Composite laminated plate curing warping control method based on multi-scale data processing

Abstract

The application discloses a composite laminate curing warping control method based on multi-scale data processing, and belongs to the technical field of data processing. The method constructs a meso-macro bidirectional multi-scale coupling simulation model, combines a curing degree related cohesive force interface and a curing coupling anisotropic constitutive, improves residual stress and warping prediction accuracy, establishes a multi-index normalization weighted quantification system, objectively and automatically divides convex and concave deformation sensitive areas, matches a partitioned differential temperature control closed-loop iterative optimization strategy, balances local deformation correction and global curing uniformity, outputs a partitioned curing process that can be directly applied, reduces physical test molds, and realizes high-precision and low-cost warping, delamination and interface debonding suppression and molding full-process automatic closed-loop management and control.

Description

Technical Field

[0001] This invention relates to a method for controlling the curing warpage of composite laminates based on multi-scale data processing, belonging to the field of data processing technology. Background Technology

[0002] Composite laminates possess excellent properties such as high specific strength, high specific modulus, and corrosion resistance, and are now widely used in load-bearing structural components of high-end equipment in aerospace, rail transportation, and other industries. The mainstream molding method is autoclave curing. During the complete curing cycle of heating, isothermal treatment, and cooling, the resin matrix undergoes cross-linking shrinkage, and the inherent differences in thermal expansion properties between fibers and resins generate unevenly distributed residual stresses within the component. This leads to molding defects such as out-of-plane warping, interlayer delamination, and micro-debonding at the fiber-resin interface, significantly reducing the dimensional accuracy and mechanical load-bearing reliability of the component. Therefore, accurate prediction and controllable process optimization of laminate curing warping have become critical issues that the industry urgently needs to address.

[0003] Current methods for predicting and controlling curing warpage in composite laminates still have many unavoidable shortcomings. Traditional simulation analysis only builds a single macroscopic model for calculation, without constructing a microscopic unit cell that can characterize the fiber and resin two-phase structure. It usually uses a constant stiffness model to describe the bonding state of the fiber-resin interface, which cannot reproduce the dynamic evolution process of interface slippage and micro-debonding throughout the resin curing process. At the same time, the simulation calculation directly uses the fixed modulus and coefficient of thermal expansion of the resin after complete curing, ignoring the objective law of the continuous nonlinear change of material properties during the transition of the resin from a viscous liquid state to a rubber state and finally to a glassy state. It also fails to capture the stress gradient in the layup thickness direction, which ultimately leads to the distortion of residual stress and molding warpage simulation results, and the prediction error for delamination and interface debonding defects is relatively large. Even if a few existing solutions introduce multi-scale simulation approaches, they can only achieve one-way transmission of microscopic parameters to the macroscopic model. They lack a two-way linkage mechanism that allows the physical field of the macroscopic high-risk region to be substituted back to the microscopic unit cell to complete parameter correction. The microscopic structural features and macroscopic forming deformation cannot be matched and coupled, further amplifying the simulation calculation deviation. In addition, the existing modeling methods require the individual configuration of material parameters for different layup angles, which is cumbersome and results in low overall simulation efficiency.

[0004] In determining the risk areas of molding, the industry has long relied on operators to identify high-risk and stratified warping areas based on their experience and stress and deformation cloud maps. There is no unified, standardized, and quantitative evaluation system. It is impossible to simultaneously coordinate the multi-dimensional defect causes such as curing degree distribution, residual stress amplitude, stress gradient, and warping deformation gradient. The dimensions of various evaluation indicators are inconsistent, making it difficult to conduct horizontal comparisons. It is also impossible to automatically distinguish between warping areas with different deformation trends, such as convex and concave areas. The risk identification results are greatly affected by subjective human factors, have poor reproducibility, and cannot provide stable and quantitative data support for precise process optimization in different zones.

[0005] In the process optimization stage, a uniform temperature profile is generally used to control the molding process. The entire component is subjected to the same heating rate, holding time, and peak temperature, which makes it impossible to implement differentiated temperature compensation for local protrusions and depressions. If the temperature control parameters of a single zone are significantly adjusted after local warping exceeds the tolerance, it is easy to cause an imbalance in the curing progress of different areas, resulting in secondary defects such as cross-zone stress concentration and uneven curing. When the component has both protrusions and depressions with combined warping, the existing technology lacks standardized and coordinated temperature control constraints, making it difficult to balance the need for local deformation correction with the synchronicity of global curing, and the process iteration convergence is difficult.

[0006] Furthermore, existing technologies heavily rely on repeated physical mold trials and adjustments to the temperature control scheme. The trial production process consumes a large amount of prepreg raw materials and occupies autoclave equipment time, resulting in high overall R&D and trial production costs. At the same time, simulation models, material calibration parameters, and optimized temperature control processes lack a standardized archiving and reuse system. New components in the same series need to undergo a complete set of simulations and process adjustments repeatedly, which lengthens the product development cycle. Moreover, optimized temperature control schemes often require changes to the layup design and reprocessing of special molds before they can be implemented, further increasing the cost of process modification.

[0007] In summary, existing technologies suffer from a series of drawbacks, including limited simulation scale, oversimplification of material parameters, strong subjectivity in deformation risk identification, the potential for secondary defects due to uniform temperature control across the entire domain, high product trial production costs, and long new product development cycles. These limitations make it difficult to achieve high-precision, low-cost automated closed-loop control of curing warpage in composite laminates. Summary of the Invention

[0008] The technical problem this invention aims to solve is to address the above-mentioned shortcomings by providing a method for controlling the curing warpage of composite laminates based on multi-scale data processing. This method achieves high-precision prediction of the forming field by constructing a bidirectional multi-scale coupled simulation model (mesoscopic-macroscopic) and a curing coupled anisotropic constitutive model. It objectively quantifies and delineates deformation-sensitive zones using a multi-index normalized weighted system, and employs a zoned differentiated temperature control closed-loop iterative optimization strategy. The final output is a zoned curing process that can be directly implemented. This method accurately suppresses component warpage, delamination, and interface debonding defects while reducing the number of physical mold trials and shortening the new product development cycle. It achieves high-precision, low-cost, automated closed-loop control of the entire laminate curing and molding warpage process.

[0009] To solve the above technical problems, the present invention adopts the following technical solution:

[0010] A method for controlling curing warpage of composite laminates based on multi-scale data processing includes the following steps:

[0011] Step S1: Calibration of basic material parameters;

[0012] Through mechanical experiments, thermal performance tests, and curing kinetics DSC tests, elastic parameters, coefficients of thermal expansion, and curing reaction kinetic parameters of fibers and resin matrices were obtained. Equivalent orthogonal coefficients of thermal expansion and curing shrinkage reference parameters were calculated at different layup angles of 0°, 45°, 90°, and -45°. A material parameter database was established, including fiber geometric characteristic parameters, dry fiber mechanical property parameters, resin matrix mechanical property parameters, thermal expansion property parameters, resin curing kinetics and curing shrinkage parameters, and cross-scale equivalent homogenization derived parameters.

[0013] Step S2: Construct a three-level multi-scale coupled simulation model with bidirectional parameter transfer between the mesoscopic, mesoscopic, and macroscopic dimensions. First, construct a mesoscopic unit cell of RVE with a solidification-related cohesive interface. After homogenization calculation, obtain the equivalent parameters of each ply and complete the mesoscopic standardized single ply model. Then, obtain the macroscopic laminate model by stacking the actual plies. Finally, construct a bidirectional data link for forward parameter transfer and downward substitution correction of equivalent physical properties by macroscopic high gradient field quantity to achieve multi-scale iterative linkage.

[0014] Step S3, Finite element multiphysics coupling setup;

[0015] Simultaneously activate the transient heat conduction field analysis, resin curing dynamic field analysis, and nonlinear mechanical field analysis. Incorporate the ply anisotropy-curing shrinkage coupled constitutive equation with curing degree as the internal variable, assign curing heat generation parameters, superimpose thermal strain and total self-generated strain of curing shrinkage, load the autoclave molding boundary and the reference temperature control curve, configure the nonlinear convergence criterion and cross-scale data interaction channel, and configure the three-field coupled solver of transient heat conduction field, resin curing dynamic field and nonlinear mechanical field.

[0016] Step S4: Multi-field coupling solution, quantization identification of deformation-sensitive areas;

[0017] A cross-scale bidirectional feedback channel is opened to solve the transient heat conduction field, resin curing dynamic field, and nonlinear mechanical field simultaneously; four indicators are extracted: curing degree spatial fluctuation amplitude, element three-dimensional equivalent residual stress peak value, adjacent mesh stress gradient, and adjacent mesh warping displacement gradient, and the dimension is eliminated by extreme value normalization; the comprehensive sensitivity score of the element is calculated by weighting, and the element is divided into three levels of partitions: non-sensitive area, secondary sensitive area, and high sensitive area, and the convex / concave deformation type of the partition is marked according to the positive and negative values ​​of the element average total strain;

[0018] Step S5: Closed-loop iterative optimization of zone-specific temperature curves;

[0019] Import partition data from non-sensitive, secondary sensitive, and highly sensitive areas to construct a partition defect database; formulate quantitative fine-tuning rules for heating rate, holding time, and peak temperature for raised and recessed partitions respectively; when multiple partitions with both raised and recessed areas coexist, implement five collaborative adjustment principles: global temperature steady state, concave-convex reverse offset, adjacent temperature difference limit, curing degree priority, and multiple rounds of small iterations; after updating the temperature control parameters, return to step S3 for recalculation; determine convergence based on three criteria: single partition warpage tolerance, global curing degree difference, and maximum iteration rounds; if convergence fails, iterate cyclically.

[0020] Step S6: Output the optimal zone curing process scheme;

[0021] Lock in the optimal temperature control parameters for each zone, plot the zone temperature-time curve, and simultaneously export the global field cloud map and quantitative verification data table. Integrate these into a process document that can be directly used for autoclave zone molding. Unify the archived material library, simulation model, and process parameters for reuse in the same series of components.

[0022] Furthermore, step 2 specifically includes the following steps:

[0023] Step S2.1: Construct a micro-fiber-resin unit cell model;

[0024] Step S2.2: Solve the micro-model and complete the positive equivalent homogenization parameter mapping;

[0025] Step S2.3, mesoscopic single-layer assembly, building a macroscopic laminate solid model;

[0026] Import the design drawings of the target composite laminate component and read the actual ply stacking sequence, thickness of each single layer, and overall shape of the component.

[0027] Call the mesoscopic orthotropic anisotropic single-layer model corresponding to different fiber angles, and complete the layer-by-layer stacking assembly according to the actual engineering layup sequence;

[0028] By merging all ply geometry features, a complete macroscopic laminate 3D solid model is constructed. Simultaneously, the mold contact surface and clamping and fixing area are divided, and the boundary constraint positions of the autoclave forming process are preset to complete the construction of the top macroscopic structural carrier.

[0029] Step S2.4: Establish a two-way parameter feedback transmission link between the microscopic, mesoscopic, and macroscopic levels.

[0030] Furthermore, step 2.1 specifically includes the following steps:

[0031] Step S2.1.1: Retrieve basic material parameters;

[0032] Retrieve the basic material parameters calibrated in step S1, including fiber diameter, actual fiber volume fraction of the component, and fiber arrangement parameters;

[0033] Step S2.1.2, RVE geometric partitioning modeling;

[0034] A periodic mesoscopic representative unit cell (RVE) geometric model was established in the finite element software platform to distinguish the independent geometric domains of the fiber phase and the resin matrix phase.

[0035] Step S2.1.3: Assigning values ​​to the two-phase material parameters;

[0036] The fiber phase and the resin matrix phase were respectively assigned the original mechanical, thermal expansion and curing kinetic parameters that had been calibrated by experiments. The fiber-resin interface contact constraint relationship was set to restore the micro-scale fiber constraint effect, resin curing shrinkage behavior and fiber / resin interface shear interaction mechanism.

[0037] S2.1.4, the fiber-resin interface contact constraint relationship is characterized by a curing degree-related cohesive interface constraint system, which characterizes the mechanical interaction between the fiber phase and the resin matrix phase. The curing degree-related dynamic cohesive interface system is constructed as follows:

[0038] First, a zero-thickness three-dimensional cohesive element is inserted between the outer surface of the fiber and the contact surface of the resin matrix. The mesh nodes on both sides are matched one-to-one through master-slave surface pairing to ensure the continuous transmission of interface stress and displacement.

[0039] The bonding behavior in the interfacial elastic stage is described by a traction-separation bilinear constitutive relation. The interfacial normal stiffness and bitangential shear stiffness are dynamically updated using an exponential evolution function with the resin's normalized degree of cure as the independent variable. The function form is as follows: ,in To fully solidify the interface reference stiffness, m is the experimentally fitted growth factor. The real-time curing degree of the resin is represented, with a value range of [value range missing]. , This refers to the resin's maximum degree of cure.

[0040] When the resin curing degree is low, the interface stiffness is small, allowing slight relative slippage between the fiber and the resin. As the curing reaction progresses, the interface stiffness increases synchronously, limiting the separation and slippage of the two phases and reflecting the evolution of the interface bonding state throughout the autoclave molding process.

[0041] Meanwhile, a secondary nominal stress combination failure criterion is introduced to determine the initiation of interface damage. The scalar damage variable is used to describe the interface stiffness degradation law. Based on the material parameters such as interface shear strength and interface fracture energy obtained by step S1, the interface compressive stress, shear stress and local micro-debonding behavior induced by resin curing shrinkage are accurately characterized.

[0042] The cohesive interface damage evolution is described by a linear degradation model. A scalar damage variable D is introduced to characterize the degree of interface stiffness degradation. The damage scalar D is calculated in segments based on the equivalent separation displacement of the interface. The value range of D is 0≤D≤1. D=0 represents no damage and complete bonding of the interface, and D=1 represents complete debonding and failure of the interface.

[0043] like , ;

[0044] like When D=0, the interface remains elastically bonded with no stiffness degradation;

[0045] like When D=1, the interface completely fails and the interface load-bearing stiffness becomes zero.

[0046] in, It is the interface equivalent separation displacement. It is the critical equivalent separation displacement at the onset of damage. It is the equivalent separation displacement due to complete interface failure and fracture. , For Macaulay brackets representing normal separation displacement, only tensile separation is involved in damage calculation; interface compression does not produce damage.

[0047] Real-time interface stiffness after introducing damage variables: .

[0048] Step S2.1.5: Apply periodic boundary conditions for each unit cell;

[0049] The periodic boundary condition assumes that the micro-fiber-resin representative volume element is a periodically repeating structure. Displacement correlation constraint equations are applied to the three sets of relatively parallel characteristic surfaces of the unit cell, including geometric pairing constraints, displacement compatibility constraints, and stress continuity constraints. Geometric pairing constraints identify the six sets of opposite faces of the unit cell and establish a one-to-one mapping relationship between corresponding nodes on the front and back faces. Displacement compatibility constraints require that the displacements of opposite face nodes in the same coordinate axis direction satisfy the linear correlation equation, allowing only uniform overall deformation of the unit cell and prohibiting non-coordinated separation, overlap, or misalignment of the unit cell boundary. Stress continuity constraints indirectly realize the continuous transmission of interface normal and tangential stresses through displacement correlation equations, simulating the real state of the infinite periodic arrangement of fibers and resins inside the composite material.

[0050] Furthermore, step 2.2 specifically includes the following steps:

[0051] Step S2.2.1, closely examine the function for solving homogeneous field quantities;

[0052] Load the RVE unit cell solution model with complete geometry, material, interface, and boundary condition settings;

[0053] Homogenized numerical solutions are performed on the mesoscopic unit cell model with periodic boundary conditions. When performing homogenized numerical solutions on the mesoscopic unit cell with applied periodic boundary conditions, the volume-averaged method is used to calculate the average stress and average strain of the unit cell. The volume integral of the stress and strain field over the mesoscopic global domain is divided by the total volume of the unit cell to obtain the equivalent average stress tensor representing the average mechanical response of the composite material. With equivalent mean strain tensor Where V is the total volume of the unit cell in the RVE, and dV is the volume of any infinitesimal element at any integration point within the unit cell. , To observe the local stress and strain tensors at specific field points;

[0054] Step S2.2.2, calculation function for equivalent physical properties of ply layup;

[0055] Based on the average stress and average strain, the orthogonal anisotropic equivalent stiffness matrix, equivalent thermal expansion coefficient, and equivalent curing shrinkage strain can be further solved to complete the positive equivalent parameter mapping of the mesoscopic to mesoscopic standardized orthogonal anisotropic single-layer model.

[0056] Step S2.2.3, Standardize the single-layer model construction function;

[0057] Based on the mapped equivalent parameters, a standardized orthogonal anisotropic single-ply model is generated. The anisotropic single-ply model can adapt to any fiber layup angle of 0°, ±45°, and 90°, accurately conveying the anisotropic characteristics of the layup brought about by the microscale interaction.

[0058] Step S2.2.4, Ply angle coordinate transformation function;

[0059] The standardized orthogonal anisotropic single-ply model incorporates a coordinate rotation transformation matrix to achieve coordinate transformation of material parameters at arbitrary ply angles. Let θ represent the angle between the principal fiber direction and the global X-axis of the model. An in-plane rotation matrix T is used to transform the principal axial stiffness, thermal expansion strain, and curing shrinkage strain of the material into the global coordinate system. ;

[0060] pass Complete the stiffness tensor transformation, through The orthogonal thermal expansion and curing shrinkage strain component conversion is completed, among which... To examine the orthogonal anisotropic stiffness matrix of the material principal axes obtained through homogenization, The equivalent stiffness matrix of the layers in the global coordinate system. The equivalent thermal expansion / curing shrinkage strain along the principal axis of the material. For strain components in the global coordinate system, inputting any 0°, ±45°, or 90° ply angle will automatically complete parameter conversion, enabling standardized batch assembly of single-layer units at different angles;

[0061] Step S2.2.5, Mesoscopic to Mesoscopic Forward Parameter Mapping Function;

[0062] Complete batch conversion and encapsulation of single-ply parameters from different angles, automatically transfer the complete set of equivalent parameters of micro-homogenization to the meso-scale, and realize the pre-assembly preparation of standardized single-ply batch assembly.

[0063] Furthermore, step 2.4 specifically includes the following steps:

[0064] Step S2.4.1: Solidify the automatic transmission link of positive parameters;

[0065] An automatic parameter calling interface is built across scales. During the simulation calculation startup phase, the equivalent parameters obtained by homogenization of the mesoscopic unit cell are automatically pushed to the mesoscopic single-layer model and the macroscopic laminate 3D solid model, realizing the automatic positive flow of material performance data between scales.

[0066] Step S2.4.2: Load the macroscopic multiphysics global result file;

[0067] After the three-dimensional solid model of the macroscopic laminate has completed the coupled solution of transient heat conduction, curing dynamics, and mechanics, the result file containing information of all mesh elements is read, and the equivalent stress of each finite element element is extracted in batches. Out-of-plane warping displacement The unit's real-time curing degree, its ply number, and its spatial coordinates are recorded. Units are grouped by component partition, and the average equivalent stress of each partition is calculated. Average warping displacement of the entire component This provides a baseline value for gradient determination;

[0068] Step S2.4.3: Automatically select high-gradient mesh cells according to the gradient determination rules;

[0069] Call the preset gradient threshold to calculate the stress relative gradient for each finite element. , is the relative gradient of deformation Finite element satisfies ≥Stress gradient threshold or If the value is greater than or equal to the deformation gradient threshold, it is initially marked as a high gradient element. Elements in stress concentration areas such as openings, fillets, and ply discontinuities are directly and forcibly marked without gradient calculation. The neighborhood radius is 3 times the single-layer thickness. Isolated single-point marked abnormal noise elements are removed, and continuous effective high gradient elements are retained. The number, spatial coordinates, ply to which they belong, element stress, out-of-plane warpage displacement, and real-time solidification degree original field quantity of all effective high gradient elements are collected as the reverse substitution data source.

[0070] Step S2.4.4: Complete the transfer of macroscopic to mesoscopic data according to the three-level hierarchical mapping matching rules;

[0071] Using the ply number and ply angle of the high gradient unit as matching indexes, the corresponding standardized orthotropic mesoscopic single-ply unit is located.

[0072] The curing degree and stress tensor of multiple adjacent high gradient units within the same layup are arithmetically averaged to reduce noise and eliminate single-point numerical oscillations.

[0073] The averaged global field quantity is stored in the data cache area of ​​the corresponding mesoscopic single-layer unit, completing the aggregation and temporary storage of macroscopic field variables to the mesoscopic scale;

[0074] Step S2.4.5: Mesoscopic to mesoscopic RVE unit cell matching and field quantity and load conversion;

[0075] Based on the fiber volume fraction and fiber arrangement of the mesoscopic single layer, retrieve periodic mesoscopic RVE unit cells of matching specifications;

[0076] Mapping assignment rules: Set the mesoscopic average degree of curing as the initial degree of curing of the entire RVE domain; convert the mesoscopic average stress into the equivalent surface force of the RVE outer contour; use the average strain obtained by converting macroscopic deformation as the input value of the periodic boundary condition of the RVE.

[0077] Batch optimization: When the parameter deviation between adjacent mesoscopic units is less than the iteration convergence threshold, the same set of mesoscopic RVEs is used to carry out homogenization solution to reduce redundant calculations;

[0078] Abnormal fallback handling: When the curing degree value exceeds the limit or there is no matching RVE model, the neighborhood mean or the same fiber volume fraction benchmark unit cell adaptation correction is called.

[0079] Homogenization is performed on the mesoscopic unit cell under the loading of macroscopic back substitution field and boundary load, and the equivalent orthogonal stiffness matrix, equivalent thermal expansion coefficient, and equivalent solidification shrinkage strain are updated by the volume averaging method to calculate and update the full set of cross-scale derived parameters.

[0080] Step S2.4.6, cross-scale iterative convergence determination and parameter backfilling;

[0081] Compare the updated parameters with the parameters of the previous iteration, calculate the relative error of each material component, and determine whether convergence has been achieved by combining the preset maximum number of iterations.

[0082] If convergence is achieved, the updated equivalent parameters will be backfilled up the mesoscopic level to the macroscopic high gradient unit, completing this reverse data backfill process.

[0083] If the convergence fails and the iteration limit is not reached, the macroscopic three-field coupling calculation is restarted using the updated equivalent parameters, and the reverse substitution module is restarted for iterative iteration.

[0084] If convergence is not achieved after reaching the maximum number of iterations, the macroscopic model is backfilled with the final microscopic homogeneous output parameters, the iteration is terminated, and a convergence indicator is retained.

[0085] Furthermore, step 3 specifically includes the following steps:

[0086] Step S3.1: Activate the multiphysics analysis module;

[0087] Transient heat conduction field analysis, resin curing dynamic field analysis, and nonlinear mechanical field analysis are simultaneously enabled on the finite element software platform. A coupled calculation framework for simultaneous solution of the three fields is established to realize the synchronous iterative calculation of temperature evolution, resin curing crosslinking reaction, and structural stress warping deformation throughout the entire component forming process.

[0088] Step S3.2: Embed the coupled constitutive model in the finite element software platform;

[0089] Implanting resin real-time curing degree into the finite element software platform The core internal variable is the ply anisotropic-curing shrinkage coupled constitutive equation; the orthogonal anisotropic stiffness matrix and the triaxial thermal expansion coefficient are defined as nonlinear updates with degree of curing; the coupled calculation relationship between thermal strain and curing shrinkage strain is established; and the laminate thickness layered integration algorithm is written into the calculation of three-dimensional residual stress at full thickness.

[0090] Step S3.3: Assign values ​​to the curing degree field and curing heat generation parameters;

[0091] Retrieve the curing kinetic parameters from the material database in step S1, define the control equation for the evolution of resin curing degree and the internal heat generation model of curing; batch assign the equivalent curing exothermic parameters of each ply obtained by micro-homogenization to the standardized orthogonal anisotropic single ply model to realize the bidirectional coupling transfer of exothermic curing reaction to the temperature field.

[0092] Step S3.4, Define the global strain field;

[0093] Two types of non-mechanical strains are superimposed: one is the orthogonal anisotropic thermal strain that varies with temperature and layup angle, and the other is the resin curing shrinkage coupled strain that is constrained by fibers and controlled in real time by the degree of curing, thus distinguishing the differential strain distribution characteristics within and between layers.

[0094] Step S3.5: Molding boundary conditions and process range loading;

[0095] Based on the actual molding conditions of the autoclave, apply the heat exchange boundary of the mold contact surface, the component clamping displacement constraint, and the heat transfer coefficient of the external surface environment; input the preset benchmark curing process range, including temperature curve parameters such as heating rate, holding temperature, holding time, and cooling rate, to complete the initial setting of the global initial temperature and initial curing degree field.

[0096] Step S3.6, Coupled solver parameter configuration;

[0097] The system sets the convergence criteria for the coupled iteration of transient heat conduction field, resin curing dynamic field, and nonlinear mechanical field, as well as the load step size and the number of nonlinear equilibrium iterations. It also opens a multi-scale data interaction channel, allowing the field variables of the high stress gradient region obtained from the macroscopic solution to be back-transferred to the mesoscopic and microscopic models, providing computational support for subsequent bidirectional parameter correction.

[0098] Furthermore, the specific constitutive equation for the ply anisotropy-curing shrinkage coupling is as follows:

[0099] Mathematical equation for the coupled evolution of degree of cure and modulus:

[0100] ;

[0101] in, It depends on the real-time curing degree of the resin The varying orthogonal anisotropic shear modulus tensor components. , To fully cure the equivalent elastic modulus and shear modulus of the underlay, The evolution coefficients for the experimental fitting were determined using the real-time curing degree of the resin. As an independent variable, the nonlinear variation law of stiffness gradually increasing from near-zero stiffness in the liquid state to the fully solidified reference stiffness is characterized by the exponential function.

[0102] Mathematical equation for the coupled evolution of degree of cure and coefficient of thermal expansion:

[0103] ;

[0104] in, The equivalent thermal expansion coefficient in the fully cured state. The evolution coefficients are the experimental fitting coefficients;

[0105] Formula for superimposed calculation of thermal strain and curing shrinkage strain:

[0106] Total nonmechanical strain along the principal axes of the local material in the layup The thermal strain caused by temperature change and the curing shrinkage strain caused by curing reaction are coupled and superimposed:

[0107] ;

[0108] in, The temperature difference between the current temperature and the reference temperature. It is the equivalent curing shrinkage strain that varies with the degree of curing;

[0109] Discrete calculation method for solving three-dimensional residual stress by layer-by-layer integration of laminate thickness:

[0110] Divide a single layer into N integral discrete thickness layers along the p-th layer thickness direction. Each layer has a thickness of micron. The equivalent residual stress of a single layer is obtained by integrating the local stress at any thickness location along the thickness and averaging the volume.

[0111] ;

[0112] Where h is the total thickness of the single-layer composite ply, the integration interval extends from the bottom surface coordinate z=0 to the top surface coordinate z=h, and p takes values ​​from 1 to N. It is the local stress tensor component calculated from the p-th discrete layer center point. It is the segment thickness corresponding to the p-th discrete layer.

[0113] Furthermore, step 4 specifically includes the following steps:

[0114] Step S4.1: Submit the multi-scale coupled model to complete the multi-physics simultaneous solution;

[0115] The three-field coupled solver for transient heat conduction field, resin curing dynamic field and nonlinear mechanical field configured in step S3 is invoked. Transient coupling calculations are carried out segment by segment according to the preset curing process load steps. The transient temperature field, resin curing degree field, three-dimensional residual stress field and out-of-plane warping displacement field of the whole domain are solved iteratively at the same time. During the solution process, the bidirectional parameter feedback channel of micro-meso-macro is kept open. The field variables of the macro high gradient region are substituted back to the micro unit cell in real time to complete the equivalent parameter iterative correction. The complete field data result file of each time step of the entire molding process is output.

[0116] Step S4.2: Extract the field data of the four quantitative evaluation indicators for the entire domain;

[0117] In the post-processing module, four types of physical field feature data of all grid elements of the model are extracted in batches: solidification degree spatial fluctuation amplitude, peak value of three-dimensional equivalent residual stress of the element, stress spatial gradient of adjacent grids of the element, and warping displacement spatial gradient of adjacent grids of the element. The extreme value normalization method is used to uniformly scale the four different dimensional indices of solidification degree spatial fluctuation amplitude, peak value of three-dimensional equivalent residual stress of the element, stress spatial gradient of adjacent grids of the element, and warping displacement spatial gradient of adjacent grids of the element to the dimensionless interval of [0,1] to eliminate the difference in the dimensionality of the indices and form a four-dimensional index dataset at the element level.

[0118] Step S4.3, Four-dimensional index weighted threshold quantization judgment calculation;

[0119] Based on the component service level, the risk weights of each item are configured, and a single-grid comprehensive sensitivity score S is obtained through linear weighted calculation, where 0≤S≤1; based on the score interval, three types of areas are automatically divided: high-sensitivity area, secondary-sensitivity area, and non-sensitivity area, generating a full-domain partitioned visualization cloud map;

[0120] In non-sensitive areas, S < 0.35, no mesh encryption is performed, and access to the macro-to-micro field quantity back-substitution link is not mandatory;

[0121] In the secondary sensitive region (0.35≤S<0.7), local mesh refinement is carried out, and cross-scale parameter iterative correction is selectively applied according to simulation accuracy requirements.

[0122] For high-sensitivity areas ≥0.7, mandatory finer mesh densification is implemented, and the macro-to-micro reverse data back-substitution link mentioned above is fixedly incorporated to carry out iterative optimization of equivalent parameters;

[0123] Step S4.4: Automatically determine the warping and deformation trends of each highly sensitive area;

[0124] The average total nonmechanical strain value of all units in the highly sensitive area is extracted, and the positive or negative value of the strain is used as the criterion: the average total strain is negative and it is determined to be an area of ​​excessive shrinkage and warping bulge; the average total strain is positive and it is determined to be an area of ​​hysteresis shrinkage and warping depression. Sensitive areas with different deformation trends are marked and classified.

[0125] Furthermore, step 5 specifically includes the following steps:

[0126] Step S5.1: Import deformation-sensitive area and molding defect data;

[0127] Read the three-level deformation sensitive area mesh partitioning results of high-sensitivity area, secondary-sensitivity area and non-sensitivity area output in step S4, and divide it into several molding temperature control sub-partitions with independent sensitive areas as units; extract the out-of-plane warping displacement, protrusion / depression defect type, three-dimensional equivalent residual stress peak value and curing degree spatial distribution data of each partition in batches, and establish a molding defect feature database for each temperature control partition;

[0128] Step S5.2: Establish rules for the quantitative fine-tuning amplitude of warpage deviation level and temperature parameter;

[0129] For warped and protruding areas with excessive resin shrinkage, select process parameters such as low heating rate, extended medium-temperature holding time, and reduced peak curing temperature; for warped and sunken areas with delayed resin shrinkage, select process parameters such as high heating rate, shortened medium-temperature holding time, and increased peak curing temperature; integrate all zone temperature parameters to form a complete global zone temperature control scheme.

[0130] Step S5.3, principle of coordinated adjustment of temperature parameters under multi-zone concave-convex coexistence operating conditions;

[0131] The principle of steady-state constraint of global temperature reference: the difference between the global average peak temperature of all zones and the original reference temperature is ≤3℃, and the fluctuation of the global average heating rate is ≤0.2℃ / min, to ensure that the overall curing rhythm of the component does not deviate significantly and to avoid the overall increase of residual stress in the whole domain;

[0132] The principle of reverse hedging adjustment for concave and convex zones: Convex zones uniformly adopt a relaxation correction strategy of slowing down, cooling down, and extending the heat preservation, while concave zones simultaneously adopt a reinforcement correction strategy of speeding up, heating up, and shortening the heat preservation; the adjustment range of each zone matches its own warp excess ratio, and the larger the excess, the higher the parameter correction weight, so as to achieve mutual hedging and cancellation of deformation in multiple areas.

[0133] Adjacent partition gradient difference constraint principle: For spatially adjacent concave and convex partitions, the peak temperature difference in a single iteration shall not exceed 8℃, and the heating rate difference shall not exceed 0.4℃ / min, to prevent excessive temperature difference and curing gradient at partition boundaries from causing new stress concentration;

[0134] The principle of prioritizing uniform curing degree: After each parameter adjustment, check the curing degree distribution across the entire domain. If the difference in normalized curing degree between different intervals exceeds 0.15, simultaneously compress the temperature adjustment range of all zones to prioritize ensuring the synchronization of curing across the entire domain, and then iteratively correct the warpage deviation step by step.

[0135] The principle of multi-round small-scale iterative correction: For composite defect conditions, a single extreme parameter tuning is not adopted. Instead, it is divided into 2 to 3 rounds of progressive iterative optimization. After each round of iteration completes the three-field coupling solution, the warping and solidification field data are updated, and the fine-tuning amplitude is dynamically corrected to avoid introducing new forming defects by changing parameters drastically at once.

[0136] Step S5.4, iterative loop linkage and parameter update mechanism;

[0137] After each round of fine-tuning and collaborative correction of the zone temperature parameters, the parameters of the global temperature process curve are updated and sent back to the multi-field coupling solution module in step S3. The three-field coupling calculation of transient heat conduction field, resin curing dynamic field and nonlinear mechanical field is carried out again. The warpage, stress gradient, curing degree fluctuation and defect status of each zone are updated again through the quantitative evaluation system in step S4.

[0138] Step S5.5, Criteria for determining the termination of global iteration convergence;

[0139] The system employs a triple termination criterion: partition compliance, global equilibrium, and iteration upper limit. For each individual partition, the warpage is considered compliant if the absolute amplitude of the simulated partition warpage deformation is less than or equal to the out-of-plane warpage displacement of the partition. The temperature curve is then locked. Global optimization convergence is achieved when the warpage of all partitions meets tolerance requirements and the maximum difference in curing degree between partitions is less than or equal to 0.1. A maximum of 6 iterations is set. If the system still does not fully meet the criteria even after reaching the iteration upper limit, the optimal compromise temperature control process curve for the entire region is output, and the residual deviation of each partition is marked, completing the overall partition temperature iteration optimization process.

[0140] Furthermore, step 6 specifically includes the following steps:

[0141] Step S6.1: Lock the convergent optimal partition temperature control parameter group;

[0142] Extract the final curing process parameters of each partition when the iteration terminates, and store the complete set of temperature control parameters, such as heating rate, medium temperature holding time, curing peak temperature, and cooling rate, for non-sensitive areas, warped protrusion sensitive areas, and warped depression sensitive areas, to form a convergent optimal parameter set.

[0143] Step S6.2: Generate complete temperature-time process curves for each region;

[0144] With molding time as the horizontal axis and temperature as the vertical axis, separate segmented curing temperature curves for three types of regions are plotted, and the temperature switching nodes and the start and end times of the constant temperature intervals are marked to form a visual partitioned temperature control curve map.

[0145] Step S6.3: Export the simulation verification data accordingly;

[0146] Simultaneously export the simulation results file under the optimal process, including the global temperature field, curing degree field, residual stress distribution cloud map, overall component warping displacement cloud map, maximum warping value, and quantitative data table of average residual stress and deformation gradient of each zone, as a basis for verifying the effectiveness of the process.

[0147] Step S6.4: Integrate and form standardized process technology documents;

[0148] The document integrates the boundaries of the partitioned areas, the temperature control curves of each area, the key points of the molding process, the simulation verification indicators, and the molding tolerance requirements into a standardized and solidified process plan document. It clarifies the on-site implementation requirements for autoclave partitioned heating, mold partitioned temperature control channels, or partitioned heat insulation pads, and can be directly used for actual molding production without modifying the mold surface and component layup structure.

[0149] Step S6.5: Archive the process plan and retain the simulation model parameters;

[0150] The optimal partitioning process documents, supporting simulation models, material parameter databases, and multi-scale equivalent mapping parameters are uniformly archived and stored, which facilitates direct reuse of composite laminate components of the same series and shortens the simulation iteration and process development cycle of similar products.

[0151] The present invention adopts the above technical solution and has the following technical effects compared with the prior art:

[0152] Existing technologies generally suffer from problems such as single scale, simplified material parameters, strong subjectivity in risk identification, easy induction of secondary defects in overall temperature control, and high trial production costs. This invention forms a complete closed loop from material testing and calibration, multi-scale refined simulation, quantitative assessment of deformation risks, iterative optimization of zoned processes to standardized process output. It connects the entire chain from microscopic interface mechanism analysis to macroscopic molding process optimization, comprehensively suppressing molding defects such as curing warpage, delamination, and fiber-resin interface debonding in composite laminates. The controllability of molding quality is significantly improved, as detailed below:

[0153] 1. Establish a three-level bidirectional coupled simulation framework of mesoscopic-mesoscopic-macroscopic to improve the accuracy of forming field quantity prediction from the root cause;

[0154] The innovative microscale model employs a curing degree-correlated dynamic cohesive interface model, where the interface stiffness evolves exponentially with the degree of resin crosslinking. This model couples interface damage degradation and failure determination mechanisms, replacing the traditional constant stiffness interface model. It accurately simulates the relative slippage and micro-debonding behavior between fibers and resin throughout the molding process, eliminating errors in the calculation of microscale equivalent material parameters. Simultaneously, periodic boundary conditions are applied to the RVE unit cell, and the volume averaging method is used to output orthogonal anisotropic equivalent physical properties at different layup angles, truly reflecting the material anisotropy characteristics caused by fiber arrangement and fiber volume fraction.

[0155] A cross-scale bidirectional parameter transfer link is established, in which the mesoscopic homogeneous parameters are forward transferred to the mesoscopic and macroscopic models in the early stage of simulation. After the macroscopic solution is completed, the stress and warpage high gradient elements are automatically selected, and the global temperature, degree of curing, stress and strain fields are substituted back to the mesoscopic unit cell to recalculate and correct the equivalent parameters. After the cross-scale iteration converges, the macroscopic model is backfilled, which solves the problems of warpage and interlayer residual stress prediction distortion caused by ignoring the microscopic interface interaction and using fixed material parameters throughout the simulation at a single macroscopic scale.

[0156] The mesoscopic layer has a built-in ply angle coordinate rotation matrix, which can quickly complete the conversion of ply stiffness, thermal expansion and curing shrinkage parameters at 0°, ±45° and 90°, realize batch rapid modeling of laminates, and balance simulation calculation efficiency and multi-anisotropic solution accuracy.

[0157] 2. A curing-coupled orthogonal anisotropic constitutive model is adopted to fully reproduce the nonlinear physical property evolution of the resin throughout the curing cycle;

[0158] By embedding the curing degree-modulus and curing degree-thermal expansion coefficient coupled evolution equations, the elasticity, shear modulus and thermal expansion coefficient in each direction are dynamically updated at each step of the three-field coupling. This accurately describes the change law of the resin from viscous liquid state, rubber state to glass state, with the stiffness increasing and the thermal expansion performance continuously decreasing. It abandons the simplified modeling method of existing technologies that only use constant parameters after complete curing.

[0159] A coupled calculation model of thermal strain and curing shrinkage strain is constructed, which simultaneously considers two types of self-generated deformation sources: temperature expansion and contraction and chemical shrinkage of resin crosslinking. This fully restores the dual causes of residual stress and significantly improves the accuracy of warpage deformation calculation.

[0160] A layered discrete integral algorithm based on layup thickness is introduced to solve the three-dimensional residual stress. The stress values ​​are sampled at multiple points along the single layer thickness and weighted averaged to capture the stress gradient between the surface and core of a single layer. This avoids underestimating the stress concentration between layers by taking values ​​at a single point, and improves the reliability of predicting delamination and interface debonding defects. A standardized nonlinear iterative convergence criterion is provided to ensure stable solution of the three coupled fields of thermo-curing-mechanical.

[0161] 3. A multi-indicator normalized weighted quantitative evaluation system enables objective and automated hierarchical identification of deformation-sensitive areas;

[0162] Meanwhile, four core evaluation indicators are selected: spatial fluctuation amplitude of curing degree, peak value of three-dimensional equivalent residual stress of unit, stress gradient of adjacent mesh, and warping displacement gradient of adjacent mesh. These indicators cover various warping defect causes such as uneven curing, stress concentration, and abrupt deformation, making the evaluation dimensions more comprehensive.

[0163] Extreme value normalization is used to eliminate the dimensional differences of the four types of indicators. The risk weight is calculated by linear weighting of the component working conditions to comprehensively score the sensitivity. The non-sensitive area, the secondary sensitive area and the high sensitive area are quantitatively divided into three levels. This replaces the traditional method of subjectively delineating risk areas based on human experience. The identification results are objective and repeatable.

[0164] The system automatically distinguishes between two types of deformation trends in the high-sensitivity area: bulge (excessive shrinkage) and depression (hysteresis) based on the positive or negative average strain of the unit. This provides a precise and directly accessible basis of partitioned data for subsequent partitioned temperature control optimization.

[0165] 4. Differentiated temperature control in different zones is optimized in a closed-loop manner, taking into account both local warpage correction and uniform curing across the entire area;

[0166] For two types of defects, namely protrusions and depressions, quantitative fine-tuning compensation rules are set for heating rate, holding time, and peak temperature to offset local deformation deviations and correct warping deviations.

[0167] For the working condition where multiple zones have composite defects of protrusions and depressions, five collaborative adjustment constraint principles are proposed to strictly limit the deviation of the temperature control benchmark across the entire domain and the temperature difference gradient between adjacent zones, prioritize ensuring uniform curing degree across the entire domain, and avoid cross-zone curing imbalance and new interlayer stress concentration caused by large parameter adjustments in a single zone.

[0168] An automated closed-loop iterative process is formed, which includes three-field coupling solution, sensitive area quantization and identification, partition temperature control parameter correction, and re-simulation verification. The process sets three convergence criteria: partition warpage tolerance, global curing degree difference, and maximum iteration rounds. The optimal partition temperature control scheme that balances deformation accuracy and curing consistency is automatically obtained. Detailed Implementation

[0169] An example of a method for controlling curing warpage of composite laminates based on multi-scale data processing includes the following steps:

[0170] Step S1: Calibration of basic material parameters;

[0171] Through mechanical tests, thermal performance tests, and curing kinetics DSC tests, the elastic parameters, coefficients of thermal expansion, and curing reaction kinetic parameters of the fiber and resin matrix were obtained. The equivalent orthogonal coefficients of thermal expansion and curing shrinkage reference parameters at different layup angles of 0°, 45°, 90°, and -45° were calculated, and a material parameter database was established. The material parameter database includes fiber geometric characteristic parameters, dry fiber mechanical property parameters, resin matrix mechanical property parameters, thermal expansion property parameters, resin curing kinetics and curing shrinkage parameters, and cross-scale equivalent homogenization derived parameters.

[0172] Fiber geometric characteristic parameters include:

[0173] Nominal diameter of fiber;

[0174] Component design fiber volume fraction;

[0175] Fiber arrangement parameters (square arrangement / hexagonal close packing, unit cell array spacing);

[0176] Nominal thickness of a single layer of unidirectional prepreg.

[0177] The mechanical properties of dry fiber filaments include:

[0178] Fiber axial elastic modulus and transverse elastic modulus;

[0179] Fiber axial Poisson's ratio, transverse Poisson's ratio;

[0180] Fiber shear modulus;

[0181] Fiber tensile strength, interfacial shear strength.

[0182] The mechanical properties of the resin matrix include:

[0183] Elastic modulus, shear modulus, and Poisson's ratio of the resin in its fully cured state;

[0184] Viscoelastic evolution parameters of resin at different degrees of curing;

[0185] Resin tensile, compressive, and shear yield properties.

[0186] Thermal expansion performance parameters include:

[0187] Coefficient of thermal expansion of pure resin matrix across the entire temperature range;

[0188] Axial and transverse coefficients of thermal expansion of dry fibers;

[0189] Orthogonal equivalent thermal expansion coefficients of unidirectional lay-up in the 0° and 90° directions as a function of temperature / degree of cure.

[0190] Resin curing kinetics and curing shrinkage parameters include:

[0191] Curing kinetics model parameters: reaction activation energy, pre-exponential factor, reaction order, and ultimate degree of cure;

[0192] Curing exothermic parameters: Evolution curves of heat release per unit mass of resin during curing and the exothermic rate;

[0193] Intrinsic cured volume shrinkage rate and shrinkage strain of resin as a function of degree of cure;

[0194] Equivalent curing shrinkage strain of monolayer composites with different fiber volume fractions.

[0195] Cross-scale equivalent homogenization derived parameters include:

[0196] Equivalent stiffness matrices for orthogonal anisotropy at 0°, 45°, 90°, and -45°;

[0197] The equivalent thermal expansion coefficients of the paving surface at various angles, both inside and outside the surface.

[0198] Reference values ​​for equivalent curing shrinkage strain of ply at various angles;

[0199] Correction coefficients for unit cell equivalent parameters used for macroscopic reverse substitution correction.

[0200] Step S2: Construct a three-level multi-scale coupled simulation model with bidirectional parameter transfer from mesoscopic to macroscopic levels;

[0201] Microscale: Taking periodic RVE unit cells composed of fiber and resin as the research object, we capture the microscale mechanism of fiber-resin interface interaction and resin curing shrinkage, and output equivalent material properties through homogenization calculation.

[0202] Mesoscale: Using unidirectional fiber single-layer layup as the unit, orthogonal anisotropic equivalent constitutive model is adopted as a standardized intermediate carrier for the transfer of mesoscopic equivalent parameters to macroscopic structure;

[0203] Macroscale: Taking the complete multi-layer stacked laminated plate component as the overall structure, global multi-field coupled simulation is carried out to obtain the overall stress and warping deformation results, and the macroscale field variables are reverse-substituted and corrected to the mesoscale.

[0204] Step S2.1: Construct a micro-fiber-resin unit cell model;

[0205] Step S2.1.1: Retrieve basic material parameters;

[0206] Retrieve the basic material parameters calibrated in step S1, including fiber diameter, actual fiber volume fraction of the component, and fiber arrangement parameters;

[0207] Step S2.1.2, RVE geometric partitioning modeling;

[0208] A periodic mesoscopic representative unit cell (RVE) geometric model was established in the finite element software platform to distinguish the independent geometric domains of the fiber phase and the resin matrix phase.

[0209] Step S2.1.3: Assigning values ​​to the two-phase material parameters;

[0210] The fiber phase and the resin matrix phase are respectively assigned the original mechanical, thermal expansion and curing kinetic parameters that have been calibrated by experiments, the fiber-resin interface contact constraint relationship is set, and the micro-scale fiber constraint effect, resin curing shrinkage behavior and fiber / resin interface shear interaction mechanism are restored.

[0211] S2.1.4, the fiber-resin interface contact constraint relationship is characterized by a curing degree-related cohesive interface constraint system, which characterizes the mechanical interaction between the fiber phase and the resin matrix phase. The curing degree-related dynamic cohesive interface system is constructed as follows:

[0212] First, a zero-thickness three-dimensional cohesive element is inserted between the outer surface of the fiber and the contact surface of the resin matrix. The mesh nodes on both sides are matched one-to-one through master-slave surface pairing to ensure the continuous transmission of interface stress and displacement.

[0213] The bonding behavior in the interfacial elastic stage is described by a traction-separation bilinear constitutive relation. The interfacial normal stiffness and bitangential shear stiffness are dynamically updated using an exponential evolution function with the resin's normalized degree of cure as the independent variable. The function form is as follows: ,in To fully solidify the interface reference stiffness, m is the experimentally fitted growth factor. The real-time curing degree of the resin is represented, with a value range of [value range missing]. , This refers to the resin's maximum degree of cure.

[0214] When the resin curing degree is low, the interface stiffness is small, allowing slight relative slippage between the fiber and the resin. As the curing reaction progresses, the interface stiffness increases synchronously, limiting the separation and slippage of the two phases and reflecting the evolution of the interface bonding state throughout the autoclave molding process.

[0215] Simultaneously, a secondary nominal stress combination failure criterion is introduced to determine the initiation of interface damage. The scalar damage variable is used to describe the interface stiffness degradation law. Based on the material parameters such as interface shear strength and interface fracture energy obtained by step S1, the interface compressive stress, shear stress and local micro-debonding behavior induced by resin curing shrinkage are accurately characterized.

[0216] The cohesive interface damage evolution is described by a linear degradation model. A scalar damage variable D is introduced to characterize the degree of interface stiffness degradation. The damage scalar D is calculated in segments based on the equivalent separation displacement of the interface. The value range of D is 0≤D≤1. D=0 represents no damage and complete bonding of the interface, and D=1 represents complete debonding and failure of the interface.

[0217] like , ;

[0218] like When D=0, the interface remains elastically bonded with no stiffness degradation;

[0219] like When D=1, the interface completely fails and the interface load-bearing stiffness becomes zero.

[0220] in, It is the interface equivalent separation displacement. It is the critical equivalent separation displacement at the onset of damage. It is the equivalent separation displacement of complete interface failure and fracture. , For Macaulay brackets representing normal separation displacement, only tensile separation is involved in damage calculation; interface compression does not produce damage.

[0221] Real-time interface stiffness after introducing damage variables: .

[0222] The effective stiffness of the interface normal and the effective stiffness of the interface tangential are simultaneously updated with the stiffness function of the curing degree evolution and the damage scalar. This can simultaneously characterize the interfacial elastic bonding, local micro-debonding and stiffness decay behavior caused by resin curing shrinkage, which is different from the existing technology that only uses a constant stiffness and uncoupled curing state interface damage model.

[0223] The fiber-resin interface contact constraint relationship adopts a dynamic cohesive interface constraint system that depends on the degree of curing. It fully reproduces the interface interaction evolution law between the fiber and the resin throughout the curing cycle, effectively eliminating the equivalent parameter calculation error caused by the traditional constant parameter interface model. It is an important technical support for realizing high-precision multi-scale coupled warp simulation.

[0224] Step S2.1.5: Apply periodic boundary conditions for each unit cell;

[0225] The periodic boundary condition is that the micro-fiber-resin representative volume element is a periodically repeating structure. Displacement correlation constraint equations are applied to the three sets of relatively parallel characteristic surfaces of the unit cell, including geometric pairing constraints, displacement compatibility constraints, and stress continuity constraints. Geometric pairing constraints identify the six sets of opposite faces of the unit cell and establish a one-to-one mapping relationship between the corresponding nodes of the front and back faces. Displacement compatibility constraints require that the displacements of the nodes on the opposite faces satisfy the linear correlation equation in the same coordinate axis direction, allowing only uniform overall deformation of the unit cell and prohibiting non-coordinated separation, overlap, or misalignment of the unit cell boundary. Stress continuity constraints indirectly realize the continuous transmission of interface normal and tangential stress through displacement correlation equations, simulating the real state of the infinite periodic arrangement of fibers and resin inside the composite material.

[0226] Step S2.2: Solve the micro-model and complete the positive equivalent homogenization parameter mapping;

[0227] Step S2.2.1, closely examine the function for solving homogeneous field quantities;

[0228] Load the RVE unit cell solution model with complete geometry, material, interface, and boundary condition settings;

[0229] Homogenized numerical solutions are performed on the mesoscopic unit cell model with periodic boundary conditions. When performing homogenized numerical solutions on the mesoscopic unit cell with applied periodic boundary conditions, the volume-averaged method is used to calculate the average stress and average strain of the unit cell. The volume integral of the stress and strain field over the mesoscopic global domain is divided by the total volume of the unit cell to obtain the equivalent average stress tensor representing the average mechanical response of the composite material. With equivalent mean strain tensor Where V is the total volume of the unit cell in the RVE, and dV is the volume of any infinitesimal element at any integration point within the unit cell. , To observe the local stress and strain tensors at specific field points.

[0230] Step S2.2.2, calculation function for equivalent physical properties of ply layup;

[0231] Based on the mean stress and mean strain, the orthogonal anisotropic equivalent stiffness matrix, equivalent thermal expansion coefficient, and equivalent curing shrinkage strain can be further solved, completing the positive equivalent parameter mapping of the mesoscopic to mesoscopic standardized orthogonal anisotropic single-layer model.

[0232] Step S2.2.3, Standardize the single-layer model construction function;

[0233] Based on the mapped equivalent parameters, a standardized orthogonal anisotropic single-ply model is generated. The anisotropic single-ply model can adapt to any fiber layup angle of 0°, ±45°, and 90°, accurately conveying the anisotropic characteristics of the layup brought about by microscale interaction.

[0234] Step S2.2.4, Ply angle coordinate transformation function;

[0235] The standardized orthogonal anisotropic single-ply model incorporates a coordinate rotation transformation matrix to achieve coordinate transformation of material parameters at arbitrary ply angles. Let θ represent the angle between the principal fiber direction and the global X-axis of the model. An in-plane rotation matrix T is used to transform the principal axial stiffness, thermal expansion strain, and curing shrinkage strain of the material into the global coordinate system. .

[0236] pass Complete the stiffness tensor transformation, through The orthogonal thermal expansion and curing shrinkage strain component conversion is completed, among which... To examine the orthogonal anisotropic stiffness matrix of the material principal axes obtained through homogenization, The equivalent stiffness matrix of the layers in the global coordinate system. The equivalent thermal expansion / curing shrinkage strain along the principal axis of the material. For strain components in the global coordinate system, input any 0°, ±45°, or 90° ply angle to automatically complete parameter conversion, enabling standardized batch assembly of single-layer units at different angles.

[0237] Step S2.2.5, Mesoscopic to Mesoscopic Forward Parameter Mapping Function;

[0238] Complete batch conversion and encapsulation of single-ply parameters from different angles, automatically transfer the complete set of equivalent parameters of micro-homogenization to the meso-scale, and realize the pre-assembly preparation of standardized single-ply batch assembly.

[0239] Step S2.3, Mesoscopic single-layer assembly, building a macroscopic laminate solid model;

[0240] Import the design drawings of the target composite laminate component and read the actual ply stacking sequence, thickness of each single layer, and overall shape of the component.

[0241] Call the mesoscopic orthotropic anisotropic single-layer model corresponding to different fiber angles, and complete the layer-by-layer stacking assembly according to the actual engineering layup sequence;

[0242] By merging all the layup geometry features, a complete macroscopic laminate 3D solid model is constructed. Simultaneously, the mold contact surface and clamping and fixing area are divided, and the boundary constraint positions of the autoclave forming process are preset to complete the construction of the top macroscopic structural carrier.

[0243] Step S2.4: Establish a two-way parameter feedback and transmission link between the mesoscopic, mesoscopic, and macroscopic levels;

[0244] Step S2.4.1: Solidify the automatic transmission link of positive parameters;

[0245] An automatic parameter calling interface is built across scales. During the simulation calculation startup phase, the equivalent parameters obtained by homogenization of the mesoscopic unit cell are automatically pushed to the mesoscopic single-layer model and the macroscopic laminate 3D solid model, realizing the automatic positive flow of material performance data between scales.

[0246] Step S2.4.2: Load the macroscopic multiphysics global result file;

[0247] After the three-dimensional solid model of the macroscopic laminate has completed the solution of the transient heat conduction, curing dynamics, and mechanical three-field coupling, the result file containing all mesh element information is read, and the equivalent stress of each finite element element is extracted in batches. Out-of-plane warping displacement The unit's real-time curing degree, its ply number, and its spatial coordinates are recorded. Units are grouped by component partition, and the average equivalent stress of each partition is calculated. Average warping displacement of the entire component This provides a baseline value for gradient determination.

[0248] Step S2.4.3: Automatically select high-gradient mesh cells according to the gradient determination rules;

[0249] Call the preset gradient threshold to calculate the stress relative gradient for each finite element. , is the relative gradient of deformation Finite element satisfies ≥Stress gradient threshold or If the value is greater than or equal to the deformation gradient threshold, it is initially marked as a high gradient element. Elements in stress concentration areas such as openings, fillets, and ply discontinuities are directly and forcibly marked without gradient calculation. The neighborhood radius is 3 times the single-layer thickness. Isolated single-point marked abnormal noise elements are removed, and continuous effective high gradient elements are retained. The number, spatial coordinates, ply to which they belong, element stress, out-of-plane warpage displacement, and real-time solidification degree original field quantity of all effective high gradient elements are collected as the reverse substitution data source.

[0250] Step S2.4.4: Complete the transfer of macroscopic to mesoscopic data according to the three-level hierarchical mapping matching rules;

[0251] Using the ply number and ply angle of the high gradient unit as matching indexes, the corresponding standardized orthotropic mesoscopic single-ply unit is located.

[0252] The curing degree and stress tensor of multiple adjacent high gradient units within the same layup are arithmetically averaged to reduce noise and eliminate single-point numerical oscillations.

[0253] The averaged global field quantity is stored in the data cache area of ​​the corresponding mesoscopic single-layer unit, completing the aggregation and temporary storage of macroscopic field variables to the mesoscopic scale.

[0254] Step S2.4.5: Mesoscopic to mesoscopic RVE unit cell matching and field quantity and load conversion;

[0255] Based on the fiber volume fraction and fiber arrangement of the mesoscopic single layer, retrieve periodic mesoscopic RVE unit cells of matching specifications;

[0256] Mapping assignment rules: Set the mesoscopic average degree of curing as the initial degree of curing of the entire RVE domain; convert the mesoscopic average stress into the equivalent surface force of the RVE outer contour; use the average strain obtained by converting macroscopic deformation as the input value of the periodic boundary condition of the RVE.

[0257] Batch optimization: When the parameter deviation between adjacent mesoscopic units is less than the iteration convergence threshold, the same set of mesoscopic RVEs is used to carry out homogenization solution to reduce redundant calculations;

[0258] Abnormal fallback handling: When the curing degree value exceeds the limit or there is no matching RVE model, the neighborhood mean or the same fiber volume fraction benchmark unit cell adaptation correction is called.

[0259] Homogenization is performed on the mesoscopic unit cell under the applied macroscopic back substitution field and boundary load, and the equivalent orthogonal stiffness matrix, equivalent thermal expansion coefficient, and equivalent solidification shrinkage strain are updated using the volume averaging method to calculate and update the full set of cross-scale derived parameters.

[0260] Step S2.4.6, cross-scale iterative convergence determination and parameter backfilling;

[0261] Compare the updated parameters with the parameters of the previous iteration, calculate the relative error of each material component, and determine whether convergence has been achieved by combining the preset maximum number of iterations.

[0262] If convergence is achieved, the updated equivalent parameters will be backfilled up the mesoscopic level to the macroscopic high gradient unit, completing this reverse data backfill process.

[0263] If the convergence fails and the iteration limit is not reached, the macroscopic three-field coupling calculation is restarted using the updated equivalent parameters, and the reverse substitution module is restarted for iterative iteration.

[0264] If convergence is not achieved after reaching the maximum number of iterations, the macroscopic model is backfilled with the final microscopic homogeneous output parameters, the iteration is terminated, and a convergence indicator is retained.

[0265] Step S3, Finite element multiphysics coupling setup;

[0266] Step S3.1: Activate the multiphysics analysis module;

[0267] Transient heat conduction field analysis, resin curing dynamic field analysis, and nonlinear mechanical field analysis are simultaneously enabled on the finite element software platform. A coupled calculation framework for simultaneous solution of the three fields is established to realize the synchronous iterative calculation of temperature evolution, resin curing crosslinking reaction, and structural stress warping deformation throughout the entire component forming process.

[0268] Step S3.2: Embed the coupled constitutive model in the finite element software platform;

[0269] Implanting resin real-time curing degree into the finite element software platform The anisotropic layup-curing shrinkage coupled constitutive equations are used as the core internal variables; the nonlinear update definitions of the orthogonal anisotropic stiffness matrix and the triaxial thermal expansion coefficient with degree of curing are completed; the coupled calculation relationship between thermal strain and curing shrinkage strain is established; and the layered integral algorithm for laminate thickness is incorporated for solving the three-dimensional residual stress of the full thickness.

[0270] The specific constitutive equation for the anisotropic plying-curing shrinkage coupling is as follows:

[0271] Mathematical equation for the coupled evolution of degree of cure and modulus:

[0272] ;

[0273] in, It depends on the real-time curing degree of the resin The varying orthogonal anisotropic shear modulus tensor components. , To fully cure the equivalent elastic modulus and shear modulus of the underlay, The evolution coefficients for the experimental fitting were determined using the real-time curing degree of the resin. As an independent variable, the nonlinear variation law of stiffness gradually increasing from near-zero stiffness in the liquid state to the fully solidified reference stiffness is characterized by the exponential function.

[0274] Fit coefficients By controlling the rate of stiffness increase, the modulus is updated in real time in each round of three-field coupling iteration within the finite element software platform, and the orthogonal anisotropic stiffness matrix is ​​dynamically corrected. This accurately captures the evolution of the mechanical properties of the resin from the viscous flow state, rubber state to the glass state, and reduces the residual stress and warpage simulation errors caused by fixed modulus modeling.

[0275] Mathematical equation for the coupled evolution of degree of cure and coefficient of thermal expansion:

[0276] ;

[0277] in, The equivalent thermal expansion coefficient in the fully cured state. The evolution coefficients are the fitting coefficients for the experiment.

[0278] In the initial stage of curing, the modulus approaches 0, and the resin is in a liquid state with no load-bearing capacity. The coefficient of thermal expansion gradually converges to the fully cured reference value as the degree of curing increases, fully reproducing the continuous evolution law of the thermal and mechanical properties of the resin from the liquid state, viscoelastic state to the glass state. The coefficient of thermal expansion in each direction is updated in real time, adapting to the objective law that the molecular thermal motion constraint becomes stronger during the transition of the resin from the liquid state and rubber state to the glass state, correcting the dynamic thermal strain value, and solving the problem of residual stress in molding and component warpage prediction deviation caused by the use of a constant coefficient of thermal expansion throughout the traditional simulation.

[0279] Formula for superimposed calculation of thermal strain and curing shrinkage strain:

[0280] Total nonmechanical strain along the principal axes of the local material in the layup The thermal strain caused by temperature change and the curing shrinkage strain caused by curing reaction are coupled and superimposed:

[0281] ;

[0282] in, The temperature difference between the current temperature and the reference temperature. It is the equivalent curing shrinkage strain that varies with the degree of curing.

[0283] The calculated total nonmechanical strain is substituted into the orthogonal anisotropic stress principal equation to solve the internal stress generated by the constraint, which fully reproduces the joint contribution of thermal expansion and contraction and resin chemical shrinkage to residual stress and molding warpage, and solves the problem of molding deformation prediction deviation caused by the solidification of traditional simulation parameters and the lack of strain terms.

[0284] Discrete calculation method for solving three-dimensional residual stress by layer-by-layer integration of laminate thickness:

[0285] Divide a single layer into N integral discrete thickness layers along the p-th layer thickness direction. Each layer has a thickness of micron. The equivalent residual stress of a single layer is obtained by integrating the local stress at any thickness location along the thickness and averaging the volume.

[0286] ;

[0287] Where h is the total thickness of the single-layer composite ply, the integration interval extends from the bottom surface coordinate z=0 to the top surface coordinate z=h, and p takes values ​​from 1 to N. It is the local stress tensor component calculated from the p-th discrete layer center point. It is the segment thickness corresponding to the p-th discrete layer.

[0288] First, a definite integral is used to average the distributed stress along the thickness direction as a function of curing degree and temperature to obtain the equivalent residual stress of a single layer. Then, relying on the discrete sampling characteristics of the finite element method, a layered summation numerical integration is used to complete the engineering solution. By splitting the sampled stress of multiple thin sheets and weighting and summing them, the stress difference between the surface and core of a single layer can be captured, accurately analyzing the three-dimensional interlayer residual stress distribution of the laminate. This avoids the error in predicting molding defects caused by representing the stress of the entire layer with a single point value, and improves the stress output capability of the entire ply anisotropic-curing shrinkage coupled constitutive model.

[0289] Step S3.3: Assign values ​​to the curing degree field and curing heat generation parameters;

[0290] Retrieve the curing kinetic parameters from the material database in step S1, define the control equation for the evolution of resin curing degree and the internal heat generation model of curing; batch assign the equivalent curing exothermic parameters of each ply obtained by micro-homogenization to the standardized orthogonal anisotropic single ply model to realize the bidirectional coupling transfer of exothermic curing reaction to the temperature field.

[0291] Step S3.4, Define the global strain field;

[0292] Two types of non-mechanical strains are superimposed: one is the orthogonal anisotropic thermal strain that varies with temperature and layup angle, and the other is the resin curing shrinkage coupled strain that is constrained by fibers and controlled in real time by curing degree, to distinguish the differential strain distribution characteristics within and between layers.

[0293] Step S3.5: Molding boundary conditions and process range loading;

[0294] Based on the actual molding conditions of the autoclave, apply the heat exchange boundary of the mold contact surface, the component clamping displacement constraint, and the heat transfer coefficient of the external surface environment; input the preset benchmark curing process range, including temperature curve parameters such as heating rate, holding temperature, holding time, and cooling rate, to complete the initial setting of the global initial temperature and initial curing degree field.

[0295] Step S3.6, Coupled solver parameter configuration;

[0296] The system sets the convergence criteria for the coupled iteration of transient heat conduction field, resin curing dynamic field, and nonlinear mechanical field, as well as the load step size and the number of nonlinear equilibrium iterations. It also opens a multi-scale data interaction channel, allowing the field variables of the high stress gradient region obtained from the macroscopic solution to be back-transferred to the mesoscopic and microscopic models, providing computational support for subsequent bidirectional parameter correction.

[0297] Step S4: Multi-field coupling solution, quantization identification of deformation-sensitive areas;

[0298] Step S4.1: Submit the multi-scale coupled model to complete the multi-physics simultaneous solution;

[0299] The three-field coupled solver configured in step S3, which solves the transient heat conduction field, resin curing dynamics field, and nonlinear mechanical field, is invoked. Transient coupling calculations are carried out segment by segment according to the preset curing process load steps. The transient temperature field, resin curing degree field, three-dimensional residual stress field, and out-of-plane warping displacement field are solved iteratively in parallel. During the solution process, the bidirectional parameter feedback channel between the micro-meso-macro scale is kept open. The field variables in the macro high gradient region are substituted back to the micro unit cell in real time to complete the equivalent parameter iterative correction. The complete field data result file of each time step of the entire molding process is output.

[0300] Step S4.2: Extract the field data of the four quantitative evaluation indicators for the entire domain;

[0301] In the post-processing module, four types of physical field feature data of all grid elements in the model are extracted in batches: solidification degree spatial fluctuation amplitude, peak value of three-dimensional equivalent residual stress of the element, stress spatial gradient of adjacent grids of the element, and warping displacement spatial gradient of adjacent grids of the element. The extreme value normalization method is used to uniformly scale the four different dimensional indices of solidification degree spatial fluctuation amplitude, peak value of three-dimensional equivalent residual stress of the element, stress spatial gradient of adjacent grids of the element, and warping displacement spatial gradient of adjacent grids of the element to the dimensionless interval of [0,1] to eliminate the difference in the dimensionality of the indices and form a four-dimensional index dataset at the element level.

[0302] Step S4.3, Four-dimensional index weighted threshold quantization judgment calculation;

[0303] Based on the component service level, the risk weights of each item are configured, and a single-grid comprehensive sensitivity score S is obtained through linear weighted calculation, where 0≤S≤1; based on the score interval, three types of areas are automatically divided: high-sensitivity area, secondary-sensitivity area, and non-sensitivity area, generating a full-domain partitioned visualization cloud map;

[0304] In non-sensitive areas, S < 0.35, no mesh encryption is performed, and access to the macro-to-micro field quantity back-substitution link is not mandatory;

[0305] In the secondary sensitive region (0.35≤S<0.7), local mesh refinement is carried out, and cross-scale parameter iterative correction is selectively applied according to simulation accuracy requirements.

[0306] For highly sensitive areas ≥0.7, mandatory finer mesh densification is implemented, and the macro-to-micro reverse data back-substitution link mentioned above is fixed, with iterative optimization of equivalent parameters carried out.

[0307] Step S4.4: Automatically determine the warping and deformation trends of each highly sensitive area;

[0308] The average total nonmechanical strain value of all units in the highly sensitive area is extracted, and the positive or negative value of the strain is used as the criterion: the average total strain is negative and it is determined to be an area of ​​excessive shrinkage and warping bulge; the average total strain is positive and it is determined to be an area of ​​hysteresis shrinkage and warping depression. Sensitive areas with different deformation trends are marked and classified.

[0309] Step S5: Iterative optimization of the zone-specific temperature curves;

[0310] Step S5.1: Import deformation-sensitive area and molding defect data;

[0311] Read the three-level deformation sensitive area mesh partitioning results of high-sensitivity area, secondary-sensitivity area and non-sensitivity area output in step S4, and divide it into several molding temperature control sub-partitions with independent sensitive areas as units; extract the out-of-plane warping displacement, convex / concave defect type, three-dimensional equivalent residual stress peak value and curing degree spatial distribution data of each partition in batches, and establish a molding defect feature database for each temperature control partition, so as to provide a quantitative input basis for subsequent fine adjustment of partition temperature parameters and multi-region collaborative optimization.

[0312] Step S5.2: Establish rules for the quantitative fine-tuning amplitude of warpage deviation level and temperature parameter;

[0313] For the two types of highly sensitive areas, a reverse compensation temperature regime is configured: for warped and protruding areas with excessive resin shrinkage, a low heating rate, extended medium-temperature holding time, and reduced curing peak temperature are selected; for warped and sunken areas with delayed resin shrinkage, a high heating rate, shortened medium-temperature holding time, and increased curing peak temperature are selected; all zone temperature parameters are integrated to form a complete global zone temperature control scheme.

[0314] Step S5.3, principle of coordinated adjustment of temperature parameters under multi-zone concave-convex coexistence operating conditions;

[0315] When multiple temperature-controlled zones of a component simultaneously exhibit combined warping defects of protrusions and depressions, it is prohibited to independently and significantly adjust the parameters of a single zone. This is to avoid disrupting the consistency of curing dynamics across the entire region and inducing abrupt changes in interlayer stress across zones. The following coordinated adjustment principles must be strictly followed:

[0316] The principle of steady-state constraint of global temperature reference: the difference between the global average peak temperature of all zones and the original reference temperature is ≤3℃, and the fluctuation of the global average heating rate is ≤0.2℃ / min, so as to ensure that the overall curing rhythm of the component does not deviate significantly and avoid the overall increase of residual stress in the whole domain.

[0317] The principle of reverse hedging adjustment for concave and convex zones: Convex zones uniformly adopt a relaxation correction strategy of slowing down, cooling down, and extending the heat preservation, while concave zones simultaneously adopt a reinforcement correction strategy of speeding up, heating up, and shortening the heat preservation; the adjustment range of each zone matches its own warp excess ratio, and the larger the excess, the higher the parameter correction weight, so as to achieve mutual hedging and cancellation of deformation in multiple areas.

[0318] Adjacent partition gradient difference constraint principle: For spatially adjacent concave and convex partitions, the peak temperature difference in a single iteration shall not exceed 8℃, and the heating rate difference shall not exceed 0.4℃ / min, to prevent excessive temperature difference and curing gradient at partition boundaries from causing new stress concentration.

[0319] The principle of prioritizing uniform curing degree is as follows: After each parameter adjustment, the curing degree distribution across the entire domain is checked. If the difference in normalized curing degree between different regions exceeds 0.15, the temperature adjustment range of all zones is compressed simultaneously to prioritize ensuring the synchronization of curing across the entire domain, and then the warpage deviation is corrected step by step in an iterative manner.

[0320] The principle of multi-round small-scale iterative correction: For composite defect conditions, single-time limit parameter tuning is not adopted. Instead, it is divided into 2 to 3 rounds of progressive iterative optimization. After each round of iteration completes the three-field coupling solution, the warping and solidification field data are updated, and the fine-tuning amplitude is dynamically corrected to avoid introducing new forming defects by changing parameters drastically at once.

[0321] Step S5.4, iterative loop linkage and parameter update mechanism;

[0322] After each round of fine-tuning and collaborative correction of the zone temperature parameters, the parameters of the global temperature process curve are updated and sent back to the multi-field coupling solution module in step S3. The three-field coupling calculation of transient heat conduction field, resin curing dynamic field and nonlinear mechanical field is carried out again. The warp, stress gradient, curing degree fluctuation and defect status of each zone are updated again through the quantitative evaluation system in step S4, forming a closed-loop iterative optimization process of zone defect identification - quantitative parameter adjustment - collaborative constraint - multi-field recalculation.

[0323] Step S5.5, Criteria for determining the termination of global iteration convergence;

[0324] The system employs a triple termination criterion: partition compliance, global equilibrium, and iteration upper limit. For each individual partition, the warpage is considered compliant if the absolute amplitude of the simulated partition warpage deformation is less than or equal to the out-of-plane warpage displacement of the partition. The temperature curve is then locked. Global optimization convergence is achieved when the warpage of all partitions meets tolerance requirements and the maximum difference in curing degree between partitions is less than or equal to 0.1. A maximum of 6 iterations is set. If the system still does not fully meet the criteria even after reaching the iteration upper limit, the optimal compromise temperature control process curve for the entire region is output, and the residual deviation of each partition is marked, completing the overall partition temperature iteration optimization process.

[0325] Step S6: Output the optimal zone curing process scheme;

[0326] Step S6.1: Lock the convergent optimal partition temperature control parameter group;

[0327] Extract the final curing process parameters of each partition at the end of the iteration, and store the complete set of temperature control parameters, such as heating rate, medium temperature holding time, curing peak temperature, and cooling rate, for the non-sensitive area, the warped convex sensitive area, and the warped concave sensitive area, to form a convergent optimal parameter set.

[0328] Step S6.2: Generate complete temperature-time process curves for each region;

[0329] With molding time as the horizontal axis and temperature as the vertical axis, separate segmented curing temperature curves for three types of regions are plotted, and the temperature switching nodes and the start and end times of the constant temperature intervals at each stage are marked to form a visual partitioned temperature control curve map.

[0330] Step S6.3: Export the simulation verification data accordingly;

[0331] Simultaneously export the simulation results file under the optimal process, including the global temperature field, curing degree field, residual stress distribution cloud map, overall component warping displacement cloud map, maximum warping value, and quantitative data table of average residual stress and deformation gradient of each zone, as a basis for verifying the effectiveness of the process.

[0332] Step S6.4: Integrate and form standardized process technology documents;

[0333] The document integrates the boundaries of the partitioned areas, the temperature control curves of each area, the key points of the molding process, the simulation verification indicators, and the molding tolerance requirements into a standardized and solidified process plan document. It clarifies the on-site implementation requirements for autoclave partitioned heating, mold partitioned temperature control channels, or partitioned heat insulation pads, and can be directly used for actual molding production without modifying the mold surface and component layup structure.

[0334] Step S6.5: Archive the process plan and retain the simulation model parameters;

[0335] The optimal partitioning process documents, supporting simulation models, material parameter databases, and multi-scale equivalent mapping parameters are uniformly archived and stored, which facilitates direct reuse of composite laminate components of the same series and shortens the simulation iteration and process development cycle of similar products.

[0336] The description of this invention is given for illustrative and descriptive purposes only and is not intended to be exhaustive or to limit the invention to the forms disclosed. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described in order to better illustrate the principles and practical application of the invention and to enable those skilled in the art to understand the invention and design various embodiments with various modifications suitable for a particular purpose.

Claims

1. A method for controlling curing warpage of composite laminates based on multi-scale data processing, characterized in that: Includes the following steps: Step S1: Calibration of basic material parameters; Through mechanical experiments, thermal performance tests, and curing kinetics DSC tests, elastic parameters, coefficients of thermal expansion, and curing reaction kinetic parameters of fibers and resin matrices were obtained. Equivalent orthogonal coefficients of thermal expansion and curing shrinkage reference parameters were calculated at different layup angles of 0°, 45°, 90°, and -45°. A material parameter database was established, including fiber geometric characteristic parameters, dry fiber mechanical property parameters, resin matrix mechanical property parameters, thermal expansion property parameters, resin curing kinetics and curing shrinkage parameters, and cross-scale equivalent homogenization derived parameters. Step S2: Construct a three-level multi-scale coupled simulation model with bidirectional parameter transfer between the mesoscopic, mesoscopic, and macroscopic dimensions. First, construct a mesoscopic unit cell of RVE with a solidification-related cohesive interface. After homogenization calculation, obtain the equivalent parameters of each ply and complete the mesoscopic standardized single ply model. Then, obtain the macroscopic laminate model by stacking the actual plies. Finally, construct a bidirectional data link for forward parameter transfer and downward substitution correction of equivalent physical properties by macroscopic high gradient field quantity to achieve multi-scale iterative linkage. Step S3, Finite element multiphysics coupling setup; Simultaneously activate the transient heat conduction field analysis, resin curing dynamic field analysis, and nonlinear mechanical field analysis. Incorporate the ply anisotropy-curing shrinkage coupled constitutive equation with curing degree as the internal variable, assign curing heat generation parameters, superimpose thermal strain and total self-generated strain of curing shrinkage, load the autoclave molding boundary and the reference temperature control curve, configure the nonlinear convergence criterion and cross-scale data interaction channel, and configure the three-field coupled solver of transient heat conduction field, resin curing dynamic field and nonlinear mechanical field. Step S4: Multi-field coupling solution, quantization identification of deformation-sensitive areas; By opening a cross-scale bidirectional feedback channel, the transient heat conduction field, the resin curing dynamic field, and the nonlinear mechanical field can be solved simultaneously. Four indicators were extracted: spatial fluctuation amplitude of curing degree, peak value of three-dimensional equivalent residual stress of unit, stress gradient of adjacent mesh, and warping displacement gradient of adjacent mesh. The extremum normalization was used to eliminate the dimensions. The weighted calculation unit comprehensively scores the sensitivity, divides the area into three levels: non-sensitive area, secondary sensitive area and high sensitive area, and marks the convex / concave deformation type of the area according to the positive and negative values ​​of the unit's average total strain; Step S5: Closed-loop iterative optimization of zone-specific temperature curves; Import partition data from non-sensitive, secondary sensitive, and highly sensitive areas to construct a partition defect database; formulate quantitative fine-tuning rules for heating rate, holding time, and peak temperature for raised and recessed partitions respectively; when multiple partitions with both raised and recessed areas coexist, implement five collaborative adjustment principles: global temperature steady state, concave-convex reverse offset, adjacent temperature difference limit, curing degree priority, and multiple rounds of small iterations; after updating the temperature control parameters, return to step S3 for recalculation; determine convergence based on three criteria: single partition warpage tolerance, global curing degree difference, and maximum iteration rounds; if convergence fails, iterate cyclically. Step S6: Output the optimal zone curing process scheme; Lock in the optimal temperature control parameters for each zone, plot the zone temperature-time curve, and simultaneously export the global field cloud map and quantitative verification data table. Integrate these into a process document that can be directly used for autoclave zone molding. Unify the archived material library, simulation model, and process parameters for reuse in the same series of components.

2. The method for controlling curing warpage of composite laminates based on multi-scale data processing as described in claim 1, characterized in that: Step 2 specifically includes the following steps: Step S2.1: Construct a micro-fiber-resin unit cell model; Step S2.2: Solve the micro-model and complete the positive equivalent homogenization parameter mapping; Step S2.3, Mesoscopic single-layer assembly, building a macroscopic laminate solid model; Import the design drawings of the target composite laminate component and read the actual ply stacking sequence, thickness of each single layer, and overall shape of the component. Call the mesoscopic orthotropic anisotropic single-layer model corresponding to different fiber angles, and complete the layer-by-layer stacking assembly according to the actual engineering layup sequence; By merging all ply geometry features, a complete macroscopic laminate 3D solid model is constructed. Simultaneously, the mold contact surface and clamping and fixing area are divided, and the boundary constraint positions of the autoclave forming process are preset to complete the construction of the top macroscopic structural carrier. Step S2.4: Establish a two-way parameter feedback transmission link between the microscopic, mesoscopic, and macroscopic levels.

3. The method for controlling curing warpage of composite laminates based on multi-scale data processing as described in claim 2, characterized in that: Step 2.1 specifically includes the following steps: Step S2.1.1: Retrieve basic material parameters; Retrieve the basic material parameters calibrated in step S1, including fiber diameter, actual fiber volume fraction of the component, and fiber arrangement parameters; Step S2.1.2, RVE geometric partitioning modeling; A periodic mesoscopic representative unit cell (RVE) geometric model was established in the finite element software platform to distinguish the independent geometric domains of the fiber phase and the resin matrix phase. Step S2.1.3: Assigning values ​​to the two-phase material parameters; The fiber phase and the resin matrix phase are respectively assigned the original mechanical, thermal expansion and curing kinetic parameters that have been calibrated by experiments, the fiber-resin interface contact constraint relationship is set, and the micro-scale fiber constraint effect, resin curing shrinkage behavior and fiber / resin interface shear interaction mechanism are restored. S2.1.4, the fiber-resin interface contact constraint relationship is characterized by a curing degree-related cohesive interface constraint system, which characterizes the mechanical interaction between the fiber phase and the resin matrix phase. The curing degree-related dynamic cohesive interface system is constructed as follows: First, a zero-thickness three-dimensional cohesive element is inserted between the outer surface of the fiber and the contact surface of the resin matrix. The mesh nodes on both sides are matched one-to-one through master-slave surface pairing to ensure the continuous transmission of interface stress and displacement. The bonding behavior in the interfacial elastic stage is described by a traction-separation bilinear constitutive relation. The interfacial normal stiffness and bitangential shear stiffness are dynamically updated using an exponential evolution function with the resin's normalized degree of cure as the independent variable. The function form is as follows: ,in To fully solidify the interface reference stiffness, m is the experimentally fitted growth factor. The real-time curing degree of the resin is represented, with a value range of [value range missing]. , This refers to the resin's maximum degree of cure. When the resin curing degree is low, the interface stiffness is small, allowing slight relative slippage between the fiber and the resin. As the curing reaction progresses, the interface stiffness increases synchronously, limiting the separation and slippage of the two phases and reflecting the evolution of the interface bonding state throughout the autoclave molding process. Meanwhile, a secondary nominal stress combination failure criterion is introduced to determine the initiation of interface damage. The scalar damage variable is used to describe the interface stiffness degradation law. Based on the material parameters such as interface shear strength and interface fracture energy obtained by step S1, the interface compressive stress, shear stress and local micro-debonding behavior induced by resin curing shrinkage are accurately characterized. The cohesive interface damage evolution is described by a linear degradation model. A scalar damage variable D is introduced to characterize the degree of interface stiffness degradation. The damage scalar D is calculated in segments based on the equivalent separation displacement of the interface. The value range of D is 0≤D≤1. D=0 represents no damage and complete bonding of the interface, and D=1 represents complete debonding and failure of the interface. like , ; like When D=0, the interface remains elastically bonded with no stiffness degradation; like When D=1, the interface completely fails and the interface load-bearing stiffness becomes zero. in, It is the interface equivalent separation displacement. It is the critical equivalent separation displacement at the onset of damage. It is the equivalent separation displacement due to complete interface failure and fracture. , For Macaulay brackets representing normal separation displacement, only tensile separation is involved in damage calculation; interface compression does not produce damage. Real-time interface stiffness after introducing damage variables: ; Step S2.1.5: Apply periodic boundary conditions for each unit cell; The periodic boundary condition assumes that the micro-fiber-resin representative volume element is a periodically repeating structure. Displacement correlation constraint equations are applied to the three sets of relatively parallel characteristic surfaces of the unit cell, including geometric pairing constraints, displacement compatibility constraints, and stress continuity constraints. Geometric pairing constraints identify the six sets of opposite faces of the unit cell and establish a one-to-one mapping relationship between corresponding nodes on the front and back faces. Displacement compatibility constraints require that the displacements of opposite face nodes in the same coordinate axis direction satisfy the linear correlation equation, allowing only uniform overall deformation of the unit cell and prohibiting non-coordinated separation, overlap, or misalignment of the unit cell boundary. Stress continuity constraints indirectly realize the continuous transmission of interface normal and tangential stress through displacement correlation equations, simulating the real state of the infinite periodic arrangement of fibers and resin inside the composite material.

4. The method for controlling curing warpage of composite laminates based on multi-scale data processing as described in claim 2, characterized in that: Step 2.2 specifically includes the following steps: Step S2.2.1, closely examine the function for solving homogeneous field quantities; Load the RVE unit cell solution model with complete geometry, material, interface, and boundary condition settings; Homogenized numerical solutions are performed on the mesoscopic unit cell model with periodic boundary conditions. When performing homogenized numerical solutions on the mesoscopic unit cell with applied periodic boundary conditions, the volume-averaged method is used to calculate the average stress and average strain of the unit cell. The volume integral of the stress and strain field over the mesoscopic global domain is divided by the total volume of the unit cell to obtain the equivalent average stress tensor representing the average mechanical response of the composite material. With equivalent mean strain tensor Where V is the total volume of the unit cell in the RVE, and dV is the volume of any infinitesimal element at any integration point within the unit cell. , To observe the local stress and strain tensors at specific field points; Step S2.2.2, calculation function for equivalent physical properties of ply layup; Based on the average stress and average strain, the orthogonal anisotropic equivalent stiffness matrix, equivalent thermal expansion coefficient, and equivalent curing shrinkage strain can be further solved to complete the positive equivalent parameter mapping of the mesoscopic to mesoscopic standardized orthogonal anisotropic single-layer model. Step S2.2.3, Standardize the single-layer model construction function; Based on the mapped equivalent parameters, a standardized orthogonal anisotropic single-ply model is generated. The anisotropic single-ply model can adapt to any fiber layup angle of 0°, ±45°, and 90°, accurately conveying the anisotropic characteristics of the layup brought about by the microscale interaction. Step S2.2.4, Ply angle coordinate transformation function; The standardized orthogonal anisotropic single-ply model incorporates a coordinate rotation transformation matrix to achieve coordinate transformation of material parameters at arbitrary ply angles. Let θ represent the angle between the principal fiber direction and the global X-axis of the model. An in-plane rotation matrix T is used to transform the principal axial stiffness, thermal expansion strain, and curing shrinkage strain of the material into the global coordinate system. ; pass Complete the stiffness tensor transformation, through The orthogonal thermal expansion and curing shrinkage strain component conversion is completed, among which... To examine the orthogonal anisotropic stiffness matrix of the material principal axes obtained through homogenization, The equivalent stiffness matrix of the layers in the global coordinate system. The equivalent thermal expansion / curing shrinkage strain along the principal axis of the material. For strain components in the global coordinate system, inputting any 0°, ±45°, or 90° ply angle will automatically complete parameter conversion, enabling standardized batch assembly of single-layer units at different angles; Step S2.2.5, Mesoscopic to Mesoscopic Forward Parameter Mapping Function; Complete batch conversion and encapsulation of single-ply parameters from different angles, automatically transfer the complete set of equivalent parameters of micro-homogenization to the meso-scale, and realize the pre-assembly preparation of standardized single-ply batch assembly.

5. The method for controlling curing warpage of composite laminates based on multi-scale data processing as described in claim 2, characterized in that: Step 2.4 specifically includes the following steps: Step S2.4.1: Solidify the automatic transmission link of positive parameters; An automatic parameter calling interface is built across scales. During the simulation calculation startup phase, the equivalent parameters obtained by homogenization of the mesoscopic unit cell are automatically pushed to the mesoscopic single-layer model and the macroscopic laminate 3D solid model, realizing the automatic positive flow of material performance data between scales. Step S2.4.2: Load the macroscopic multiphysics global result file; After the three-dimensional solid model of the macroscopic laminate has completed the solution of the transient heat conduction, curing dynamics, and mechanical three-field coupling, the result file containing all mesh element information is read, and the equivalent stress of each finite element element is extracted in batches. Out-of-plane warping displacement The unit's real-time curing degree, its ply number, and its spatial coordinates are recorded. Units are grouped by component partition, and the average equivalent stress of each partition is calculated. Average warping displacement of the entire component This provides a baseline value for gradient determination; Step S2.4.3: Automatically select high-gradient mesh cells according to the gradient determination rules; Call the preset gradient threshold to calculate the stress relative gradient for each finite element. , is the relative gradient of deformation Finite element satisfies ≥Stress gradient threshold or If the value is greater than or equal to the deformation gradient threshold, it is initially marked as a high gradient element. Elements in stress concentration areas such as openings, fillets, and ply discontinuities are directly and forcibly marked without gradient calculation. The neighborhood radius is 3 times the single-layer thickness. Isolated single-point marked abnormal noise elements are removed, and continuous effective high gradient elements are retained. The number, spatial coordinates, ply to which they belong, element stress, out-of-plane warpage displacement, and real-time solidification degree original field quantity of all effective high gradient elements are collected as the reverse substitution data source. Step S2.4.4: Complete the transfer of macroscopic to mesoscopic data according to the three-level hierarchical mapping matching rules; Using the ply number and ply angle of the high gradient unit as matching indexes, the corresponding standardized orthotropic mesoscopic single-ply unit is located. The curing degree and stress tensor of multiple adjacent high gradient units within the same layup are arithmetically averaged to reduce noise and eliminate single-point numerical oscillations. The averaged global field quantity is stored in the data cache area of ​​the corresponding mesoscopic single-layer unit, completing the aggregation and temporary storage of macroscopic field variables to the mesoscopic scale; Step S2.4.5: Mesoscopic to mesoscopic RVE unit cell matching and field quantity and load conversion; Based on the fiber volume fraction and fiber arrangement of the mesoscopic single layer, retrieve periodic mesoscopic RVE unit cells of matching specifications; Mapping assignment rules: Set the mesoscopic average degree of curing as the initial degree of curing of the entire RVE domain; convert the mesoscopic average stress into the equivalent surface force of the RVE outer contour; use the average strain obtained by converting macroscopic deformation as the input value of the periodic boundary condition of the RVE. Batch optimization: When the parameter deviation between adjacent mesoscopic units is less than the iteration convergence threshold, the same set of mesoscopic RVEs is used to carry out homogenization solution to reduce redundant calculations; Abnormal fallback handling: When the curing degree value exceeds the limit or there is no matching RVE model, the neighborhood mean or the same fiber volume fraction benchmark unit cell adaptation correction is called. Homogenization is performed on the mesoscopic unit cell under the loading of macroscopic back substitution field and boundary load, and the equivalent orthogonal stiffness matrix, equivalent thermal expansion coefficient, and equivalent solidification shrinkage strain are updated by the volume averaging method to calculate and update the full set of cross-scale derived parameters. Step S2.4.6, cross-scale iterative convergence determination and parameter backfilling; Compare the updated parameters with the parameters of the previous iteration, calculate the relative error of each material component, and determine whether convergence has been achieved by combining the preset maximum number of iterations. If convergence is achieved, the updated equivalent parameters will be backfilled up the mesoscopic level to the macroscopic high gradient unit, completing this reverse data backfill process. If the convergence fails and the iteration limit is not reached, the macroscopic three-field coupling calculation is restarted using the updated equivalent parameters, and the reverse substitution module is restarted for iterative iteration. If convergence is not achieved after reaching the maximum number of iterations, the macroscopic model is backfilled with the final microscopic homogeneous output parameters, the iteration is terminated, and a convergence indicator is retained.

6. The method for controlling curing warpage of composite laminates based on multi-scale data processing as described in claim 1, characterized in that: Step 3 specifically includes the following steps: Step S3.1: Activate the multiphysics analysis module; Transient heat conduction field analysis, resin curing dynamic field analysis, and nonlinear mechanical field analysis are simultaneously enabled on the finite element software platform. A coupled calculation framework for simultaneous solution of the three fields is established to realize the synchronous iterative calculation of temperature evolution, resin curing crosslinking reaction, and structural stress warping deformation throughout the entire component forming process. Step S3.2: Embed the coupled constitutive model in the finite element software platform; Implanting resin real-time curing degree into the finite element software platform The core internal variable is the ply anisotropic-curing shrinkage coupled constitutive equation; the orthogonal anisotropic stiffness matrix and the triaxial thermal expansion coefficient are defined as nonlinear updates with degree of curing; the coupled calculation relationship between thermal strain and curing shrinkage strain is established; and the laminate thickness layered integration algorithm is written into the calculation of three-dimensional residual stress at full thickness. Step S3.3: Assign values ​​to the curing degree field and curing heat generation parameters; Retrieve the curing kinetic parameters from the material database in step S1, define the control equation for the evolution of resin curing degree and the internal heat generation model of curing; batch assign the equivalent curing exothermic parameters of each ply obtained by micro-homogenization to the standardized orthogonal anisotropic single ply model to realize the bidirectional coupling transfer of exothermic curing reaction to the temperature field. Step S3.4, Define the global strain field; Two types of non-mechanical strains are superimposed: one is the orthogonal anisotropic thermal strain that varies with temperature and layup angle, and the other is the resin curing shrinkage coupled strain that is constrained by fibers and controlled in real time by the degree of curing, thus distinguishing the differential strain distribution characteristics within and between layers. Step S3.5: Molding boundary conditions and process range loading; Based on the actual molding conditions of the autoclave, apply the heat exchange boundary of the mold contact surface, the component clamping displacement constraint, and the heat transfer coefficient of the external surface environment; input the preset benchmark curing process range, including temperature curve parameters such as heating rate, holding temperature, holding time, and cooling rate, to complete the initial setting of the global initial temperature and initial curing degree field. Step S3.6, Coupled solver parameter configuration; The system sets the convergence criteria for the coupled iteration of transient heat conduction field, resin curing dynamic field, and nonlinear mechanical field, as well as the load step size and the number of nonlinear equilibrium iterations. It also opens a multi-scale data interaction channel, allowing the field variables of the high stress gradient region obtained from the macroscopic solution to be back-transferred to the mesoscopic and microscopic models, providing computational support for subsequent bidirectional parameter correction.

7. The method for controlling curing warpage of composite laminates based on multi-scale data processing as described in claim 6, characterized in that: The specific constitutive equation for the anisotropic plying-curing shrinkage coupling is as follows: Mathematical equation for the coupled evolution of degree of cure and modulus: ； in, It depends on the real-time curing degree of the resin The varying orthogonal anisotropic shear modulus tensor components. , To fully cure the equivalent elastic modulus and shear modulus of the underlay, The evolution coefficients for the experimental fitting were determined using the real-time curing degree of the resin. As an independent variable, the nonlinear variation law of stiffness gradually increasing from near-zero stiffness in the liquid state to the fully solidified reference stiffness is characterized by the exponential function. Mathematical equation for the coupled evolution of degree of cure and coefficient of thermal expansion: ； in, The equivalent thermal expansion coefficient in the fully cured state. The evolution coefficients are the experimental fitting coefficients; Formula for superimposed calculation of thermal strain and curing shrinkage strain: Total nonmechanical strain along the principal axes of the local material in the layup The thermal strain caused by temperature change and the curing shrinkage strain caused by curing reaction are coupled and superimposed: ； in, The temperature difference between the current temperature and the reference temperature. It is the equivalent curing shrinkage strain that varies with the degree of curing; Discrete calculation method for solving three-dimensional residual stress by layer-by-layer integration of laminate thickness: Divide a single layer into N integral discrete thickness layers along the p-th layer thickness direction. Each layer has a thickness of micron. The equivalent residual stress of a single layer is obtained by integrating the local stress at any thickness location along the thickness and averaging the volume. ； Where h is the total thickness of the single-layer composite ply, the integration interval extends from the bottom surface coordinate z=0 to the top surface coordinate z=h, and p takes values ​​from 1 to N. It is the local stress tensor component calculated from the p-th discrete layer center point. It is the segment thickness corresponding to the p-th discrete layer.

8. The method for controlling curing warpage of composite laminates based on multi-scale data processing as described in claim 1, characterized in that: Step 4 specifically includes the following steps: Step S4.1: Submit the multi-scale coupled model to complete the multi-physics simultaneous solution; The three-field coupled solver for transient heat conduction field, resin curing dynamic field and nonlinear mechanical field configured in step S3 is invoked. Transient coupling calculations are carried out segment by segment according to the preset curing process load steps. The transient temperature field, resin curing degree field, three-dimensional residual stress field and out-of-plane warping displacement field of the whole domain are solved iteratively at the same time. During the solution process, the bidirectional parameter feedback channel of micro-meso-macro is kept open. The field variables of the macro high gradient region are substituted back to the micro unit cell in real time to complete the equivalent parameter iterative correction. The complete field data result file of each time step of the entire molding process is output. Step S4.2: Extract the field data of the four quantitative evaluation indicators for the entire domain; In the post-processing module, four types of physical field feature data of all grid elements of the model are extracted in batches: solidification degree spatial fluctuation amplitude, peak value of three-dimensional equivalent residual stress of the element, stress spatial gradient of adjacent grids of the element, and warping displacement spatial gradient of adjacent grids of the element. The extreme value normalization method is used to uniformly scale the four different dimensional indices of solidification degree spatial fluctuation amplitude, peak value of three-dimensional equivalent residual stress of the element, stress spatial gradient of adjacent grids of the element, and warping displacement spatial gradient of adjacent grids of the element to the dimensionless interval of [0,1] to eliminate the difference in the dimensionality of the indices and form a four-dimensional index dataset at the element level. Step S4.3, Four-dimensional index weighted threshold quantization judgment calculation; Based on the component service level, the risk weights of each item are configured, and a single-grid comprehensive sensitivity score S is obtained through linear weighted calculation, where 0≤S≤1; based on the score interval, three types of areas are automatically divided: high-sensitivity area, secondary-sensitivity area, and non-sensitivity area, generating a full-domain partitioned visualization cloud map; In non-sensitive areas, S < 0.35, no mesh encryption is performed, and access to the macro-to-micro field quantity back-substitution link is not mandatory; In the secondary sensitive region (0.35≤S<0.7), local mesh refinement is carried out, and cross-scale parameter iterative correction is selectively applied according to simulation accuracy requirements. For high-sensitivity areas ≥0.7, mandatory finer mesh densification is implemented, and the macro-to-micro reverse data back-substitution link mentioned above is fixedly incorporated to carry out iterative optimization of equivalent parameters; Step S4.4: Automatically determine the warping and deformation trends of each highly sensitive area; The average total nonmechanical strain value of all units in the highly sensitive area is extracted, and the positive or negative value of the strain is used as the criterion: the average total strain is negative and it is determined to be an area of ​​excessive shrinkage and warping bulge; the average total strain is positive and it is determined to be an area of ​​hysteresis shrinkage and warping depression. Sensitive areas with different deformation trends are marked and classified.

9. The method for controlling curing warpage of composite laminates based on multi-scale data processing as described in claim 1, characterized in that: Step 5 specifically includes the following steps: Step S5.1: Import deformation-sensitive area and molding defect data; Read the three-level deformation sensitive area mesh partitioning results of high-sensitivity area, secondary-sensitivity area and non-sensitivity area output in step S4, and divide it into several molding temperature control sub-partitions with independent sensitive areas as units; extract the out-of-plane warping displacement, protrusion / depression defect type, three-dimensional equivalent residual stress peak value and curing degree spatial distribution data of each partition in batches, and establish a molding defect feature database for each temperature control partition; Step S5.2: Establish rules for the quantitative fine-tuning amplitude of warpage deviation level and temperature parameter; For warped and protruding areas with excessive resin shrinkage, select process parameters such as low heating rate, extended medium-temperature holding time, and reduced peak curing temperature; for warped and sunken areas with delayed resin shrinkage, select process parameters such as high heating rate, shortened medium-temperature holding time, and increased peak curing temperature; integrate all zone temperature parameters to form a complete global zone temperature control scheme. Step S5.3, principle of coordinated adjustment of temperature parameters under multi-zone concave-convex coexistence operating conditions; The principle of steady-state constraint of global temperature reference: the difference between the global average peak temperature of all zones and the original reference temperature is ≤3℃, and the fluctuation of the global average heating rate is ≤0.2℃ / min, to ensure that the overall curing rhythm of the component does not deviate significantly and to avoid the overall increase of residual stress in the whole domain; The principle of reverse hedging adjustment for concave and convex zones: Convex zones uniformly adopt a relaxation correction strategy of slowing down, cooling down, and extending the heat preservation, while concave zones simultaneously adopt a reinforcement correction strategy of speeding up, heating up, and shortening the heat preservation; the adjustment range of each zone matches its own warp excess ratio, and the larger the excess, the higher the parameter correction weight, so as to achieve mutual hedging and cancellation of deformation in multiple areas. Adjacent partition gradient difference constraint principle: For spatially adjacent concave and convex partitions, the peak temperature difference in a single iteration shall not exceed 8℃, and the heating rate difference shall not exceed 0.4℃ / min, to prevent excessive temperature difference and curing gradient at partition boundaries from causing new stress concentration; The principle of prioritizing uniform curing degree: After each parameter adjustment, check the curing degree distribution across the entire domain. If the difference in normalized curing degree between different intervals exceeds 0.15, simultaneously compress the temperature adjustment range of all zones to prioritize ensuring the synchronization of curing across the entire domain, and then iteratively correct the warpage deviation step by step. The principle of multi-round small-scale iterative correction: For composite defect conditions, a single extreme parameter tuning is not adopted. Instead, it is divided into 2 to 3 rounds of progressive iterative optimization. After each round of iteration completes the three-field coupling solution, the warping and solidification field data are updated, and the fine-tuning amplitude is dynamically corrected to avoid introducing new forming defects by changing parameters drastically at once. Step S5.4, iterative loop linkage and parameter update mechanism; After each round of fine-tuning and collaborative correction of the zone temperature parameters, the parameters of the global temperature process curve are updated and sent back to the multi-field coupling solution module in step S3. The three-field coupling calculation of transient heat conduction field, resin curing dynamic field and nonlinear mechanical field is carried out again. The warpage, stress gradient, curing degree fluctuation and defect status of each zone are updated again through the quantitative evaluation system in step S4. Step S5.5, Criteria for determining the termination of global iteration convergence; The system employs a triple termination criterion: partition compliance, global equilibrium, and iteration upper limit. For each individual partition, the warpage is considered compliant if the absolute amplitude of the simulated partition warpage deformation is less than or equal to the out-of-plane warpage displacement of the partition. The temperature curve is then locked. Global optimization convergence is achieved when the warpage of all partitions meets tolerance requirements and the maximum difference in curing degree between partitions is less than or equal to 0.

1. A maximum of 6 iterations is set. If the system still does not fully meet the criteria even after reaching the iteration upper limit, the optimal compromise temperature control process curve for the entire region is output, and the residual deviation of each partition is marked, completing the overall partition temperature iteration optimization process.

10. The method for controlling curing warpage of composite laminates based on multi-scale data processing as described in claim 1, characterized in that: Step 6 specifically includes the following steps: Step S6.1: Lock the convergent optimal partition temperature control parameter group; Extract the final curing process parameters of each partition when the iteration terminates, and store the complete set of temperature control parameters, such as heating rate, medium temperature holding time, curing peak temperature, and cooling rate, for non-sensitive areas, warped protrusion sensitive areas, and warped depression sensitive areas, to form a convergent optimal parameter set. Step S6.2: Generate complete temperature-time process curves for each region; With molding time as the horizontal axis and temperature as the vertical axis, separate segmented curing temperature curves for three types of regions are plotted, and the temperature switching nodes and the start and end times of the constant temperature intervals are marked to form a visual partitioned temperature control curve map. Step S6.3: Export the simulation verification data accordingly; Simultaneously export the simulation results file under the optimal process, including the global temperature field, curing degree field, residual stress distribution cloud map, overall component warping displacement cloud map, maximum warping value, and quantitative data table of average residual stress and deformation gradient of each zone, as a basis for verifying the effectiveness of the process. Step S6.4: Integrate and form standardized process technology documents; The document integrates the boundaries of the partitioned areas, the temperature control curves of each area, the key points of the molding process, the simulation verification indicators, and the molding tolerance requirements into a standardized and solidified process plan document. It clarifies the on-site implementation requirements for autoclave partitioned heating, mold partitioned temperature control channels, or partitioned heat insulation pads, and can be directly used for actual molding production without modifying the mold surface and component layup structure. Step S6.5: Archive the process plan and retain the simulation model parameters; The optimal partitioning process documents, supporting simulation models, material parameter databases, and multi-scale equivalent mapping parameters are uniformly archived and stored, which facilitates direct reuse of composite laminate components of the same series and shortens the simulation iteration and process development cycle of similar products.

Images