Simulation Method for Asymmetric Aluminum Profile Extrusion Process

Abstract

This invention discloses a simulation method for the asymmetric aluminum profile extrusion process, specifically relating to the field of metal forming simulation technology. The method involves: constructing an asymmetric profile geometric model and extracting key structural parameters; establishing a multi-cavity welding die thermo-mechanical coupling simulation model; constructing a dynamic recrystallization microstructure evolution model; setting initial extrusion process conditions and performing finite element simulation; comparing the simulation results with historical conditions to identify abnormal regions and extract flow trajectory feature vectors; combining microstructure parameters to invert the microstructure evolution path and generate a defect prediction matrix; solving for the optimal process parameters using a numerical optimization algorithm and re-simulating to verify whether the microstructure and stress field distribution meet the standards. This method can achieve prediction of microstructure defects and intelligent optimization of extrusion parameters, improving the quality control capability of the asymmetric profile extrusion process.

Description

Technical Field

[0001] This invention relates to the field of metal forming simulation technology, specifically to a method for simulating the extrusion process of asymmetric aluminum profiles. Background Technology

[0002] With the increasing demand for high-performance, lightweight structures, asymmetric complex cross-section aluminum profiles are widely used in rail transportation, aerospace, and high-end equipment manufacturing. 6xxx series aluminum alloys, due to their excellent comprehensive properties, are the most commonly used materials for such structural components. However, compared to symmetrical cross-section profiles, asymmetric aluminum profiles experience extremely uneven stress distribution during extrusion, leading to non-coordinated metal flow paths. This easily causes serious quality problems such as uneven internal structure, edge warping, deformation distortion, and dimensional deviations.

[0003] Especially during the welding die extrusion process, asymmetric profiles often experience a significant amplification of flow velocity differences due to unequal flow resistance in each flow chamber. This generates turbulence and vortex structures in the welding chamber, leading to incomplete welding or microstructure delamination. Furthermore, this type of profile is extremely sensitive to equivalent strain, grain size, and equivalent stress under different process parameters (such as temperature, extrusion speed, and extrusion ratio). Traditional linear parameter scanning methods are insufficient to accurately predict its deformation behavior under complex boundary conditions.

[0004] In the existing technology, there is no dynamic prediction method for the microstructure evolution and residual stress distribution during the extrusion of asymmetric aluminum profiles, let alone a systematic solution for identifying and simulating the location of local structural weakening. Summary of the Invention

[0005] The purpose of this invention is to provide a method for simulating the asymmetric aluminum profile extrusion process to address the shortcomings of the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] Simulation methods for asymmetric aluminum profile extrusion processes include:

[0008] S100. Establish a two-dimensional or three-dimensional geometric model of the target asymmetric aluminum profile, identify the features of its cross-section, and obtain the key structural parameter set P.

[0009] S200. Based on the parameter set P, construct a multi-cavity welding mold model, and combine material properties and thermo-mechanical boundary conditions to establish a thermo-mechanical coupling constitutive simulation model of the multi-cavity welding mold that includes equivalent strain, equivalent stress and thermal conduction coupling.

[0010] S300, input the target aluminum alloy grade, heat treatment state and initial grain size, construct a dynamic recrystallization model and grain refinement evolution equation, and generate a set of microstructure parameters O;

[0011] S400, Set the initial extrusion conditions, including extrusion speed. Mold temperature billet temperature and extrusion ratio In conjunction with the multi-cavity welding mold thermo-mechanical coupling constitutive simulation model obtained in step S200, a full-process thermo-mechanical coupling simulation system is constructed using finite element simulation analysis.

[0012] S500. Compare the initial simulation results obtained in step S400 with historical stable operating conditions, identify abnormal regions, and extract the local flow trajectory feature vectors of the abnormal regions. ;

[0013] S600, the feature vector F a By coupling the mapping with the set of organizational parameters O, the microscopic organizational evolution path leading to the anomaly is deduced, and a defect prediction matrix M is formed.

[0014] S700. Based on the defect prediction matrix M, perform numerical inversion to optimize the input extrusion parameter set. For the optimal parameter set ;

[0015] S800: Rerun the simulation process to verify whether the stress-strain field and grain size of each region of the asymmetric aluminum profile meet the forming standards under the optimal parameter set, and locate potential microstructure weakening locations for pretreatment or online quality inspection reference.

[0016] Preferably, S100 includes:

[0017] S101. Based on actual product design drawings or scanned modeling data, generate a two-dimensional cross-sectional view or three-dimensional solid structure of the target asymmetric aluminum profile, and perform meshing processing on the geometric model.

[0018] S102. An edge gradient-based contour recognition algorithm is used to identify the curvature and detect corners of the cross-section of a two-dimensional or three-dimensional geometric model, and to extract the set of feature regions in the profile structure, including flanges, stiffeners, thickened areas and cavity boundaries.

[0019] S103. Based on the extracted feature region set, principal component analysis and spatial projection analysis are used to calculate quantitative indices of asymmetry, including geometric eccentricity. Structural imbalance coefficient and the difference in flow resistance ΔR;

[0020] S104, will , and The key structural parameter set P is composed of quantitative structural indicators.

[0021] Preferably, S200 includes:

[0022] S201. Based on the profile structure parameters extracted from parameter set P, construct a multi-cavity welding mold structure model with multiple independent flow channels and welding zones.

[0023] S202. Based on the thermo-mechanical properties of the selected aluminum alloy material, including density, thermal conductivity, specific heat capacity, stress-strain relationship and dynamic recrystallization parameters, establish a material database model.

[0024] S203. Set thermal-mechanical boundary conditions, including mold temperature. billet temperature Friction factor f, interfacial thermal conductivity h, radiation coefficient r, and contact type;

[0025] S204. Based on the rigid-plastic constitutive relation, a thermo-mechanical coupled constitutive simulation model of a multi-cavity welding mold is established for the full-process simulation of material flow, microstructure evolution and heat conduction within the multi-cavity welding mold region.

[0026] Preferably, S300 includes:

[0027] S301. Input the target aluminum alloy grade, heat treatment state, and initial grain size;

[0028] S302. Construct a dynamic recrystallization critical strain model based on equivalent strain rate, temperature, and material constants, and express the critical strain in exponential function form. With temperature T, equivalent strain rate The relationship between them is defined as follows: , where a, n, and Q are aluminum alloy material constants, and R is a gas constant;

[0029] S303. Establish the grain size evolution equation, and express the dynamic recrystallized grain size d_drex as the initial grain size d. o The function of strain stress state, in the form: ,in m These are the empirical fitting coefficients. is the grain scaling factor, reflecting the material type; m is the strain rate influence coefficient; It is the activation energy for grain evolution.

[0030] Preferably, S400 includes:

[0031] S401. Set the initial parameter set for the target extrusion process, including the extrusion speed. Mold temperature billet temperature and extrusion ratio ;

[0032] S402. Perform finite element simulation on the constructed multi-cavity welded mold thermo-mechanical coupling constitutive simulation model, and complete mesh generation, property mapping and initial state loading. Use a rigid-plastic solver to perform material-mold system initialization.

[0033] S403. Construct a three-dimensional thermo-mechanical coupling analysis system, in which the thermal field is obtained by calculating the heat transfer equation and the internal heat source term, and the mechanical field updates the material deformation behavior by using the equivalent stress-strain relationship, and realizes thermo-mechanical iterative coupling.

[0034] S404. Set the simulation control strategy, including incremental time step Δt, termination condition, contact algorithm and convergence criterion.

[0035] Preferably, S500 includes:

[0036] S501. Qualified extrusion simulation data generated under similar materials, molds and process parameters will be divided into regions and stored in a standardized manner, including information on the distribution of stable strain field, stress field, temperature field and flow velocity field.

[0037] S502. Compare the initial simulation results with the historical stable working condition database. Use the spatial point-to-point error mapping and morphological matching algorithm to identify abnormal areas with deviations. The judgment indicators include strain deviation Δε, velocity disturbance Δv, temperature change ΔT, and welding pressure anomaly Δp.

[0038] S503. The flow trajectory of the identified abnormal area is tracked. The particle tracking method is used to simulate the path of multiple tracer points along the metal flow direction to form a local flow trajectory line group.

[0039] S504. Extract the flow trajectory feature vector V for each trajectory line, including trajectory length L and average velocity. Flow velocity gradient The path curvature κ and relative entrance offset Δx are used as behavioral characteristic data to represent the abnormal region.

[0040] Preferably, S600 includes:

[0041] S601, Transfer the flow trajectory feature vector of the abnormal region Spatially aligned with the corresponding set of organizational parameters O, data coupling is achieved through grid node mapping;

[0042] S602. Based on the grain size evolution model and recrystallization volume fraction model, a trajectory-driven microstructure inversion function set is constructed to predict the microstructure evolution trend under different strain paths.

[0043] S603. Set the threshold conditions for microstructure evolution, including the grain coarsening threshold d_max, the incomplete recrystallization threshold X_min, and the local non-uniformity σ_d, as the basis for defect judgment.

[0044] S604. Generate a defect prediction matrix M based on the prediction results. Each element in M ​​represents the potential defect level of the corresponding trajectory at different time steps.

[0045] Preferably, S700 includes:

[0046] S701. Using the parameters corresponding to the abnormal trajectories marked in the defect prediction matrix M as the objective function input, construct a multi-objective optimization-based extrusion parameter control model.

[0047] S702. Select extrusion speed, die temperature, billet temperature and extrusion ratio as variables to be optimized, and define process boundaries and constraints.

[0048] S703: Apply genetic algorithms, particle swarm optimization, or gradient descent algorithms to iteratively search for the parameter combination with the lowest defect risk;

[0049] S704, Output the optimal extrusion parameter set.

[0050] Preferably, S800 includes:

[0051] S801. Using the optimal parameter set as the initial input for the simulation, rerun the thermo-mechanical coupling simulation process to obtain the global stress, strain and temperature field distribution.

[0052] S802. Using the microstructure evolution model, calculate the grain size and recrystallization volume fraction of each region to assess whether the established forming quality standards are met.

[0053] S803. Perform microstructure and property scanning on complex structural parts of the profile, extract indicators of insufficient recrystallization rate and abnormal grain coarsening, and locate potential microstructure weakening areas.

[0054] S804, Output the spatial coordinates of the weakened area and the organization evaluation data.

[0055] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0056] 1. This invention constructs a multi-scale, full-process simulation system for asymmetric aluminum profile extrusion, integrating a material microstructure evolution model, thermo-mechanical coupling simulation analysis, and a data-driven anomaly identification mechanism. This achieves closed-loop control from process parameter setting to microstructure defect prediction and parameter optimization. Compared to traditional experience-based process formulation methods, this invention can accurately identify potential microstructure weakening areas in the early stages of extrusion, significantly improving the accuracy and controllability of forming quality prediction.

[0057] 2. The trajectory feature vector and organizational parameter coupling inversion method proposed in this invention solves the problem of difficulty in quantifying and tracing the evolution behavior of microstructure. Combined with the defect prediction matrix and numerical optimization algorithm, it can realize the automatic reverse correction of key process parameters such as extrusion speed, temperature, and ratio, providing an intelligent and transferable process optimization platform for asymmetric complex profile extrusion. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0059] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] For examples, please refer to Figure 1 As shown, the asymmetric aluminum profile extrusion process simulation method described in this embodiment includes:

[0062] S100. Establish a two-dimensional or three-dimensional geometric model of the target asymmetric aluminum profile, identify the features of its cross-section, and obtain the key structural parameter set P.

[0063] S200. Based on the parameter set P, construct a multi-cavity welding mold model, and combine material properties and thermo-mechanical boundary conditions to establish a thermo-mechanical coupling constitutive simulation model of the multi-cavity welding mold that includes equivalent strain, equivalent stress and thermal conduction coupling.

[0064] S300, input the target aluminum alloy grade, heat treatment state and initial grain size, construct a dynamic recrystallization model and grain refinement evolution equation, and generate a set of microstructure parameters O;

[0065] S400, Set the initial extrusion conditions, including extrusion speed. Mold temperature billet temperature and extrusion ratio In conjunction with the multi-cavity welding mold thermo-mechanical coupling constitutive simulation model obtained in step S200, a full-process thermo-mechanical coupling simulation system is constructed using finite element simulation analysis.

[0066] S500. Compare the initial simulation results obtained in step S400 with historical stable operating conditions, identify abnormal regions, and extract the local flow trajectory feature vectors of the abnormal regions. ;

[0067] S600, the feature vector F a By coupling the mapping with the set of organizational parameters O, the microscopic organizational evolution path leading to the anomaly is deduced, and a defect prediction matrix M is formed.

[0068] S700. Based on the defect prediction matrix M, perform numerical inversion to optimize the input extrusion parameter set. For the optimal parameter set ;

[0069] S800: Rerun the simulation process to verify whether the stress-strain field and grain size of each region of the asymmetric aluminum profile meet the forming standards under the optimal parameter set, and locate potential microstructure weakening locations for pretreatment or online quality inspection reference.

[0070] In this invention, the geometric information of the target asymmetric aluminum profile can be obtained using one of the following three methods:

[0071] Import structural design drawings of profiles using CAD software (such as SolidWorks, AutoCAD, etc.) and extract two-dimensional cross-sections or three-dimensional structural diagrams;

[0072] If there are no design drawings, use 3D scanning equipment (such as laser scanners or industrial CT) to collect point cloud data of the physical profile surface;

[0073] The collected point cloud data is subjected to noise filtering, boundary repair, and patch fitting to convert it into a geometric model that can be used for numerical simulation. Common formats include STEP, IGES, or STL.

[0074] After modeling is completed, CAE software (such as DEFORM, Abaqus, or ANSYS) is used for model preprocessing to set an appropriate meshing strategy. Typically, a uniform 6mm mesh is used for the entire profile, and the mesh is refined to 2mm to 3mm at boundary details and welded areas to improve simulation accuracy.

[0075] To accurately identify the local structural features of asymmetric profiles, a composite algorithm based on edge gradient detection and curvature analysis is employed. This algorithm mainly includes the following processing steps:

[0076] Boundary extraction: Obtain the outer contour of a two-dimensional cross-sectional image using image edge operators (such as Canny or Sobel);

[0077] Curvature calculation: Sample the boundary curve at equal intervals, calculate the curvature value at each point, and identify inflection points and curvature change zones based on the magnitude of change;

[0078] Region Classification: Based on the curvature variation pattern and closed boundaries, the following typical characteristic regions were identified:

[0079] Flange region: The convex structural region that typically affects the local cooling rate of the material;

[0080] Reinforcing ribs: Slender structural strips used to enhance the rigidity of profiles;

[0081] Cavity boundary: The outline that forms the closed space, affecting the welding pressure field of the mold;

[0082] Variable thickness zone: Locations where the cross-sectional thickness changes abruptly, which may lead to stress concentration;

[0083] Corner detection: The Harris corner detection algorithm is used to detect sharp corners and connection points in the cross section where stress transfer may be significant.

[0084] To support subsequent simulation and model control, the following three core structural parameters are extracted to form parameter set P:

[0085] Geometric eccentricity : Defined as the distance between the centroid of a cross section and the center of its circumscribed rectangle, in millimeters, used to quantify the degree of overall asymmetry. It is calculated by projecting the cross section profile along the X and Y directions and calculating the distance between the centroid and the axis of symmetry respectively.

[0086] Structural Imbalance Coefficient This indicates the asymmetry of the area of ​​a typical characteristic region. The calculation method is as follows: divide the cross-section into two regions, left and right, or top and bottom, and calculate the area of ​​each region separately. and ,calculate The result is a dimensionless ratio.

[0087] The flow resistance difference ratio ΔR is defined as the ratio of the metal flow resistance formed by different combinations of path lengths and local thicknesses. First, the path length L_i and the corresponding thickness H_i from the inlet to each typical structural region are extracted, and the unit resistance R_i = L_i / H_i is calculated. The ratio of the maximum to the minimum is ΔR.

[0088] The above parameters are all automatically calculated by writing feature recognition scripts (such as those based on Python or Matlab) and exported as standard structured data formats (such as XML or JSON) for subsequent modules to call.

[0089] Finally, the parameter set P is encapsulated into the following data structure: P = { Feature region coordinate set: }

[0090] Based on the three parameters in the key structural parameter set P extracted in step S100: geometric eccentricity ( ), structural imbalance coefficient ( The ratio of flow resistance difference (ΔR) is used to guide structural compensation and cavity design during the mold modeling process.

[0091] Mold construction method:

[0092] Use 3D modeling software (such as UG NX or SolidWorks) to create the basic mold geometry;

[0093] According to the structural imbalance coefficient The cross-sectional area of ​​each flow branch cavity is modified to match the metal flow requirements of the target profile cross-sectional area, so as to avoid local overflow or underflow in a certain cavity;

[0094] The welding chamber is designed with a symmetrical guiding platform and an intermediate pressure converging cavity to avoid welding quality problems caused by high strain rate;

[0095] Set the path length L and local thickness H between the inlet and outlet of each cavity to ensure path resistance. The maximum ratio ΔR between the cavities is kept below 3.5 to reduce velocity differences.

[0096] Parameter compensation rules:

[0097] If the ΔR value of a certain region is more than twice the average value, the flow channel width or the length of the welding chamber buffer zone should be appropriately increased in that direction.

[0098] If e s If the deviation exceeds 5 mm, the mold centerline offset needs to be adjusted to ensure that the welding pressure direction is consistent with the flow vector.

[0099] Finally, the mold structure was exported in STEP format and imported into the finite element simulation software.

[0100] The thermo-mechanical properties of the selected aluminum alloy (such as 6A02) under high-temperature extrusion environment are input into the simulation platform to form a standardized material parameter library.

[0101] Material parameters include:

[0102] Density (ρ): 2785 kg per cubic meter;

[0103] Thermal conductivity (λ): 175.8 W / m / Kelvin;

[0104] Specific heat capacity (c): 960 joules per kilogram per Kelvin;

[0105] Poisson's ratio (ν): 0.33;

[0106] Emissivity (r): 0.7;

[0107] Elastic modulus (E): omitted, used only for contact calculations;

[0108] The relationship between flow stress σ and strain ε (i.e., σ-ε relationship): using the Cowper-Symonds model or experimental regression model.

[0109] Dynamic recrystallization parameter model:

[0110] The following empirical formula is used to construct the grain evolution model:

[0111] The dynamic recrystallization grain size d_drex is equal to the initial grain size. Multiplied by an exponential factor, the strain rate being negative. A function of temperature T;

[0112] The recrystallization volume fraction X_drex is represented by the Sigmoid function and depends on the critical strain. ;

[0113] All model coefficients were determined with reference to experimental data from the literature, specifically: Q = 43680 joules per mole, h = 0.1023, a = 0.00015, etc.

[0114] Define all thermo-mechanical boundary inputs in the simulation process, including mold temperature, initial billet temperature, and friction and heat exchange parameters.

[0115] Temperature setting: Mold temperature 425 degrees Celsius;

[0116] billet temperature 510 degrees Celsius;

[0117] Ambient temperature 25 degrees Celsius;

[0118] All temperatures were converted to Kelvin and then input into the simulation system.

[0119] Friction factor f: set using the Shear Friction model; the value of f is set to 0.3, which is suitable for the contact surface between aluminum and cemented carbide molds. In the simulation, it is expressed by the ratio of interface shear force to normal force.

[0120] Thermal conductivity h: The interfacial thermal conductivity between the billet and the mold is set to 10,500 W / m² / Kelvin; thermal contact adopts the "contact thermal conduction" model, which supports pressure sensitivity adjustment.

[0121] Radiation term setting: According to the Stefan-Boltzmann law, radiative heat flux density Where σ is the Stefan constant, The absolute temperature of the high-temperature surface. denoted as absolute temperature of the low-temperature surface; e is the surface emissivity, which is the ratio of the material's radiation capacity to that of an ideal blackbody, ranging from 0 to 1, depending on the material and surface condition.

[0122] Contact type: The relationship between the mold and the blank is defined using "rigid-plastic body contact"; the contact criterion is that the contact feedback is triggered when the node penetration depth is less than 0.05 mm.

[0123] After completing the above preparations, a multiphysics coupled simulation model is constructed using the rigid-plastic finite element method. The constitutive relations used are as follows:

[0124] Stress-strain model: The Levy-Mises plastic flow law is adopted, where the equivalent stress σ is equal to two-thirds of the stress deviator multiplied by the square root of one-half;

[0125] Equivalent strain rate calculation: It is obtained by calculating the derivative of the velocity field. The partial derivative of the equivalent strain rate is the sum of the partial derivatives of the velocity vector with respect to the spatial coordinates.

[0126] Heat conduction equation: constructed based on the law of conservation of energy, considering internal heat sources. (Generated by plastic work and frictional work), and the temperature field is solved using explicit integration.

[0127] Tissue evolution module: Enable the dynamic recrystallization sub-module in the simulation platform to control grain size changes according to the evolution equations in the aforementioned material library;

[0128] Coupling strategy:

[0129] At each time step, the strain and stress field are calculated first;

[0130] Then update the temperature based on the internal heat source calculation;

[0131] Finally, the temperature is fed back to the material model to update the yield strength and flow stress;

[0132] Complete one coupled iterative loop.

[0133] The mesh density in the simulation area is dynamically adjusted, and the mesh accuracy of key parts such as the welding chamber, mold entrance and profile corner is controlled within 2 mm. The calculation step size is adaptive, and the maximum step size does not exceed 0.1 mm.

[0134] This step first requires inputting the following three types of parameters as the basis for model construction:

[0135] Aluminum alloy grades, such as 6A02 and 6061, must be consistent with the material database to index their thermal properties and kinetic constants.

[0136] Heat treatment states include annealed state (O), T6, T4, etc., which correspond to different initial microstructures and recrystallization behaviors;

[0137] Initial grain size The unit is micrometer (μm), obtained by metallographic image analysis or scanning electron microscopy (SEM), and the average value and standard deviation are calculated after image segmentation and particle recognition algorithms (such as threshold segmentation + connected component analysis).

[0138] Among them, the initial grain size It will serve as one of the key variables in the evolution equation, significantly influencing the final grain size variation trend.

[0139] Dynamic recrystallization (DRX) is a process in which a material, during high-temperature plastic deformation, alleviates internal stress and energy accumulation by generating new equiaxed grains. To accurately describe this process, this invention employs a strain-driven model, combined with the Arrhenius thermal activation principle, to establish the following mathematical model:

[0140] The initiation condition for dynamic recrystallization is determined by the critical strain. The critical strain is characterized by the point at which recrystallization begins when the local strain of the material exceeds a certain threshold. The empirical formula for the critical strain is as follows: ;in: : Critical strain for dynamic recrystallization; a: Material constant, in dimensionless units; : Equivalent strain rate, in seconds; n: Equivalent strain rate sensitivity index; Q: Recrystallization activation energy, in joules per mole; R: Ideal gas constant, with a value of 8.314 joules per mole per Kelvin; T: Absolute temperature, in Kelvin.

[0141] In practical applications, material constants a, n, and Q need to be obtained through thermal simulation experiments (such as the Gleeble compression test) or literature databases. For example, for 6A02 aluminum alloy, Q is approximately 43680 joules per mole, and n is 0.1023.

[0142] During recrystallization, new equiaxed grains form, and their size is controlled by the deformation rate and thermal conditions. This invention employs an exponential grain evolution model, expressed as follows: Where: d_drex: average grain size after recrystallization; Initial grain size; : Grain scaling factor, reflecting material type; m: Strain rate influence coefficient; : Activation energy for grain evolution; other symbols are the same as before.

[0143] This model has practical physical significance, reflecting the grain refinement trend; that is, the higher the strain rate or the lower the temperature, the finer the recrystallized grains. For 6A02 aluminum alloy, this model can be selected. m=0.0247 Joules per mole.

[0144] To express the extent of recrystallization, this invention introduces a volume fraction model X_drex constructed using the Sigmoid function, expressed as follows: Where: X_drex: recrystallization volume fraction under the current deformed state; ε: current equivalent strain; : Critical strain, determined by the above formula; k: Transition coefficient, reflecting the rate of microstructure transformation; p: Exponential adjustment parameter, typically ranging from 1 to 3. This expression provides a smooth description of the recrystallization process, avoiding simulation non-convergence issues caused by discontinuities.

[0145] The real-time output of the above model is recorded according to the simulation time step, and a set of tissue parameters O is generated, which specifically includes the following structure:

[0146] O = { Initial grain size, Critical strain, d_drex(t): Grain size evolution function over time, X_drex(t): Recrystallization volume fraction evolution function, σ_d(t): Grain size standard deviation evolution function over time, K: Structure hardening coefficient, Mapping_T-ES: Temperature-strain-stress mapping table}.

[0147] The aforementioned dataset will be integrated with the thermo-mechanical coupling simulation module to achieve simultaneous prediction of microstructure and mechanical properties. Mapping_T-ES represents the grain variation range under different temperature T, strain ε, and stress σ conditions, and can be used for subsequent defect identification and online monitoring standard development.

[0148] To ensure the integrity of the boundary inputs of the simulation system and their clear engineering significance, the initial parameters of the target extrusion process need to be uniformly encapsulated to form a parameter set S, which includes the following elements:

[0149] Extrusion speed (Unit: meters per minute): The typical range is 1.0 to 2.0 m / min, used to simulate the mainstream speed of metal along the mold direction, which affects the deformation rate and the intensity of internal heat source formation.

[0150] mold temperature (Unit: degrees Celsius or Kelvin): Set to 425℃, which needs to be converted to absolute temperature K as the thermal boundary condition input to calculate the heat transfer rate at the contact surface between the billet and the mold.

[0151] billet temperature (Unit: Kelvin): The initial heating temperature is set at 510℃, which affects the initial plasticity of the metal and the level of initial microstructure evolution.

[0152] Extrusion ratio (Unit: dimensionless): Defined as the ratio of the cross-sectional area of ​​the billet to the cross-sectional area of ​​the mold outlet. In this invention, the range is set to 8.34 to 17.38.

[0153] The above parameters must meet the material's plasticity window limit, that is, they must be within the equivalent strain rate-temperature space that allows the material to deform; otherwise, it will lead to metal cracking or excessive recrystallization.

[0154] All parameters are input into the simulation platform in a structured format.

[0155] The constructed multi-cavity welding mold thermo-mechanical coupling constitutive simulation model is invoked, and the preprocessing settings in the finite element platform are completed, including the following three key tasks:

[0156] Mesh generation: A volume control algorithm is used to perform non-uniform mesh generation on the billet region, with the following settings: Initial stage of billet: mesh size 6 mm; shunt cavity and welding chamber: fine mesh 3 mm; welding surface and profile outlet: set to 2 mm to improve simulation accuracy. Tetrahedral or hexahedral elements are used, and local adaptive remeshing is supported.

[0157] Property mapping includes loading data from the material database, including flow stress models, thermal properties (thermal conductivity, specific heat, emissivity), and dynamic recrystallization parameters, into the simulation system. All property values ​​are mapped to the mesh model according to node distribution. Initial state loading includes loading according to the set... and The values ​​are assigned to the blank and mold units respectively; the initial stress of the material is zero; the initial displacement value is set to static, and the velocity is constantly applied to the rear boundary at v0.

[0158] After completing the above settings, call the Rigid-Plastic Solver to initialize the system and perform preliminary calculations for determining the contact relationship between the material and the mold and for thermal field equilibrium.

[0159] To accurately reflect the multiphysics interactions during the extrusion process, a three-dimensional thermo-mechanical coupling analysis system was constructed, specifically divided into two sub-modules that are iteratively coupled:

[0160] Based on the principle of energy conservation, the following heat conduction governing equations are established: Where: ρ is density; c is specific heat; λ is thermal conductivity; T is temperature; This is an internal heat source term, generated by plastic deformation work and friction work.

[0161] in The definition of is: Where σ is the equivalent stress, η is the equivalent strain rate, and η is the thermal conversion efficiency (typically taken as 0.9).

[0162] A rigid-plastic flow model is adopted, neglecting elastic stress, and the stress-strain relationship is updated based on the Levy-Mises flow rule and the von Mises yield criterion.

[0163] The expression for the equivalent rate of change is: Where u is the velocity vector and x is the spatial coordinate.

[0164] The coupling method is as follows: at each time step, first solve the current stress field; then, based on... Update the temperature field; back-calculate the temperature to the material constitutive model to update the yield strength; iterate until convergence.

[0165] This process is completed automatically by the coupled solver of the finite element platform.

[0166] To ensure a stable and efficient simulation process, the following control parameters are set:

[0167] An adaptive time integration strategy is adopted, with an initial step size of 0.1 seconds. If the temperature gradient or deformation rate changes drastically, the step size is reduced to 0.01 seconds, with a maximum of 0.2 seconds.

[0168] Termination conditions include: the extrusion length reaches the set value of the profile (e.g., 1.5 meters), or the simulation time reaches the limit value (e.g., 100 seconds), or all the exit metal flows out of the mold.

[0169] The penalty function method is used to handle the contact between the billet and the mold; the contact stiffness coefficient k is taken as 1000 N / mm; the maximum allowable penetration distance is 0.01 mm.

[0170] Both the thermal field and the force field must satisfy the residual convergence condition, defined as follows: Where n is the number of iterations. If the condition is still not met after 5 iterations, Δt is automatically reduced and recalculated.

[0171] Establish a historical simulation condition database containing typical stable extrusion behavior characteristics for quantitative comparison with current simulation results.

[0172] The "historical operating condition data" are derived from simulation results that share the following characteristics:

[0173] The materials are the same or belong to the same aluminum alloy series (such as 6xxx series);

[0174] The mold structure, cavity distribution, and welding method are similar;

[0175] The extrusion parameters (extrusion speed, extrusion ratio, temperature) are within the recommended process window range;

[0176] The simulation results were verified to be consistent with the actual profile quality, with no obvious welding defects or structural abnormalities.

[0177] The complete extrusion model is divided into several functional areas, including the flow distribution chamber, welding chamber, sizing zone, die exit and profile forming area;

[0178] Within each region, spatial discretization is performed using a voxel grid to establish a local field distribution index.

[0179] Each historical operating condition instance contains the following standard field information:

[0180] The stable strain field ε_stable(x, y, z);

[0181] Stable stress field σ_stable(x, y, z);

[0182] Temperature field T_stable(x, y, z);

[0183] Local velocity field v_stable(x, y, z);

[0184] All data is stored in a three-dimensional grid format with a unified unit system, organized and stored in HDF5 or JSON format for easy retrieval and difference analysis.

[0185] By spatially mapping the current simulation results with the stable operating condition database and comparing the field difference, regions with significant deviations are identified and judged as possible "abnormal behavior areas".

[0186] The point-to-point field comparison algorithm is adopted, and the specific process is as follows:

[0187] Using the current simulation grid as a reference, historical data is mapped to the same coordinate system through a three-dimensional interpolation algorithm (such as trilinear interpolation);

[0188] For each grid node, calculate the error terms for the following physical quantities:

[0189] ;

[0190] ;

[0191] ;

[0192] (Welding pressure);

[0193] in: Indicates the current operating conditions. Under historical stable operating conditions, Δε represents the equivalent strain deviation; Δv represents the flow velocity modulus deviation; ΔT represents local temperature abrupt change; and Δp represents abnormal pressure fluctuations, especially at the weld surface or weld zone.

[0194] Set the following recognition thresholds:

[0195] ;

[0196] ;

[0197] ;

[0198] ;

[0199] If any condition is met, the corresponding region is determined to be an abnormal region, denoted as region Ω_abn.

[0200] To eliminate false positive errors caused by mesh disturbances, a shape matching algorithm is further introduced. The shape of Ω_abn is compared with the shape of historical weld quality areas, and the IoU (Intersection over Union) index is used for shape matching. If IoU < 0.4, it is retained as a true abnormal area.

[0201] Metal flow path tracing was performed within the identified anomalous region Ω_abn to reveal the root cause of its flow behavior differences.

[0202] Equally spaced tracing particles are arranged along the metal flow front end in the Ω_abn region, with the number of particles z ranging from 50 to 100.

[0203] Each particle receives a local velocity vector according to the simulation time step and updates its position vector r(t);

[0204] The complete path of each particle from the entrance to the exit is accumulated to form a trajectory line τ_i (i∈1~z);

[0205] The algorithm can be embedded in simulation software such as DEFORM or processed offline using Matlab / Python.

[0206] For each τ_i, record its three-dimensional coordinates at each time step;

[0207] Trajectory denoising is performed using curve smoothing algorithms (such as the Savitzky-Golay filter) to preserve the main flow direction information.

[0208] For each trajectory τ_i, a set of features reflecting the flow behavior can be extracted to form a feature vector V_i, which specifically includes:

[0209] Trajectory length L: Path integral length of τ_i, in millimeters, reflecting the overall path distance;

[0210] average speed : The average velocity modulus of the particle across all time steps along the path;

[0211] velocity gradient The rate of change of velocity between any two adjacent points on the path is used to determine acceleration / deceleration segments.

[0212] Path curvature κ: the rate of change of direction of path per unit length; the larger the value, the stronger the path disturbance.

[0213] Entrance offset Δx: The difference in horizontal displacement between the initial position and the stable trajectory, in millimeters.

[0214] The final feature vector is formed: .

[0215] The obtained abnormal region trajectory feature vector F a Spatial mapping is performed with the generated set of tissue parameters O, and a coupled data structure is constructed. .

[0216] Let w flow trajectory lines τ_i (i = 1, 2, ..., w) originate from the anomaly region, and their eigenvectors be: O: represents the set of microstructure parameter fields within the entire profile extrusion domain, specifically including: .

[0217] F is achieved using the trajectory node interpolation projection method. a The specific steps for data coupling with O are as follows:

[0218] Discretize each trajectory τ_i into m temporal nodes p_ij (j = 1, 2, ..., m);

[0219] For each node p_ij, a three-dimensional linear interpolation is performed in the organization parameter field O based on its spatial coordinates (x, y, z);

[0220] Obtain the organizational parameter O_ij corresponding to time t_ij at this node;

[0221] Combining the trajectory line τ_i with its along-process organizational changes O_ij into a composite data structure, the following mapping is formed: This data structure provides the basic input for subsequent organizational evolution path inversion and defect identification.

[0222] Based on the already obtained coupled data structure, this step takes each trajectory τ_i as the main evolution path and combines the local equivalent strain rate, temperature and velocity gradient information on the path to invert the corresponding micro-organization evolution behavior.

[0223] Combining the dynamic recrystallization model and the grain refinement equation, the following evolution function is established for each trajectory node p_ij:

[0224] Grain size evolution equation: ;

[0225] recrystallization volume fraction function: ;

[0226] in: Initial grain size; : Equivalent strain rate of the j-th node; : Local temperature of the node; a', m, Q', k, p: Material constants, retrieved from the database; R: Gas constant, value ; Critical strain, which has been calculated using equivalent strain rate and temperature.

[0227] Solving the above model node by node on the trajectory yields the grain size evolution sequence. and recrystallization fraction sequence This forms a trajectory-level organizational evolution path: .

[0228] To identify anomalous behaviors in evolutionary pathways, thresholds for key organizational parameters need to be set as the basis for judgment. Commonly used indicators are as follows:

[0229]

[0230] Set threshold: X_min = 0.3; σ_d critical value = 15 μm.

[0231] In each trajectory Γ_i, the above indicators are compared node by node. If any indicator exceeds the threshold for k consecutive nodes (k ≥ 3), it is determined that there is an organizational anomaly in the trajectory.

[0232] Based on the organizational anomaly determination results of all trajectories, a defect prediction matrix M is constructed to describe the organizational risk distribution within the entire anomaly region Ω_abn.

[0233] The matrix construction method includes: M is an n×m size matrix; rows represent trajectory numbers. ; Columns represent time steps on the path. Each element M_ij is defined as follows: M_ij = 0: normal tissue; 1: slight abnormality (1 indicator exceeds the limit); 2: moderate abnormality (2 indicators exceed the limit); 3: severe abnormality (3 indicators exceed the limit).

[0234] The M matrix can be visualized using a heatmap, with the color transitioning from green (0) to red (3), which visually indicates the intensity of the anomaly.

[0235] Using the defect prediction matrix M as the objective function constraint input, a parameter optimization model is established. Its core is to minimize the number of abnormal trajectories and the sum of defect levels, while simultaneously satisfying the process rationality constraint.

[0236] Assuming the defect prediction matrix M is a two-dimensional array of n trajectories and m time steps, the objective function is defined as follows: Where: x is the extrusion parameter vector to be optimized; N_abn represents the total number of abnormal trajectories; ΣM_ij is the cumulative defect level of all trajectories; α and β are weighting factors, with α / β generally set to 1 / 10~1 / 5, reflecting the emphasis on the balance between the frequency and intensity of abnormalities. This objective function allows for the reduction of the risk of concentrated distribution of severe abnormalities while retaining some acceptable minor abnormalities.

[0237] Define the variables to be optimized and the process boundary conditions to form the optimization variable space.

[0238] Definition of variable set:

[0239] Extrusion speed v: range 0.5~3.0 m / min;

[0240] mold temperature :scope ;

[0241] billet temperature :scope ;

[0242] Extrusion ratio λ: range 6~18.

[0243] Define the optimization vector as: .

[0244] Material plasticity range constraint: Ensure that the deformation rate and temperature fall within the allowable processing range of the material;

[0245] Mold and equipment limits: Ensure the temperature resistance of the heating system and mold materials;

[0246] Process redundancy boundary: The integrity of profile forming shall not be reduced due to excessively low speed or excessively high ratio.

[0247] To efficiently solve the above multivariable nonlinear optimization problem, the following types of algorithms can be used in this implementation:

[0248] Genetic Algorithm (GA): Initialize the population (each individual is a group x); evaluate fitness based on the objective function f(x); generate a new generation through crossover and mutation operations; repeat the iteration until the objective function converges or the maximum number of generations is reached. It is suitable for solving non-convex, multimodal, and complex objective functions.

[0249] Particle Swarm Optimization (PSO): Each particle represents a parameter solution x; particles fly in the parameter space, guided by historical best positions and swarm best positions; it can quickly find local or global optimal solutions.

[0250] Gradient descent: It can be used for rapid descent when the objective function is continuously differentiable; however, it is prone to getting trapped in local optima and is suitable for simplifying models and quickly evaluating them.

[0251] After the algorithm converges, it outputs the optimal squeezing parameter vector: x_opt is used as the input for the next round of simulation or actual process, thus forming a complete optimization closed-loop path: "simulation - prediction - optimization - update".

[0252] Using the optimal set of extrusion parameters as the initial input for updating the simulation process, the complete thermo-mechanical coupling simulation model process is re-executed.

[0253] The optimal parameter set is represented as: ;in: Optimized extrusion speed (unit: m / min); Optimized mold temperature (unit: °C); Optimized billet temperature Optimized extrusion ratio (dimensionless).

[0254] Update the above parameters to the finite element simulation platform constructed in step S400 to ensure that the thermal boundary conditions, friction model, contact conditions, etc. remain consistent, so as to facilitate the comparison of results before and after.

[0255] Based on the rigid-plastic constitutive model, start the thermo-mechanical coupled solver and execute the following solution process:

[0256] Iterative updates of material flow velocity field, stress-strain field, and thermal field;

[0257] Output the material state parameters at each time step;

[0258] Record key data such as flow path, pressure distribution, and weld surface pressure.

[0259] The output simulation results include: equivalent stress σ_eq(x, y, z), equivalent strain ε_eq(x, y, z), temperature field T(x, y, z), and trajectory path τ_i(t).

[0260] Based on the dynamic recrystallization model and grain refinement equation established in step S300, the evolution of microstructure parameters is analyzed on the simulation results.

[0261] Grain size evolution:

[0262] .

[0263] Recrystallization volume fraction:

[0264] .

[0265] Set the following judgment threshold:

[0266]

[0267] Any area that meets any of the above three criteria is considered an unqualified area and must proceed to the next step of scanning analysis.

[0268] Extract areas that do not meet forming standards in three-dimensional space, focusing on areas with complex structures, drastic changes in flow paths, and sensitive welding pressure, and perform organizational risk scanning and spatial positioning.

[0269] For each node (x, y, z), construct the organization evaluation vector: U(x, y, z) = {d, X_drex, ΔT}.

[0270] Using rule-based judgment to mark: Flag(x, y, z) = 1: meets any abnormal condition; Flag(x, y, z) = 0: normal tissue area.

[0271] Spatial clustering algorithms (such as DBSCAN or 3D connected component analysis) are used to integrate and aggregate adjacent outliers to identify a potential weakened region Ω_weak_i.

[0272] The following metrics will be output for each region:

[0273] The coordinates of the center of space are (x_c, y_c, z_c);

[0274] Anomaly type identifier (coarsening, non-crystallization, heat concentration);

[0275] Region volume V_i (used to determine the degree of influence).

[0276] Information on all identified potential areas of organizational weakness is output as structured data for subsequent processing.

[0277] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method of simulation of the process of extrusion of non-symmetrical aluminium profiles, characterised by the fact that: Comprise: S100, establish a two-dimensional or three-dimensional geometric model of the target asymmetric aluminum profile, and perform feature recognition on the cross section to obtain a set of key structural parameters P; S200, according to the parameter set P, construct a multi-cavity welding die model, and combine material properties and thermal-mechanical boundary conditions to establish a multi-cavity welding die thermal-mechanical coupling constitutive simulation model containing equivalent strain, equivalent stress and thermal conduction coupling; S300, input the target aluminum alloy grade, heat treatment state and initial grain size, construct a dynamic recrystallization model and grain refinement evolution equation, and generate a set of organization parameters O; S400, set the initial conditions of extrusion, including extrusion speed , mold temperature , billet temperature and extrusion ratio , and combine the multi-cavity welding die thermal-mechanical coupling constitutive simulation model obtained in step S200 to use finite element simulation analysis to build a full-process thermal-mechanical coupling simulation system; S500, compare the initial simulation result obtained in step S400 with the historical stable working condition, identify an abnormal region, and extract a local flow trajectory feature vector of the abnormal region ; S600, the feature vector F a The mapping is coupled with the set of organization parameters O, backtracks the microstructure evolution path that leads to the anomaly, and forms a defect prediction matrix M; S700, according to the defect prediction matrix M, numerical inversion is performed to optimize the input extrusion parameter set for the optimal parameter set ; S800, re-run the simulation process to verify whether the stress and strain fields and grain size of the asymmetric aluminum profile in each region meet the forming standard under the optimal parameter set, and locate the potential organizational weakening position for pretreatment or online quality detection reference.

2. The asymmetric aluminum profile extrusion process simulation method of claim 1, characterized in that: The S100 comprises: S101, based on the actual product design drawings or scanning modeling data, generate a two-dimensional cross section or three-dimensional entity structure of the target asymmetric aluminum profile, and perform mesh division processing on the geometric model; S102, adopt an edge gradient-based contour recognition algorithm to perform curvature recognition and corner detection on the cross section of the two-dimensional or three-dimensional geometric model, and extract a set of feature regions in the profile structure, including flanges, stiffeners, thickening regions and cavity boundaries; S103、based on the extracted feature region set, using principal component analysis and spatial projection analysis method, calculate the quantitative index of asymmetry, including geometric eccentricity , structure imbalance coefficient and flow resistance difference ratio ΔR; S104、to 、 and The structural quantitative index constitutes the key structural parameter set P.

3. The asymmetric aluminum profile extrusion process simulation method of claim 1, wherein: The S200 comprises: S201, according to the profile structure parameters extracted in the parameter set P, construct a multi-cavity welding die structure model with multiple independent split cavities and welding zones; S202, according to the thermodynamic physical property parameters of the selected aluminum alloy material, including density, thermal conductivity, specific heat capacity, stress-strain relationship and dynamic recrystallization parameters, establish a material database model; S203, setting thermal-force boundary conditions, including mold temperature , blank temperature , friction factor f, interface heat conduction coefficient h, radiation coefficient r, and contact type; S204, based on the rigid-plastic constitutive relationship, establish a multi-cavity welding die thermal-mechanical coupling constitutive simulation model for the whole process simulation of material flow, organization evolution and heat conduction in the multi-cavity welding die region.

4. The asymmetric aluminum profile extrusion process simulation method of claim 1, wherein: The S300 comprises: S301, input the target aluminum alloy grade, heat treatment state and initial grain size; S302, construct a dynamic recrystallization critical strain model based on equivalent strain rate, temperature and material constant, adopt exponential function form to express the critical strain The relationship between temperature T, equivalent strain rate , and is defined as: Wherein a, n, Q are aluminum alloy material constants, and R is the gas constant; S303、Establish the grain size evolution equation, and express the dynamic recrystallization grain size d_drex as the initial grain size d o as a function of the strain stress state, in the form: wherein , m, are empirical fitting coefficients, is a grain scaling factor, reflecting the material type; m is a strain rate influence coefficient; is the grain evolution activation energy.

5. The asymmetric aluminum profile extrusion process simulation method of claim 1, wherein: The S400 comprises: S401, set an initial parameter set of a target extrusion process, including an extrusion speed , a die temperature , a billet temperature , and an extrusion ratio ; S402, perform finite element simulation on the constructed multi-cavity welding die thermal-mechanical coupling constitutive simulation model, complete mesh division, physical property mapping and initial state loading, and use a rigid-plastic solver to perform material-die system initialization; S403, construct a three-dimensional thermal-mechanical coupling analysis system, wherein the thermal field is calculated through the heat conduction equation and internal heat source term, the mechanical field uses the equivalent stress-strain relationship to update the material deformation behavior, and the thermal-mechanical iterative coupling is realized; S404, set the simulation control strategy, including the incremental time step Δt, the termination condition, the contact algorithm and the convergence criterion.

6. The asymmetric aluminum profile extrusion process simulation method of claim 1, wherein: The S500 comprises: S501, divide the qualified extrusion simulation data formed under similar materials, dies and process parameters by region and store them in a standardized manner, containing stable strain field, stress field, temperature field and flow velocity field distribution information; S502, difference comparison between the obtained initial simulation results and the historical stable working condition database, using a space point-to-point error mapping and shape matching algorithm to identify abnormal areas with deviations, and the judgment indexes include strain deviation Δε, velocity disturbance Δv, temperature mutation ΔT and abnormal welding pressure Δp; S503, flow trajectory tracking of the identified abnormal areas, using particle tracking method to simulate the path of multiple tracer points along the metal flow direction to form a local flow trajectory line group; S504, extract the flow trajectory feature vector V of each trajectory line, including trajectory length L, average speed , flow rate gradient , path curvature κ, relative inlet offset Δx, as the behavior feature representation data of the abnormal area.

7. The asymmetric aluminum profile extrusion process simulation method of claim 1, wherein: The S600 includes: S601, Transfer the flow trajectory feature vector of the abnormal region Spatially aligned with the corresponding set of organizational parameters O, data coupling is achieved through grid node mapping; S602, based on the grain size evolution model and the recrystallization volume fraction model, a track-driven microstructure inversion function group is constructed to predict the microstructure evolution trend under different strain paths; S603, set the microstructure evolution threshold conditions, including the grain coarsening threshold d_max, the incomplete recrystallization threshold X_min and the local non-uniformity σ_d, as the defect judgment basis; S604, generate a defect prediction matrix M according to the prediction results, and each element in M represents the potential defect level of the corresponding trajectory at different time steps.

8. The asymmetric aluminum profile extrusion process simulation method of claim 1, wherein: The S700 includes: S701, taking the abnormal trajectory corresponding parameters marked in the defect prediction matrix M as the objective function input, and constructing an extrusion parameter regulation model based on multi-objective optimization; S702, select the extrusion speed, mold temperature, billet temperature and extrusion ratio as the optimization variables, and define the process boundary and constraint conditions; S703, apply genetic algorithm, particle swarm optimization or gradient descent algorithm to iteratively search for the parameter combination with the minimum defect risk; S704, output the optimal extrusion parameter set.

9. The asymmetric aluminum profile extrusion process simulation method of claim 1, wherein: The S800 includes: S801, taking the optimal parameter set as the initial input of simulation, re-running the thermal-mechanical coupling simulation process to obtain the global stress, strain and temperature field distribution; S802, combined with the microstructure evolution model, calculate the grain size and recrystallization volume fraction of each region to evaluate whether the established forming quality standard is met; S803, implement microstructure performance scanning on the complex structure parts of the profile, extract the recrystallization rate and grain abnormal coarsening indexes, and locate the potential microstructure weakening area; S804, output the weakening area spatial coordinates and microstructure evaluation data.

Images