Training data generation method and device, electronic equipment and storage medium

Abstract

The invention discloses a training data generation method and device, electronic equipment and a storage medium, and belongs to the technical field of data processing. The method is realized through a driving module, and comprises the following steps: driving pretreatment software to perform pretreatment on first input data to obtain first sample model data; driving a calculation server to calculate the first sample model data to obtain first calculation data; driving post-processing software to perform post-processing on the first calculation data to obtain first output data; and generating training data of a simulation model based on a mapping relationship between the first input data and the first output data. According to the embodiment of the invention, the pre-processing software, the computing server and the post-processing software are driven by the driving module, compared with manual parameter adjustment, a large amount of training data can be quickly and automatically generated, and the generation efficiency of the training data is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a training data generation method and device, electronic equipment and storage medium. BACKGROUND

[0002] In recent years, with the rapid development of the automobile industry, the number of cars has increased rapidly, and the data of automobile traffic accidents has also increased year by year. Finite element simulation, as an indispensable means in the development of vehicle safety performance, plays a crucial role in shortening the development cycle. How to improve the development efficiency of the passenger protection has become the focus of research of major automobile companies. The introduction of AI simulation can greatly improve the development efficiency of the passenger protection, but the AI simulation of the passenger protection needs a large amount of training data to obtain high prediction accuracy. The existing technology mainly collects the input and output of the passenger protection driver by artificial means, but high-precision AI models require tens of thousands of sample data, and the current scheme is inefficient and cannot meet the training data requirements of the passenger protection AI model. SUMMARY

[0003] The embodiments of the present application provide a training data generation method and device, electronic equipment and storage medium, which can quickly and automatically generate a large amount of training data by driving the pre-processing software, the computing server and the post-processing software, thereby improving the generation efficiency of the training data.

[0004] In a first aspect, the embodiments of the present application provide a training data generation method, which is realized by a driving module, and the method comprises the following steps: driving a pre-processing software to pre-process first input data to obtain first sample model data; driving a computing server to calculate the first sample model data to obtain first calculation data; driving a post-processing software to post-process the first calculation data to obtain first output data; generating training data of a simulation model based on a mapping relationship between the first input data and the first output data.

[0005] Optionally, before the step of driving the pre-processing software to pre-process the first input data to obtain the first sample model data, the method further comprises the following steps: determining key parameters of the simulation model and value ranges corresponding to each of the key parameters; based on the key parameters and the value ranges, using a mathematical sampling strategy to expand the key parameters, and generating a corresponding sample file based on each set of expanded key parameters, and taking the sample file as the first input data.

[0006] Optionally, the simulation model is a passenger protection driver AI model. The key parameters include at least one of the following: airbag parameters, steering column parameters, seatbelt module parameters, aiming point parameters, pretensioner parameters, instrument and seat parameters, and test condition parameters.

[0007] Optionally, the driving pre-processing software pre-processes the first input data to obtain first sample model data, including: Obtaining a node ID number preset for a seatbelt model; Updating position information corresponding to the node ID number according to coordinate values of key parameters in each sample file; Based on the updated position information, automatically generating a seatbelt sample model corresponding to each sample file by pre-programming the pre-processing software, and taking the seatbelt sample model as the first sample model data.

[0008] Optionally, the driving computing server calculates the first sample model data to obtain first calculation data, including: Monitoring usage of computing resources of the computing server; If idle resources of the computing server are monitored, the driving module automatically submits the first sample model data to the computing server, so that the computing server calculates the simulation model according to the first sample model data, and takes the calculated result data as the first calculation data.

[0009] Optionally, the driving post-processing software post-processes the first calculation data to obtain first output data, including: Based on the first calculation data, the driving module drives the post-processing software to determine first output data corresponding to the first calculation data according to a preset rule; the first output data is a damage value corresponding to each part of the driver simulation model; Saving the first output data in the form of a file to a folder in which the first input data is stored.

[0010] Optionally, based on a mapping relationship between the first input data and the first output data, training data of a simulation model is generated, including: Taking key parameters in each sample file as input parameters of a simulation model, taking damage values corresponding to each part of a corresponding driver simulation model as output parameters, and integrating the input parameters and the output parameters into training parameters of the simulation model.

[0011] In a second aspect, the embodiments of the present application further provide a training data generation device, including: A driving module; Preprocessing software configured to preprocess the first input data under the driving of the driving module to obtain first sample model data; A computing server configured to calculate the first sample model data under the driving of the driving module to obtain first calculation data; Postprocessing software configured to postprocess the first calculation data under the driving of the driving module to obtain first output data; A training data generation module configured to generate training data of a simulation model based on a mapping relationship between the first input data and the first output data.

[0012] In a third aspect, an electronic device is provided, including a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to implement the method in any of the above embodiments.

[0013] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, and the computer program is loaded by a processor to perform the steps in the method in any of the above embodiments.

[0014] The training data generation method provided by the embodiments of the present application can drive the preprocessing software to preprocess the first input data by the driving module to obtain the first sample model data, drive the computing server to calculate the first sample model data by the driving module to obtain the first calculation data, and further drive the postprocessing software to postprocess the first calculation data by the driving module to obtain the first output data. Then, the training data of the simulation model is generated based on the mapping relationship between the first input data and the first output data. Compared with manual parameter adjustment, the embodiments of the present application can quickly and automatically generate a large amount of training data, and improve the generation efficiency of the training data by driving the preprocessing software, the computing server and the postprocessing software by the driving module. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0016] Figure 1 A flowchart of the training data generation method provided by the embodiments of the present application; Figure 2 A data list diagram of the first input data provided by the embodiments of the present application; Figure 3A safety belt sample model provided by an embodiment of the present application exists in a deformed illustration; Figure 4 An automatically generated safety belt sample model illustration provided by an embodiment of the present application; Figure 5 A training data chart of a passenger main driver AI model provided by an embodiment of the present application; Figure 6 A chart of injury values of each part of a main driver dummy provided by an embodiment of the present application. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person skilled in the art without creative labor fall within the scope of protection of the present application.

[0018] In the description of the present application, it should be understood that the orientations or positional relationships indicated by the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements indicated must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second" are only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features limited by "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0019] "A and / or B" includes the following three combinations: only A, only B, and a combination of A and B.

[0020] The use of "applicable to" or "configured to" in the present application means open and inclusive language, which does not exclude devices applicable to or configured to perform additional tasks or steps. In addition, the use of "based on" means open and inclusive, because the process, step, calculation or other action "based on" one or more stated conditions or values can be based on additional conditions or values beyond those stated in practice.

[0021] In this application, the term "exemplary" is used to mean "used as an example, illustration, or description." Any embodiment described as "exemplary" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be made without using these specific details. In other instances, well-known structures and processes are not described in detail to avoid obscuring the description of this application with unnecessary detail. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0022] In recent years, with the rapid development of the automotive industry and the rapid increase in car ownership, the number of traffic accidents has also risen year by year. my country's automotive market has both entry-level regulations and evaluations of vehicle crash safety performance by C-NCAP, C-IASI, and various third-party testing organizations. Simultaneously, as major domestic automakers expand internationally, ECE regulations and Euro-NCAP evaluation rules related to crash safety, particularly in Europe, have become a key focus for them. Finite element simulation, as an indispensable tool in vehicle safety performance development, plays a crucial role in shortening the development cycle. Improving the development efficiency of passenger safety systems (SFSS) has become a key research focus for major automakers. Introducing AI simulation can greatly improve the development efficiency of SFSS, but SFSS AI simulation requires a large amount of training data to achieve high prediction accuracy. Current technologies mainly rely on manually collecting the inputs and outputs of the driver / passenger safety system or manually adjusting the model to obtain a set of inputs and outputs. This method is inefficient and cannot meet the data requirements of the SFSS driver / passenger safety system AI model.

[0023] Based on this, embodiments of this application provide a training data generation method, which is implemented by a driver module. The driver module can be implemented using a programming language and is used to execute driving operations during the training data generation process to achieve automated generation of training data. The driver module has different functions corresponding to different driving operations, and these different functions are implemented through programming. For example, the driver module in this application embodiment can be a Python script programmed using the Python programming language.

[0024] Please see Figure 1 , Figure 1 A flowchart illustrating a training data generation method provided in this application embodiment. The method includes: S110: The driver preprocessing software preprocesses the first input data to obtain the first sample model data.

[0025] The first input data consists of the main influencing parameters of the driver simulation model in the passenger safety AI model, along with the corresponding value ranges for each parameter. For example, the first input data may include airbag-related parameters, steering column-related parameters, and test condition-related parameters. The first input data may include multiple sets of parameter samples, each with different values. These multiple sets of samples can be obtained by sampling the value ranges of each influencing parameter. The driver simulation model corresponds to the driver dummy in the simulation model.

[0026] The first sample model data is the data required for collision calculation. When the simulation model is the passenger safety driver AI model, the first sample model data is the seat belt sample model calculated by the pre-processing software.

[0027] The preprocessing software is driven by a driver module to automatically generate a seatbelt sample model based on the first input data. For example, the preprocessing software in this application can be Primer software. The first input data is imported into Primer software one by one, and the driver module drives the preprocessing software to automatically generate the seatbelt sample model based on the corresponding parameter values ​​in the parameter sample and the preset node ID number of the seatbelt model.

[0028] Each set of parameter samples in the first input data can be used to calculate a corresponding seat belt sample model, and the obtained seat belt sample model is used as the first sample model data.

[0029] S120. Drive the computing server to perform calculations on the first sample model data to obtain the first calculation data.

[0030] In this embodiment, after obtaining the first sample model data, the driver module automatically submits the first sample data to the computing server for computation. For example, the driver module can be a Python script that performs the automatic submission function. This script can automatically submit the first sample model data to the computing server based on the resource load on the computing server. For instance, if it detects that the current computing server has a large amount of idle resources, one or more seatbelt sample models from the first sample model data can be uploaded to the computing server to improve the utilization efficiency of computing resources.

[0031] The first calculation data is the result file obtained from the simulation of each seat belt sample model. The calculation server solves the first sample model data to obtain the calculation data representing the collision result. For example, the first calculation data may include the force-time curve representing the force exerted by the seat belt, steering wheel, etc. on the driver's dummy.

[0032] S130: The driver post-processing software performs post-processing on the first calculated data to obtain the first output data.

[0033] The first output data is the output result obtained after collision simulation of the first sample model data, including the damage values ​​corresponding to each part of the driver dummy. The first output data includes multiple sets of damage values, and each set of damage values ​​corresponds to each parameter sample in the first sample model.

[0034] After obtaining the first calculation data, the post-processing module drives the automatic processing of the first calculation data, sequentially reads the curves of each part of the driver dummy, calculates the damage value corresponding to each part according to the preset rules, and uses the damage value corresponding to each part as the first output data.

[0035] The pre-set rules correspond to different regulatory requirements in different countries' automotive markets. For example, China's C-NCAP and C-IASI crash safety evaluation rules, and Europe's ECE regulations and Euro-NCAP evaluation rules related to crash safety.

[0036] S140. Based on the mapping relationship between the first input data and the first output data, generate training data for the simulation model.

[0037] The first input data and the first output data correspond to each other, that is, each set of parameter samples corresponds to a set of damage values. By summarizing and organizing the first input data and the first output data, the training data of the simulation model can be obtained.

[0038] This application embodiment uses a driver module to drive preprocessing software to preprocess the first input data, obtaining first sample model data; the driver module also drives a computing server to perform calculations on the first sample model data, obtaining first calculated data; furthermore, the driver module drives postprocessing software to postprocess the first calculated data, obtaining first output data; and then, based on the mapping relationship between the first input data and the first output data, training data for the simulation model is generated. By utilizing a driver module to drive the preprocessing software, computing server, and postprocessing software, compared to manual parameter tuning, a large amount of training data can be generated quickly and automatically, improving the efficiency of training data generation.

[0039] In some embodiments, before the driver preprocessing software preprocesses the first input data to obtain the first sample model data, the method further includes: determining the key parameters of the simulation model and the value range corresponding to each key parameter; expanding the key parameters using a mathematical sampling strategy based on the key parameters and the value range; generating a corresponding sample file based on each expanded set of key parameters; and using the sample file as the first input data.

[0040] The simulation model in this application embodiment is a passenger safety driver AI model. The key parameters are obtained by statistically analyzing historical vehicle models and combining expert experience, including at least one of the following: airbag parameters, steering column parameters, seat belt module parameters, aiming point parameters, pretensioning and force limiting device parameters, instrument and seat parameters, and test condition parameters.

[0041] The above key parameters include a total of 35 parameters. The specific parameters for each category of key parameters and their value ranges are shown in Tables 1 to 7: Table 1: Airbag Parameters and Their Value Ranges

[0042] Table 2: Steering column parameters and their value ranges

[0043] Table 3: Seatbelt Module Parameters and Their Value Ranges

[0044] Table 4: Anchor point parameters and their value ranges

[0045] Table 5: PLP parameters and their value ranges

[0046] Table 6: Instrument and Seat Parameters and Their Value Ranges

[0047] Table 7: Test Condition Parameters and Their Value Ranges

[0048] In Table 5, PLP (Pyrotechnic Lap Pretensioner) refers to the configuration parameters of the pretensioning and limiting device.

[0049] After determining the key parameters and their corresponding value ranges, a mathematical sampling strategy is used to expand the key parameters to obtain multiple sets of parameter samples. Specifically, in this embodiment, by using a Latin square design to perform a Design of Experiments (DOE) sampling on 35 key parameters, tens of thousands of sample files corresponding to the key parameters can be obtained at once. These generated sample files are then exported one by one as the first input data. See further details. Figure 2 , Figure 2 This is a data list diagram of the first input data provided in an embodiment of this application. Figure 2 Each line in the document corresponds to a set of sample files.

[0050] This application embodiment uses a Latin square DOE point-scattering design for key parameters, resulting in a more uniform distribution of sample files in multidimensional space. This improves the coverage efficiency and representativeness of key parameters in the parameter space. At the same time, this method can reflect the relationship between input and driver dummy injury using a smaller number of sample files, thereby improving the generalization ability of the passenger insurance driver AI model.

[0051] In some embodiments, the driver preprocessing software preprocesses the first input data to obtain first sample model data, including: obtaining a preset node ID number of the seat belt model; updating the position information corresponding to the node ID number according to the coordinate values ​​of key parameters in each group of sample files; and automatically generating a seat belt sample model corresponding to each sample file based on the updated position information by driving the preprocessing software through preset programming, and using the seat belt sample model as the first sample model data.

[0052] Please see Figure 3 After importing the first input data into the Primer software, the seatbelt sample model generated based on the first input data will appear as follows: Figure 3 The mesh is distorted at the fixed points A, B, and C of the seat belt. This distortion is caused by a change in the parameters of the seat belt coordinates, but the mesh does not change with the coordinates of the fixed points. Therefore, it is necessary to regenerate the seat belt sample model under the new coordinates.

[0053] The ID number of the fixed point includes preset ID numbers corresponding to each node, such as the ID number of the seat belt retractor point, the starting fixed point, the shoulder slip ring, the driver's dummy's chest, the buckle point, the driver's dummy's abdomen, and the end fixed point. Since nodes in finite element simulations consist of ID numbers and coordinate values, changes in coordinate values ​​do not cause changes in ID numbers. The generation of the seat belt is linked by the ID numbers of the coordinate points. The preset node ID numbers can be saved in a CSV file. When generating a seat belt with new coordinates, the fixed ID numbers are read, and the seat belt is naturally tightened according to the new position. Finally, the model is exported to generate a seat belt sample model that meets the requirements for collision calculation. The regenerated seat belt sample model can be found in [reference needed]. Figure 4 The data in each set of DOE sample files is processed in this way until all DOE sample files are processed, thereby obtaining the seat belt sample model corresponding to each set of DOE sample files.

[0054] This application embodiment uses a driver module to drive preprocessing software to process the first input data, which can realize the automated generation of seat belt sample models, reduce model calculation errors caused by human operation errors, and improve the quality of data generation.

[0055] In some embodiments, the driving computing server performs calculations on the first sample model data to obtain first calculation data, including: monitoring the usage of computing resources of the computing server; if idle resources are detected on the computing server, the driving module automatically submits the first sample model data to the computing server so that the computing server can perform calculations on the simulation model based on the first sample model data, and uses the calculated result data as the first calculation data.

[0056] This application embodiment uses a Python script with automatic submission functionality for job submission. For example, this Python script uses the Python paramiko library to automatically submit the first sample model data. It connects to a high-performance computing (HPC) server via SSH, checks the server's resource status every minute, and submits one or more first sample model data jobs to the server when there are available resources; otherwise, it waits. This automatic submission Python script can fully utilize the idle resources of the HPC computing server for simulation calculations. The entire calculation process requires no manual intervention, greatly improving the data computation efficiency.

[0057] In some embodiments, the drive post-processing software performs post-processing on the first calculated data to obtain first output data, including: based on the first calculated data, driving the post-processing software through the drive module to determine the first output data corresponding to the first calculated data according to preset rules; the first output data is the injury value corresponding to each part of the driver simulation model; and saving the first output data in the form of a file to the folder where the first input data is stored.

[0058] Once all seatbelt sample model data is completed and the first calculated data is obtained, a Python script-driven post-processing software is used to post-process the first calculated data. For example, the first calculated data can be stored as a Bionut data file. First, the Bionut data file is imported into the post-processing software. Then, curves for each body part are read sequentially from the Bionut data file according to the following order: head, neck, chest, thigh, and calf. The injury value for each body part is then calculated according to regulatory requirements. Finally, the injury value for each body part is saved as an Excel file in the folder where the first input data was stored. The post-processing results are as follows: Figure 6 As shown, Figure 6 The last column in the table shows the damage value for each body part.

[0059] The embodiments of this application use Python-driven post-processing software to automatically perform post-processing operations, reducing the workload of manual data reading and improving work efficiency.

[0060] In some embodiments, generating training data for the simulation model based on the mapping relationship between the first input data and the first output data includes: The key parameters in each set of sample files are used as input parameters for the simulation model, and the corresponding injury values ​​of each part of the driver simulation model are used as output parameters. The input parameters and the output parameters are integrated into the training parameters of the simulation model.

[0061] Specifically, each set of input parameters and its corresponding output parameters are organized and saved in an Excel file, such as... Figure 5 The first 35 rows of the table contain input parameters, and the last 19 rows contain the output damage results of the driver's dummy. Each column represents a set of training data. This integrated Excel table can be used to train the driver's AI model for passenger insurance, generating an AI model of the damage values ​​of various parts of the driver's dummy under the influence of 35 input parameters.

[0062] This application also provides a training data generation device, which includes a driver module, preprocessing software, a computing server, postprocessing software, and a training data generation module.

[0063] The preprocessing software is configured to preprocess the first input data under the drive of the driving module to obtain the first sample model data. The computing server is configured to perform calculations on the first sample model data under the drive of the driving module to obtain first calculated data; The post-processing software is configured to perform post-processing on the first calculated data under the drive of the driving module to obtain the first output data; The training data generation module is configured to generate training data for the simulation model based on the mapping relationship between the first input data and the first output data.

[0064] In some embodiments, the above apparatus further includes: a first input data generation module, configured to: Determine the key parameters of the simulation model and the value range of each key parameter; Based on the key parameters and the range of values, a mathematical sampling strategy is used to expand the key parameters. A corresponding sample file is generated based on each expanded set of key parameters, and the sample file is used as the first input data.

[0065] In some embodiments, the simulation model is a passenger safety driver AI model; The key parameters include at least one of the following categories: airbag parameters, steering column parameters, seat belt module parameters, aiming point parameters, pretensioning and force limiting device parameters, instrument and seat parameters, and test condition parameters.

[0066] In some embodiments, the preprocessing software is further configured to: obtain the preset node ID number of the seat belt model; Update the location information corresponding to the node ID number based on the coordinate values ​​of the key parameters in each set of sample files; Based on the updated location information, the preprocessing software is automatically generated to generate a seat belt sample model for each sample file through a preset programming driver, and the seat belt sample model is used as the first sample model data.

[0067] In some embodiments, the computing server is further configured as follows: Monitor the usage of computing resources on the computing server; If idle resources are detected on the computing server, the first sample model data is automatically submitted to the computing server through the driver module, so that the computing server can perform calculations on the simulation model based on the first sample model data and use the calculated result data as the first calculation data.

[0068] In some embodiments, the post-processing software is further configured as follows: Based on the first calculated data, the post-processing software is driven by the driving module to determine the first output data corresponding to the first calculated data according to preset rules; the first output data is the injury value corresponding to each part of the driver simulation model. Save the first output data as a file to the folder where the first input data is stored.

[0069] The training data generation module is specifically used to: take the key parameters of each set of sample files as input parameters of the simulation model, take the damage values ​​corresponding to each part of the driver simulation model as output parameters, and integrate the input parameters and the output parameters into the training parameters of the simulation model.

[0070] This application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the method as described in any of the above embodiments.

[0071] This embodiment also provides a computer-readable storage medium having a computer program stored thereon, the computer program being loaded by a processor to perform the steps of any of the methods in the above embodiments.

[0072] In the embodiments of this application, the storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0073] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0074] The foregoing has provided a detailed description of a training data generation method, apparatus, electronic device, and storage medium provided in the embodiments of this application. Specific examples have been used 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 method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for generating training data, characterized in that, Implemented via a driver module, the method includes: The driver preprocessing software preprocesses the first input data to obtain the first sample model data; The driving computing server performs calculations on the first sample model data to obtain the first calculated data; The driver post-processing software performs post-processing on the first calculated data to obtain the first output data; Training data for the simulation model is generated based on the mapping relationship between the first input data and the first output data.

2. The method according to claim 1, characterized in that, Before the driver preprocessing software preprocesses the first input data to obtain the first sample model data, the method further includes: determining the key parameters of the simulation model and the value range corresponding to each key parameter; Based on the key parameters and the range of values, a mathematical sampling strategy is used to expand the key parameters. A corresponding sample file is generated based on each expanded set of key parameters, and the sample file is used as the first input data.

3. The method according to claim 2, characterized in that, The simulation model is a passenger safety driver AI model; The key parameters include at least one of the following categories: airbag parameters, steering column parameters, seat belt module parameters, aiming point parameters, pretensioning and force limiting device parameters, instrument and seat parameters, and test condition parameters.

4. The method according to claim 3, characterized in that, The driver preprocessing software preprocesses the first input data to obtain the first sample model data, including: Obtain the preset node ID number of the seat belt model; Update the location information corresponding to the node ID number based on the coordinate values ​​of the key parameters in each set of sample files; Based on the updated location information, the preprocessing software is automatically generated to generate a seat belt sample model for each sample file through a preset programming driver, and the seat belt sample model is used as the first sample model data.

5. The method according to claim 1, characterized in that, The driving computing server performs calculations on the first sample model data to obtain first calculation data, including: Monitor the usage of computing resources on the computing server; If idle resources are detected on the computing server, the first sample model data is automatically submitted to the computing server through the driver module, so that the computing server can perform calculations on the simulation model based on the first sample model data and use the calculated result data as the first calculation data.

6. The method according to claim 3, characterized in that, The driver post-processing software performs post-processing on the first calculated data to obtain first output data, including: Based on the first calculated data, the post-processing software is driven by the driving module to determine the first output data corresponding to the first calculated data according to preset rules; the first output data is the injury value corresponding to each part of the driver simulation model. Save the first output data as a file to the folder where the first input data is stored.

7. The method according to claim 6, characterized in that, The step of generating training data for the simulation model based on the mapping relationship between the first input data and the first output data includes: The key parameters in each set of sample files are used as input parameters for the simulation model, and the corresponding injury values ​​of each part of the driver simulation model are used as output parameters. The input parameters and the output parameters are integrated into the training parameters of the simulation model.

8. A training data generation device, characterized in that, include: Driver module; The preprocessing software is configured to preprocess the first input data under the drive of the driving module to obtain the first sample model data; The computing server is configured to perform calculations on the first sample model data under the drive of the driving module to obtain first calculated data; The post-processing software is configured to perform post-processing on the first calculated data under the drive of the drive module to obtain the first output data; The training data generation module is configured to generate training data for the simulation model based on the mapping relationship between the first input data and the first output data.

9. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which is loaded by a processor to perform the steps of the method according to any one of claims 1 to 7.

Images