A multi-scale finite element simulation method based on EBSD grain orientation bolt roll-R forming

Abstract

The application discloses a kind of multi-scale finite element simulation methods for EBSD grain orientation bolt roll R forming, and the method belongs to the intersection field of precision plastic forming and computational materials science.For the problems such as the real microstructure of material being ignored by existing macro finite element simulation, it is difficult to accurately predict the mechanical response of local deformation zone of roll R forming, etc., the application provides a whole-process optimization scheme.The system first collects bolt material EBSD data to obtain grain orientation, reconstructs micro polycrystal model, then constructs cross-scale constitutive model based on Taylor factor and critical shear stress correction, then sets up macro-micro displacement field transmission coupling mechanism and introduces dynamic contact stiffness and closed-loop optimization strategy, finally the system completes multi-scale simulation and accurate prediction of residual compressive stress in roll forming process.This method can effectively show grain slip, texture evolution and residual stress gradient distribution in roll R forming, and is mainly applied to high-strength bolt roll R process parameter optimization and digital control of forming quality.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of precision plastic forming and computational materials science, specifically involving a multi-scale finite element simulation method for EBSD grain-oriented bolt rolling R-forming. Background Technology

[0002] Bolts are the most widely used fasteners in mechanical structures. For TC4 titanium alloy bolts, their load-bearing capacity and fatigue life directly affect the overall service safety of equipment. Under alternating loads, the thread root and shank transition fillets of bolts are often areas of severe stress concentration and weak points where fatigue cracks are most likely to initiate. The roll forming process introduces high residual compressive stress into the bolt fillet surface, inducing plastic deformation of the surface material and forming continuous metal flow lines. This improves the bolt's fatigue resistance and stress corrosion resistance, making it an indispensable core process in the manufacturing of high-strength bolts such as TC4 titanium alloy bolts.

[0003] With the rapid development of digital manufacturing technology, finite element simulation has become an important means to optimize rolling radius process parameters and predict forming quality. However, most current simulations of bolt rolling are limited to the macroscopic mechanical level, based on the continuum assumption and classical elastoplastic constitutive model. They neglect the influence of grain morphology, orientation, and grain boundary distribution within the material on local deformation, making it difficult to accurately describe microscopic processes such as crystal slip, lattice rotation, and texture evolution in large-gradient deformation regions. This results in insufficient accuracy of simulation results in certain aspects. Electron backscatter diffraction (EBSD) technology can obtain real crystallographic information inside the bolt, but how to effectively integrate microscopic orientation data into finite element calculations and establish a multi-scale coupling mapping between macroscopic processes and microscopic deformations remains a bottleneck. This invention introduces real grain orientation information on the existing basis to construct a cross-scale simulation model, which can better improve the understanding of the material deformation mechanism during the rolling radius forming of TC4 titanium alloy bolts. This can also provide more rigorous theoretical support for achieving high-performance manufacturing of fasteners. Summary of the Invention

[0004] In view of this, this invention addresses the shortcomings of existing macroscopic finite element simulation methods that are detached from the real microscopic crystal structure and have difficulty in accurately predicting the influence of grain morphology changes and texture evolution on the mechanical response of local deformation zones. It proposes a multi-scale finite element simulation method based on EBSD grain orientation bolt rolling R-forming.

[0005] A multi-scale finite element simulation method based on EBSD grain orientation bolt rolling R-forming is proposed, and its specific implementation steps are summarized as follows:

[0006] Step 1: Establish the macroscopic geometric model and mesh generation for bolt roll forming.

[0007] Step 2: Acquire EBSD data and characterize the microstructure of the bolt material.

[0008] Step 3: Reconstructing the microscopic polycrystalline model based on EBSD grain orientation

[0009] Step 4: Constructing a multi-scale material constitutive model and calibrating parameters

[0010] Step 5: Establish a cross-scale information transmission and coupling mechanism between macro and micro scales.

[0011] Step Six: Set machining process parameters and kinematic boundary conditions

[0012] Step 7: Perform multi-scale finite element analysis and process simulation.

[0013] Step 8: Analysis of Macro- and Micro-level Multiphysics Fields and Grain Evolution Results

[0014] Step 9: Closed-loop optimization of process parameters for optimal residual compressive stress.

[0015] Furthermore, in step one, a macroscopic three-dimensional solid model is established based on the actual geometric dimensions of the bolt and the design parameters of the rolling wheel. The local mesh of the R-angle region at the root of the thread is refined. The macroscopic finite element mesh type and element attributes are defined in the software, and the friction contact pairs of the macroscopic interface are set.

[0016] Furthermore, in step two, the system uses electron backscatter diffraction (EBSD) technology to obtain data on the microstructure, grain boundary distribution, and initial grain orientation of the bolt material to be processed. After optimization, these data are used to characterize key topological and crystallographic information of microstructure features. The obtained Euler angles are converted into rotation matrices to describe the spatial mapping relationship between the crystal coordinate system and the sample coordinate system.

[0017] Furthermore, in step three, the preprocessed EBSD data is imported into a finite element system. Using the Voronoi polygon algorithm, the centroid extracted by EBSD is used as the lattice seed point to generate a polycrystalline model within a defined representative volume element (RVE) region. The system uniformly assigns the average orientation Euler angle of each independent grain measured by EBSD to all geometric mesh elements contained within that grain, treating the initial orientation within the grain as a uniform distribution. Additionally, the crystal plasticity finite element analysis (CPFEM) theory is introduced to represent the total crystal deformation gradient. By performing a multiplicative decomposition of elasticity and plasticity, and mapping the spatial rotation matrix to each Gaussian integration point using the slip system matrix, the following microscopic dynamic evolution control equations are constructed:

[0018] In the formula, and These represent the elastic and plastic deformation gradients, respectively. For plastic velocity gradient, The total number of slip systems activated by the target material. , and The first Shear strain rate, slip direction vector, and slip surface normal vector of a slip system For shear stress, For slip resistance, The strain rate sensitivity index The reference shear strain rate is used.

[0019] Furthermore, in step four, the system establishes a cross-scale constitutive model based on the correction of EBSD grain orientation within the bolt. Taylor factors are extracted from the EBSD data. With critical shear stress By replacing the static yield term in the macroscopic constitutive equation with microscopic physical quantities, a modified multi-scale flow stress model is established:

[0020] In the formula, For equivalent flow stress, The strain hardening coefficient is... The hardening index, For equivalent plastic strain, The strain rate sensitivity coefficient, For equivalent plastic strain rate, For reference strain rate, Dimensionless temperature, which has a clear definition And constrained to the interval established within, The current temperature. The ambient room temperature, The melting point temperature of the material. This represents the thermal softening index. This method enables cross-scale mapping of microscopic grain orientation parameters to macroscopic mechanical responses.

[0021] Furthermore, in step five, a cross-scale boundary condition driving mechanism is established using sub-model technology. Historical displacement or strain field data of the macroscopic model during processing are extracted and transferred to the boundary nodes of the microscopic polycrystalline model through shape function interpolation, thereby realizing cross-scale data transfer.

[0022] Furthermore, in step six, processing parameters and kinematic boundary conditions are set, and a dynamic boundary mechanics solution model considering the equipment-workpiece coupling stiffness is established. The entire system adopts an incremental linearization approximation method to approximate the system's mechanical stiffness coefficients. Dynamic contact stiffness coefficient that is considered constant with respect to the workpiece material within a single computational increment step. Series coupling, its comprehensive yield equivalent stiffness coefficient is denoted as... The dynamic contact stiffness coefficient In each computational increment step, based on the current transient contact pressure With indentation depth Following the micro-business control equation Perform incremental step dynamic updates; ultimately combine with the process feed rate curve. An explicit Euler integral algorithm is used to iteratively update the transient force field boundary load acting on the rolling contact surface in the next incremental step at each time step. :

[0023] In the formula, Overall yield equivalent stiffness coefficient, For the current transient time, For a single computational increment time step, The contact force calculation equation is used as the mechanical boundary condition input into the finite element solver for dynamic loading, representing the transient force field boundary load of the current step.

[0024] Furthermore, in step seven, a multi-scale solution task is submitted, and the solver iteratively calculates within discrete time steps to simultaneously or stepwise solve the process of macroscopic plastic deformation and microscopic crystal slip and lattice rotation, recording the historical variables of the entire process.

[0025] Furthermore, in step eight, multi-scale physical field data is extracted to construct a residual compressive stress volume integral objective function for performance evaluation. For the high stress gradient region of the bolt's R-angle, the residual stress evaluation model is constructed as a three-dimensional residual principal stress tensor. median of volume fraction :

[0026] In the formula, Volume of the plastic deformation zone of the bolt radius (R-angle). It is a spatial volume element. The first-order principal stress... The algebraic properties of the compressive stress state are uniformly agreed to be negative, and the objective function is... The smaller the algebraic value, the greater the absolute value of the introduced residual compressive stress.

[0027] The entire reverse optimization process takes the objective function Minimization is used as the target metric to construct a modified iterative equation:

[0028] The formula includes the rolling speed. Rolling time and rolling pressure Including process parameter vectors . For the first The processing parameter configuration for the next iteration. To relax the step size vector, This is the sensitivity gradient operator for the objective function with respect to the processing parameters. The solver uses the gradient backpropagation algorithm dynamically to update the kinematic boundary conditions of the simulation model until the increment of the process parameters in two adjacent iterations is less than a preset convergence threshold. Only then can it be considered that the inverse optimization of the processing parameters and the closed-loop optimization of the entire process have been completed. Attached Figure Description

[0029] Figure 1 This is an overall flowchart of a multi-scale finite element simulation method for EBSD grain-oriented bolt rolling R-forming based on the present invention;

[0030] Figure 2 This is the geometric model and assembly diagram of the bolt rolling R-forming of the present invention;

[0031] Figure 3 This is a detailed implementation step and data flow diagram of the multi-scale simulation of the bolt rolling R process of the present invention;

[0032] Figure 4 This is a schematic diagram of the reconstructed bolt microscopic polycrystalline finite element mesh model of the present invention;

[0033] Explanation of reference numerals in the attached diagram: 1-bolt; 2-rolling roller; 3-transition fillet (R angle). Specific implementation methods

[0034] The invention will now be described in detail with reference to specific examples.

[0035] The steps for implementing this invention are as follows:

[0036] I. Establishing the macroscopic geometric model and mesh generation for bolt roll forming.

[0037] First, the three-dimensional solid structure models of the target titanium alloy bolt and rolling wheel are imported using a three-dimensional digital modeling system. Then, the transition fillet (R-corner) region connecting the bolt head and the shank is isolated using a geometric topology segmentation tool. This region is defined as the core deformation subdomain, providing a clear and distinct geometric boundary for subsequent embedding of microcrystalline structure data and cross-scale coupling.

[0038] In the mesh generation module, the core deformation subdomain with R-angle is locally meshed with high density adaptive refinement using an eight-node linear reduced integral three-dimensional solid element (C3D8R). This effectively overcomes the volumetric self-locking effect while capturing local large plastic strain gradients. For bolt shanks and threaded sections far from the main deformation zone of rolling strengthening, a non-uniform transition meshing scheme with gradually changing mesh size is used. This maximizes the reduction of the overall stiffness matrix solution scale while ensuring the accuracy of macroscopic stress field calculation.

[0039] In the boundary conditions and contact interaction settings, the properties of the rolling roller assembly model are defined as an analytical rigid body. A dynamic surface-to-surface contact mechanism is established between the main outer profile of the rolling roller and the R-angle surface of the bolt. This mechanism can accurately describe the interfacial mechanical behavior under the extreme high pressure stress of rolling and introduce a modified Coulomb-shear composite friction constitutive model into the contact properties.

[0040] II. Acquisition of EBSD data and microstructure characterization of bolt materials

[0041] At the experimental end, a blank substrate sample of the same material as the workpiece to be simulated (TC4 two-phase crystal or TB9 single-phase body-centered cubic crystal) was cut out. After removing the micro-work hardening layer by multiple passes of mechanical polishing and electrolytic polishing, it was placed in a scanning electron microscope to collect full-field electron backscatter diffraction (EBSD) patterns.

[0042] After data acquisition, the original grain boundary topology and lattice orientation data are exported as a text matrix file in a common format. A preset orientation analysis script is then used to perform neighborhood averaging denoising and smooth reconstruction on the lattice data.

[0043] III. Reconstructing a Microscopic Polycrystalline Model Based on EBSD Grain Orientation

[0044] Run the microcrystalline polycrystalline finite element mesh generator, read the feature database file output from step two, and extract the geometric equivalent diameter, aspect ratio, and spatial center point coordinates of each grain in the sample. Within the geometric boundary of a three-dimensional representative volume element (RVE) of a preset spatial dimension, using the extracted grain center points as spatial growth seeds, call the Voronoi Thiessen polygon growth algorithm to perform spatial topological segmentation, thereby constructing a microcrystalline polycrystalline non-uniform finite element mesh model containing the geometric features of real phase boundaries and grain boundaries.

[0045] To impart realistic anisotropic crystalline rheological behavior to the microscopic model, a user-defined material subroutine (UMAT or VUMAT) is written and loaded into the solver kernel. The governing equations of Crystal Plasticity Finite Element Analysis (CPFEM) are written into the subroutine, and the total deformation gradient tensor at each Gaussian integration point is expressed as... Multiplicative decomposition into elastic deformation gradients describing the elastic elongation and rotation of the lattice And the plastic deformation gradient describing the dislocation shear slip of the slip system. During the initial step, the subroutine retrieves the texture feature file generated in step two and sets the rotation matrix corresponding to each element. Write the predefined state variable field to complete the initial assembly of the microcrystalline plastic finite element model.

[0046] IV. Construction of Multi-Scale Material Constitutive Models and Parameter Calibration

[0047] To enable the analysis of macroscopic continuous medium physical fields to dynamically correlate work hardening and anisotropic response introduced by microscopic crystal orientation, a macroscopic flow stress model based on microscopic statistical parameter correction is constructed.

[0048] The EBSD full-field data obtained in step two is read using a matrix analysis script. After crystal orientation space integration, the average Taylor factor representing the overall texture intensity of this batch of fastener materials is calculated. and the initial critical shear stress of the matrix slip system Substituting these two quantities, which have microscopic physical mechanisms, into the macroscopic multiscale hardening flow constitutive equation:

[0049] In the formula, Equivalent flow stress; hardening coefficient; The hardening index; Macroscopic equivalent plastic strain; It is the strain rate strengthening constant; and These are the equivalent plastic strain rate and the reference strain rate, respectively. The operating temperature is dimensionless. The thermal softening index is used. By conducting quasi-static and dynamic mechanical tests on this batch of titanium alloy materials under multi-physics environment, and using a global optimization calibration algorithm, the hardening parameters and strain rates in the above flow equations were fitted and calibrated to establish a cross-scale quantitative mapping relationship between macro- and micro-rheological behaviors.

[0050] V. Establish a mechanism for cross-scale information transmission and coupling between macro and micro scales.

[0051] This method is based on the finite element distributed data interface and sub-model driven algorithm to establish a cross-scale unidirectional or bidirectional solution channel for the processing deformation field from macroscopic to microscopic.

[0052] First, configure the macroscopic bolt rolling R-deformation simulation task described in step one of the core solver operations. The solver writes the displacement vectors, stress tensors, and transient deformation rates of each mesh node in the global and critical reinforcement zones into the spatial global results database file in real time within discrete time steps.

[0053] After the macroscopic dynamic analysis step is completed, the microscopic RVE polycrystalline plastic finite element simulation constructed in step three is initiated. The spatial external topological boundary nodes of the microscopic model are defined as boundary-driven master nodes. At the start of the time step integration, the microscopic solver automatically reads the macroscopic result database file by calling the sub-model data chain, retrieves and locates the spatial isoparametric elements containing the geometric coordinates of the current microscopic model, and calls the corresponding macroscopic element geometric shape function matrix. Perform multidimensional interpolation to calculate the transient driving displacement field of the boundary nodes:

[0054] In the formula, The total number of nodes in the macroscopic unit that encloses the current microscopic local spatial location; This represents the transient displacement component matrix output by the macroscopic model in this unit. For the spatial coordinates of micro-nodes The geometric element shape function values ​​within the corresponding macroscopic isoparametric element. Periodic boundary conditions are applied to the outer boundary of the microscopic RVE polycrystalline mesh model. The macroscopic displacement field is interpolated by isoparametric shape functions and then applied to the master nodes of the microscopic mesh. The displacement of the slave nodes is dynamically determined by the displacement difference of the corresponding master nodes, thereby realizing a continuous and smooth mapping of mechanical loads from the continuous medium model to the microscopic polycrystalline control mesh.

[0055] VI. Setting machining process parameters and kinematic boundary conditions

[0056] Within the load and working condition control definition section of the macroscopic finite element input file, the kinematic chains of the workpiece and tooling are configured. The bolt model is constrained in all spatial degrees of freedom except for rotation along the axis, and a feed boundary along the radial direction of the workpiece center is defined at the rolling roller reference point. To introduce the influence of elastic deformation relief of actual mechanical equipment on the rolling pressure field in the simulation and to avoid localized stress singularities caused by rigid interference, this embodiment loads a nonlinear stiffness contact pair matrix subroutine in solving the boundary conditions.

[0057] The subroutine presets the equivalent stiffness matrix of the current machine tool main servo control system. During the solution process, the indentation normal contact stiffness tensor of each Gaussian point on the contact surface at the current time step is dynamically extracted. This embodiment employs an incremental linearization approximation method, that is, in each implicit / explicit calculation increment step, the transient normal contact stress output in the current step is used as the approximation. With transient local indentation depth Based on contact mechanics theory, stiffness The discrete derivative expression for performing dynamic incremental updates is specified as follows:

[0058] Workpiece contact stiffness coefficient The mechanical stiffness coefficient of the rolling mill itself Perform a series summation to obtain the comprehensive yield equivalent stiffness coefficient of the interface within the current increment step. Combined with the process feed rate curve The transient force field boundary load acting on the rolling contact surface in the next incremental step is updated using explicit Euler integrals. This eliminates the mathematical inconsistency problem of not being able to derive an integral sign from time-varying stiffness, and accurately restores the physical processing state:

[0059] VII. Perform multi-scale finite element analysis and process simulation.

[0060] The integrated input control batch file, which includes the macroscopic spatial mesh assembly tree, contact amplitude control curves, cross-scale sub-model data interfaces, and the compiled dynamic link library file of crystal plastic materials, is submitted to the solver core of the high-performance computing server.

[0061] The solver employs a nonlinear dynamic explicit integrator to process files within discrete time steps and submits them to the solver core of the high-performance computing server. Sequential solving is then performed. The solver core first calculates the macroscopic elastoplastic large deformation and full-field contact response in the first macroscopic solver step. Subsequently, through the aforementioned cross-scale unidirectional data channel, the local strain matrix within the critical reinforcement zone is smoothly pushed down to the micro-constrained polycrystalline RVE model mesh as displacement history conditions. This independently triggers the micromaterial subroutine to iteratively calculate crystal plastic slip motion and texture rotation in the second solver step, completing the cross-scale full-physics sequential iterative solution.

[0062] VIII. Analysis of Macro- and Micro-level Multiphysics Fields and Grain Evolution Results

[0063] After the simulation calculation is completed, the control script automatically loads the post-processing analysis interface to extract the historical files of the generated solution results across scales.

[0064] At the microscopic level, the script extracts predefined lattice variables at each integration point of the polycrystalline material, calculates the lattice orientation deviation matrix before and after deformation, and outputs the altered pole figure and inverse pole figure to characterize and quantitatively evaluate the microscopic surface texture evolution characteristics introduced by the rolling strengthening process. Simultaneously, the post-processing script extracts the spatial strain gradient tensor of each element to calculate the geometrically required dislocation (GND) density distribution within the ultrafine-grained region of the surface layer. Used to evaluate the depth of microhardening:

[0065] In the formula, The Burgers vector modulus of the target titanium alloy; The nine discrete physical components of the Nye dislocation tensor in a three-dimensional Cartesian coordinate system are extracted by scalarization approximation through the sum of squares and square root operations.

[0066] On a macroscopic level, the system automatically selects the complete three-dimensional rheological region of the bolt root radius affected by extrusion deformation, and obtains the volumetric region. Volume of all grid cells and the residual first principal stress characterizing local deformation (Here, the physical properties of the beneficial compressive stress state are uniformly defined and denoted as negative values ​​at the algebraic numerical level.) Subsequently, a discrete triple integral is performed throughout the entire space to calculate the median objective function of the residual compressive stress volume integral, which characterizes the overall strengthening effect of the fillet fillet throughout the entire space. :

[0067] In the formula, This represents the total volume of the plastic deformation zone of the bolt's radius (R-angle).

[0068] IX. Closed-loop optimization of process parameters for optimal residual compressive stress

[0069] To eliminate the blindness of relying on manual trial and error to select process parameters and to achieve direct closed-loop feedback and precise reverse control of on-site physical manufacturing parameters by virtual numerical simulation, a seamless process closed-loop optimization framework is built on an external system using scripting languages ​​(such as Python or MATLAB).

[0070] The optimization control script encapsulates the core controllable parameters in the bolt rolling process into process boundary vectors. The system utilizes a dynamic global optimization algorithm with the objective function... Minimization is the direction of process optimization, aiming to minimize the algebraic value of residual compressive stress in the corresponding strengthening zone and maximize the absolute value of compressive stress strengthening effect.

[0071] After each simulation, the external control script reads the current... The numerical value automatically calculates its spatial variational partial derivative with respect to the current process vector, and constructs a transient process sensitivity partial derivative gradient operator. Subsequently, the system executes the following gradient-adjusted process vector correction iterative equation:

[0072] In the formula, For the first The configuration status of process parameters during each iteration; The relaxation adjustment step size array is used. Based on the calculated updated values, the external optimization control program automatically rewrites the finite element loading input control file for the next cycle, and automatically resubmits it to the solution core for recalculation using the system's underlying command-line instructions. The entire system autonomously cycles in an unattended state until the process vector offset increment between two adjacent iterations falls below the preset convergence limit. At this point, the optimization is terminated, and the globally optimal combination of rolling process parameters for this titanium alloy fastener model is output on the terminal screen, realizing the complete implementation from virtual multi-scale calculation to physical processing technology closed-loop design.

Claims

1. A multi-scale finite element simulation method based on EBSD grain orientation bolt roll-R forming, characterized in that, Includes the following steps: Step 1: Establish the macroscopic geometric model and mesh generation for bolt roll forming; Step 2: Collect EBSD data and characterize the microstructure of the bolt material; Step 3: Reconstruct the microscopic polycrystalline model based on EBSD grain orientation; Step 4: Construct a multi-scale material constitutive model and calibrate its parameters; Step 5: Establish a cross-scale information transmission and coupling mechanism between macro and micro scales; Step Six: Set the machining process parameters and kinematic boundary conditions; Step 7: Perform multi-scale finite element analysis and process simulation; Step 8: Analysis of macroscopic and microscopic multi-physical field quantities and grain evolution results; Step 9: Closed-loop optimization of process parameters for optimal residual compressive stress.

2. The multi-scale finite element simulation method of claim 1, wherein, The method for establishing the macroscopic geometric model and mesh generation of the bolt rolling R-shape in step one includes: establishing a macroscopic three-dimensional solid model based on the actual geometric dimensions of the bolt and the design parameters of the rolling wheel; refining the local mesh in the R-angle region at the root of the thread; defining the macroscopic finite element mesh type and element attributes in the software, and setting the friction contact pairs of the macroscopic interface.

3. The multiscale finite element emulation method of claim 1 or 2, wherein, The method for acquiring EBSD data and microstructure characterization of bolt materials in step two includes: using electron backscatter diffraction technology to obtain the micromorphology, grain boundary distribution and initial grain orientation data of the bolt material to be processed; converting the acquired Euler angles into a rotation matrix to describe the spatial mapping relationship between the crystal coordinate system and the sample coordinate system.

4. The multi-scale finite element simulation method of claim 1, wherein, The method for reconstructing the microscopic polycrystalline model in step three includes: using the Voronoi polygon algorithm, taking the centroid extracted by EBSD as the lattice seed point, and generating a polycrystalline model within a defined representative volume element region; uniformly assigning the measured average orientation Euler angle of each independent grain to all geometric mesh elements contained within the grain; introducing the crystal plasticity finite element analysis theory, and performing elastic and plastic multiplicative decomposition of the total crystal deformation gradient to construct the microscopic dynamic evolution control equation.

5. The multi-scale finite element simulation method according to claim 1, characterized in that, The method for constructing a multi-scale material constitutive model in step four includes: extracting the Taylor factor and critical shear stress from the EBSD data, replacing the static yield term of the macroscopic constitutive equation with microscopic physical quantities; establishing a modified multi-scale flow stress model to achieve cross-scale mapping of microscopic grain orientation parameters to macroscopic mechanical responses.

6. The multi-scale finite element simulation method according to claim 1, characterized in that, The method for establishing a cross-scale information transmission and coupling mechanism in step five includes: using sub-model technology to establish a cross-scale boundary condition driving mechanism; extracting and transmitting historical data of displacement or strain fields of the macroscopic model during processing through shape function interpolation to the boundary nodes of the microscopic polycrystalline model.

7. The multi-scale finite element simulation method according to claim 1, characterized in that, The method for setting kinematic boundary conditions in step six includes: establishing a dynamic boundary mechanics solution model that considers the coupling stiffness of the equipment and the workpiece; using an incremental linearization approximation method to couple the system mechanical stiffness coefficient and the dynamic contact stiffness coefficient in series to obtain a comprehensive yielding equivalent stiffness coefficient; and combining the process feed rate curve, using an explicit Euler integral algorithm to iteratively update the transient force field boundary load acting on the rolling contact surface at each time step, which is then used as the mechanical boundary condition input to the finite element solver for dynamic loading.

8. The multi-scale finite element simulation method of claim 1, wherein, The method for result analysis in step eight includes: constructing a residual compressive stress volume integral objective function for performance evaluation; constructing the residual stress evaluation model as the median of the volume integral of the three-dimensional residual first principal stress tensor for the high stress gradient region of the bolt R-angle; the algebraic value of the compressive stress state of the first principal stress is conventionally negative, and the smaller the algebraic value of the objective function, the larger the absolute value of the introduced residual compressive stress.

9. The multi-scale finite element simulation method of claim 8, wherein, The method for closed-loop optimization of process parameters in step nine includes: minimizing the objective function as the target index and constructing a modified iterative equation; using the gradient backpropagation algorithm to dynamically update the kinematic boundary conditions of the simulation model until the increment of process parameters in two adjacent iterations is less than a preset convergence threshold, thereby completing the reverse optimization and closed-loop optimization of processing parameters.

10. A computer apparatus comprising a processor, characterised in that, The processor is used to calculate the optimal combination of rolling process parameters used in bolt machining using the multi-scale finite element simulation method as described in any one of claims 1-9.

Images