A finite element analysis result data automatic processing and screening method for proxy model construction
By using Python API scripts in Abaqus software to automatically extract and filter finite element analysis results and generate standardized datasets, the problem of low data processing efficiency in composite material structure design is solved, and the efficiency and accuracy of design optimization are improved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHAANXI UNIV OF SCI & TECH
- Filing Date
- 2026-02-06
- Publication Date
- 2026-06-05
AI Technical Summary
In the design of composite material structures, the post-processing of finite element analysis results in the existing technology suffers from low efficiency, error-proneness and difficulty in automating the batch processing of valid data, resulting in the dataset being mixed with invalid data, which affects the accuracy and reliability of the surrogate model.
Using Python API scripts based on Abaqus software, specific field variable data in finite element analysis results are automatically extracted, and invalid data is automatically filtered out by material failure criteria to generate standardized datasets. This includes creating data extraction scripts, batch control programs, and data conversion modules, achieving fully automated processing.
It achieves efficient and automated processing of finite element simulation post-processing data, ensuring the reliability and consistency of the dataset, and improving the efficiency and accuracy of composite material structure design optimization.
Smart Images
Figure CN122154289A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of computer-aided engineering and data preprocessing technology, and relates to an automated processing and screening method for finite element analysis results data used in proxy model construction. Background Technology
[0002] In high-end equipment manufacturing fields such as aerospace and new energy vehicles, composite materials are widely used due to their excellent specific strength, specific stiffness, and designability. Computer simulation based on the finite element method has become a core tool for the design, performance analysis, and optimization of composite material structures. A typical optimization process usually involves parametric modeling, simulation calculation, and post-processing analysis of dozens or even hundreds of design samples to construct a surrogate model that can quickly predict the structural response, thereby replacing time-consuming direct simulation for design exploration.
[0003] The post-processing data extraction stage faces significant challenges. First, extracting key physical field data (such as stress, strain, and user-defined state variables) at integration points under specific analysis steps and frames (e.g., peak load time) from the results files generated by finite element software currently relies heavily on engineers manually querying, filtering, and exporting data through the software interface. This manual approach is extremely inefficient when processing large numbers of samples and is prone to errors due to operator fatigue or negligence. Second, failure analysis of composite material structures is crucial, and their validity is typically determined by a set of user-defined state variables. Traditional methods struggle to automatically and batch-filter data for validity based on these failure criteria while extracting data, resulting in invalid or non-physical data points mixed in the dataset used to build surrogate models, severely impacting the model's accuracy and reliability. Finally, manually processed data often has inconsistent formats, requiring further tedious sorting and formatting to meet the input data standardization requirements for machine learning or surrogate model training.
[0004] Therefore, existing technologies lack a systematic solution capable of automatically and in batches extracting effective field variable data from finite element analysis results, integrating material failure criteria for intelligent filtering, and ultimately outputting a standardized dataset. This has become a key bottleneck restricting the efficiency and accuracy of simulation-based composite material structure design optimization. Summary of the Invention
[0005] The purpose of this invention is to provide an efficient, automatic, and reliable method for automated processing and filtering of finite element analysis result data for surrogate model construction. This method can automatically extract, filter, and process specific field variable data from finite element analysis result files. It is applicable to the post-processing of composite material structure simulation and provides a high-quality, standardized training dataset for surrogate model construction.
[0006] This invention is achieved through the following technical solution: An automated method for processing and filtering finite element analysis results data used in proxy model construction includes the following steps: Step 1: Create a data extraction script for a single finite element analysis result file, configure the data extraction parameters, perform automated data extraction on the single finite element analysis result file, automatically filter out the valid integration point data, and output a single structured report file; Step 2: Construct batch automated control program scripts; The batch automated control program includes a directory traversal and dynamic script generation module, a task execution module, a data conversion and verification module, and a key value extraction module; The directory traversal and dynamic script generation module automatically traverses multiple samples and dynamically modifies the file path parameters and target frame number parameters to generate a temporary script for the current result file. The task execution module calls the data extraction script created in step one to complete the data extraction and generate structured report files corresponding to multiple finite element analysis results; The traversal and dynamic script generation module and the task execution module enable unattended batch processing of multiple result files; The data conversion and verification module converts the generated structured report file into a common data format file; The key value extraction module is used to extract the feature values of preset physical quantities from the valid data of each result file; Step 3: Collect all feature values and corresponding valid integration point data obtained in Step 2; organize the collected data according to the preset data structure to generate a standardized dataset file for training the surrogate model.
[0007] The present invention also has the following technical features: Preferably, the data extraction script in step one is a Python API script file based on the Abaqus software. The script file imports the necessary module libraries, including abaqus, abaqusConstants, caeModules, and driverUtils, to establish a runtime environment that interacts with Abaqus and to process multiple finite element analysis result files in a loop through a batch control program.
[0008] Preferably, the method for configuring data extraction parameters in step one is as follows: using the writeFieldReport() function, data is extracted from a specified analysis step and a specific result frame, the output position is specified as the integration point, and the sorting basis is the cell label.
[0009] Preferably, the automated data extraction process in step one includes opening a specified result file through the script and extracting multiple field variable data at the integration point positions of the user-specified frame; the extracted field variable data includes response variable data for constructing the surrogate model and one or more user-defined state variable data for judging the validity of the data; based on the values of the user-defined state variable data, valid integration point data are automatically filtered out; and the valid integration point data is output as a structured report file.
[0010] Preferably, the user-defined state variables include: a first state variable for identifying fiber tensile failure, a second state variable for identifying fiber compressive failure, a third state variable for identifying matrix tensile failure, and a fourth state variable for identifying matrix compressive failure. The automatic filtering process includes: retaining the data of the integration point when the user-defined status variable data indicates that the corresponding integration point material is not invalid; and removing the data of the integration point when the indication is invalid.
[0011] Preferably, the characteristic value of the preset physical quantity mentioned in step two is the maximum value, including at least one of the following: the maximum value of equivalent stress, the maximum value of displacement, structural weight, and the maximum value of user-defined state variables.
[0012] Preferably, the standardized dataset file described in step three contains multiple key output variables; The key output variables include: the maximum equivalent stress of the structure under tensile load, the maximum displacement of the structure under tensile load, the maximum equivalent stress of the structure under internal compressive load, the maximum displacement of the structure under internal compressive load, and the weight of the structure.
[0013] Compared with existing technologies, the present invention has the following superior effects: This invention achieves highly efficient and automated processing of the entire process of finite element simulation post-processing data preparation by scripting data extraction rules, automating validity judgment, batching processing procedures, and standardizing output results. It overcomes the inefficient manual file-by-file and integration point-by-integral operation mode and can complete the data processing of dozens or even hundreds of samples with one click, greatly improving efficiency. This invention automatically filters invalid data through built-in material failure criteria (state variables), ensuring the reliability of the dataset used for subsequent modeling; at the same time, the automated process eliminates human error, ensures the consistency of data processing rules and the repeatability of results, and guarantees data quality and consistency. This invention standardizes and modularizes the conversion process from the original result file to the final modeling dataset, forming a standardized data pipeline with a unified output data format, which greatly facilitates the connection with downstream proxy model training, optimization design and other stages. Attached Figure Description
[0014] Figure 1 The flowchart shows the data automated extraction and processing method for the invention. Detailed Implementation
[0015] The present invention will be further described in detail below with reference to specific embodiments. These descriptions are for explanation purposes only and are not intended to limit the scope of the invention. This invention addresses the problem of inefficiency and error-proneness in manually extracting and filtering valid field variable data from a large number of result files during the post-processing of finite element analysis of composite material pipe structures. It provides an automated, batch method for extracting specific frame field variable data. This method not only automatically extracts data but also automatically filters invalid data based on material failure criteria, ultimately outputting a standardized dataset for constructing a high-precision surrogate model.
[0016] Please see Figure 1 The complete method provided in this embodiment of the invention includes the following steps: S1: Create and configure the single-result file data extraction script (Output_rpt_data.py).
[0017] First, create a script file based on the Abaqus software Python API. This script first imports the necessary module libraries, including abaqus, abaqusConstants, caeModules, and driverUtils, to establish a runtime environment for interacting with Abaqus.
[0018] Next, specify the path to the ODB result file to be processed in the script. For example, for the sample numbered i (i=1,2,3…50), the corresponding two result file paths are D: / Sample_i / iJob-Force.odb and D: / Sample_i / iJob-Pressure.odb. Open the specified ODB file using the openOdb() function to obtain basic model information, such as assemblies, part instances, number of analysis steps, etc., and establish an ODB object handle for data connection.
[0019] Next, configure the data extraction parameters. Using the `writeFieldReport()` function, specify the data extraction source for the specified analysis step (analysis step 1) and a specific result frame (frame 200, representing the moment the load reaches steady state or peak). The output location is specified as the integration point, and the sorting is based on the cell label.
[0020] The single-result-file data extraction script performs automated data extraction on a single finite element analysis result file, automatically filters out valid integration point data, and outputs a single structured report file; The automated data extraction process includes: opening a specified result file via a script and extracting multiple field variable data at integration point locations from the user-specified frame; the extracted field variable data includes response variable data for constructing a proxy model and one or more user-defined state variable data for determining data validity; automatically filtering out valid integration point data based on the values of the user-defined state variable data; and outputting the valid integration point data as a structured report file.
[0021] User-defined state variables include: a first state variable for identifying fiber tensile failure, a second state variable for identifying fiber compressive failure, a third state variable for identifying matrix tensile failure, and a fourth state variable for identifying matrix compressive failure. The automatic filtering process includes: retaining the data of the integration point when the user-defined status variable data indicates that the corresponding integration point material is not invalid; and removing the data of the integration point when the indication is invalid.
[0022] The core of this step is to extract the following two types of field variables: 1. Response variables: Key physical quantities used for subsequent surrogate model construction, including: S.Mises (equivalent stress), U (displacement deformation), and structural weight; 2. State variables: User-defined variables used to determine the validity of data. In this embodiment, there are four criteria: SDV1: fiber tensile failure, SDV2: fiber compression failure, SDV3: matrix tensile failure, and SDV4: matrix compression failure.
[0023] In composite material failure analysis, if any SDV value at an integration point is equal to 1, it indicates that the material has failed at that point, and all data at that point are considered invalid; only when all SDV values are 0 are the data at that integration point considered valid.
[0024] Finally, the script writes the extracted data of the above 6 field variables (S.Mises, U, SDV1-SDV4) into a single structured report file in text format.
[0025] S2: Build a batch automated processing framework.
[0026] To improve the processing efficiency for multiple samples (e.g., 50 samples with different design parameters), this invention constructs an external batch control program (written using Python scripts). This program includes the following modules: Directory traversal and dynamic script generation module: The program automatically traverses all sample calculation directories numbered from 1 to 50. For the two files Job-Force.odb and Job-Pressure.odb in each sample directory, it dynamically modifies the file path parameters and target frame number parameters in the template script (Output_rpt_data.py) prepared in step S1 to generate a temporary Python script for the current result file.
[0027] Task execution module: Calls the Abaqus command-line interface to execute the temporary script generated in the previous step, thereby driving the Abaqus kernel to complete data extraction and generate a corresponding structured report for each result file.
[0028] Data Conversion and Validation Module: Converts the generated structured report file into a more universal CSV (comma-separated values) format file for easier reading and processing with Excel. Data format validation is performed simultaneously during the conversion process.
[0029] Key value extraction module: From each valid CSV file (i.e., after SDV judgment, the file contains at least some valid integration point data), extract the maximum value of preset physical quantities, which typically include: the maximum value of equivalent stress, the maximum value of displacement, structural weight, and the maximum value of user-defined state variables. In this embodiment, it specifically includes: the maximum value of S.Mises equivalent stress, the maximum value of U displacement deformation, and the maximum values of SDV1-SDV4 (used to monitor the failure state of the most dangerous point), and saves these maximum values as structured text files (such as .txt or .csv).
[0030] S3: Effective dataset organization and output.
[0031] This step aims to integrate the aforementioned results to form the final dataset used to train the Kriging agent model.
[0032] First, the program archives and stores all intermediate process files (RPT and CSV files) and result files (maximum value files) of all samples according to the original sample directory structure, generating a standardized dataset file for easy tracking and review. The standardized dataset file contains several key output variables; Key output variables include: the maximum equivalent stress of the structure under tensile load, the maximum displacement of the structure under tensile load, the maximum equivalent stress of the structure under internal compressive load, the maximum displacement of the structure under internal compressive load, and the weight of the structure.
[0033] Then, the program reads all the valid data extracted from the samples. For each sample, under both tensile and internal pressure load conditions, the input and output variables used for the surrogate model are extracted or calculated from its valid integration point data. Finally, the data from all samples are compiled into a master table (an Excel file named surrogate model dataset.xlsx).
[0034] The core output variables of this dataset are shown in the table below: Table 1 Output Variables These output variables, together with the sample's design parameters (ply angle, thickness, etc., as input variables), constitute a complete "input-output" dataset, which is used to train a high-precision Kriging surrogate model for rapid design optimization and performance prediction.
[0035] In summary, the embodiments of the present invention achieve efficient and reliable preparation of surrogate model training data from massive finite element results through automated script extraction, data filtering based on failure criteria, batch process control, and standardized data integration, significantly improving the efficiency and reliability of composite material structure simulation analysis and optimization design.
Claims
1. A method for automated processing and filtering of finite element analysis results data for surrogate model construction, characterized in that, Includes the following steps: Step 1: Create a data extraction script for a single finite element analysis result file, configure the data extraction parameters, perform automated data extraction on the single finite element analysis result file, automatically filter out the valid integration point data, and output a single structured report file; Step 2: Construct batch automated control program scripts; The batch automated control program includes a directory traversal and dynamic script generation module, a task execution module, a data conversion and verification module, and a key value extraction module; The directory traversal and dynamic script generation module automatically traverses multiple samples and dynamically modifies the file path parameters and target frame number parameters to generate a temporary script for the current result file. The task execution module calls the data extraction script created in step one to complete the data extraction and generate structured report files corresponding to multiple finite element analysis results; The traversal and dynamic script generation module and the task execution module enable unattended batch processing of multiple result files; The data conversion and verification module converts the generated structured report file into a common data format file; The key value extraction module is used to extract the feature values of preset physical quantities from the valid data of each result file; Step 3: Collect all feature values and corresponding valid integration point data obtained in Step 2; organize the collected data according to the preset data structure to generate a standardized dataset file for training the surrogate model.
2. The method for automated processing and filtering of finite element analysis results data for surrogate model construction according to claim 1, characterized in that, The data extraction script mentioned in step one is a Python API script file based on the Abaqus software. The script file imports the necessary module libraries including abaqus, abaqusConstants, caeModules, and driverUtils to establish a runtime environment that interacts with Abaqus, and processes multiple finite element analysis result files in a loop through a batch control program.
3. The method for automated processing and filtering of finite element analysis results data for surrogate model construction according to claim 1, characterized in that, The method for configuring data extraction parameters described in step one is as follows: using the writeFieldReport() function, data is extracted from a specified analysis step and a specific result frame, with the output position specified as the integration point and the sorting basis as the cell label.
4. The method for automated processing and filtering of finite element analysis results data for proxy model construction according to claim 1, characterized in that, The automated data extraction process described in step one includes: opening the specified result file through the script and extracting multiple field variable data at the integration point positions of the user-specified frame; the extracted field variable data includes response variable data for constructing the surrogate model and one or more user-defined state variable data for judging the validity of the data; automatically filtering out valid integration point data based on the values of the user-defined state variable data; and outputting the valid integration point data as a structured report file.
5. The method for automated processing and filtering of finite element analysis results data for surrogate model construction according to claim 1 or 4, characterized in that, The user-defined state variables include: a first state variable for identifying fiber tensile failure, a second state variable for identifying fiber compressive failure, a third state variable for identifying matrix tensile failure, and a fourth state variable for identifying matrix compressive failure. The automatic filtering process includes: retaining the data of the integration point when the user-defined status variable data indicates that the corresponding integration point material is not invalid; and removing the data of the integration point when the indication is invalid.
6. The method for automated processing and filtering of finite element analysis results data for proxy model construction according to claim 1, characterized in that, The characteristic value of the preset physical quantity mentioned in step two is the maximum value, including at least one of the following: the maximum value of equivalent stress, the maximum value of displacement, structural weight, and the maximum value of user-defined state variables.
7. The method for automated processing and filtering of finite element analysis results data for surrogate model construction according to claim 1, characterized in that, The standardized dataset file mentioned in step three contains several key output variables; The key output variables include: the maximum equivalent stress of the structure under tensile load, the maximum displacement of the structure under tensile load, the maximum equivalent stress of the structure under internal compressive load, the maximum displacement of the structure under internal compressive load, and the weight of the structure.