Simulation pre-processing method and device, storage equipment and medium

Abstract

The invention discloses a simulation pretreatment method and device, storage equipment and a medium, firstly, mechanical property data of a to-be-simulated part at different spatial positions is obtained, and the data comprises a hardening curve defining material plasticity behaviors and fracture strain defining material failure; then, the mechanical property data are endowed to each finite element analysis grid unit of the simulation model through a mapping algorithm, so that each unit has independent mechanical property parameters; then, according to a preset partitioning criterion, all the units are automatically graded, the units belonging to the same level are merged into a logic sub-part, and a uniform material attribute is allocated to each logic sub-part; and finally, outputting a simulation model file containing the sub-part partitions and the corresponding material attributes thereof. According to the method, refined and automatic simulation pretreatment of parts with spatial heterogeneous mechanical properties is realized, and the accuracy and simulation efficiency of a simulation model are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of engineering simulation, in particular to a simulation preprocessing method and device, a storage device and a medium. BACKGROUND

[0002] In the performance analysis of parts based on finite element simulation, especially in parts formed by processes such as die-casting aluminum and injection molding, due to the non-uniformity of the temperature field and flow field in the manufacturing process, the mechanical properties of different parts of the same part often show significant differences. In addition, due to process fluctuations, the mechanical properties of parts between different production batches are also discrete. Traditional simulation methods usually rely on material cards obtained by sampling and testing after the parts are completed, but this method has obvious limitations: the material cards have poor reusability and are difficult to apply to simulation requirements of different regions of the same part or different batches; testing needs to be performed after the parts are completed, which causes simulation to lag far behind the design process; at the same time, it is difficult and costly to obtain standard samples from the parts, which further restricts the timeliness and accuracy of simulation analysis.

[0003] When applying material mechanical property data to collision simulation, it is usually necessary to map the material properties obtained by experiments or other simulations (such as mold flow simulation) to the grid model of the finite element software (such as LSDYNA). However, due to the inconsistency of the mold flow simulation grid and the collision simulation grid in terms of cell type, size, number, and topological structure, the material property mapping process across platforms and grid types is complex and has poor universality. Existing mapping methods usually rely on manual adjustment or simplified approximation, which makes it difficult to ensure accurate transmission of mechanical property distribution information in the simulation model, thereby affecting the reliability of the simulation results.

[0004] In addition, in the simulation considering the heterogeneous mechanical properties of parts, it is usually necessary to divide the parts into regions and assign different material properties to different regions. The traditional method relies on manual partitioning based on human experience, which is rough and subjective, and cannot accurately depict the continuous gradient change of the mechanical properties inside the part. This manual partitioning method not only has low efficiency, but also easily introduces human errors, which makes the simulation model unable to truly reflect the non-uniform deformation and failure behavior of the part under actual load, limiting the further improvement of the accuracy of collision simulation. SUMMARY

[0005] The present application aims to overcome the existing technical defects and provides a simulation preprocessing method and device, a storage device and a medium, which realizes fine and automatic simulation preprocessing of parts with spatially heterogeneous mechanical properties, and improves the accuracy of the simulation model and the simulation efficiency.

[0006] The present application achieves the above-mentioned purpose by means of the following technical solutions: In a first aspect, the application provides a simulation preprocessing method, comprising: S1: obtaining mechanical property data of the part to be simulated at different spatial positions, the mechanical property data including a hardening curve for defining the plastic behavior of the material and a fracture strain for defining the failure of the material; S2: mapping the mechanical property data to each finite element analysis grid of the simulation model through a mapping algorithm, so that each element has independent hardening curve and fracture strain parameters; S3: automatically classifying all elements of the simulation model according to a preset partition criterion, and merging elements belonging to the same class into a logical sub-part, and assigning a uniform material property to each logical sub-part; S4: outputting a simulation model file containing sub-part partition and corresponding material properties.

[0007] In a possible implementation, the method for obtaining mechanical property data includes any one or more of the following: obtained by sampling and physical testing of the part to be simulated; obtained based on the material performance chart output by the mold flow simulation software; obtained based on a mechanical property prediction model constructed through machine learning.

[0008] In a possible implementation, the step of obtaining mechanical property data by sampling and physical testing of the part to be simulated includes: preparing a flat plate reference part and performing a standard uniaxial tensile test to obtain a reference stress-strain curve and calculate the yield ratio; cutting a sample from the part to be simulated for mechanical testing, and simultaneously obtaining a force-time curve and a strain field data based on digital image correlation method; synchronously processing the force-time curve and the strain field data to obtain an engineering stress-gauge length curve for each sample; determining the tensile strength according to the engineering stress-gauge length curve, and calculating the corresponding yield strength using the yield ratio, and retaining the section after the yield strength on the curve; based on the reference stress-strain curve, obtaining a reference equivalent plastic strain-equivalent stress curve through reverse calculation; adjusting the vertical coordinate scaling factor of the reference equivalent plastic strain-equivalent stress curve through simulation iteration, so that the force-gauge length curve output by the simulation best coincides with the retained section, and using the optimal scaling factor to correct the reference curve to obtain the hardening curve corresponding to the sample position; based on the strain field data, obtaining the local fracture strain of the sample center, and comparing it with the reference fracture strain of the reference part to obtain the fracture strain scaling factor corresponding to the sample position; For the positions on the part which are not directly sampled, the inverse distance weighted spatial interpolation algorithm is used to obtain the hardening curve and fracture strain information based on the data of the measured points.

[0009] In a possible implementation, the step of obtaining the mechanical property data based on the mechanical property prediction model constructed through machine learning comprises: The multi-source feature data of the target material micro-zone is input into the mechanical property prediction model, and the mechanical property data of the corresponding micro-zone is output, the multi-source feature data comprising: the temperature history of the micro-zone in the forming process, the average porosity of the micro-zone, the average grain size, the average chemical composition, and the grain orientation distribution function, and the output mechanical property data comprising the hardening curve and / or the fracture strain.

[0010] In a possible implementation, the partition criterion in step S3 is set based on any one or more of the following indicators related to the mechanical property parameters independent of the unit: Based on the energy absorbed by the unit when it fractures, the energy absorbed by the unit when it fractures is calculated by integrating the constitutive curve of the unit from zero strain to its fracture strain; Based on the initial yield strength of the unit, the stress value corresponding to zero strain or a preset proportion of strain on the constitutive curve of the unit is determined; Based on the fracture strain value independent of the unit.

[0011] In a possible implementation, the simulation model file format is an input file format suitable for any one of LS-DYNA, Abaqus or ANSYS finite element analysis software.

[0012] In a second aspect, the application provides a simulation preprocessing device, the device comprising: The acquisition module is configured to acquire the mechanical property data of the part to be simulated at different spatial positions, the mechanical property data comprising a hardening curve for defining the plastic behavior of the material and a fracture strain for defining the failure of the material; The mapping module is configured to map the mechanical property data to each finite element analysis grid of the simulation model through a mapping algorithm, so that each unit has independent hardening curve and fracture strain parameters; The hierarchical module is configured to automatically classify all units of the simulation model according to a preset partition criterion, and merge units belonging to the same level into a logical sub-part, and assign a uniform material property to each logical sub-part; The output module is configured to output a simulation model file containing the partition of the sub-part and the corresponding material property.

[0013] In a third aspect, the present application also provides a computer storage device, comprising a processor and a memory, wherein the memory stores a computer program, and the computer program is loaded and executed by the processor to implement the pre-simulation processing method according to any one of the first aspect.

[0014] In a fourth aspect, the present application also provides a computer readable storage medium, wherein the storage medium stores a computer program, and the computer program is loaded and executed by a processor to implement the pre-simulation processing method according to any one of the first aspect.

[0015] The above-mentioned main scheme of the present application and each further selected scheme can be freely combined to form multiple schemes, which are all the schemes that can be adopted and claimed by the present application; and the present application can also be freely combined between each non-conflicting selection and between and other selections. Those skilled in the art can understand that there are many combinations according to the prior art and common knowledge after understanding the schemes of the present application, which are all the technical schemes claimed by the present application, and are not listed here.

[0016] The present application discloses a pre-simulation processing method, device, storage device and medium. First, the mechanical property data of a part to be simulated at different spatial positions is obtained, which includes a hardening curve defining the plastic behavior of the material and a fracture strain defining the failure of the material. Then, the mechanical property data is assigned to each finite element analysis grid cell of the simulation model through a mapping algorithm, so that each cell has independent mechanical property parameters. Next, according to a preset partition criterion, all cells are automatically classified, cells belonging to the same level are merged into a logical sub-part, and a uniform material property is assigned to each logical sub-part. Finally, a simulation model file containing the sub-part partition and the corresponding material property is output. This method realizes fine and automatic pre-simulation processing of parts with spatially inhomogeneous mechanical properties, and improves the accuracy and simulation efficiency of the simulation model. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical schemes of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. Those skilled in the art can also obtain other related drawings without making creative efforts on the basis of these drawings.

[0018] Figure 1 A flowchart of a pre-simulation processing method according to an embodiment of the present application is shown.

[0019] Figure 2 A schematic diagram showing the differences in the division results that may be produced when automatically classifying cells according to different partition criteria is shown.

[0020] Figure 3 A schematic diagram of different sample yield strength and tensile strength is shown.

[0021] Figure 4 A pore distribution cloud chart measured by industrial CT is shown.

[0022] Figure 5 A schematic diagram of implementation effect in engineering software (LS-PrePost) of the application is shown. DETAILED DESCRIPTION

[0023] The embodiments of the application are described below by way of specific examples, and those skilled in the art can easily understand other advantages and effects of the application from the disclosure. The application can also be implemented or applied by other different specific embodiments, and various modifications or changes can be made to the details in the specification without departing from the spirit of the application. It should be noted that the following examples and features in the examples can be combined with each other without conflict.

[0024] Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0025] In the prior art, the following problems mainly exist: first, the mechanical properties of the same part are different at different positions due to the process characteristics of die-cast aluminum, injection molding and other parts, and the mechanical properties of different batches of parts are also different. The material card has poor reusability. The material card needs to be tested after the part is offline, and the simulation timeliness is poor. The material card needs to be obtained by sampling and testing on the part, and the sampling difficulty is high. Second, collision simulation is generally performed in LSDYNA, but the material mechanical properties are generally obtained through other ways, and the method of mapping the material mechanical properties to the LSDYNA simulation grid has poor universality. Third, in collision simulation, the part is generally artificially subdivided according to the mechanical property difference, but the manual method is not fine enough, and the gradient difference of the mechanical properties of the part cannot be accurately described.

[0026] Therefore, in order to solve the above technical problems, the embodiments of the application propose a simulation preprocessing method, device, storage equipment and medium, which realizes high-resolution and high-efficiency characterization of the spatial inhomogeneous mechanical properties of the part by using small-size samples and digital image correlation method for in-situ tensile testing to simultaneously and simultaneously obtain the hardening curve and fracture strain of the material in one integrated test.

[0027] Please refer to Figure 1 , Figure 1 A schematic diagram of the flow of a simulation preprocessing method according to an embodiment of the application is shown, and the method comprises: S1: Obtain the mechanical property data of the part to be simulated at different spatial locations, including hardening curves for defining the plastic behavior of the material and fracture strain for defining the failure of the material.

[0028] First, based on a special sampling tool, a micro specimen is cut from a specific area of interest of the part to be simulated. The specimen has a small size feature of about 25mm x 30mm, which is significantly smaller than a standard tensile specimen, allowing multiple specimens to be obtained on the same part, thereby greatly improving the density of spatial sampling points. Then, a precise directional tensile test is performed on the obtained specimen (aligning the specimen symmetry axis with the tensile direction). During the test, load-time data is recorded synchronously by a high-precision mechanical sensor, and at the same time, a digital image correlation (DIC) visual measurement system is used to collect and process high-speed images of the speckle pattern on the surface of the gauge section, thereby non-contact obtaining full-field strain distribution and evolution data during the entire deformation process.

[0029] The measured force-time curve is combined with the average strain of the gauge section or the true strain of a specific point obtained by DIC to draw the engineering or true stress-strain curve at that location. Through material constitutive model back-calculation and iterative simulation calibration, the accurate hardening curve for simulation can be determined. By analyzing the full-field strain data provided by DIC, the local true strain at the fracture instant, fracture initiation point or critical section of the specimen is accurately identified, which is the fracture strain of that material point.

[0030] The method for obtaining mechanical property data includes any one or more of the following: obtained by sampling and physical testing of the part to be simulated; obtained based on the material property contour map output by the mold flow simulation software; obtained based on a mechanical property prediction model constructed through machine learning.

[0031] The method for obtaining spatial mechanical property data of the part includes the following three complementary technical approaches: first, direct sampling and physical testing, using a special tool to cut a small size specimen from the part, and simultaneously obtaining high-precision local hardening curve and fracture strain through tensile testing combined with digital image correlation method; second, based on the output of process simulation, directly reading the material property spatial distribution contour map generated by the mold flow simulation software to quickly obtain the performance gradient information reflecting the process impact; third, intelligent prediction based on machine learning, by constructing a prediction model with material composition, process history and microstructure features as input and mechanical properties as output, to accurately infer the performance of untested areas.

[0032] The step of obtaining mechanical property data by sampling and physical testing of the part to be simulated includes: Prepare flat die reference parts and conduct standard uniaxial tensile test to obtain reference stress-strain curve and calculate yield ratio; Cut samples from the part to be simulated for mechanical test, and obtain force-time curve and strain field data based on digital image correlation method synchronously; Synchronously process force-time curve and strain field data to obtain engineering stress-gauge length curve of each sample; Determine tensile strength according to engineering stress-gauge length curve, and calculate corresponding yield strength using yield ratio, and retain the section after yield strength on the curve; Based on the reference stress-strain curve, obtain the reference equivalent plastic strain-equivalent stress curve by reverse method; Adjust the vertical coordinate scaling factor of the reference equivalent plastic strain-equivalent stress curve through simulation iteration to make the simulation output force-gauge length curve best coincide with the retained section, and use the optimal scaling factor to correct the reference curve to obtain the hardening curve corresponding to the sample position; Based on the strain field data, obtain the local fracture strain of the sample center, and compare it with the reference fracture strain of the reference part to obtain the fracture strain scaling factor of the corresponding sample position; For positions on the part that are not directly sampled, use inverse distance weighted spatial interpolation algorithm to obtain hardening curve and fracture strain information based on the data of measured points.

[0033] First, establish material performance benchmark. Prepare flat die standard reference parts with the same material and forming process as the part to be simulated, and conduct standard uniaxial tensile test to obtain the reference stress-strain curve of the material and calculate its yield ratio r. At the same time, use digital image correlation method (DIC) to measure the average strain along the tensile direction in a standard small area (about 1mm×1mm) on the reference part as the reference fracture strain of the material .

[0034] Secondly, conduct high-density sampling test of the part. Use special test tooling to cut small size samples (about 25mm×30mm) at different key positions of the part to be analyzed for tensile test, and synchronously collect force-time curve obtained through mechanical sensor and full-field strain evolution data covering the sample surface obtained through DIC system. In particular, set virtual extensometer in the center area of the sample in DIC analysis to obtain gauge length change, and define a small area in the center to monitor local strain concentration.

[0035] Afterwards, data synchronization and key feature extraction are performed. By time stamp alignment, the force-time curve is combined with the gauge length change data obtained by DIC to calculate the engineering stress-gauge length curve of the specimen. The tensile strength is determined from the curve, and the yield strength is calculated using the yield ratio r. Then, the section of the curve from the yield point to the fracture is intercepted, which reflects the plastic deformation behavior of the material.

[0036] Then, the hardening curve is accurately back-calculated. The effective experimental curve obtained above is taken as the target, and the back-calculation method is used for calibration. First, an initial reference equivalent plastic strain-equivalent stress curve is derived based on the reference stress-strain curve. This curve is used as input to establish a tensile simulation model of the specimen in finite element software such as LS-DYNA. By iteratively adjusting the vertical coordinate scaling factor of the reference curve , the simulation is run, and the simulation force-gauge length curve consistent with the DIC virtual extensometer position and length is extracted. When the simulation curve and the experimental intercepted section reach the best coincidence, the optimal scaling factor is recorded. The initial reference curve is scaled by this factor to obtain the high-fidelity equivalent plastic strain-equivalent stress curve corresponding to the spatial position of the specimen, i.e., the required hardening curve.

[0037] Then, the fracture strain parameter is determined. The average strain in the tensile direction of a small region at the center of the specimen at the time of fracture is extracted from the local strain field data obtained by DIC testing of the specimen, serving as the measured local fracture strain at this position. By comparing this measured value with the reference fracture strain of the reference part, a fracture strain scaling coefficient is calculated, which is used to adjust the fracture criterion at this position when the fracture model is known.

[0038] Finally, the construction of the spatial performance field is realized. Through the above steps, the accurate hardening curve and fracture strain information of multiple sampled discrete points on the part are obtained. For any position on the part that is not directly sampled, spatial interpolation algorithms such as inverse distance weighting are used to calculate the corresponding hardening curve and fracture strain information based on the data of these discrete points, thereby constructing a continuous and non-homogeneous mechanical performance database covering the entire part space.

[0039] The step of obtaining mechanical property data based on a mechanical property prediction model constructed through machine learning includes: Inputting multi-source feature data of a target material micro-region into the mechanical property prediction model to output mechanical property data of the corresponding micro-region, the multi-source feature data including: temperature history of the micro-region in the forming process, average porosity, average grain size, average chemical composition, and grain orientation distribution function of the micro-region, and the output mechanical property data including a hardening curve and / or a fracture strain.

[0040] A machine learning model is constructed, which, when applied, takes as input a plurality of source feature data of a micro-zone of a target material, including: a key temperature history experienced by the micro-zone during a forming process, its average porosity, average grain size, average chemical composition, and a grain orientation distribution function describing texture characteristics, etc. After training, the model can directly output the corresponding key mechanical property data of the micro-zone, mainly including a hardening curve for defining plastic behavior and / or a fracture strain for defining failure. In the training data of the model, the input features can be obtained through mold flow simulation, industrial CT or infrared temperature measurement, etc., and the output performance data as labels is preferably obtained by high-precision direct sampling and physical testing methods.

[0041] S2: Map the mechanical property data to each finite element analysis grid of the simulation model through a mapping algorithm, so that each element has an independent hardening curve and fracture strain parameter.

[0042] A mapping algorithm is used, which can intelligently process the spatial relationship between source data and target simulation grids. Its core mechanism is: based on spatial search and interpolation technology, find the corresponding contribution point in the source data space for each target simulation grid element, and according to distance, shape function or other weighting criteria, reasonably distribute or interpolate the mechanical property parameters of the source data points to the target element.

[0043] S3: According to a predetermined partitioning criterion, automatically classify all elements of the simulation model, and merge elements belonging to the same level into a logical sub-part, and assign a uniform material property to each logical sub-part.

[0044] According to one or more partitioning criteria, automatically analyze and classify all elements in the simulation model that have completed mapping. These criteria are directly and objectively derived from the independent mechanical property parameters that each element has been assigned, and then according to the numerical size of the selected criteria, all elements are automatically classified into several discrete performance levels.

[0045] Then perform the core logical merging: all elements classified into the same performance level, whether continuous in geometric space or not, will be merged into a logical sub-part, which is a virtual, functional collection representing a group of material regions with high similarity in specific mechanical behavior. Finally, a set of uniform material properties is assigned to each logical sub-part.

[0046] The partitioning criterion in step S3 is based on any one or more of the following indicators related to the independent mechanical property parameters of the element: Based on the energy absorbed by the element when it fractures, calculated by integrating the element's constitutive curve from zero strain to its fracture strain. Based on the initial yield strength of the element, the stress value corresponding to zero strain or a preset proportion of strain on the element constitutive curve is determined; Based on the element independent fracture strain value.

[0047] The partition criterion can be set based on any one or more of the following physical indicators directly related to the element independent parameters: first, based on the energy absorbed by the element when it breaks. Under the premise that the complete stress-strain constitutive curve of each element and its corresponding fracture strain are known, the area enclosed by integrating each curve from zero strain to its fracture strain can be calculated by numerical integration method, and the area value is the total energy that the element can absorb during the fracture process. By calculating the energy values of all elements and statistically classifying them, element clustering according to energy absorption capacity can be achieved. Second, based on the initial yield strength of the element. The initial segment of the hardening curve of each element independently determines its yield behavior. The stress value corresponding to the zero plastic strain point or the 0.002 equivalent plastic strain point on the curve can be taken as the initial yield strength representation value of the element. By comparing and classifying the yield strengths of all elements, regions with similar material strength levels can be automatically classified into one category. Third, based on the element independent fracture strain value. Directly use the fracture strain parameter mapped to each element itself for classification. Sorting and classifying the fracture strain values of all elements in the model can identify and merge element regions with similar fracture toughness or ductility.

[0048] S4: Output a simulation model file containing sub-partition and its corresponding material properties.

[0049] After completing the automatic classification and logical merging of elements, all elements belonging to the same performance level are defined as a component of a logical sub-part. Then the system assigns a unified material property identifier to each logical sub-part, and encapsulates and writes the geometric information, connection relationship and corresponding unified material properties of all elements in the sub-part in the format of the preset finite element solver. A typical implementation is to output an input file format (.key or.k file) suitable for LS-DYNA software. The file contains complete information of the reconstructed part logical partition and the corresponding material parameters of each partition, thereby generating a pre-processing model file.

[0050] Figure 2The figure shows the difference in the division results that may be generated when the cells are automatically classified according to different partition criteria. The four curves in the figure respectively represent the independent stress-strain responses and the fracture points (end points of the curves) of the four finite element cells. If the classification is based on the fracture absorbed energy (area under the curve), cells 1, 2, and 4 can be classified into one logical sub-part because their energies are similar, and cell 3 is classified into another class. If the classification is based on the initial yield strength (the initial stress value of the curve), cells 2 and 3 can be combined because their strengths are similar, and cells 1 and 4 are combined into another class. If the classification is based on the fracture strain (the strain value of the end point of the curve), cells 1 and 3 are classified into one class, and cells 2 and 4 are classified into another class. According to different engineering analysis objectives, the corresponding mechanical indicators can be flexibly selected as the partition basis, and the logical sub-part division that best meets the analysis requirements can be intelligently generated, so that the heterogeneous characteristics of the part performance can be accurately and efficiently reflected in the simulation.

[0051] Figure 3 The figure shows the different sample yield strength and tensile strength diagrams proposed in the embodiments of the present application, including two engineering stress-gauge length curves obtained from different position sample tests. The highest point (peak stress) of each curve is the tensile strength of the sample and ). First, by testing the reference part of the standard flat plate mold, the stable yield strength to tensile strength ratio (r) of the material is obtained, which is defined as the ratio of the yield strength to the tensile strength. Based on the "fixed reference yield strength to tensile strength ratio" setting, that is, the formula , under the premise of knowing the tensile strength of each sample and ) and the reference yield strength to tensile strength ratio r, the yield strength of each sample for simulation and ) can be directly calculated.

[0052] Figure 4 The figure shows the pore distribution cloud map measured by industrial CT. Different color areas in the figure represent the pore volume (unit: mm³) of the material micro area at different spatial positions calculated after three-dimensional reconstruction of the industrial CT scanning. The continuous distribution and gradient change from low porosity (material density) to high porosity (material porosity) are clearly indicated. The micro area porosity information represented by the cloud map can be directly used as one of the input features of the machine learning prediction model to establish a quantitative correlation between process defects and macro mechanical properties, thereby supporting accurate prediction of the spatial mechanical properties of the part.

[0053] Figure 5The implementation effect schematic diagram of the application in engineering software (LS-PrePost) is shown. (a) illustrates a part model with symmetrically gradient distribution of mechanical properties along the length direction, the yield strength of both ends is 440 MPa, and the yield strength of the middle part is 72 MPa. (b) represents the dilemma of traditional pre-simulation: if no partition is performed, the whole part can only be defined as a component (Part ID 200), at this time, assigning a single material property (whether 72 MPa, 440 MPa or some average value therebetween) to it cannot accurately depict its real inhomogeneous performance, resulting in simulation distortion. (c) and (d) show the effect after applying the method of the application: according to the 'equal yield strength' criterion, the system automatically divides the part into 8 (c) and 13 (d) logical sub-parts (Part I, Part II…). Each sub-part contains units with similar mechanical properties and can therefore be assigned an accurate and uniform material property.

[0054] Compared with the prior art, the embodiments of the application have the following beneficial effects: Firstly, by fusing physical testing, process simulation and machine learning prediction, a high-resolution spatial inhomogeneous mechanical property database is constructed, so that the simulation input parameters truly reflect the performance gradient inside the part, and the simulation distortion problem caused by the traditional homogenization material card is fundamentally overcome.

[0055] Secondly, an intelligent mapping algorithm is adopted, which effectively solves the data transfer problem under the condition that the source data and the target collision simulation grid are inconsistent in type, size and topological structure, and ensures the accuracy and universality of the performance distribution information mapping.

[0056] Thirdly, by pre-setting objective criteria based on physical meaning, the system automatically completes intelligent grading and logical merging of a large number of grid units, completely replacing the traditional rough manual partitioning which relies on experience, and the processing efficiency is improved by orders of magnitude.

[0057] A possible implementation mode of a simulation pre-processing device for executing each execution step and corresponding technical effect of the simulation pre-processing method shown in the above embodiments and possible implementation modes is given below, and the device comprises: an acquisition module configured to acquire mechanical property data of a part to be simulated at different spatial positions, the mechanical property data comprising a hardening curve for defining plastic behavior of a material and a fracture strain for defining failure of the material; A mapping module is configured to map the mechanical property data to each finite element analysis grid of a simulation model by a mapping algorithm, so that each unit has independent hardening curve and fracture strain parameters; a hierarchical module, configured to automatically classify all units of the simulation model according to preset partition criteria, and merge units belonging to the same level into a logical subpart, and assign a uniform material attribute to each logical subpart; an output module, configured to output a simulation model file containing the subpart partition and the corresponding material attribute.

[0058] The preferred embodiment provides a computer storage device, which can implement the steps in any of the simulation preprocessing methods provided in the embodiments of the present application, and thus can achieve the beneficial effects of the simulation preprocessing methods provided in the embodiments of the present application. Details are described in the foregoing embodiments, which will not be repeated here.

[0059] Those skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructions, or by controlling relevant hardware by the instructions, which can be stored in a computer readable storage medium and loaded and executed by a processor. Therefore, the embodiments of the present application provide a storage medium, which stores a plurality of instructions that can be loaded by a processor to execute the steps in any of the simulation preprocessing methods provided in the embodiments of the present application.

[0060] The storage medium can include a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0061] Since the instructions stored in the storage medium can execute the steps in any of the simulation preprocessing methods provided in the embodiments of the present application, the beneficial effects that can be achieved by any of the simulation preprocessing methods provided in the embodiments of the present application can be achieved. Details are described in the foregoing embodiments, which will not be repeated here.

[0062] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A simulation preprocessing method, characterized in that, include: S1: Obtain the mechanical property data of the part to be simulated at different spatial locations. The mechanical property data includes the hardening curve used to define the plastic behavior of the material and the fracture strain used to define the failure of the material. S2: The mechanical property data is mapped to each finite element analysis mesh of the simulation model through a mapping algorithm, so that each element has an independent hardening curve and fracture strain parameters; S3: Based on the preset partitioning criteria, all units of the simulation model are automatically classified, and units belonging to the same level are grouped into a logical sub-part, and a uniform material property is assigned to each logical sub-part. S4: Outputs a simulation model file containing sub-part partitions and their corresponding material properties.

2. The simulation preprocessing method as described in claim 1, characterized in that, Methods for obtaining mechanical property data include one or more of the following: Obtained by sampling and physical testing of the parts to be simulated; Obtained based on material property cloud maps output by model flow simulation software; It is obtained based on a mechanical performance prediction model built through machine learning.

3. The simulation preprocessing method as described in claim 2, characterized in that, The steps for obtaining mechanical property data by sampling and physical testing of the parts to be simulated include: A flat plate mold reference part was prepared and a standard uniaxial tensile test was performed to obtain the reference stress-strain curve and calculate the yield strength ratio. A sample is cut from the part to be simulated for mechanical testing, and force-time curves and strain field data based on digital image correlation are obtained simultaneously. Synchronously process force-time curves and strain field data to obtain engineering stress-gauge length curves for each specimen; The tensile strength is determined based on the engineering stress-gauge length curve, and the corresponding yield strength is calculated using the yield ratio. The section after the yield strength on the curve is retained. Based on the reference stress-strain curve, the equivalent plastic strain-equivalent stress curve of the reference is obtained by inverse method; By iteratively adjusting the scaling factor of the ordinate of the equivalent plastic strain-equivalent stress curve of the reference, the force-gauge length curve output by the simulation is optimally coincident with the retained section. The reference curve is then corrected using the optimal scaling factor to obtain the hardening curve of the corresponding sample position. The local fracture strain at the center of the specimen is obtained based on the strain field data, and compared with the reference fracture strain of the reference part to obtain the fracture strain reduction factor at the corresponding specimen position. For locations on the part that were not directly sampled, an inverse distance weighted spatial interpolation algorithm was used to obtain the hardening curve and fracture strain information based on the data from the measured points.

4. The simulation preprocessing method as described in claim 2, characterized in that, The steps for obtaining mechanical performance data based on a mechanical performance prediction model built through machine learning include: The multi-source characteristic data of the target material micro-region is input into the mechanical property prediction model, and the mechanical property data of the corresponding micro-region is output. The multi-source characteristic data includes: the temperature history of the micro-region during the forming process, the average porosity, average grain size, average chemical composition, and grain orientation distribution function of the micro-region. The output mechanical property data includes hardening curve and / or fracture strain.

5. The simulation preprocessing method as described in claim 2, characterized in that, In step S3, the partitioning criteria are set based on any one or more of the following indices related to the element's independent mechanical performance parameters: Based on the energy absorbed when the element fractures, the strain is calculated from zero strain to its fracture strain using the integral element constitutive curve. Based on the initial yield strength of the element, the stress value at the corresponding zero strain or preset proportional strain on the element constitutive curve is determined. Based on the fracture strain value of the individual element.

6. The simulation preprocessing method as described in claim 1, characterized in that, The simulation model file format is an input file format suitable for any of the LS-DYNA, Abaqus, or ANSYS finite element analysis software.

7. A simulation preprocessing device, characterized in that, The device includes: The acquisition module is used to acquire the mechanical property data of the part to be simulated at different spatial locations. The mechanical property data includes the hardening curve used to define the plastic behavior of the material and the fracture strain used to define the failure of the material. The mapping module is used to map mechanical property data to each finite element analysis mesh of the simulation model through a mapping algorithm, so that each element has an independent hardening curve and fracture strain parameters. The grading module is used to automatically grade all units of the simulation model according to preset zoning criteria, group units belonging to the same level into a logical sub-part, and assign uniform material properties to each logical sub-part. The output module is used to output simulation model files containing sub-part partitions and their corresponding material properties.

8. A computer storage device, characterized in that, The computer storage device includes a processor and a memory, wherein the memory stores a computer program, which is loaded and executed by the processor to implement the simulation preprocessing method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which is loaded and executed by a processor to implement the simulation preprocessing method as described in any one of claims 1-6.

Images