A Simulation Method and System for Moldless Seal Molding Based on AI

Abstract

This application relates to the field of computer simulation technology, and provides an AI-based simulation method and system for moldless sealing component molding. The method includes: acquiring an initial set of geometric topological information and an initial set of material property information of a target 3D model; then constructing an adaptive physical field model associated with the spatial coordinate system of the target 3D model; next, inputting the initial information into the model for spatiotemporally coupled molding process simulation processing, synchronously iteratively solving the multi-physics interaction to generate dynamic evolution trajectory data; then, based on the characteristics of material particles in the trajectory data, performing joint mechanism analysis to obtain the cross-sectional profile distortion mode of the extrusion swell effect influence area, the spatial distribution of vulcanization shrinkage deformation, and the location information of residual stress concentration areas; finally, integrating the above information to perform molding defect source analysis and obtain the molding process simulation results. In this way, accurate simulation and defect root source tracing of the entire molding process of moldless sealing components are achieved.

Description

Technical Field

[0001] This application belongs to the field of computer simulation technology, specifically relating to an AI-based simulation method and system for moldless sealing component molding. Background Technology

[0002] In high-end manufacturing, especially in industries such as automotive and aerospace, the molding quality of rubber seals with complex cross-sectional shapes (such as rubber sealing strips used in engine compartments) directly affects the sealing performance and service life of the product. These seals are typically produced using a dieless extrusion molding process, which involves the coupling of multiple physicochemical processes, including complex material flow under high temperature and pressure, cross-linking reactions, and cooling shrinkage.

[0003] For a long time, the engineering and academic communities have developed various simulation analysis methods to predict and control product quality during the molding process. Early technologies mainly focused on simulating single physical fields, such as using computational fluid dynamics to analyze the flow behavior of rubber materials at the extrusion die to predict extrusion swell; or using the finite element method to analyze the thermal shrinkage and stress distribution of fully vulcanized rubber parts during the cooling process. With the development of technology, sequentially coupled simulation methods have emerged, in which flow simulation is performed first, and then the results of the flow analysis are used as initial conditions to pass to subsequent vulcanization or structural analysis.

[0004] The existing simulation technologies generally break down the originally tightly coupled molding process into several independent stages, which cannot truly reflect the influence of flow-induced molecular orientation on the subsequent vulcanization reaction kinetics, nor can they accurately simulate the dynamic interaction between the exothermic reaction and the curing shrinkage, and it is even more difficult to trace the entire formation history of the final defect throughout the entire process. Summary of the Invention

[0005] This application provides an AI-based simulation method and system for moldless sealing component molding.

[0006] This application provides an AI-based simulation method for moldless seal molding, applied to a moldless seal molding simulation system. The method includes: Obtain the initial set of geometric topological information and the initial set of material property information of the target 3D model; An adaptive physical field model is constructed that is associated with the spatial coordinate system of the target three-dimensional model. The adaptive physical field model includes a flow stress field calculation module, a reaction progress field evolution module, and a solidification deformation field prediction module configured based on the initial material property information. The initial geometric topology information set and the initial material property information set are input into the adaptive physical field model for spatiotemporal coupled molding process simulation. The multi-physics interaction between each module is solved iteratively in a synchronous manner to generate dynamic evolution trajectory data of the target three-dimensional model from the initial shape of the extrusion die to the fully vulcanized molding shape. Based on the characteristics of the spatial distribution of material particles, velocity vector direction, and local reaction completion in the dynamic evolution trajectory data, a joint mechanism analysis of material flow and deformation behavior is performed to obtain the cross-sectional profile distortion mode, the spatial distribution of vulcanization shrinkage deformation, and the location information of residual stress concentration areas in the region affected by the extrusion swell effect. By integrating the cross-sectional profile distortion mode, the spatial distribution of shrinkage deformation, and the location information of the residual stress concentration area, a forming defect tracing analysis is performed to generate forming process simulation results that include defect type identification, defect location coordinates, and defect evolution path.

[0007] This application provides a moldless seal molding simulation system, which includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the above method.

[0008] This application provides a computer-readable storage medium including a computer program. When the computer program is run on a moldless seal molding simulation system, the computer program is used to cause the moldless seal molding simulation system to perform the steps of the above-described method.

[0009] This application embodiment breaks through the barrier of isolated analysis of flow, vulcanization, and curing deformation in traditional simulations by creating an adaptive physics field model directly associated with the spatial coordinate system of the target 3D model. It realizes spatiotemporal coupling integrated simulation of the entire forming process from extrusion, vulcanization to cooling. Unlike simply simulating a single physical process, this application embodiment inputs the initial geometric topology information set and the initial material property information set into an adaptive physics field model that includes a flow stress field calculation module, a reaction progress field evolution module, and a curing deformation field prediction module. It simultaneously iterates and solves the multi-physics field interaction, thereby generating dynamic evolution trajectory data that can completely trace the material particles from the initial shape at the extrusion die to the fully vulcanized and formed shape. The above-mentioned holistic simulation strategy enables subsequent joint mechanism analysis based on the spatial distribution of material particles, velocity vector direction, and local reaction completion characteristics in the dynamic evolution trajectory data to accurately reveal the intrinsic physical relationship between the cross-sectional profile distortion mode of the extrusion expansion effect area, the spatial distribution of vulcanization shrinkage deformation, and the location information of the residual stress concentration area.

[0010] Finally, by integrating the above multi-dimensional analysis results for defect tracing analysis, the generated molding process simulation results not only include defect type identifiers and defect location coordinates, but also creatively include defect evolution paths. This fundamentally elevates the simulation results from static defect descriptions to dynamic, traceable defect formation histories. Therefore, this application's embodiments, starting from a global logic, can systematically and fundamentally reveal the complex causes and evolution patterns of molding defects in moldless seals, providing precise guidance for process optimization. Attached Figure Description

[0011] Figure 1 This is a flowchart illustrating an AI-based simulation method for moldless sealing component molding, provided in an embodiment of this application.

[0012] Figure 2 This is a schematic diagram of a moldless seal molding simulation system provided in an embodiment of this application.

[0013] Figure 3 This is a functional block diagram of a moldless seal forming simulation system provided in an embodiment of this application. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this application. Obviously, the described embodiments are only some embodiments of the technical solutions of this application, and not all embodiments. Based on the embodiments recorded in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the technical solutions of this application.

[0015] See Figure 1 This is an AI-based simulation method for moldless sealing component molding provided in the embodiments of this application. This method can be applied to a moldless sealing component molding simulation system. The specific process is as follows: steps 110-150.

[0016] Step 110: Obtain the initial set of geometric topology information and the initial set of material property information of the target 3D model.

[0017] In this embodiment, the target 3D model corresponds to a rubber seal with a complex cross-sectional shape to be simulated, such as a rubber sealing strip with a complex hollow structure used in the engine compartment of an automobile. The moldless seal molding simulation system (or simply the system) first parses the target 3D model of the seal from the computer-aided design (CAD) model file. The parsing process extracts the geometric topological information constituting the surface and interior of the model. This information is organized in the form of a voxelized mesh, with each voxel mesh cell assigned a unique global index identifier and recording the spatial rectangular coordinates of its eight vertices in millimeters. Simultaneously, the system reads the initial set of material property information corresponding to the seal formulation from the material property database. This set exists in the form of a structured parameter file, containing viscoelastic constitutive parameters of the rubber substrate in the uncured state, such as rubber constitutive equation coefficients based on tubular model theory, and crosslinking reaction kinetic parameters during the curing process, such as the activation energy and pre-exponential factor based on the autocatalytic reaction model, as well as hyperelastic constitutive parameters in the fully cured state, such as the shear modulus and incompressibility parameter in the Ogden constitutive model.

[0018] Step 120: Construct an adaptive physical field model associated with the spatial coordinate system of the target 3D model. This adaptive physical field model includes a flow stress field calculation module, a reaction progress field evolution module, and a solidification deformation field prediction module configured based on the initial material property information.

[0019] The core of this step lies in constructing a highly coupled virtual testing platform capable of adapting to the geometry and material properties of the target model. First, the system analyzes the viscoelastic parameters of the material in its uncured state, constructs constitutive equations describing its flow behavior, and couples these equations with the laws of conservation of mass and momentum, laying the theoretical foundation for the flow stress field module. Next, the system spatially discretizes this set of equations, generating mesh elements that fit the solid structure, and allocates memory space for each element to store key transient variables such as pressure, velocity, and stress. Similarly, the system constructs a reaction progress field module based on curing reaction kinetic parameters to describe the interaction between the crosslinking reaction and the temperature field, and allocates storage areas for temperature, reactivity, and exothermic rate to its mesh elements. Finally, a curing deformation field module is constructed based on the hyperelastic parameters after complete curing to predict shrinkage and stress evolution during the cooling process, and allocates displacement, strain, and stress storage space to its nodes and elements.

[0020] After the independent construction of each module is completed, the system encapsulates the three into a multiphysics coupled solver by establishing the topological mapping relationship and data transfer interface between the meshes. This solver can realize bidirectional data feedback from the flow field to the reaction field, from the reaction field to the curing field, and from the curing field to the flow field, and applies initial and boundary conditions that conform to the actual process. Finally, it generates an adaptive physics model that can completely describe the behavior of the seal from extrusion, vulcanization to cooling throughout its entire life cycle and is closely related to the target 3D model.

[0021] Step 121: Analyze the viscoelastic constitutive parameters of the rubber substrate in the unvulcanized state from the initial material property information. Based on the viscoelastic constitutive parameters, construct a constitutive equation describing the dynamic balance between viscous dissipation and elastic energy storage of the rubber substrate during shear flow. Couple the constitutive equation with the mass conservation equation and the momentum conservation equation to obtain the basic governing equations of the flow stress field calculation module.

[0022] For example, the system reads parameters from the initial material property information used to describe the viscoelasticity of uncured rubber. These parameters include the relaxation time spectrum G_i and the corresponding shear modulus ω_i based on the generalized Maxwell model. The system substitutes these parameters into the integral viscoelastic constitutive equation, which describes the relationship between the stress tensor σ(t) and the deformation history. Subsequently, the system combines this constitutive equation with the continuity equation describing the conservation of fluid mass and the Navier-Stokes equation describing the conservation of momentum. The combined equations constitute the basic governing equations of the flow stress field calculation module. This set of equations is used to solve for the transient distribution of pressure p, velocity vector u, and stress tensor σ on the computational grid cells.

[0023] Step 122: Spatial discretization is performed on the basic governing equations of the flow stress field calculation module to generate flow stress field calculation grid cells corresponding to the solid filling structure inside the target three-dimensional model, and storage space is allocated for each flow stress field calculation grid cell to store transient pressure values, transient velocity vectors and transient stress tensor components.

[0024] In this step, the system uses the finite volume method to discretize the coupled governing equations. The discretization process divides the continuous spatial region occupied by the target 3D model into a series of non-overlapping control volumes, each of which is a computational grid cell. The generated grid cells are precisely fitted to the boundaries of the target 3D model; for example, in the hollow region of the seal, the grid cells only fill the solid portion. The system assigns a data structure to each generated flow stress field computational grid cell. This data structure includes a floating-point variable P_flow for storing scalar pressure values, a velocity vector storage area containing three floating-point components (x-direction component u_x, y-direction component u_y, z-direction component u_z), and a tensor storage area for storing six independent stress components (normal stresses σ_xx, σ_yy, σ_zz and shear stresses τ_xy, τ_xz, τ_yz).

[0025] Step 123: Analyze the crosslinking reaction kinetic parameters of the rubber substrate in the initial material property information during the vulcanization process. Based on the crosslinking reaction kinetic parameters, construct a kinetic differential equation describing the dependence of the crosslinking network formation rate of the rubber substrate on temperature and time during the vulcanization process. Couple the kinetic differential equation with the energy conservation equation to obtain the basic control equation set of the reaction progress field evolution module.

[0026] The system extracts kinetic parameters of the sulfurization reaction from initial material property information, such as the reaction rate constants k_1 and k_2 based on the Kamal-Sourour model, and the reaction orders m and n. Based on these parameters, the system constructs a differential equation describing the rate of change of reaction completion α over time: dα / dt = (k_1 + k_2 * α^m) * (1 - α)^n. This equation reflects the influence of temperature T on k_1 and k_2 via the Arrhenius equation. Next, the system couples this kinetic differential equation with the energy conservation equation describing heat transfer and generation. The energy conservation equation introduces an exothermic term for the sulfurization reaction, which equals the reaction rate dα / dt multiplied by the total enthalpy change ΔH. The coupled equation set forms the fundamental governing equations for the reaction progress field evolution module, used to solve for the temperature field T and the reaction completion field α.

[0027] Step 124: Spatial discretization is performed on the basic control equations of the reaction progress field evolution module to generate reaction progress field evolution calculation grid cells corresponding to the solid filling structure inside the target three-dimensional model, and storage space is allocated to each reaction progress field evolution calculation grid cell for storing transient temperature values, transient reaction completion degree and transient heat release rate.

[0028] Similar to step 122, the system uses the finite volume method to discretize the governing equations of the reaction progress field evolution module, generating a set of computational grid cells for the reaction progress field evolution. Considering the accuracy and efficiency of the physical field calculation, this set of grid cells can share the same meshing scheme as the flow stress field calculation grid cells, or it can use a more refined mesh that adapts to temperature gradient changes. The system allocates storage space for each computational grid cell for the reaction progress field evolution, including a floating-point variable T_cure for storing transient temperature values ​​(in Kelvin), a floating-point variable α_cure for storing transient reaction completion (range 0 to 1), and a floating-point variable Q_cure for storing the instantaneous exothermic rate (in watts per cubic meter) generated by the crosslinking reaction.

[0029] Step 125: Analyze the hyperelastic constitutive parameters of the rubber substrate in the fully vulcanized state from the initial material property information. Based on the hyperelastic constitutive parameters, construct a constitutive model describing the nonlinear stress-strain response of the fully vulcanized rubber substrate during cooling and shrinkage. Couple the constitutive model with the momentum conservation equation and the heat conduction equation to determine the basic governing equations of the curing deformation field prediction module.

[0030] The system extracts hyperelastic parameters describing the mechanical behavior of fully vulcanized rubber from initial material property information, such as material constants C_10, C_20, C_30 and bulk modulus K based on the Yeoh model. These parameters are then substituted into a hyperelastic constitutive model, which defines the relationship between the strain energy density function W and the deformation gradient tensor F, thus deriving the second-kind Piola-Kirchhoff stress tensor S. The system then couples this constitutive model with the momentum conservation equation describing force equilibrium (which simplifies to the equilibrium equation div(S·F^T) + f=0 under quasi-static settings) and the heat conduction equation describing heat conduction and thermal strain. The heat conduction equation includes a thermal strain term caused by the temperature change ΔT, which is related to the deformation gradient tensor through the material's thermal expansion coefficient. The coupled equations constitute the fundamental governing equations of the curing deformation field prediction module, used to solve for the displacement field u and the stress field σ.

[0031] Step 126: Spatial discretization is performed on the basic control equations of the solidification deformation field prediction module to generate solidification deformation field prediction calculation grid cells corresponding to the solid filling structure inside the target three-dimensional model, and storage space is allocated for each solidification deformation field prediction calculation grid cell to store transient displacement vector, transient strain tensor and transient stress tensor.

[0032] The system employs the finite element method to discretize the governing equations of the solidification deformation field prediction module, generating a set of computational mesh elements for solidification deformation field prediction with nodes as the basic unit. This set of mesh elements closely matches the geometry of the target 3D model. The system allocates storage space for each node in the computational mesh element, including a displacement vector storage area containing three components (u_x_disp, u_y_disp, u_z_disp) to store the node's displacement in the x, y, and z directions. The system also calculates the strain at the Gaussian point within each element based on the node displacements and element shape functions. For each Gaussian point, a tensor storage area is allocated to store the six independent strain components (normal strain ε_xx, ε_yy, ε_zz and shear strain γ_xy, γ_xz, γ_yz), and a tensor storage area is allocated to store the corresponding six independent stress components.

[0033] Step 127: After constructing the basic governing equations of each physics module and performing spatial discretization, establish the topological correspondence between the computational grid cells of each module, and configure the data transfer interface and encapsulate the multiphysics coupling solver.

[0034] Step 1271: Establish the topological correspondence between the computational grid cells of the flow stress field calculation module, the reaction progress field evolution module, and the solidification deformation field prediction module, so that the computational grid cells of the same spatial location in different physics field modules can be mutually indexed and exchange data through spatial coordinates.

[0035] Specifically, the system creates a global spatial index mapping table. Since the three modules may use different meshing strategies (e.g., the flow stress field uses a structured mesh, and the solidification deformation field uses an unstructured tetrahedral mesh), this mapping table records the identifier of the reaction progress field mesh cell corresponding to the center point coordinates of each flow stress field mesh cell, as well as the identifier of the solidification deformation field mesh node closest to that center point. Conversely, for each node of the solidification deformation field, its corresponding flow stress field mesh cell and reaction progress field mesh cell can also be found through the mapping table. Thus, physical quantities calculated by any module at a certain spatial location can be accurately transferred to the calculation units of other modules at the same spatial location through this mapping relationship.

[0036] Step 1272: Encapsulate the flow stress field calculation module, the reaction progress field evolution module, and the solidification deformation field prediction module into a multiphysics coupling solver, and configure a data transfer interface in the multiphysics coupling solver. The data transfer interface is used to realize the one-way transfer of the flow stress field calculation results to the reaction progress field evolution module, the one-way transfer of the reaction progress field evolution calculation results to the solidification deformation field prediction module, and the feedback transfer of the solidification deformation field prediction calculation results to the flow stress field calculation module on a continuous simulation time step.

[0037] The system encapsulates the aforementioned modules within a unified computational framework, which includes three standardized data transfer interfaces. The first interface, after each simulation time step, interpolates the transient velocity vector field data calculated by the flow stress field module and transfers it to the corresponding mesh element of the reaction progress field module, serving as input for the convection term in the energy conservation equation. The second interface, after the reaction progress field module completes its calculations, interpolates the calculated transient temperature field data and transfers it to the mesh nodes of the solidification deformation field module, serving as the driving condition for thermal strain calculations. The third interface, after the solidification deformation field module completes its calculations, feeds back the calculated transient displacement vector field data to the flow stress field module via a mapping table, updating the computational mesh coordinates of the flow stress field module and simulating geometric nonlinear effects.

[0038] Step 1273: Apply an initial condition set and a boundary condition set corresponding to the spatial boundary of the target 3D model to each computational grid cell in the multiphysics coupled solver. The initial condition set includes the temperature value, reaction completion value and velocity vector value of each computational grid cell at the initial time. The boundary condition set includes the mass flow rate boundary at the mold inlet, the no-slip boundary at the mold wall and the stress free boundary at the free surface.

[0039] The system sets a uniform initial state for the solver. For example, the initial temperature of all computational mesh elements is set to room temperature, such as 293 Kelvin; the initial reaction completion value is 0, indicating no vulcanization; and the initial velocity vector value is 0. Next, the system applies boundary conditions. At the die inlet boundary in the simulated extrusion molding, a time-varying mass flow rate boundary condition is applied, specifying the mass flowing into the inlet boundary surface per unit time, determined by the extrusion speed of the extruder. At the boundary in contact with the die wall, a no-slip boundary condition is applied, specifying the velocity vector of the fluid particles on the wall as 0. On the free surface of the extruded seal, i.e., the boundary in contact with air, a stress-free boundary condition is applied, specifying the stress vector on this boundary as 0.

[0040] Step 1274: Store and call the multiphysics coupled solver configured with initial and boundary conditions as an adaptive physics model associated with the spatial coordinate system of the target 3D model.

[0041] At this point, an adaptive physics model for a specific 3D sealing target model has been completed. This model includes three core physics modules and their coupling relationships used to describe the entire process of rubber material from extrusion, vulcanization to cooling. It has been saved as an executable calculation model file, waiting to be called for simulation.

[0042] Step 130: Input the initial geometric topology information set and the initial material property information set into the adaptive physics model for spatiotemporal coupling molding process simulation processing, synchronously iteratively solve the multiphysics interaction between each module, and generate dynamic evolution trajectory data of the target three-dimensional model from the initial shape of the extrusion die to the fully vulcanized molding shape.

[0043] This step is the core execution phase of the simulation, transforming the static model and data into a dynamic, physically rich evolutionary process. The system initiates the pre-built adaptive physics model, loading information describing the initial geometry and material properties of the seal, and assigning initial spatial position and material properties to each computational unit. Subsequently, the system enters a refined, multi-physics synchronous iterative solution loop. Within each time step, the system first calls the flow stress field module to solve for the instantaneous velocity and stress distribution based on the current boundary conditions and the geometric deformation inherited from the previous time step. During the solution process, the system tracks the flow front in real time and updates the material filling state.

[0044] Next, the system invokes the reaction progress field evolution module, using the newly calculated velocity field as the convection condition, and the current temperature field and reaction degree as initial values ​​to solve for the updated temperature and reaction completion distribution, and marks the units that have begun vulcanization according to the reaction degree. Then, the system invokes the curing deformation field prediction module, using the newly obtained temperature field as the thermal load, to solve the equilibrium equations on the vulcanized units, calculating the displacement, strain, and stress caused by thermal contraction. Finally, the system feeds back the calculated displacement field to the flow stress field module to correct the geometry for the next time step. Through this iterative process, the system completely records the spatial position, flow velocity, and reaction degree of each material particle at every moment, ultimately weaving this massive amount of data into a dynamic evolution trajectory dataset that can trace the entire molding history.

[0045] Step 131: At the beginning of the first simulation time step, the voxelized meshing parameters in the initial geometric topology information set are loaded into each computational grid cell of the flow stress field calculation module, reaction progress field evolution module, and solidification shape prediction module of the adaptive physical field model, so that each computational grid cell obtains the spatial coordinate position and initial volume occupancy corresponding to the initial extrusion shape of the target 3D model.

[0046] The system reads mesh generation parameters from the initial geometric topology information. These parameters define the spatial positions of all computational mesh elements at the initial moment. For the Eulerian meshes of the flow stress field and reaction progress field modules, each mesh element is assigned a fixed spatial coordinate range. Based on the initial geometry of the target 3D model at the extrusion die, the system calculates the proportion of material filling each Eulerian mesh element, i.e., the initial volume occupancy. For elements at the die, if their spatial range is entirely within the model, their initial volume occupancy is set to 1; if partially within the model, a floating-point number between 0 and 1 is calculated using an intersection algorithm as the initial volume occupancy; for elements completely outside the model, the initial volume occupancy is set to 0. For the Lagrangian meshes of the solidification deformation field module, their node coordinates are set to precisely correspond to the geometric positions of the initial shape of the target 3D model.

[0047] Step 132: Call the flow stress field calculation module. Based on the boundary condition configuration of the current simulation time step and the displacement vector data in the solidification deformation field prediction calculation results passed from the previous simulation time step, solve the mass conservation equation and momentum conservation equation coupled with the constitutive equation on each calculation grid cell to obtain the transient velocity vector field distribution data and transient stress tensor field distribution data of the current simulation time step.

[0048] At the current nth simulation time step, the system first reads the boundary conditions configured in step 1273, such as the inlet mass flow rate corresponding to the current time. Simultaneously, the system receives the displacement vector field calculated at the (n-1)th time step from the solidification deformation field module. Using this displacement vector field, the system updates the geometry of the flow stress field calculation grid, i.e., moves the grid nodes according to the displacement vector, reflecting the impact of geometric changes caused by solidification shrinkage on subsequent flow. Then, on the updated grid, the system iteratively solves the mass and momentum conservation equations using a semi-implicit method (SIMPLE algorithm) for example, the pressure coupling equations, and combines this with the viscoelastic constitutive equations constructed in step 121 to calculate the transient pressure value p_n, the velocity vector u_n (containing u_x_n, u_y_n, u_z_n components), and the transient stress tensor σ_n (containing six independent components) at the center of each computational grid cell. The solution process continues until the residuals of all equations fall below a preset convergence criterion, such as 10 to the power of -4.

[0049] Step 133: Extract the velocity vectors of the computational grid cells located near the flow front from the transient velocity vector field distribution data, store them as new entries in the material particle instantaneous flow velocity vector sequence at the current simulation time step, and update the occupancy status of each computational grid cell according to the transient velocity vector field distribution data. Computational grid cells with a volume occupancy rate exceeding the preset filling threshold are marked as filled cells.

[0050] The system defines a flow front identification algorithm, for example, identifying grid cells whose volume occupancy changes from 0 to greater than 0 in the current time step, or those cells with a volume occupancy greater than 0 and adjacent to cells with a volume occupancy of 0. The system extracts the velocity vector u_n from the center point of these leading-edge cells and appends it, along with the current time step identifier t_n, as a new entry to a global velocity vector time-series database table. This entry contains the time identifier t_n, the cell identifier CellID_flow, and the velocity vector components u_x_n, u_y_n, and u_z_n. Simultaneously, the system iterates through all flow stress field calculation grid cells, comparing the volume occupancy calculated for each cell in the current time step with a preset filling threshold, such as 0.98. If the current volume occupancy of a cell is greater than 0.98, the system updates the cell's status flag from "unfilled" to "filled" and stores this status information in the cell's corresponding data structure.

[0051] Step 134: When the flow stress field solution and state update of the current time step are completed, iterative solution and data transfer are performed through each module until the iterative loop of all simulation time steps is completed.

[0052] Step 1341: Call the reaction progress field evolution module. Based on the boundary condition configuration of the current simulation time step and the transient velocity vector field distribution data calculated by the flow stress field calculation module, take the reaction completion value of the previous simulation time step as the initial value, solve the energy conservation equation coupled with the dynamic differential equation on each filled cell, and obtain the transient temperature field distribution data and transient reaction completion field distribution data of the current simulation time step.

[0053] The system uses the velocity vector field u_n calculated in step 132 as the known convection velocity and passes it to the reaction progress field evolution module. For each grid cell marked as "filled," the system uses the reaction completion degree α_n-1 and temperature T_n-1 calculated in the previous time step t_{n-1} as the initial values ​​for this iteration. Then, the system uses the finite volume method to discretize and solve the energy conservation equation coupled with the sulfurization reaction kinetics equation established in step 123. During the solution process, the convection term in the equation is calculated using the velocity field u_n, and the reaction heat source term is updated in real time according to α and T in the current iteration step. After iterative solution, the system obtains the transient temperature value T_n and transient reaction completion degree value α_n for each filled cell in the current time step t_n.

[0054] Step 1342: Extract the reaction completion value of each filled cell from the transient reaction completion field distribution data, store it as a new entry in the local reaction completion percentage sequence of material particles in the current simulation time step, and identify the computational grid cells whose reaction completion value exceeds the preset gelation threshold based on the transient reaction completion field distribution data, and mark them as vulcanized cells.

[0055] The system iterates through all filled reaction progress field computational grid cells, appending the reaction completion value α_n calculated for each cell at the current time step, along with the cell identifier CellID_cure and time identifier t_n, as a new entry to a global reaction completion time-series database table. Next, the system compares each cell's α_n with a preset gelation threshold, such as 0.05 (indicating the beginning of cross-linking network formation). When a cell's α_n first exceeds 0.05, the system updates the cell's state from "filled" to "cured" and records this moment t_n_gel in its data structure. This marker is crucial for subsequent analysis of the starting point of curing shrinkage.

[0056] Step 1343: Call the solidification deformation field prediction module. Based on the boundary condition configuration of the current simulation time step and the transient temperature field distribution data calculated by the reaction progress field evolution module, solve the heat conduction equation and momentum conservation equation coupled with the constitutive model on each vulcanized unit to obtain the transient displacement vector field distribution data and transient stress tensor field distribution data of the current simulation time step.

[0057] The system uses the temperature field T_n calculated in step 1341 as a known thermal load and passes it to the curing deformation field prediction module. For each element marked as "cured," the system calculates the thermal strain based on the temperature change ΔT = T_n - T_ref at its nodes (where T_ref is the reference stress-free temperature, such as the temperature t_n_gel at the time of gelation). Then, the system uses the finite element method to solve the equilibrium equations of the coupled thermal strain and hyperelastic constitutive model established in step 125, considering geometric and material nonlinearities. The solution process aims to find the nodal displacement field u_disp_n that satisfies the force equilibrium condition. Based on the obtained nodal displacements, the strain tensor ε_n and stress tensor σ_n at the integration point of each element are further calculated.

[0058] Step 1344: Superimpose the displacement vector of each vulcanized unit in the transient displacement vector field distribution data onto the original spatial coordinate position of the unit to generate the updated spatial coordinate position as a new entry in the material particle spatial position coordinate sequence in the current simulation time step and store it. Calculate the principal stress difference of each vulcanized unit based on the transient stress tensor field distribution data as the residual stress indicator value.

[0059] The system iterates through all nodes of the computational mesh for the solidified deformation field of the vulcanized element. For each node, its original coordinates (X_0, Y_0, Z_0) are added to the displacement vector u_disp_n (containing u_x_disp_n, u_y_disp_n, u_z_disp_n components) calculated at the current time step to obtain the updated coordinates (X_n, Y_n, Z_n) of the node at the current time step t_n. The system appends the identifier NodeID_solid and the updated coordinates (X_n, Y_n, Z_n) of each node as a new entry to a global spatial location coordinate time-series database table. Simultaneously, for each vulcanized element, the system calculates the three principal stresses (σ1_n, σ2_n, σ3_n) based on the stress tensor σ_n at its integration point by solving an eigenvalue problem. Then, the system calculates the difference between the maximum principal stress and the minimum principal stress, i.e., σ1_n-σ3_n, uses this difference as the residual stress indicator value S_res_n of the element at the current time step, and stores it in the element's data structure.

[0060] Step 1345: The transient displacement vector field distribution data calculated by the solidification deformation field prediction module is fed back to the flow stress field calculation module through the data transmission interface, which is used to correct the spatial coordinate position and volume occupancy of the calculation grid cell in the next simulation time step.

[0061] The system, through the data transfer interface configured in step 1272, packages and transmits the displacement vectors u_disp_n of all vulcanized nodes calculated at the current time step t_n, along with the node identifiers, to the flow stress field calculation module. Before proceeding to the next time step t_{n+1}, the flow stress field module moves the boundaries of its Eulerian mesh or updates the coordinates of its internal mesh nodes using dynamic meshing techniques based on the received displacement vectors. Simultaneously, based on the updated mesh, it recalculates the volume occupancy of each Eulerian element relative to the updated geometry, preparing for the flow calculation at the next time step.

[0062] Step 1346: Repeat the solution and data transfer process of each module until all computational grid cells are marked as vulcanized cells and the temperature values ​​of all vulcanized cells drop below the preset room temperature stability threshold, and end the iterative loop of the simulation time step.

[0063] The system proceeds to the next simulation time step t_{n+1}, repeating steps 132 to 1345. At the beginning of each time step, the state of all computational grid cells is checked. This iterative loop continues until two termination conditions are met: first, all computational grid cells for the flow / reaction progress field have been marked as "cured"; second, the temperature value T_n of all "cured" cells is below a preset room temperature stability threshold, such as 300 Kelvin (slightly above room temperature to account for computational tolerance), indicating that the seal has cooled sufficiently and the deformation is basically stable. When both conditions are met simultaneously, the system stops iterating.

[0064] Step 1347: Arrange and combine the material particle spatial position coordinate sequence, material particle instantaneous flow velocity vector sequence, and material particle local response completion percentage sequence stored in each simulation time step according to the time sequence to generate dynamic evolution trajectory data.

[0065] The system reads all data from three time-series database tables. First, it organizes the coordinates (X_n, Y_n, Z_n) of each node in the spatial location coordinate sequence into a four-dimensional array (time, node identifier, x, y, z) in ascending order of time identifier t_n. Similarly, it organizes the velocities (u_x_n, u_y_n, u_z_n) of each leading edge unit in the velocity vector sequence in chronological order into a four-dimensional array (time, unit identifier, vx, vy, vz). Finally, it organizes the α_n of each unit in the reaction completion sequence in chronological order into a three-dimensional array (time, unit identifier, α). Finally, the system links these three types of data through a unified data structure. For example, through the mapping relationship between unit identifiers and node identifiers, it can find the flow velocity and reaction completion of the material near a certain spatial location at a specific time t_n. This complete dataset, containing the spatial location, flow velocity, and reaction degree of each material particle at each time step, is the desired dynamic evolution trajectory data.

[0066] Step 140: Based on the characteristics of the spatial distribution of material particles, velocity vector direction and local reaction completion in the dynamic evolution trajectory data, perform joint mechanism analysis of material flow and deformation behavior to obtain the cross-sectional profile distortion mode, vulcanization shrinkage deformation spatial distribution and residual stress concentration area location information of the extrusion swell effect affected area.

[0067] This step involves in-depth mining and knowledge extraction of massive amounts of dynamic evolution trajectory data, aiming to transform raw simulation data into physical mechanism information that engineers can intuitively understand and apply. The system first focuses on the area near the extrusion die, identifying the actual profile of the extruded material through cluster analysis of the spatial trajectories of material particles. By comparing this profile with design standards, the magnitude of the extrusion swell effect can be quantified, and the area affected by this effect and its distortion pattern can be located based on the geometric characteristics of the profile distortion (such as uniform expansion, local protrusions, or wavy patterns). Next, the system shifts its focus to the material's curing process. By tracking the change in the reaction completion rate of each material particle over time, the acceleration start point and completion point of the vulcanization reaction can be accurately determined.

[0068] Based on this, by comparing the spatial coordinates of material particles at the start and end of the reaction, the net shrinkage deformation caused by the combined effects of vulcanization cross-linking network formation and thermal shrinkage can be calculated, and a three-dimensional spatial distribution map of the shrinkage deformation across the entire seal can be plotted. The system further identifies areas of abnormal deformation from the shrinkage distribution map and determines whether the shrinkage is isotropic or anisotropic by analyzing their geometry. Finally, the system analyzes the evolution of residual stress, summing the accumulated stress indicators of each unit throughout the cooling process to obtain the final residual stress distribution, and using a peak detection algorithm to locate the critical areas of stress concentration. By superimposing the abnormal shrinkage areas with the stress concentration areas, the system can accurately identify the risk areas most likely to produce serious defects such as cracking and warping.

[0069] Step 141: Extract a subset of material particle spatial position coordinate sequences within a preset spatial range near the extrusion die outlet from the dynamic evolution trajectory data. Perform spatial cluster analysis on the subset of material particle spatial position coordinate sequences to identify clustered regions where the distribution density of coordinate points in the plane perpendicular to the extrusion direction is significantly higher than the average distribution density. Compare the boundary contour of the clustered region with the standard cross-sectional contour of the corresponding position of the target 3D model and calculate the maximum distance of the outward bulge of the contour boundary as the extrusion swell effect amplitude value.

[0070] The system sets the extrusion direction as the positive z-axis of the coordinate system. The extrusion die exit plane is defined as the z=0 plane. From the dynamic evolution trajectory data generated in step 1347, the system filters out spatial coordinate points whose z-coordinate values ​​are within the range of 0 to a preset distance D_bulge (e.g., 10 mm) at all time steps. For these coordinate points, the system ignores their z-coordinates and only focuses on their projected coordinates (x, y) on the xy plane. Next, the system uses a density-based spatial clustering algorithm (DBSCAN) to cluster these two-dimensional projected points. The algorithm identifies densely distributed point cloud regions using two parameters: the scanning radius Eps and the minimum number of points MinPts. These dense regions correspond to the physical cross-section of the extruded material. For each identified cross-section cluster, the system calculates its convex hull as the actual profile of the cross-section. Then, the system reads the standard design cross-sectional profile of the seal at the corresponding z-position from the CAD model. After aligning the actual profile with the standard profile, the system calculates the outward offset distance of the actual profile points relative to the standard profile in multiple radial directions. The maximum value of the offset distance in all directions is taken as the extrusion bulge effect amplitude value A_bulge(z) at that cross section. The system repeats this process for multiple cross sections with z values ​​in the range of 0 to D_bulge to obtain the amplitude value variation curve A_bulge(z) along the extrusion direction.

[0071] Step 142: Based on the continuous change trend of the extrusion expansion effect amplitude value in the extrusion direction, the spatial interval where the extrusion expansion effect amplitude value continuously increases and exceeds the preset distortion threshold is divided into the extrusion expansion effect influence area. Within the extrusion expansion effect influence area, the cross-sectional contour distortion mode is classified into the overall uniform expansion mode, the local asymmetric protrusion mode, and the edge wavy undulation mode according to the shape characteristics of the contour boundary protrusion.

[0072] The system analyzes the curve A_bulge(z) obtained in step 141 and sets a distortion threshold Th_bulge, for example, 0.5 mm. Starting from z=0, the system scans along the positive z-axis, identifying the first z-position z_start where A_bulge(z) continuously exceeds Th_bulge, and the last z-position z_end where A_bulge(z) falls back to below Th_bulge. The interval [z_start, z_end] is the region affected by the extrusion bulging effect. Within this region, for the actual contour of each cross-section, the system further analyzes its shape characteristics. By calculating the Fourier descriptor of the actual contour, it decomposes it into components of different frequencies. If the low-frequency components of the contour (such as 0th and 1st order) are absolutely dominant, and the contour is approximately concentrically enlarged, it is classified as a global uniform expansion mode. If significant asymmetric components appear in the low-frequency components of the contour, such as a significant protrusion within a specific azimuth angle range, it is classified as a local asymmetric protrusion mode. If the high-frequency components of the profile show a significant increase in energy, indicating periodic fluctuations at the profile edges, it is classified as an edge-wavy undulation pattern. The system records each cross-sectional position z and its corresponding distortion pattern.

[0073] Step 143: Based on the cross-sectional profile distortion pattern obtained from the extrusion swell effect analysis, calculate the vulcanization shrinkage deformation and identify its spatial distribution, and calculate the residual stress and locate the concentrated area.

[0074] Step 1431: Extract the percentage of reaction completion for each vulcanized unit at different simulation time steps from the dynamic evolution trajectory data. Calculate the derivative of the percentage of reaction completion for each vulcanized unit along the time dimension to obtain the reaction rate curve of each vulcanized unit as a function of time. Determine the acceleration start time and completion time of the vulcanization reaction for that unit based on the peak time of the reaction rate curve as a function of time.

[0075] The system iterates through all units marked as "vulcanized" in step 1342. For each unit, it extracts a series of discrete points (t_i, α_i) from the reaction completion time-series data stored in step 1342, showing the change of its reaction completion α over time t. The system uses numerical differentiation methods, such as the finite difference method, to calculate the reaction rate v_i = (α_{i+1} - α_{i-1}) / (t_{i+1} - t_{i-1}) at each time point t_i. This yields the reaction rate curve v(t) for that unit. The system searches for the first local maximum point on this curve; the time t_acc corresponding to this point is defined as the starting time of the vulcanization reaction acceleration, before which the reaction rate is low. The system continues searching forward along the time axis; when the reaction rate v(t) decreases and first falls below the preset completion threshold v_complete (e.g., one percent of the maximum reaction rate), the corresponding time t_complete is defined as the completion time of the vulcanization reaction.

[0076] Step 1432: For each vulcanized unit, calculate the Euclidean distance between its spatial coordinate position at the start of the vulcanization reaction acceleration and its spatial coordinate position at the completion of the vulcanization reaction. This distance is taken as the shrinkage deformation of the unit due to the formation of the vulcanization crosslinking network. Arrange the shrinkage deformation of all vulcanized units according to their spatial coordinate positions to generate a spatial distribution map of shrinkage deformation.

[0077] For each vulcanized unit, the system retrieves the coordinates (x_acc, y_acc, z_acc) of the unit (or its corresponding node) at time t_acc and the coordinates (x_comp, y_comp, z_comp) at time t_comp from the spatial coordinate sequence stored in step 1347. Then, it calculates the Euclidean distance between the two points, D_shrink = sqrt((x_comp - x_acc)^2 + (y_comp - y_acc)^2 + (z_comp - z_acc)^2), where D_shrink represents the net shrinkage deformation of the unit due to the combined effects of the vulcanization crosslinking network formation and subsequent thermal shrinkage. The system iterates through all vulcanized units, obtaining a shrinkage deformation value for each unit. Then, it maps these values ​​back to their spatial coordinate positions as field variables, forming a three-dimensional scalar field. Using three-dimensional visualization techniques, such as color mapping, the magnitude of the shrinkage deformation at each location is displayed in the geometric space of the target three-dimensional model, thus generating a spatial distribution map of the shrinkage deformation.

[0078] Step 1433: In the spatial distribution map of shrinkage deformation, identify the spatial regions where the shrinkage deformation value exceeds the preset shrinkage threshold and is continuously distributed, mark them as abnormal shrinkage deformation regions, and determine whether the dominant mechanism of shrinkage deformation is isotropic volumetric shrinkage or anisotropic directional shrinkage based on the geometry and spatial extension direction of the abnormal shrinkage deformation regions.

[0079] The system sets a shrinkage threshold Th_shrink, for example, 0.1 mm. In the three-dimensional scalar field generated in step 1432, a region growing algorithm is used to aggregate all spatially adjacent units where D_shrink > Th_shrink, forming connected regions. Each connected region is marked as a shrinkage deformation anomaly region, and its contained unit identifiers and spatial extent are recorded. For each marked anomaly region, the system analyzes its shape. The three-dimensional covariance matrix of the region's point cloud is calculated, and its eigenvalues ​​are decomposed to obtain three eigenvalues ​​λ1 ≥ λ2 ≥ λ3. If the three eigenvalues ​​are similar in magnitude, it indicates that the region extends to a similar degree in the three directions, approximating a sphere, and the dominant shrinkage mechanism is determined to be isotropic volumetric shrinkage. If λ1 is much larger than λ2 and λ3, it indicates that the region is significantly elongated in one direction, appearing as a thin strip, and this elongation direction is consistent with or perpendicular to the extrusion direction (z-axis), and the dominant shrinkage mechanism is determined to be anisotropic directional shrinkage, possibly related to molecular orientation or flow-induced anisotropy.

[0080] Step 1434: Extract the residual stress indication value of each vulcanized unit at different simulation time steps from the dynamic evolution trajectory data, accumulate and sum the residual stress indication value of each vulcanized unit along the time dimension to obtain the final residual stress value of the unit after complete cooling, and arrange the final residual stress values ​​of all vulcanized units according to their spatial coordinate positions to generate a residual stress spatial distribution map.

[0081] For each vulcanized element, the system extracts the residual stress indicator sequence S_res_i from the residual stress indicator value S_res stored in step 1344, starting from the time t_cure_start when the element is marked as "vulcanized" until the simulation ends at time t_end. Since residual stress gradually accumulates and freezes during the cooling process, the system performs time integration or summation on this sequence to obtain the final residual stress value S_res_final = ΣS_res_i * Δt_i for the element, where Δt_i is the step size of each time step, and S_res_final represents the final accumulated residual stress level within the element. The system iterates through all vulcanized elements to obtain S_res_final for each element. Then, the above values ​​are mapped back to their spatial coordinate positions in the form of field variables, forming a three-dimensional scalar field, i.e., a spatial distribution map of residual stress.

[0082] Step 1435: In the residual stress spatial distribution map, the peak detection algorithm is used to identify the local maximum points of the residual stress values, and the area where the residual stress values ​​in the surrounding preset neighborhood are all higher than the preset stress threshold is located as the residual stress concentration area. The center coordinates and coverage area of ​​each residual stress concentration area are marked.

[0083] The system runs a three-dimensional peak detection algorithm on the residual stress spatial distribution map generated in step 1434. This algorithm identifies local maxima points whose values ​​are greater than those of all neighboring elements by comparing the S_res_final value of each element with those of all its neighboring elements. For each detected local maximum point, a cube neighborhood of a preset size is defined centered on its coordinates, for example, a cube with a side length of 2 mm. The system extracts all elements within this neighborhood and checks whether the S_res_final value of these elements is higher than a preset stress threshold Th_stress, for example, 1 MPa. If this condition is met, all elements within the neighborhood are collectively identified as a residual stress concentration region. The system records the center coordinates of this region (i.e., the coordinates of the local maximum point) and the spatial extent covered by the region (e.g., the minimum and maximum x, y, z coordinates). If the condition is not met, it may simply be an isolated high-stress point and does not constitute a concentration region.

[0084] Step 1436: Perform an overlay analysis of the spatial distribution of the shrinkage deformation anomaly region and the spatial distribution of the residual stress concentration region to identify the overlapping region that belongs to both the shrinkage deformation anomaly region and the residual stress concentration region, and mark the overlapping region as the potential occurrence area of ​​risk defects in material deformation behavior.

[0085] The system performs an intersection operation on the set of elements in all shrinkage deformation anomaly regions marked in step 1433 and the set of elements in all residual stress concentration regions marked in step 1435. This identifies computational mesh elements that exist within both a shrinkage deformation anomaly region and a residual stress concentration region. These elements, simultaneously subjected to abnormal shrinkage deformation and high residual stress, represent the most likely risk areas for severe defects such as cracking and warping during the forming process. The system separately lists the element identifiers and spatial coordinate ranges of these overlapping regions as key focus areas for subsequent defect source tracing analysis.

[0086] Step 150: Integrate the cross-sectional profile distortion mode, spatial distribution of shrinkage deformation, and location information of residual stress concentration areas to perform a molding defect tracing analysis, and generate molding process simulation results that include defect type identification, defect location coordinates, and defect evolution path.

[0087] This step synthesizes the aforementioned analysis results, aiming to construct a complete knowledge system that traces the entire lifecycle of a defect from its final form. The system uses spatial logic operations to superimpose problem areas caused by different physical mechanisms, thereby accurately identifying composite defects. For example, by spatially intersecting the extrusion swell-affected area and the shrinkage deformation anomaly area, the system can locate areas where dimensional accuracy is simultaneously affected by both mechanisms and mark them as "dimensional accuracy composite defects." Similarly, intersecting the extrusion swell-affected area and the residual stress concentration area can identify "stress cracking composite defects" that are highly prone to cracking due to complex flow histories and stress concentration.

[0088] After spatially identifying the defect types, the system initiates a defect backtracking analysis. For each identified defect, the system traces its corresponding material particles backward along the time axis, extracting the time and location when the extrusion swell effect begins to deviate, the time and location when the shrinkage begins due to accelerated vulcanization, and the time and location when residual stress begins to freeze as the temperature drops below the glass transition temperature. These times and locations together constitute the starting point of the defect evolution. Subsequently, the system connects the spatial trajectories of the material particles between the starting point and the final defect location, constructing a complete three-dimensional defect evolution path. Finally, the system integrates the defect type, the final location, and this evolution path containing time and spatial information, arranging them in spatial order to form a structured and traceable simulation result.

[0089] Step 151: Perform an intersection operation between the spatial coordinate range of the extrusion bulging effect region marked in the cross-sectional profile distortion mode and the spatial coordinate range of the shrinkage deformation anomaly region marked in the shrinkage deformation spatial distribution. If there is a first overlapping region between the extrusion bulging effect region and the shrinkage deformation anomaly region, then mark the defect type in the first overlapping region as a dimensional accuracy composite defect, and record that the dimensional accuracy composite defect is caused by the combined effect of extrusion bulging and vulcanization shrinkage.

[0090] The system first obtains the spatial coordinate range of the region affected by the extrusion bulge effect from step 142, denoted as Region_bulge. From step 1433, it obtains the spatial coordinate range of all abnormal shrinkage deformation regions, denoted as the set {Region_shrink_k}, k=1..M. The system calculates the spatial overlap between Region_bulge and each Region_shrink_k. If there exists a k such that Region_bulge∩Region_shrink_k is not empty, this non-empty region is defined as the first overlapping region, denoted as Overlap_1. The system identifies the defect type of all spatial points within Overlap_1 as the type code DIM_001, which corresponds to "composite dimensional accuracy defect". Simultaneously, the description field of this defect type records that its cause is the spatial coupling and superposition of geometric distortion caused by the extrusion bulge effect and the abnormal deformation generated by subsequent vulcanization shrinkage, resulting in a significant deviation of the final dimension from the design value.

[0091] Step 152: Perform an intersection operation between the spatial coordinate range of the extrusion expansion effect region marked in the cross-sectional profile distortion mode and the spatial coordinate range of the residual stress concentration region marked in the residual stress concentration region location information. If there is a second overlapping region between the extrusion expansion effect region and the residual stress concentration region, then mark the defect type in the second overlapping region as a stress cracking composite defect, and record that the stress cracking composite defect is caused by the combined effect of extrusion expansion and residual stress.

[0092] The system performs an intersection operation between Region_bulge and the set of spatial coordinate ranges {Region_stress_p}, p=1..Q, of all residual stress concentration regions obtained in step 1435. If there exists a p such that Region_bulge∩Region_stress_p is not empty, this non-empty region is defined as the second overlapping region, denoted as Overlap_2. The system identifies the defect type of all spatial points within Overlap_2 as the type code CRK_002, which corresponds to "stress cracking composite defect". In the description field of this defect type, it is recorded that its cause is that the extrusion swelling effect causes the material in this region to undergo a complex flow history and molecular orientation, forming a potential weak point in strength. At the same time, this region happens to be the location where residual stress is highly concentrated after cooling. Under the combined effect of the two, it is very easy to induce and propagate microcracks.

[0093] Step 153: After identifying the defect types, trace the evolution starting point and construct the evolution path for each type of defect that has been identified.

[0094] Step 1531: For each spatial location marked in the extrusion bulging effect influence area in the cross-sectional profile distortion mode, trace the flow path of the corresponding material particle in the dynamic evolution trajectory data, extract the earliest moment when the velocity direction deviates from the extrusion direction on the flow path as the starting time point of the extrusion bulging effect, and take the spatial location corresponding to the starting time point as the evolution starting coordinate of the extrusion bulging defect.

[0095] For each spatial point P belonging to the extrusion bulge effect region in Overlap_1 and Overlap_2, the system first uses spatial indexing to find the particle (or a group of particles) representing that point in the dynamic evolution trajectory data. Then, the system traces back the historical trajectory of that particle from the die extrusion (z≈0) and extracts its velocity vector sequence. The system analyzes the magnitude of the velocity vector components perpendicular to the extrusion direction (xy plane) in this sequence. A deviation threshold is defined, for example, the lateral velocity component exceeds 5% of the axial velocity component. The system searches along the time axis from front to back and finds the first moment t_bulge_start, at which point the lateral velocity component of the particle first continuously exceeds the threshold, marking the beginning of extrusion bulge deformation. The system records the spatial coordinates (x_bulge_start, y_bulge_start, z_bulge_start) of the particle at this moment and uses these coordinates as the evolution starting coordinates of the extrusion bulge defect associated with that spatial point P.

[0096] Step 1532: For each spatial location marked in the spatial distribution of shrinkage deformation anomaly region, trace the sulfidation reaction process of the corresponding material particle in the dynamic evolution trajectory data, extract the moment when the reaction rate of the particle first exceeds the preset reaction rate threshold as the starting time point of sulfidation shrinkage, and take the spatial location corresponding to the starting time point as the evolution starting point coordinate of shrinkage deformation defect.

[0097] For each spatial point Q belonging to the shrinkage deformation anomaly region in Overlap_1, the system finds its corresponding material particle and traces back the reaction completion history of that particle. The reaction rate curve v(t) for that particle has already been obtained from step 1431. A reaction rate threshold v_threshold is set, for example, 10% of the maximum reaction rate. The system searches along the time axis from front to back, finding the moment t_shrink_start when the reaction rate v(t) first rises from a low level and exceeds v_threshold. This moment marks the beginning of the acceleration of the vulcanization crosslinking reaction, and the resulting shrinkage effect is about to become significant. The system records the spatial coordinates (x_shrink_start, y_shrink_start, z_shrink_start) of the particle at this moment as the starting coordinates of the evolution of the shrinkage deformation defect associated with that spatial point Q.

[0098] Step 1533: For each spatial location marked in the residual stress concentration area in the residual stress concentration area location information, trace the temperature change history of the corresponding material particle in the dynamic evolution trajectory data, extract the moment when the temperature of the particle first drops below the preset glass transition temperature as the starting time point of residual stress freezing, and take the spatial location corresponding to the starting time point as the evolution starting point coordinate of the residual stress defect.

[0099] For each spatial point R belonging to the residual stress concentration region in Overlap_2, the system finds its corresponding material particle and traces the temperature change history T(t) of that particle. The glass transition temperature T_g of the rubber material is obtained from the initial material property information. The system searches along the time axis from front to back, finding the moment t_stress_start when the temperature T(t) first drops from above T_g to below T_g. Before this moment, the material is in a highly elastic state, the molecular chains can move, and the stress is easily relaxed; after this moment, the material enters the glassy state, the molecular chains are frozen, and the stress is frozen to form residual stress. Therefore, t_stress_start is considered the starting point for the freezing of residual stress. The system records the spatial coordinates (x_stress_start, y_stress_start, z_stress_start) of the particle at this moment as the starting coordinates for the evolution of residual stress defects associated with that spatial point R.

[0100] Step 1534: Based on the evolution starting point coordinates of each defect type and the subsequent flow path and reaction path of the corresponding material particles in the dynamic evolution trajectory data, construct a continuous spatial curve that starts from the evolution starting point coordinates and extends along the extrusion direction and radial direction to the final defect occurrence location, as a description of the evolution path of the defect.

[0101] This step aims to connect discrete start and end point information into a continuous, traceable spatiotemporal trajectory. For each specific defect identified in steps 151 and 152 (e.g., a specific region within Overlap_1), the system already possesses the final location coordinates P_final (e.g., the center point of the region) of the defect, as well as the evolutionary starting point coordinates P_start obtained through step 1531 or 1532. The system extracts the complete spatial trajectory of the particle corresponding to P_start from time t_start to the end of the simulation from the dynamic evolution trajectory data. This trajectory is a series of spatial points (x(t), y(t), z(t)) ordered by time. Since the evolution of the defect is closely related to the motion of the material, this trajectory itself describes the migration path of the material particles from the defect's starting point to its final location. The system performs spline interpolation on this trajectory to generate a smooth, continuous spatial curve Curve_defect, which starts from P_start, meanders, and finally reaches P_final. For composite defects in dimensional accuracy, this curve may start from a point near the die opening and gradually develop to the position where the contour distortion is most severe as the material flows and vulcanization shrinks. This curve describes the evolution path of the defect.

[0102] Step 1535: Integrate the defect type identifier, defect location coordinates, and defect evolution path description to generate defect description information for each defect. Then, sort all the defect description information according to the order of defect location coordinates along the extrusion direction to generate structured molding process simulation results.

[0103] The system creates a structured data record for each identified defect. This record contains a unique defect identifier (e.g., DFCT_001), its type identifier (e.g., DIM_001 or CRK_002), the coordinates of the defect's location (e.g., P_final), and the evolution path description Curve_defect generated in step 1534. The evolution path is stored as an ordered list of spatial coordinate points. After collecting all defect records, the system sorts them in ascending order of the z-component of the defect's location coordinates (i.e., along the extrusion direction). The sorted list of defect records, together with the various physical field distribution maps (e.g., shrinkage deformation distribution, residual stress distribution, etc.) obtained in step 140, constitutes the final structured and traceable simulation result dataset of the molding process.

[0104] Following step 150, the method further includes: Step 210: After generating the molding process simulation results containing defect type identifiers, defect location coordinates, and defect evolution paths, obtain the preset set of process parameters and parse out the extrusion speed control curve, vulcanization temperature control curve, and cooling rate control curve. Map the marked defect location coordinates in the molding process simulation results to the spatiotemporal trajectory segments of the corresponding material particles in the dynamic evolution trajectory data. Backtrack from the spatiotemporal trajectory segments to extract the instantaneous extrusion speed value experienced by each defect location coordinate at the start time of the extrusion swelling effect, the reaction temperature value experienced at the start time of the vulcanization shrinkage, and the cooling rate value experienced at the start time of the residual stress freezing. Generate a defect-process parameter correlation comparison set containing defect identifiers and corresponding key process parameter anomalies.

[0105] The system reads a preset set of process parameters from the process design file. This set contains three key control curves: one is the extrusion speed control curve V_extrude(t), which describes the change of the extruder screw speed over time, in millimeters per second; another is the vulcanization temperature control curve T_cure_zone(z), which describes the temperature settings of each section of the vulcanization channel, as a function of the position z along the extrusion direction; and the third is the cooling rate control curve R_cool(t), which describes the temperature gradient of the cooling zone, in Kelvin per second. For each defect in the simulation results generated in step 1535, for example, defect identified as DFCT_001, its final position is P_final, and the evolution starting coordinates are P_bulge_start (extrusion bulge start), P_shrink_start (vulcanization shrinkage start), and P_stress_start (stress freezing start). The system uses dynamic evolution trajectory data to find the mass point located at P_bulge_start at time t_bulge_start and extracts the instantaneous extrusion speed value V_bulge given by the control curve V_extrude(t) at that time.

[0106] Similarly, the particle located at P_shrink_start at time t_shrink_start is found, and the reaction temperature value T_shrink at that location is extracted from the T_cure_zone(z) curve based on its z-coordinate. The particle located at P_stress_start at time t_stress_start is found, and the cooling rate value R_stress at that time is extracted from the R_cool(t) curve. The system associates the defect identifier DFCT_001 with the extracted three key process parameter values ​​(V_bulge, T_shrink, R_stress) to form a record. This process is repeated for all defects, ultimately generating a defect-process parameter association set, where each record specifies the exact value of a particular process parameter experienced by a defect when it occurred.

[0107] Step 220: Compare and calculate the difference between the abnormal values ​​of each key process parameter recorded in the defect-process parameter correlation comparison set and the standard process parameter values ​​at the corresponding positions in the preset process parameter set. Based on the process parameter types whose difference exceeds the preset tolerance range and their occurrence time periods in the comparison calculation results, determine the root causes of process parameter fluctuations that lead to the formation of each defect type. Use the root causes of process parameter fluctuations as input to drive the preset process parameter optimization model to perform backpropagation correction. Iteratively generate the local smoothing correction amount for the extrusion speed control curve, the heating rate adjustment amount for the vulcanization temperature control curve, and the segmented gradient cooling strategy for the cooling rate control curve. After integrating all correction amounts and adjustment strategies, output the output as a process parameter optimization instruction set for the moldless sealing part molding process.

[0108] For each record in the correlation comparison set, the system compares its V_bulge value with the standard extrusion speed value V_std at the corresponding time point t_bulge_start in the preset process parameter set, calculating the difference ΔV = |V_bulge - V_std| / V_std. Similarly, it calculates the temperature difference ΔT and the cooling rate difference ΔR. The system presets a set of tolerance ranges, such as speed tolerance δV = 0.05, temperature tolerance δT = 0.02, and cooling rate tolerance δR = 0.1. If the ΔV of a certain defect is greater than δV, it is considered that there is an abnormal fluctuation in the extrusion speed at that moment. The system analyzes the statistical results of all defects, for example, it finds that multiple dimensional accuracy composite defects (DIM_001) all correspond to a continuously high ΔV within a certain time period [t1, t2]. Therefore, it is determined that the fluctuation of the extrusion speed within this time period is the root cause of the process parameters leading to the formation of the DIM_001 type defect.

[0109] Next, the system inputs this information into a pre-defined process parameter optimization model, which can be a backpropagation neural network based on gradient descent. The model takes the current process parameter curves V_extrude(t), T_cure_zone(z), and R_cool(t) as input, and performs backpropagation calculations with the objective of minimizing the number of predicted defects. The model outputs corrections to the input curves. For example, a negative local smoothing correction ΔV_smooth(t) is applied to V_extrude(t) over the time interval [t1, t2] to reduce its fluctuations; a heating rate adjustment ΔT_rate is output for T_cure_zone(z) in a certain segment to change the start time of the vulcanization reaction; and a piecewise gradient cooling strategy is output for R_cool(t), such as slow cooling in the high-temperature range and rapid cooling near T_g. The system integrates these corrections and strategies to generate a structured process parameter optimization instruction set, which contains the modified curve key point data for the production control system to read and execute.

[0110] Following step 150, the method further includes: Step 310: After generating the molding process simulation results containing defect type identifiers, defect location coordinates, and defect evolution paths, obtain the historical molding process simulation results of multiple different batches. Each historical molding process simulation result contains a set of defect type identifiers marked for the corresponding batch, a defect location coordinate distribution map, and a defect evolution path network.

[0111] The system reads historical simulation result files from multiple batches, such as batches B001 to B050, from the simulation database. Each file contains all defect information identified in that simulation. The system loads this data into memory to create a dataset containing multiple experimental samples. Each sample (i.e., each batch) contains a list of defect type identifiers, a 3D point cloud dataset of defect location coordinates, and a network structure data consisting of all defect evolution paths, where each path is a spatial curve.

[0112] Step 320: Classify and aggregate the historical molding process simulation results of multiple different batches according to the defect type identifier. Perform spatial distribution density cluster analysis on the coordinates of all defects under the same defect type. Extract the hot spot areas of frequent occurrence in the extrusion direction and cross-sectional radial direction of each defect type. Perform skeleton extraction and path topology alignment processing on all defect evolution paths under the same defect type to generate a standard propagation path template that represents the defect type from the evolution starting point to the final position.

[0113] The system first groups all historical defects into groups based on their type (e.g., DIM_001, CRK_002). For type DIM_001, the system collects the location coordinates of all defects belonging to this type, forming a 3D point set. Then, the system uses a 3D kernel density estimation method to calculate the distribution density of the point set within the target 3D model space. Regions with density values ​​higher than a preset threshold Th_hotspot are identified as frequent hotspots for this defect type; for example, it was found that DIM_001 defects occur most frequently at the corners of seals and in the middle of long straight sections. Next, the system collects the evolution path curves of all DIM_001 defects. For each curve, the system uses a Laplace contraction-based skeletonization algorithm to extract its central skeleton line to reduce detail noise. Then, the system uses the Dynamic Time Warping (DTW) algorithm to align and average all skeleton lines in space, generating a standard path template Curve_template_DIM001 that can represent the common propagation law of this type of defect. This template curve reflects that the DIM_001 type defect usually starts from a specific area (such as the position near the die opening) and then develops to the final hot spot area along a certain spatial trajectory.

[0114] Step 330: Map the spatial coordinates of each path node on the standard propagation path template to the corresponding simulation time step in the dynamic evolution trajectory data. Extract the transient velocity vector field distribution data, transient temperature field distribution data, and transient response completion field distribution data of each path node in the simulation time step. Generate a multidimensional feature vector sequence describing the physical field evolution characteristics during defect evolution. Associate and store the multidimensional feature vector sequence as the dynamic feature identifier of the defect propagation process with the corresponding defect type identifier to generate a defect feature knowledge base for real-time monitoring of defect initiation and expansion trends during the molding process.

[0115] The system discretizes the standard path template Curve_template_DIM001 generated in step 320 into a series of sequentially arranged node coordinates P_template_1, P_template_2, ..., P_template_K. For each node P_template_i, the system finds the simulation time step t_i corresponding to the first arrival of a material particle at node position P_template_i in the dynamic evolution trajectory data of a reference, defect-free standard simulation process (or a representative historical batch). Then, the system extracts the transient velocity vector field data at time t_i from the database of the standard simulation process, interpolates it to obtain the velocity vector u_i at position P_template_i; extracts the transient temperature field data at time t_i, interpolates it to obtain the temperature T_i at that position; and extracts the transient response completion field data at time t_i, interpolates it to obtain the response completion α_i at that position. The system combines (u_i, T_i, α_i) into a multidimensional feature vector F_i. This process is performed sequentially on K nodes to obtain a feature vector sequence (F_1, F_2, ..., F_K). This sequence describes the key physical field evolution characteristics experienced by the DIM_001 type defect as it propagates along its standard path. The system associates this feature vector sequence with the defect type identifier DIM_001 and stores it in a dedicated defect feature knowledge base. In subsequent real-time production monitoring, the system can collect data from online sensors, extract the physical field characteristics on the material particle trajectory, and compare them with the standard sequence in the knowledge base, thereby providing early warnings of potential defect types and their evolution stages.

[0116] This application embodiment breaks through the barrier of isolated analysis of flow, vulcanization, and curing deformation in traditional simulations by creating an adaptive physics field model directly associated with the spatial coordinate system of the target 3D model. It realizes spatiotemporal coupling integrated simulation of the entire forming process from extrusion, vulcanization to cooling. Unlike simply simulating a single physical process, this application embodiment inputs the initial geometric topology information set and the initial material property information set into an adaptive physics field model that includes a flow stress field calculation module, a reaction progress field evolution module, and a curing deformation field prediction module. It simultaneously iterates and solves the multi-physics field interaction, thereby generating dynamic evolution trajectory data that can completely trace the material particles from the initial shape at the extrusion die to the fully vulcanized and formed shape. The above-mentioned holistic simulation strategy enables subsequent joint mechanism analysis based on the spatial distribution of material particles, velocity vector direction, and local reaction completion characteristics in the dynamic evolution trajectory data to accurately reveal the intrinsic physical relationship between the cross-sectional profile distortion mode of the extrusion expansion effect area, the spatial distribution of vulcanization shrinkage deformation, and the location information of the residual stress concentration area.

[0117] Finally, by integrating the above multi-dimensional analysis results for defect tracing analysis, the generated molding process simulation results not only include defect type identifiers and defect location coordinates, but also creatively include defect evolution paths. This fundamentally elevates the simulation results from static defect descriptions to dynamic, traceable defect formation histories. Therefore, this application's embodiments, starting from a global logic, can systematically and fundamentally reveal the complex causes and evolution patterns of molding defects in moldless seals, providing precise guidance for process optimization.

[0118] It is understood that those skilled in the art, based on the above content of the embodiments of this application, can clearly understand and implement the present invention. The embodiments of this application involve numerous technical terms such as "tubular model theory," "autocatalytic reaction model," and "generalized Maxwell model," all of which are standard theories in the fields of polymer physics and rubber processing. Their specific mathematical forms (such as constitutive equations and kinetic differential equations) are publicly recorded in academic literature and materials databases. Those skilled in the art can select applicable equation forms from existing technologies or materials handbooks and program them according to the parameter types given in the scheme (such as relaxation time spectrum, reaction activation energy, and Ogden model coefficients). This is a conventional model selection and parameter configuration task. Computational mechanics methods such as "voxelated mesh," "finite volume discretization," and "SIMPLE algorithm" are mature technologies in the fields of computational fluid dynamics and finite element analysis. Numerous open-source libraries and commercial software (such as OpenFOAM and Abaqus) are available for reference or secondary development, and their implementation paths are clear.

[0119] Secondly, the coupling relationship between the three physics modules in this embodiment is clear: the velocity field calculated by the flow field serves as the input to the convection term of the reaction field, the temperature field output by the reaction field serves as the thermal load of the solidification field, and the displacement field calculated by the solidification field is fed back to the flow field to update the geometry. This causal chain conforms to the physical reality of rubber extrusion molding. Although the three modules may use different types of meshes (Eulerian meshes and Lagrange meshes), this embodiment clearly establishes a "global spatial index mapping table" to achieve data exchange between different meshes. This inter-mesh interpolation mapping technique (such as node-based search algorithms and radial basis function interpolation) is a standard practice for handling multiphysics coupling problems and does not have any insurmountable logical defects. The unidirectional and feedback relationships of data transmission between modules are also clearly defined in this embodiment, ensuring the convergence and stability of the solution process.

[0120] Regarding dimensional consistency, the embodiments of this application have explicitly or implicitly ensured the consistency of physical quantity units. For example, coordinate values ​​are explicitly expressed in "millimeters," temperature in "Kelvin," and reaction completion is a dimensionless number. When constructing governing equations, such as the equations for conservation of mass, momentum, and energy, these are themselves mathematical expressions with homogeneous dimensions. When inputting specific numerical values ​​for calculation, as long as all input parameters (such as density, viscosity, specific heat capacity, coefficient of thermal expansion, etc.) use derived units consistent with the basic units (such as millimeters, kilograms, seconds, Kelvin), dimensional errors will not occur in the calculation process. For example, velocity is calculated in millimeters per second, and stress is calculated in Newtons per square millimeter (i.e., megapascals). These units are commonly used and compatible in engineering simulation software. Even if unit mismatches occur, they can be resolved by introducing unit conversion factors or forcing the use of a unified unit system during software implementation, which is a standard engineering practice in the field of simulation. Therefore, those skilled in the art have sufficient ability to fully and clearly implement the technical solutions of the embodiments of this application by following the guidance of the specification and combining common general knowledge and conventional technical means.

[0121] Based on the same inventive concept, this application also provides a moldless seal molding simulation system. (See also...) Figure 2 As shown, it is a schematic diagram of a possible moldless seal molding simulation system provided in an embodiment of this application. Figure 2 The moldless seal forming simulation system 200 includes a processor 210 and a memory 220. The processor 210 and the memory 220 are connected to each other via a communication bus. The memory 220 stores computer programs that can be executed by the processor 210. By executing the instructions stored in the memory 220, the processor 210 can perform the steps of the aforementioned AI-based moldless seal forming simulation method.

[0122] Based on the same inventive concept, embodiments of this application provide a computer-readable storage medium including a computer program. When the computer program runs on a moldless seal molding simulation system, it causes the moldless seal molding simulation system to perform the steps of the aforementioned AI-based moldless seal molding simulation method. In some possible implementations, various aspects of the AI-based moldless seal molding simulation method provided in this application can also be implemented as a program product, including a computer program. When the program product runs on a moldless seal molding simulation system, the computer program causes the moldless seal molding simulation system to perform the steps of the aforementioned AI-based moldless seal molding simulation method. For example, the moldless seal molding simulation system can perform actions such as... Figure 1The steps are shown in the diagram. The computer-readable storage medium includes volatile or non-volatile or a combination thereof, and may be removable or non-removable. Examples of computer-readable storage media include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random-access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), flash memory or other memory technologies, CD-ROM, Digital Video Disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium.

[0123] like Figure 3 The diagram shown is a functional block diagram of the moldless seal molding simulation system provided in this embodiment of the application. The moldless seal molding simulation system includes a moldless seal molding simulation device, which includes: The model information acquisition module is used to acquire the initial geometric topology information set and the initial material property information set of the target 3D model; The physical field model construction module is used to construct an adaptive physical field model associated with the spatial coordinate system of the target three-dimensional model. The adaptive physical field model includes a flow stress field calculation module, a reaction progress field evolution module, and a solidification deformation field prediction module configured based on the initial material property information. The molding process simulation module is used to input the initial geometric topology information set and the initial material property information set into the adaptive physical field model for spatiotemporal coupled molding process simulation processing, synchronously iteratively solve the multi-physics interaction between each module, and generate dynamic evolution trajectory data of the target three-dimensional model from the initial shape of the extrusion die to the fully vulcanized molding shape. The joint mechanism analysis module is used to perform joint mechanism analysis of material flow and deformation behavior based on the characteristics of the spatial position distribution, velocity vector direction and local reaction completion of material particles in the dynamic evolution trajectory data, and to obtain the cross-sectional profile distortion mode, vulcanization shrinkage deformation spatial distribution and residual stress concentration area location information of the extrusion swell effect affected area. The forming defect tracing module is used to integrate the cross-sectional profile distortion mode, the spatial distribution of shrinkage deformation, and the location information of the residual stress concentration area to perform forming defect tracing analysis and generate forming process simulation results that include defect type identifier, defect location coordinates, and defect evolution path.

[0124] Accordingly, this application also provides a computer program product, which includes a computer program or instructions that, when executed by a processor, cause the processor to implement the steps in the above method embodiments. It should be understood that each step or combination of steps in the above method flow can be implemented by the computer program or instructions. Furthermore, these computer programs or instructions can be applied to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device, enabling the processor of the general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to function as an apparatus for implementing the corresponding functions in the above method embodiments.

[0125] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0126] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0127] Finally, it should be noted that the above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A simulation method for moldless sealing component molding based on AI, characterized in that, include: Obtain the initial set of geometric topological information and the initial set of material property information of the target 3D model; An adaptive physical field model is constructed that is associated with the spatial coordinate system of the target three-dimensional model. The adaptive physical field model includes a flow stress field calculation module, a reaction progress field evolution module, and a solidification deformation field prediction module configured based on the initial material property information. The initial geometric topology information set and the initial material property information set are input into the adaptive physical field model for spatiotemporal coupled molding process simulation. The multi-physics interaction between each module is solved iteratively in parallel to generate dynamic evolution trajectory data of the target three-dimensional model from the initial shape of the extrusion die to the fully vulcanized molding shape. Based on the characteristics of the spatial distribution of material particles, velocity vector direction, and local reaction completion in the dynamic evolution trajectory data, a joint mechanism analysis of material flow and deformation behavior is performed to obtain the cross-sectional profile distortion mode, the spatial distribution of vulcanization shrinkage deformation, and the location information of residual stress concentration areas in the region affected by the extrusion swell effect. By integrating the cross-sectional profile distortion mode, the spatial distribution of shrinkage deformation, and the location information of the residual stress concentration area, a forming defect tracing analysis is performed to generate forming process simulation results that include defect type identification, defect location coordinates, and defect evolution path.

2. The method according to claim 1, characterized in that, The method integrates the cross-sectional profile distortion mode, the spatial distribution of shrinkage deformation, and the location information of residual stress concentration areas to perform forming defect tracing analysis, generating forming process simulation results that include defect type identifiers, defect location coordinates, and defect evolution paths, including: The spatial coordinate range of the extrusion bulging effect region marked in the cross-sectional profile distortion mode is intersected with the spatial coordinate range of the shrinkage deformation anomaly region marked in the shrinkage deformation spatial distribution. If there is a first overlapping region between the extrusion bulging effect region and the shrinkage deformation anomaly region, the defect type in the first overlapping region is identified as a dimensional accuracy composite defect, and the dimensional accuracy composite defect is recorded as being caused by the combined effect of extrusion bulging and vulcanization shrinkage. The spatial coordinate range of the extrusion bulging effect influence area marked in the cross-sectional profile distortion mode is intersected with the spatial coordinate range of the residual stress concentration area marked in the residual stress concentration area location information. If there is a second overlapping area between the extrusion bulging effect influence area and the residual stress concentration area, the defect type in the second overlapping area is identified as a stress cracking composite defect, and the stress cracking composite defect is recorded as being caused by the combined effect of extrusion bulging and residual stress. After identifying the defect types, the evolution starting point and evolution path are traced for each type of defect that has been identified.

3. The method according to claim 2, characterized in that, The process of tracing the evolutionary origins and constructing evolutionary paths for the identified defects includes: For each spatial location marked in the extrusion bulging effect influence area in the cross-sectional profile distortion mode, trace the flow path of the corresponding material particle in the dynamic evolution trajectory data, extract the earliest moment when the velocity direction deviates from the extrusion direction on the flow path as the starting time point of the extrusion bulging effect, and take the spatial location corresponding to the starting time point as the evolution starting coordinate of the extrusion bulging defect. For each spatial location marked in the abnormal shrinkage deformation area in the spatial distribution of shrinkage deformation, trace the sulfidation reaction process of the corresponding material particle in the dynamic evolution trajectory data, extract the moment when the reaction rate of the particle first exceeds the preset reaction rate threshold as the starting time point of sulfidation shrinkage, and take the spatial location corresponding to the starting time point as the evolution starting point coordinate of the shrinkage deformation defect. For each spatial location marked in the residual stress concentration area in the residual stress concentration area location information, trace the temperature change history of the corresponding material particle in the dynamic evolution trajectory data, extract the moment when the temperature of the particle first drops below the preset glass transition temperature as the starting time point of residual stress freezing, and take the spatial location corresponding to the starting time point as the evolution starting point coordinate of the residual stress defect. Based on the evolution starting coordinates of each defect type and the subsequent flow path and reaction path of the corresponding material particles in the dynamic evolution trajectory data, a continuous spatial curve is constructed from the evolution starting coordinates, extending along the extrusion direction and radial direction to the final defect occurrence location, as a description of the evolution path of the defect. The defect type identifier, defect location coordinates, and defect evolution path description are associated and integrated to generate defect description information for each defect. The defect description information of all defects is then sorted according to the order of defect location coordinates along the extrusion direction to generate structured molding process simulation results.

4. The method according to claim 1, characterized in that, The initial geometric topology information set and the initial material property information set are input into the adaptive physics model for spatiotemporal coupled molding process simulation. Simultaneously, the multi-physics interactions between modules are iteratively solved to generate dynamic evolution trajectory data of the target 3D model from its initial extrusion die shape to its fully vulcanized molded shape. This includes: At the start of the first simulation time step, the voxelized meshing parameters in the initial geometric topology information set are loaded into each computational grid cell of the flow stress field calculation module, reaction progress field evolution module, and solidification deformation field prediction module of the adaptive physical field model, so that each computational grid cell obtains the spatial coordinate position and initial volume occupancy corresponding to the initial extrusion shape of the target three-dimensional model. The flow stress field calculation module is invoked. Based on the boundary condition configuration of the current simulation time step and the displacement vector data in the solidification deformation field prediction calculation results passed from the previous simulation time step, the mass conservation equation and momentum conservation equation coupled with the constitutive equation are solved on each calculation grid cell to obtain the transient velocity vector field distribution data and transient stress tensor field distribution data of the current simulation time step. The velocity vectors of the computational grid cells located near the flow front in the transient velocity vector field distribution data are extracted and stored as new entries in the instantaneous flow velocity vector sequence of material particles in the current simulation time step. The occupancy status of each computational grid cell is updated according to the transient velocity vector field distribution data, and computational grid cells with a volume occupancy rate exceeding a preset filling threshold are marked as filled cells. Once the flow stress field solution and state update for the current time step are completed, iterative solutions and data transfer are performed through each module until the iterative loop for all simulation time steps is completed.

5. The method according to claim 4, characterized in that, The iterative solution and data transfer through each module, until the iterative loop of all simulation time steps is completed, includes: The reaction progress field evolution module is invoked. Based on the boundary condition configuration of the current simulation time step and the transient velocity vector field distribution data calculated by the flow stress field calculation module, the reaction completion value of the previous simulation time step is used as the initial value. The energy conservation equation coupled with the dynamic differential equation is solved on each filled cell to obtain the transient temperature field distribution data and transient reaction completion field distribution data of the current simulation time step. The reaction completion value of each filled cell in the transient reaction completion field distribution data is extracted and stored as a new entry in the local reaction completion percentage sequence of material particles in the current simulation time step. The computational grid cells whose reaction completion values ​​exceed the preset gelation threshold are identified based on the transient reaction completion field distribution data and marked as vulcanized cells. The solidification deformation field prediction module is invoked, and the transient temperature field distribution data calculated by the reaction progress field evolution module based on the boundary condition configuration of the current simulation time step and the transient temperature field distribution data is obtained. The heat conduction equation and momentum conservation equation coupled with the constitutive model are solved on each vulcanized unit to obtain the transient displacement vector field distribution data and transient stress tensor field distribution data of the current simulation time step. The displacement vector of each vulcanized unit in the transient displacement vector field distribution data is superimposed on the original spatial coordinate position of the unit to generate the updated spatial coordinate position as a new entry in the material particle spatial position coordinate sequence at the current simulation time step and stored. The principal stress difference of each vulcanized unit is calculated based on the transient stress tensor field distribution data as a residual stress indicator value. The transient displacement vector field distribution data calculated by the solidification deformation field prediction module is fed back to the flow stress field calculation module through the data transmission interface, which is used to correct the spatial coordinate position and volume occupancy of the calculation grid cell in the next simulation time step. Repeat the solution and data transfer process of each module until all computational grid cells are marked as vulcanized cells and the temperature values ​​of all vulcanized cells drop below the preset room temperature stability threshold, and end the iterative loop of the simulation time step. The dynamic evolution trajectory data is generated by arranging and combining the sequence of spatial position coordinates of material particles, the sequence of instantaneous flow velocity vectors of material particles, and the sequence of local reaction completion percentages of material particles stored at each simulation time step in chronological order.

6. The method according to claim 1, characterized in that, The method involves a joint mechanism analysis of material flow and deformation behavior based on the spatial distribution, velocity vector direction, and local response completion characteristics of material particles in the dynamic evolution trajectory data. This analysis yields information on the cross-sectional profile distortion mode of the extrusion swell effect influence region, the spatial distribution of vulcanization shrinkage deformation, and the location of residual stress concentration areas. Extract a subset of material particle spatial position coordinate sequences within a preset spatial range near the extrusion die outlet from the dynamic evolution trajectory data. Perform spatial cluster analysis on the subset of material particle spatial position coordinate sequences to identify clustered regions where the distribution density of coordinate points in the plane perpendicular to the extrusion direction is significantly higher than the average distribution density. Compare the boundary contour of the clustered region with the standard cross-sectional contour of the corresponding position of the target three-dimensional model and calculate the maximum distance of the outward bulge of the contour boundary as the extrusion bulging effect amplitude value. Based on the continuous change trend of the extrusion swell effect amplitude value in the extrusion direction, the spatial interval where the extrusion swell effect amplitude value continuously increases and exceeds the preset distortion threshold is divided into the extrusion swell effect influence area. Within the extrusion swell effect influence area, the cross-sectional contour distortion mode is classified into the overall uniform expansion mode, the local asymmetric protrusion mode, and the edge wavy undulation mode according to the shape characteristics of the contour boundary protrusion. Based on the cross-sectional profile distortion pattern obtained from the extrusion swell effect analysis, the vulcanization shrinkage deformation amount is calculated and its spatial distribution is identified, and residual stress is calculated and concentrated areas are located.

7. The method according to claim 6, characterized in that, The calculation and spatial distribution identification of vulcanization shrinkage deformation based on the cross-sectional profile distortion pattern obtained from the extrusion swell effect analysis, as well as the calculation of residual stress and location of concentrated areas, include: Extract the percentage of reaction completion of each vulcanized unit at different simulation time steps from the dynamic evolution trajectory data, and perform derivative calculations on the percentage of reaction completion of each vulcanized unit along the time dimension to obtain the reaction rate change curve of each vulcanized unit over time. Then, determine the vulcanization reaction acceleration start time and vulcanization reaction completion time of the unit based on the peak occurrence time of the reaction rate change curve over time. For each vulcanized unit, the Euclidean distance between its spatial coordinate position at the start of the vulcanization reaction acceleration and its spatial coordinate position at the end of the vulcanization reaction is calculated. This distance is used as the shrinkage deformation of the unit due to the formation of the vulcanization cross-linking network. The shrinkage deformation of all vulcanized units is arranged according to their spatial coordinate positions to generate a spatial distribution map of shrinkage deformation. In the spatial distribution map of shrinkage deformation, spatial regions where the shrinkage deformation value exceeds the preset shrinkage threshold and is continuously distributed are identified and marked as abnormal shrinkage deformation regions. Based on the geometry and spatial extension direction of the abnormal shrinkage deformation regions, it is determined whether the dominant mechanism of shrinkage deformation is isotropic volumetric shrinkage or anisotropic directional shrinkage. The residual stress indication value of each vulcanized unit at different simulation time steps is extracted from the dynamic evolution trajectory data. The residual stress indication value of each vulcanized unit is accumulated and summed along the time dimension to obtain the final residual stress value of the unit after complete cooling. The final residual stress values ​​of all vulcanized units are arranged according to their spatial coordinate positions to generate a residual stress spatial distribution map. In the residual stress spatial distribution map, a peak detection algorithm is used to identify local maxima of residual stress values, and each local maxima and the area within a preset neighborhood where the residual stress values ​​are all higher than a preset stress threshold are located as residual stress concentration areas, and the center coordinates and coverage area of ​​each residual stress concentration area are marked. By superimposing the spatial distribution of the shrinkage deformation anomaly region with the spatial distribution of the residual stress concentration region, the overlapping region that simultaneously belongs to the shrinkage deformation anomaly region and the residual stress concentration region is identified, and the overlapping region is marked as the potential occurrence area of ​​risk defects in material deformation behavior.

8. The method according to claim 1, characterized in that, The adaptive physical field model, which is constructed in relation to the spatial coordinate system of the target 3D model, includes a flow stress field calculation module, a reaction progress field evolution module, and a solidification deformation field prediction module configured based on the initial material property information. The viscoelastic constitutive parameters of the rubber substrate in the unvulcanized state in the initial material property information are analyzed. Based on the viscoelastic constitutive parameters, a constitutive equation describing the dynamic balance between viscous dissipation and elastic energy storage of the rubber substrate during shear flow is constructed. The constitutive equation is coupled with the mass conservation equation and the momentum conservation equation to obtain the basic control equation set of the flow stress field calculation module. The basic control equations of the flow stress field calculation module are spatially discretized to generate flow stress field calculation grid cells corresponding to the solid filling structure inside the target three-dimensional model. Storage space is allocated to each flow stress field calculation grid cell to store transient pressure values, transient velocity vectors, and transient stress tensor components. The crosslinking reaction kinetic parameters of the rubber substrate during the vulcanization process are analyzed from the initial material property information. Based on the crosslinking reaction kinetic parameters, a kinetic differential equation is constructed to describe the dependence of the crosslinking network formation rate of the rubber substrate on temperature and time during the vulcanization process. The kinetic differential equation is coupled with the energy conservation equation to obtain the basic control equation set of the reaction progress field evolution module. The basic control equations of the reaction progress field evolution module are spatially discretized to generate reaction progress field evolution calculation grid cells corresponding to the solid filling structure inside the target three-dimensional model. Storage space is allocated to each reaction progress field evolution calculation grid cell for storing transient temperature values, transient reaction completion degree and transient heat release rate. The hyperelastic constitutive parameters of the rubber substrate in the fully vulcanized state are analyzed from the initial material property information. Based on the hyperelastic constitutive parameters, a constitutive model describing the nonlinear stress-strain response of the fully vulcanized rubber substrate during cooling and shrinkage is constructed. The constitutive model is coupled with the momentum conservation equation and the heat conduction equation to determine the basic control equation set of the curing deformation field prediction module. The basic control equations of the solidification deformation field prediction module are spatially discretized to generate solidification deformation field prediction calculation grid units corresponding to the solid filling structure inside the target three-dimensional model. Storage space is allocated for each solidification deformation field prediction calculation grid unit to store transient displacement vector, transient strain tensor and transient stress tensor. After constructing the basic governing equations for each physics module and performing spatial discretization, the topological correspondence between the computational grid cells of each module is established, and the data transfer interface is configured and the multiphysics coupling solver is encapsulated.

9. A moldless sealing component molding simulation system, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program that, when executed by the processor, causes the processor to perform the steps of the AI-based moldless seal molding simulation method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a computer program that, when run on a moldless seal molding simulation system, causes the moldless seal molding simulation system to perform the steps of any of the AI-based moldless seal molding simulation methods described in claims 1 to 8.

Images