figure 2 A user graphical interface illustration provided in an embodiment of this application Figure 1 ; Figure 3 A user graphical interface illustration provided in an embodiment of this application Figure 2 ; Figure 4 A user graphical interface illustration provided in an embodiment of this application Figure 3 ; Figure 5 A schematic diagram of the functional modules of a simulation data generation device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0019] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the following detailed description of this application is provided in conjunction with the accompanying drawings and specific embodiments.

[0020] In one exemplary embodiment, such as Figure 1 As shown, this application provides a simulation data generation method, which is executed by a computer device. In this embodiment, the simulation data generation method includes the following steps S101 to S105. Wherein: Step S101: Obtain the target simulation parameters and target file name input by the user through the user graphical interface.

[0021] In the example embodiment, the target files for this application embodiment are XMF files and H5 files. For example, target parameter data: the simulation parameter matrix is ​​extracted from the CSV file, which includes parameter combinations under different working conditions (such as Mach number, angle of attack, Reynolds number, etc.); XMF file: an XML format file containing data structure information, mesh topology, and field variable definitions; H5 file: a binary format file containing actual numerical data, storing snapshot data of field variables (such as pressure, velocity, temperature, etc.).

[0022] Step S102: Parse the target file name to obtain the extraction path, simulation dataset name, and simulation data structure of the simulation data file associated with the target file name.

[0023] In the example implementation, by performing deep parsing of the XML structure of the XMF file, the path, dataset name, and data organization method of the referenced H5 data files can be accurately identified. For H5 files, a chunked reading and lazy loading strategy is adopted to avoid loading the entire large dataset into memory at once, thereby significantly improving processing efficiency and reducing memory usage. The module also supports concurrent reading. For multiple H5 files or multiple independent datasets within H5 files, parallel reading can be performed using multi-threading or multi-processing to quickly extract parameter and field variable snapshot data (npz format). Users can select field variables with specific names for extraction. The field variable names are determined by the naming of attributes in the XML, such as pressure, velocity, etc.

[0024] Optionally, by performing deep parsing of the XML structure of the XMF file, block reading and lazy loading are used to identify the path of a single H5 file, the dataset name, and the simulation data structure of the XMF file; for multiple H5 files, parallel reading is performed using multi-threading or multi-processing to obtain parameter and field variable snapshot data.

[0025] Step S103: Based on the extraction path and simulation dataset name of the simulation data file, use block reading, lazy loading or concurrent reading strategies to extract the field variable data associated with the target simulation parameters from the simulation data file.

[0026] Step S104: Call the pre-trained reduced-order model to predict the high-dimensional data of the field variable data, and obtain the high-dimensional field variable prediction data.

[0027] Optionally, the obtained target simulation parameters are input into a pre-trained reduced-order model, which predicts the coefficients of the target simulation parameters in the low-dimensional space. The coefficients of the target simulation parameters in the low-dimensional space are then combined with the POD mode set and reconstructed through linear combination to obtain the high-dimensional field variable prediction data corresponding to the target simulation parameters.

[0028] Step S105: Reconstruct the high-dimensional field variable prediction data according to the format of the simulation data to generate high-dimensional field variable data with the same structure as the simulation data.

[0029] Optionally, the high-dimensional field variable prediction data is reconstructed according to the simulation data structure to generate reconstructed field variable data; a new XMF file is created according to the structure of the simulation data, and the new XMF file contains a reference to the new H5 data file; the reconstructed field variable data is written into the new H5 file, and the data structure, dataset name and attributes of the reconstructed field variable data are kept consistent with the target file to generate high-dimensional field variable data.

[0030] The prediction results file can be directly opened and analyzed in three dimensions by mainstream scientific visualization software, achieving a seamless transition from prediction to visualization.

[0031] Furthermore, in this embodiment, the training process of the pre-trained reduced-order model includes: acquiring simulation parameters, XMF files, and H5 files input by the user through a graphical user interface; parsing the XMF files to establish a mapping relationship between field variable names and the H5 dataset; extracting numerical data from the H5 files to obtain a snapshot data set associated with the simulation parameter data; performing POD decomposition on the snapshot data set to obtain a set of POD modes and low-dimensional coefficients corresponding to the snapshot data; using the simulation parameters as input and the corresponding low-dimensional coefficients as output, training a surrogate model using the user-selected RBF, GPR, or ANN algorithm to learn the mapping relationship from parameters to low-dimensional coefficients; the trained reduced-order model includes POD modes, a machine learning model, and corresponding metadata.

[0032] Optionally, the surrogate model is a reduced-order model built based on the EZyRB library. The surrogate model includes any combination of POD and RBF, GPR and ANN.

[0033] gather Figures 2 to 4 As can be understood from the above embodiments, in different application scenarios, different simulation parameters and field variable data (snapshot data) input by the user can be used to train and generate reduced-order models with different parameters to establish a mapping relationship between simulation parameters and field variable data. This may include: S1, in Figure 2 The interface retrieves the following data: 1. Simulation parameter data: The simulation parameter matrix is ​​extracted from the CSV file, containing parameter combinations under different operating conditions (such as Mach number, angle of attack, Reynolds number, etc.). 2. XMF file: An XML format file containing data structure information, mesh topology, and field variable definitions. 3. H5 file: A binary format file containing actual numerical data, storing snapshot data of field variables (such as pressure, velocity, temperature, etc.). 4. Mesh data: Geometric mesh information and topology extracted from the XMF / H5 file.

[0034] The specific implementation process is as follows: parse the XMF file, establish the mapping relationship between the field variable names and the H5 dataset, extract numerical data from the H5 file, and load the parameter matrix.

[0035] S2, process the acquired data: 1. Data format verification: check the dimensionality matching of the parameter matrix and the snapshot matrix; 2. Data normalization: MinMax normalization: scale the data to the [0,1] interval; Z-score standardization: standardize the data to a distribution with a mean of 0 and a variance of 1; 3. Data reconstruction: reconstruct the snapshot data into the format required by the reduced-order modeling library (n_dof×n_samples); 4. Data consistency check: verify the consistency between the number of parameter samples and the number of snapshot samples.

[0036] S3, in Figure 3The following parameters can be obtained from the interface: Dimensionality reduction parameters: Number of POD modes (n_modes): controls the dimension after dimensionality reduction, usually 10-50; Energy retention ratio: determines the main modes to be retained, usually retaining more than 99% of the energy; Approximation method parameters: RBF kernel function type: gaussian, multiquadric, linear, etc.; GPR kernel function parameters: length scale, noise variance; ANN network structure: number of hidden layers, number of neurons, activation function type.

[0037] S4, in Figure 3 Training control parameters in the interface: training / validation data split ratio (default 8:2); cross-validation folds; convergence criterion and maximum number of iterations.

[0038] It should be noted that the embodiments of this application employ a data-driven reduction model, specifically composed of: a dimensionality reduction part: Intrinsic Orthogonal Decomposition (POD) – a linear dimensionality reduction method based on singular value decomposition; and an approximation part: Radial Basis Function (RBF) / Gaussian Process Regression (GPR) / Artificial Neural Network (ANN) – a nonlinear regression model. For example, taking fluid simulation as an example, the input parameters are [Mach number = 0.8, angle of attack = 5°], and the output is pressure field snapshot data. The model establishes a mapping relationship from the parameter space to the snapshot space: f([0.8, 5°]) → pressure field data. Users train the basic model framework included in the simulation data generation tool, rather than creating a completely new model. The tool provides a combined framework of POD+RBF / GPR / ANN, and users can adjust the parameters to adapt to specific application requirements.

[0039] The reduced-order model involved in the embodiments of this application is essentially a composite function mapping: for example, input: simulation parameter vector μ∈R^p; output: field variable snapshot u∈R^n; mapping relationship: u≈Φ×f(μ), where Φ is the POD basis matrix and f(μ) is the nonlinear mapping from parameters to POD coefficients.

[0040] The process of adjusting relevant parameters in the model can be implemented as follows: 1. POD dimensionality reduction parameter adjustment: Adjust the number of retained POD modes; determine the main modes through singular value decomposition; automatically calculate the cumulative energy ratio to determine the optimal number of modes. 2. RBF approximation parameter adjustment: Adjust the kernel function type; automatically optimize the shape parameters and weight coefficients internally. 3. Training process monitoring: POD decomposition: Calculate the singular value decomposition of the snapshot matrix; Dimensionality reduction mapping: Project the high-dimensional snapshot onto the low-dimensional POD space; Approximate training: Establish the mapping from parameters to POD coefficients in the low-dimensional space; Error evaluation: Calculate the training error and cross-validation error.

[0041] exist Figure 4In this process, the steps of obtaining the target simulation parameters input by the user and generating simulation data using the trained model can include: obtaining the target simulation parameters, namely: 1. The combination of parameters of the operating conditions to be predicted input by the user; for example: new Mach number, angle of attack, Reynolds number and other parameter values; supporting single-point prediction and batch prediction; 2. Output configuration parameters: specifying the output file path and file name format; selecting the type of field variable to be output (pressure, velocity, temperature, etc.); 3. Mesh information: reusing the mesh structure information of the training data; 4. File format settings: XMF / H5 file format configuration parameters.

[0042] Specific implementation: Obtain the parameter values ​​input by the user through the parameter table component; verify whether the parameter range is within the parameter space of the training data; configure the naming rules and storage path of the output file.

[0043] Next, the trained model is used to process the acquired data and generate simulation data; specifically: 1. Parameter preprocessing: The new input is standardized using the normalized parameters from training; ensuring the parameter format is consistent with the training data; 2. Model prediction calculation: Parameter normalization (using the normalized parameters from training); the reduced-order model is used to predict the POD coefficients; the POD is reconstructed to obtain the complete field variables; 3. File generation process: An H5 file is created, and the predicted field variable data is written into it; an XMF file is created, and a link is established with the H5 file; 4. Data post-processing: The prediction results are reconstructed according to the original data format; a data structure compatible with the original XMF / H5 file is generated; ensuring that the generated file can be recognized by visualization tools such as ParaView and Tecplot.

[0044] It should be noted that the prediction process in this application is a mapping from the parameter space to the field variable space, that is, given a new parameter μ_new, the corresponding field variable u_new=f(μ_new) is calculated; file generation: the predicted field variable data u_new is converted into the standard XMF / H5 format.

[0045] By implementing steps S101 to S105 above, the target simulation parameters and target file name input by the user are obtained through a user graphical interface; by parsing the target file name, the extraction path, simulation dataset name, and simulation data structure of the simulation data file associated with the target file name are obtained; using the extraction path and simulation dataset name of the simulation data file, the field variable data associated with the target simulation parameters in the simulation data file are extracted using block reading, lazy loading, or concurrent reading strategies; by calling a pre-trained reduced-order model to predict the high-dimensional data of the field variable data, high-dimensional field variable prediction data is obtained; the high-dimensional field variable prediction data is reconstructed according to the simulation data format to generate high-dimensional field variable data with the same structure as the simulation data; this can reduce the modeling complexity of high-dimensional simulation data, significantly shorten the data generation cycle, solve the problem of long processing time in traditional methods, and effectively improve data processing efficiency; the visual user interface can effectively reduce the dependence on the user's professional programming ability, improve the ease of use of the tool, and achieve the goal of user-friendliness.

[0046] Based on the same inventive concept, this application also provides a simulation data generation apparatus for implementing the simulation data generation method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the simulation data generation apparatus provided below can be found in the limitations of the simulation data generation method described above, and will not be repeated here.

[0047] In one exemplary embodiment, such as Figure 5 As shown, a simulation data generation device 500 is provided, comprising: The acquisition module 501 is used to acquire the target simulation parameters and target file name input by the user through the user's graphical interface; The parsing module 502 is used to parse the target file name to obtain the extraction path, simulation dataset name and simulation data structure of the simulation data file associated with the target file name; The extraction module 503 is used to extract field variable data associated with the target simulation parameters from the simulation data file based on the extraction path and simulation dataset name of the simulation data file, using block reading, lazy loading or concurrent reading strategies. Prediction module 504 is used to call a pre-trained reduced-order model to predict high-dimensional field variable data and obtain high-dimensional field variable prediction data. The generation module 505 is used to reconstruct the high-dimensional field variable prediction data according to the simulation data format, and generate high-dimensional field variable data with the same structure as the simulation data.

[0048] As an optional implementation, the target file is an XMF file; the parsing module 502 is specifically used to identify the path, dataset name and simulation data structure of a single H5 file of the XMF file by performing deep parsing of the XML structure of the XMF file, and using block reading and lazy loading; for multiple H5 files, parallel reading is performed using multi-threading or multi-processing to obtain parameter and field variable snapshot data.

[0049] As an optional implementation, the prediction module 504 is specifically used to input the acquired target simulation parameters into a pre-trained reduced-order model, and the pre-trained reduced-order model predicts the coefficients of the target simulation parameters in the low-dimensional space; the coefficients of the target simulation parameters in the low-dimensional space are combined with the POD mode set and reconstructed by linear combination to obtain the high-dimensional field variable prediction data corresponding to the target simulation parameters.

[0050] As an optional implementation, the generation module 505 is specifically used to: reconstruct the high-dimensional field variable prediction data according to the simulation data structure to generate reconstructed field variable data; create a new XMF file according to the structure of the simulation data, the new XMF file containing a reference to a new H5 data file; write the reconstructed field variable data into the new H5 file, and ensure that the data structure, dataset name, and attributes of the reconstructed field variable data are consistent with the target file, so as to generate high-dimensional field variable data.

[0051] As an optional implementation, the simulation data generation device further includes a training module. Specifically, the training module is used to: acquire simulation parameters, XMF files, and H5 files input by the user through a graphical user interface; parse the XMF files to establish a mapping relationship between field variable names and the H5 dataset; extract numerical data from the H5 files to obtain a snapshot data set associated with the simulation parameters; perform POD decomposition on the snapshot data set to obtain a set of POD modes and low-dimensional coefficients corresponding to the snapshot data; use the simulation parameters as input and the corresponding low-dimensional coefficients as output to train a surrogate model using the user-selected RBF, GPR, or ANN algorithm to learn the mapping relationship from parameters to low-dimensional coefficients; the trained reduced-order model includes POD modes, a machine learning model, and corresponding metadata.

[0052] As an optional implementation, the above proxy model is a reduced-order model built based on the EZyRB library. The proxy model includes any combination of POD and RBF, GPR and ANN.

[0053] This implementation method involves obtaining the target simulation parameters and target file name input by the user through a graphical user interface; parsing the target file name to obtain the extraction path, simulation dataset name, and simulation data structure of the simulation data file associated with the target file name; using the extraction path and simulation dataset name, extracting the field variable data associated with the target simulation parameters from the simulation data file using block reading, lazy loading, or concurrent reading strategies; predicting the high-dimensional data of the field variable data by calling a pre-trained reduced-order model to obtain high-dimensional field variable prediction data; reconstructing the high-dimensional field variable prediction data according to the simulation data format to generate high-dimensional field variable data with the same structure as the simulation data; this reduces the modeling complexity of high-dimensional simulation data, significantly shortens the data generation cycle, solves the problem of long processing times in traditional methods, and effectively improves data processing efficiency; the visual user interface effectively reduces the dependence on users' professional programming skills, improves tool usability, and achieves the goal of user-friendliness.

[0054] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 6 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores simulation data generation data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a simulation data generation method.

[0055] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0056] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0057] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0058] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0059] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0060] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0061] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0062] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0063] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for generating simulation data, characterized in that, The simulation data generation method includes: The target simulation parameters and target file name input by the user are obtained through a user graphical interface; the target file is an XMF file. Parsing the target file name yields the extraction path, simulation dataset name, and simulation data structure of the simulation data file associated with the target file name. This includes: performing deep parsing of the XML structure of the XMF file, using block reading and lazy loading to identify the path, dataset name, and simulation data structure of a single H5 file of the XMF file; and using multi-threading or multi-processing to read multiple H5 files in parallel to obtain parameter and field variable snapshot data. Based on the extraction path and simulation dataset name of the simulation data file, the field variable data associated with the target simulation parameters in the simulation data file are extracted using block reading, lazy loading, or concurrent reading strategies. A pre-trained reduced-order model is invoked to predict the high-dimensional data of the field variable data, thereby obtaining high-dimensional field variable prediction data. Reconstructing the high-dimensional field variable prediction data according to the simulation data format to generate high-dimensional field variable data with the same structure as the simulation data includes: reconstructing the high-dimensional field variable prediction data according to the simulation data structure to generate reconstructed field variable data; creating a new XMF file according to the structure of the simulation data, wherein the new XMF file contains a reference to a new H5 data file; writing the reconstructed field variable data into the new H5 file, and ensuring that the data structure, dataset name, and attributes of the reconstructed field variable data are consistent with the target file to generate the high-dimensional field variable data.

2. The simulation data generation method according to claim 1, characterized in that, The step of calling a pre-trained reduced-order model to predict the high-dimensional data of the field variable data, resulting in high-dimensional field variable prediction data, includes: The obtained target simulation parameters are input into the pre-trained reduced-order model, and the pre-trained reduced-order model predicts the coefficients of the target simulation parameters in the low-dimensional space. The coefficients of the target simulation parameters in the low-dimensional space are combined with the POD mode set and reconstructed by linear combination to obtain the high-dimensional field variable prediction data corresponding to the target simulation parameters.

3. The simulation data generation method according to claim 1, characterized in that, The training process of the pre-trained reduced-order model includes: Acquire simulation parameters, XMF files, and H5 files input by the user through the user's graphical interface; The XMF file is parsed to establish a mapping relationship between field variable names and H5 datasets; Extract the numerical data from the H5 file to obtain a snapshot data set associated with the simulation parameters; The snapshot data set is decomposed into PODs to obtain a set of POD modes and low-dimensional coefficients corresponding to the snapshot data. The simulation parameters are used as inputs and the corresponding low-dimensional coefficients are used as outputs. A surrogate model is trained using the user-selected RBF, GPR, or ANN algorithm to learn the mapping relationship from parameters to low-dimensional coefficients. The trained reduced-order model includes POD modalities, machine learning models, and corresponding metadata.

4. The simulation data generation method according to claim 3, characterized in that, The surrogate model is a reduced-order model built based on the EZyRB library, and the surrogate model includes any combination of POD and RBF, GPR and ANN.

5. A simulation data generation device, characterized in that, The simulation data generation device includes: The acquisition module is used to acquire the target simulation parameters and target file name input by the user through a user graphical interface; the target file is an XMF file. The parsing module is used to parse the target file name to obtain the extraction path, simulation dataset name, and simulation data structure of the simulation data file associated with the target file name. This includes: performing deep parsing of the XML structure of the XMF file, using block reading and lazy loading to identify the path, dataset name, and simulation data structure of a single H5 file of the XMF file; and using multi-threading or multi-processing to read multiple H5 files in parallel to obtain parameter and field variable snapshot data. The extraction module is used to extract field variable data associated with the target simulation parameters from the simulation data file based on the extraction path and simulation dataset name of the simulation data file, using block reading, lazy loading, or concurrent reading strategies. The prediction module is used to call a pre-trained reduced-order model to predict the high-dimensional data of the field variable data, and obtain high-dimensional field variable prediction data. The generation module is used to reconstruct the high-dimensional field variable prediction data according to the simulation data format, generating high-dimensional field variable data with the same structure as the simulation data. This includes: reconstructing the high-dimensional field variable prediction data according to the simulation data structure to generate reconstructed field variable data; creating a new XMF file according to the simulation data structure, the new XMF file containing a reference to a new H5 data file; writing the reconstructed field variable data into the new H5 file, ensuring that the data structure, dataset name, and attributes of the reconstructed field variable data are consistent with the target file, thereby generating the high-dimensional field variable data.

6. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of the simulation data generation method according to any one of claims 1-4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the simulation data generation method according to any one of claims 1-4.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the simulation data generation method according to any one of claims 1-4.

Images