Method and system for calculating mold structure strength based on digital design

Abstract

The present invention provides a mold structure strength calculation method and system based on data-based design, which relates to the technical field of mold design, including: obtaining a mold assembly and converting it into a three-dimensional digital model, performing adaptive segmentation, generating a semantic feature tree, setting a parameterized mapping table, generating a sub-region finite element model and constructing a mold finite element model; constructing a material microstructure evolution model, extracting equivalent mechanical performance parameters, establishing a stress-strain analysis model, solving the stress-strain analysis model to obtain a full-field dynamic response and constructing a sub-region stress characteristic parameter database; constructing a mold structure strength evaluation model, generating a mold structure strength knowledge map, performing multi-index fusion calculation, obtaining a mold strength score and judging whether the mold strength meets the use requirements, and if not, performing a global sensitivity analysis, determining a key sub-region, generating a topology optimization scheme and a size optimization scheme, performing associated modification, repeating iterations, and outputting the three-dimensional digital model.

Description

Technical Field

[0001] The present invention relates to the technical field of mold design, and in particular to a mold structure strength calculation method and system based on digital design. Background Art

[0002] Injection die-casting molds are key process equipment in the production of plastic aluminum alloy products. Their structural strength directly affects the quality and production efficiency of injection die-casting products. At present, the design of mold structural strength mainly relies on the experience of designers and finite element simulation analysis. Traditional empirical design methods are difficult to cope with the increasingly complex product structure and process requirements. Finite element simulation analysis has problems such as high modeling difficulty, high calculation cost, and long cycle, which makes it difficult to meet the rapid iteration and optimization needs of mold design.

[0003] Although the existing technology has solved the problem of calculating the strength of mold structures to a certain extent, it still lacks the fusion analysis of multi-source heterogeneous data such as materials, structures, and processes, making it difficult to comprehensively and accurately evaluate the strength of mold structures, as well as the lack of correlation analysis between mold failure mechanisms and key influencing factors, making it difficult to accurately diagnose the causes of failure and provide targeted optimization measures.

[0004] Therefore, a solution is urgently needed to solve the problems existing in the prior art. Summary of the invention

[0005] The embodiments of the present invention provide a method and system for calculating the strength of a mold structure based on digital design, which can at least solve some of the problems existing in the prior art.

[0006] A first aspect of an embodiment of the present invention provides a method for calculating mold structure strength based on digital design, comprising:

[0007] The mold assembly to be analyzed is obtained and converted into a three-dimensional digital model, the three-dimensional digital model is adaptively segmented by intelligent feature recognition technology based on deep learning, the high-level features of the mold structure are extracted and a semantic feature tree is generated, the model sub-regions are dynamically divided based on the semantic feature tree and a parameterized mapping table is set, the model sub-regions are discretized by an improved octree adaptive grid, the grid unit size and type are adjusted in combination with the model characteristic curvature and stress gradient, a sub-region finite element model is generated, and a mold finite element model is constructed in combination with a mold assembly tolerance and an assembly gap nonlinear contact algorithm;

[0008] Based on the nonlinear mechanical behavior of the mold material, the material microstructure evolution model is constructed in combination with the crystal plasticity finite element method. The mold material behavior is simulated by characterizing the grain orientation, dislocation density and slip system evolution, and the equivalent mechanical performance parameters of the mold material are extracted to construct a cross-scale mechanical performance parameter database. A multi-condition load spectrum is generated according to the mold working conditions, and the multi-physical field in the coupled injection and die-casting molding process is determined. The multi-condition load spectrum and the boundary conditions of the multi-physical field are applied to the mold finite element model. The deformation of the cavity wall is tracked in combination with the improved adaptive dynamic mesh technology and a stress-strain analysis model is established. The stress-strain analysis model is solved in combination with the multi-grid method to obtain the full-field dynamic response and to construct a sub-region stress characteristic parameter database for failure modes in combination with the stress-strain evolution history and key stress distribution characteristic parameters of the sub-region of the model.

[0009] Based on the analysis of the mold failure mechanism, a mold structure strength assessment model is constructed, the failure mode is associated with the influencing factors and a mold structure strength knowledge map is generated, the parameterized mapping table and the sub-region stress characteristic parameter database are added to the strength assessment model for multi-index fusion calculation, the mold strength score is obtained, and whether the mold strength meets the use requirements is judged according to a preset strength score threshold. If not, a global sensitivity analysis is performed based on the parameterized mapping table to determine the key sub-regions and the corresponding weak links, and a topology optimization scheme and a size optimization scheme are generated in combination with a pre-built optimization design knowledge base and intelligent reasoning technology. The topology optimization scheme and the size optimization scheme are mapped to the corresponding sub-regions in the three-dimensional digital model, and the associated modification is performed through the parameterized mapping table. The iteration is repeated until the mold structure strength meets the use requirements, and the three-dimensional digital model is output.

[0010] In an optional embodiment,

[0011] The mold assembly to be analyzed is obtained and converted into a three-dimensional digital model, the three-dimensional digital model is adaptively segmented by intelligent feature recognition technology based on deep learning, the high-level features of the mold structure are extracted and a semantic feature tree is generated, the model sub-regions are dynamically divided based on the semantic feature tree and a parameterized mapping table is set, the model sub-regions are discretized by an improved octree adaptive grid, the grid unit size and type are adjusted in combination with the model characteristic curvature and stress gradient, a sub-region finite element model is generated, and a mold finite element model is constructed in combination with the mold assembly tolerance and the assembly gap nonlinear contact algorithm, including:

[0012] Acquire the two-dimensional drawings, three-dimensional CAD models and assembly data of the mold assembly to be analyzed, combine them as heterogeneous design data, perform digital processing and format conversion on the heterogeneous design data, extract the geometric information, topological information and attributes of the mold parts, assemble the mold parts through parametric constraints and assembly interface definitions according to the structural hierarchy and assembly relationship of the mold assembly through a bottom-up assembly modeling strategy, and obtain the three-dimensional digital model;

[0013] For the three-dimensional digital model, a mold feature data set is constructed and manually annotated, a three-dimensional convolutional neural network is constructed by using an intelligent feature recognition technology based on deep learning, the mold feature data set with annotations is converted into a voxel representation and added to the three-dimensional convolutional neural network for training, and a feature category probability distribution corresponding to each voxel is obtained, the training is repeated and the hyperparameters of the three-dimensional convolutional neural network are adjusted to obtain a trained feature recognition network, and feature extraction is performed on the three-dimensional digital model through the feature recognition network to obtain the high-level features, and based on the high-level features, a hierarchical semantic feature tree is constructed with the mold part as a root node and the high-level features as child nodes by determining the topological relationship and semantic attributes between different high-level features;

[0014] Based on the hierarchical semantic feature tree, the three-dimensional digital model is divided into a plurality of sub-regions according to geometric attributes and topological relationships, and a parameterized mapping table between the sub-regions and the three-dimensional digital model is established;

[0015] Based on the improved octree adaptive meshing algorithm, the characteristic curvature and stress gradient of the sub-region corresponding to the current node are calculated at each leaf node of the octree. If the characteristic curvature exceeds the preset curvature threshold, the current node is subdivided according to the direction and size of the characteristic curvature in combination with the anisotropic subdivision strategy until the error requirement is met. If the stress gradient is greater than the preset gradient threshold, the current node is subdivided in combination with the tetrahedral unit, and the size and type of the mesh unit are adjusted.

[0016] Based on the mesh unit, an initial finite element model is generated in combination with material properties and boundary conditions. The mating surfaces and gap areas between parts are identified according to the mold assembly drawing and tolerance standards. Face-to-face contact boundary conditions are applied on the mating surfaces. The contact boundary conditions are added to the initial finite element model through a penalty function. The contact problem is solved in combination with the augmented Lagrange multiplier method to obtain the contact pressure and gap size distribution on the mating surfaces. The contact pressure is applied as a load condition to the initial finite element model for bidirectional coupling to obtain the mold finite element model.

[0017] In an optional embodiment,

[0018] The contact boundary condition is added to the initial finite element model through a penalty function as shown in the following formula:

[0019] ;

[0020] in, P c represents the penalty function value, G c represents the contact boundary, e N represents the normal penalty factor, g N represents the normal gap, l N represents the normal Lagrange multiplier, e T represents the tangential penalty factor, g T represents the tangential relative slip, l T represents the tangential Lagrange multiplier, dS Represents the area differential used to describe the range of integration.

[0021] In an optional embodiment,

[0022] Based on the nonlinear mechanical behavior of mold materials, a material microstructure evolution model is constructed in combination with the crystal plasticity finite element method. The mold material behavior is simulated by characterizing the grain orientation, dislocation density and slip system evolution, and the equivalent mechanical performance parameters of the mold material are extracted. A cross-scale mechanical performance parameter database is constructed. A multi-condition load spectrum is generated according to the mold working conditions, and the multi-physical field in the coupled injection and die-casting process is determined. The multi-condition load spectrum and the boundary conditions of the multi-physical field are applied to the mold finite element model. The deformation of the cavity wall is tracked in combination with the improved adaptive dynamic mesh technology and a stress-strain analysis model is established. The stress-strain analysis model is solved in combination with the multi-grid method to obtain the full-field dynamic response and the stress-strain evolution history and key stress distribution characteristic parameters of the sub-region of the model are combined to construct a sub-region stress characteristic parameter database for failure modes, including:

[0023] Based on the nonlinear mechanical behavior of the mold material and in combination with the crystal plasticity finite element method, the velocity gradient tensor in the crystal with multiple slip systems is decomposed into an elastic part and a plastic part. For the elastic part, the slip rate is calculated by the flow law of the thermal activation effect and the dislocation density is determined in combination with the critical shear stress. The dislocation density evolution process is determined in combination with the kinetic equation and dislocation multiplication and annihilation. Based on the dislocation density evolution process, the material microstructure evolution model is constructed.

[0024] Based on the material microstructure evolution model, volume units are randomly selected and periodic boundary conditions are applied to simulate the microstructure evolution process of the material under different conditions, and the yield strength, hardening modulus and plastic strain of the polycrystalline polymerization model are extracted as equivalent mechanical performance parameters by numerical homogenization method to construct the cross-scale mechanical performance parameter database;

[0025] Based on the cross-scale mechanical performance parameter database, combined with the load data of the mold under actual working conditions collected by the sensor, the load characteristic parameters under different working conditions are extracted through statistical analysis, and a probability distribution model of the load-time series is established based on the random process theory. A random load sequence is sampled from the probability distribution model through the Monte Carlo method, and a multi-condition load spectrum is synthesized by combining the frequency and duration of different working conditions.

[0026] Establish a three-dimensional injection die-casting model corresponding to the injection die-casting molding process, wherein the three-dimensional injection die-casting model includes a mold cavity, a pouring system and a cooling system, establish a fluid control equation based on the constitutive model of the plastic aluminum alloy material and the three-dimensional injection die-casting model, establish an energy conservation equation in the solidification cooling stage in combination with heat transfer theory, discretize the fluid control equation and the energy conservation equation through a finite volume method, combine multi-physical field coupling conditions, solve the flow field, temperature field and stress field through a single coupling solution strategy, and extract the boundary conditions of each physical field respectively;

[0027] The multi-condition load spectrum and the boundary conditions of each physical field are applied to the mold finite element model, the deformation of the cavity wall is tracked and the grid coordinates are updated through the improved adaptive dynamic mesh technology, based on the updated grid coordinates, the material microstructure evolution model is embedded into the mold finite element model in combination with the crystal plasticity finite element method, a stress-strain analysis model is established and a nonlinear stress-strain equation is constructed, the nonlinear stress-strain equation is solved by an incremental iterative method and accelerated by a multi-grid method, the full-field dynamic response of the mold under the load is obtained, the stress-strain evolution history of the mold in different sub-regions and the key stress distribution characteristic parameters including the maximum principal stress and stress triaxiality are extracted, and a sub-region stress characteristic parameter database oriented to failure modes is constructed.

[0028] In an optional embodiment,

[0029] The slip rate is calculated by the flow law of thermal activation effect as shown in the following formula:

[0030] ;

[0031] in, is the slip rate, represents the reference slip rate, represents the equivalent strain rate, represents the reference strain rate, k B represents the Boltzmann constant, T represents absolute temperature, Q represents the activation energy, t α Indicates α The shear stress on the slip system is represents the critical shear stress, represents the reference stress, m represents the strain rate sensitive parameter, p represents the stress index, q represents the shape factor, sgn Used to determine the sign of the shear stress.

[0032] In an optional embodiment,

[0033] Based on the analysis of the mold failure mechanism, a mold structure strength assessment model is constructed, the failure mode is associated with the influencing factors and a mold structure strength knowledge map is generated, the parameterized mapping table and the sub-region stress characteristic parameter database are added to the strength assessment model for multi-index fusion calculation, the mold strength score is obtained, and the mold strength is judged whether it meets the use requirements according to the preset strength score threshold. If not, a global sensitivity analysis is performed based on the parameterized mapping table to determine the key sub-regions and the corresponding weak links, and a topology optimization scheme and a size optimization scheme are generated in combination with the pre-built optimization design knowledge base and intelligent reasoning technology. The topology optimization scheme and the size optimization scheme are mapped to the corresponding sub-regions in the three-dimensional digital model, and the association modification is performed through the parameterized mapping table. It is repeated until the mold structure strength meets the use requirements, and the three-dimensional digital model output includes:

[0034] Obtain and analyze mold failure cases, determine the failure mode and failure mechanism corresponding to the mold, build a mold structure strength assessment model, determine the key factors affecting the mold structure strength based on the failure mechanism, wherein the key factors include the mechanical properties of the material, the structural design of the mold, and the injection molding and die-casting process parameters, associate the failure mode with the key factors, take the failure mode, material properties, structural parameters, and process parameters as core knowledge elements, define the categories, attributes, and associations of the core knowledge elements through ontology modeling language, build an ontology model, extract knowledge element instances from the knowledge in this field and perform semantic annotations according to the ontology model, determine the semantic associations and build link relationships through natural language processing technology, and obtain the mold structure strength knowledge graph;

[0035] The parameterized mapping table and the sub-region stress characteristic parameter database are added to the strength evaluation model, and multi-index calculation is performed through weighted fusion to obtain a mold strength score corresponding to the overall structure of the mold and compare it with a preset strength score threshold. If the mold strength score is greater than the preset strength score threshold, it is considered that the current mold meets the use requirements, otherwise, it is considered that the current mold does not meet the use requirements;

[0036] If it does not meet the use requirements, then based on the parameterized mapping table, the influence of each sub-region and failure index on the mold strength score is calculated by a global sensitivity analysis method. According to the sensitivity analysis result, the sub-region with the greatest influence on the mold strength score is taken as the key sub-region, and the failure index with the greatest influence on the mold strength score is taken as the weak link. The weak link is matched with the failure mode in the mold structure strength knowledge graph, and the failure cause is determined in combination with intelligent reasoning technology. Based on the failure cause, a topology optimization plan and a size optimization plan are generated;

[0037] For the key sub-area, based on the topology optimization scheme, the variable density method is used to optimize and solve it to obtain the density distribution result. For the weak link, based on the size optimization scheme, the local structure size is optimized through parametric modeling to obtain the local structure size change. The density distribution result and the local structure size change are applied to the corresponding area of ​​the three-dimensional digital model and are associated and modified through a parametric mapping table. The optimization is repeated until the preset maximum number of iterations is reached, and the three-dimensional digital model corresponding to the last iteration and the mold strength score corresponding to the mold are output.

[0038] In an optional embodiment,

[0039] The variable density method is used to optimize and solve the density distribution, and the density distribution result is shown in the following formula:

[0040] ;

[0041] in, r Represents the relative density of the unit, tanh represents the hyperbolic tangent function, β represents the magnification factor, or e represents a continuous slack variable, H represents the penalty factor, Represents the solid element stiffness matrix.

[0042] A second aspect of an embodiment of the present invention provides a mold structure strength calculation system based on digital design, comprising:

[0043] The first unit is used to obtain the mold assembly to be analyzed and convert it into a three-dimensional digital model, adaptively segment the three-dimensional digital model through intelligent feature recognition technology based on deep learning, extract high-level features of the mold structure and generate a semantic feature tree, dynamically divide the model sub-regions based on the semantic feature tree and set a parameterized mapping table, discretize the model sub-regions through an improved octree adaptive grid, adjust the grid unit size and type in combination with the model characteristic curvature and stress gradient, generate a sub-region finite element model, and construct a mold finite element model in combination with the mold assembly tolerance and assembly gap nonlinear contact algorithm;

[0044] The second unit is used to construct a material microstructure evolution model based on the nonlinear mechanical behavior of the mold material in combination with the crystal plasticity finite element method, simulate the mold material behavior by characterizing the grain orientation, dislocation density and slip system evolution, extract the equivalent mechanical performance parameters of the mold material, and construct a cross-scale mechanical performance parameter database. A multi-condition load spectrum is generated according to the mold working conditions, and the multi-physical fields in the coupled injection and die-casting process are determined. The multi-condition load spectrum and the boundary conditions of the multi-physical fields are applied to the mold finite element model, and the deformation of the cavity wall is tracked and a stress-strain analysis model is established in combination with the improved adaptive dynamic mesh technology. The stress-strain analysis model is solved in combination with the multi-grid method to obtain the full-field dynamic response and construct a sub-region stress characteristic parameter database for failure modes in combination with the stress-strain evolution history and key stress distribution characteristic parameters of the sub-region of the model;

[0045] The third unit is used to build a mold structure strength evaluation model based on mold failure mechanism analysis, associate failure modes with influencing factors and generate a mold structure strength knowledge map, add the parameterized mapping table and the sub-region stress characteristic parameter database to the strength evaluation model for multi-index fusion calculation, obtain the mold strength score, and judge whether the mold strength meets the use requirements according to a preset strength score threshold. If not, a global sensitivity analysis is performed based on the parameterized mapping table to determine the key sub-regions and corresponding weak links, and a topology optimization scheme and a size optimization scheme are generated in combination with a pre-built optimization design knowledge base and intelligent reasoning technology. The topology optimization scheme and the size optimization scheme are mapped to the corresponding sub-regions in the three-dimensional digital model, and the associated modifications are performed through the parameterized mapping table. The iteration is repeated until the mold structure strength meets the use requirements, and the three-dimensional digital model is output.

[0046] According to a third aspect of the embodiments of the present invention,

[0047] An electronic device is provided, comprising:

[0048] processor;

[0049] a memory for storing processor-executable instructions;

[0050] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0051] According to a fourth aspect of the embodiments of the present invention,

[0052] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0053] In the present invention, the interpretability and operability of the model are enhanced through the semantic feature tree, which is convenient for subsequent analysis and optimization. The discretization of the model sub-regions by the improved octree adaptive grid helps to improve the flexibility and accuracy of the grid division, ensure the calculation efficiency and accuracy of the finite element model, and simulate the behavior of the mold material by characterizing the grain orientation, dislocation density and slip system evolution, so as to ensure the accuracy of the material model and improve the analysis accuracy of the mold under different working conditions. The stress-strain evolution history and key stress distribution characteristic parameters of the model sub-region are extracted, and a sub-region stress characteristic parameter database for failure mode is constructed to provide comprehensive stress-strain data, providing a reliable basis for failure mode analysis and subsequent optimization. The parameterized mapping table and the sub-region stress characteristic parameter database are added to the strength assessment model, and multi-index fusion calculation is performed to provide a systematic and structured strength assessment method, thereby improving the comprehensiveness and accuracy of the strength assessment. In summary, the present invention can realize accurate sensitivity analysis, improve optimization efficiency, obtain the mold structure strength that meets the task requirements, and output a three-dimensional digital model that meets the use requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a flow chart of a method for calculating mold structure strength based on digital design according to an embodiment of the present invention;

[0055] Figure 2 It is a schematic diagram of the structure of a mold structure strength calculation system based on digital design according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0058] Figure 1 FIG. 1 is a flow chart of a method for calculating mold structure strength based on digital design according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0059] S1. Obtain the mold assembly to be analyzed and convert it into a three-dimensional digital model, adaptively segment the three-dimensional digital model through intelligent feature recognition technology based on deep learning, extract high-level features of the mold structure and generate a semantic feature tree, dynamically divide the model sub-regions based on the semantic feature tree and set a parameterized mapping table, discretize the model sub-regions through an improved octree adaptive grid, adjust the grid unit size and type in combination with the model characteristic curvature and stress gradient, generate a sub-region finite element model, and construct a mold finite element model in combination with the mold assembly tolerance and assembly gap nonlinear contact algorithm;

[0060] The intelligent feature recognition technology uses machine learning and artificial intelligence algorithms to automatically identify and extract key features in the model. The adaptive segmentation is a method of dynamically adjusting mesh division to optimize the mesh according to the geometric complexity and physical properties of the model. The semantic feature tree is a structure that organizes and represents model features, and can manage and describe various features in the model in a hierarchical manner. The model sub-region refers to dividing the entire model into multiple smaller regions, each of which can be analyzed and optimized independently. The parameterized mapping table is used to describe the relationship between model parameters and their actual physical properties. The octree adaptive grid is a grid division method based on the octree structure. Through recursive subdivision, each node can be divided into eight sub-nodes to achieve adaptive grid division of complex geometric bodies. The model feature curvature describes the model surface in The degree of bending at different points is an important indicator of the geometric characteristics of the model. The stress gradient represents the rate of change of stress distribution in the material and is an important parameter for evaluating stress concentration and fatigue life. The sub-region finite element model refers to finite element analysis in the divided model sub-region. By establishing a finite element model in the sub-region, the calculation efficiency can be improved, the analysis results can be refined, and more detailed physical phenomena can be captured in local optimization. The mold assembly tolerance refers to the allowable dimensional error range of each part of the mold during the assembly process. The assembly gap nonlinear contact algorithm is used to simulate the contact and gap changes between parts during the assembly process. It takes into account nonlinear contact mechanical properties such as friction and contact stiffness, and can accurately predict the contact behavior and stress distribution during the assembly process. The mold finite element model is a mathematical model for simulating and analyzing the mold structure through the finite element method.

[0061] In an optional embodiment,

[0062] The mold assembly to be analyzed is obtained and converted into a three-dimensional digital model, the three-dimensional digital model is adaptively segmented by intelligent feature recognition technology based on deep learning, the high-level features of the mold structure are extracted and a semantic feature tree is generated, the model sub-regions are dynamically divided based on the semantic feature tree and a parameterized mapping table is set, the model sub-regions are discretized by an improved octree adaptive grid, the grid unit size and type are adjusted in combination with the model characteristic curvature and stress gradient, a sub-region finite element model is generated, and a mold finite element model is constructed in combination with the mold assembly tolerance and the assembly gap nonlinear contact algorithm, including:

[0063] Acquire the two-dimensional drawings, three-dimensional CAD models and assembly data of the mold assembly to be analyzed, combine them as heterogeneous design data, perform digital processing and format conversion on the heterogeneous design data, extract the geometric information, topological information and attributes of the mold parts, assemble the mold parts through parametric constraints and assembly interface definitions according to the structural hierarchy and assembly relationship of the mold assembly through a bottom-up assembly modeling strategy, and obtain the three-dimensional digital model;

[0064] For the three-dimensional digital model, a mold feature data set is constructed and manually annotated, a three-dimensional convolutional neural network is constructed by using an intelligent feature recognition technology based on deep learning, the mold feature data set with annotations is converted into a voxel representation and added to the three-dimensional convolutional neural network for training, and a feature category probability distribution corresponding to each voxel is obtained, the training is repeated and the hyperparameters of the three-dimensional convolutional neural network are adjusted to obtain a trained feature recognition network, and feature extraction is performed on the three-dimensional digital model through the feature recognition network to obtain the high-level features, and based on the high-level features, a hierarchical semantic feature tree is constructed with the mold part as a root node and the high-level features as child nodes by determining the topological relationship and semantic attributes between different high-level features;

[0065] Based on the hierarchical semantic feature tree, the three-dimensional digital model is divided into a plurality of sub-regions according to geometric attributes and topological relationships, and a parameterized mapping table between the sub-regions and the three-dimensional digital model is established;

[0066] Based on the improved octree adaptive meshing algorithm, the characteristic curvature and stress gradient of the sub-region corresponding to the current node are calculated at each leaf node of the octree. If the characteristic curvature exceeds the preset curvature threshold, the current node is subdivided according to the direction and size of the characteristic curvature in combination with the anisotropic subdivision strategy until the error requirement is met. If the stress gradient is greater than the preset gradient threshold, the current node is subdivided in combination with the tetrahedral unit, and the size and type of the mesh unit are adjusted.

[0067] Based on the mesh unit, an initial finite element model is generated in combination with material properties and boundary conditions. The mating surfaces and gap areas between parts are identified according to the mold assembly drawing and tolerance standards. Face-to-face contact boundary conditions are applied on the mating surfaces. The contact boundary conditions are added to the initial finite element model through a penalty function. The contact problem is solved in combination with the augmented Lagrange multiplier method to obtain the contact pressure and gap size distribution on the mating surfaces. The contact pressure is applied as a load condition to the initial finite element model for bidirectional coupling to obtain the mold finite element model.

[0068] The mold feature data set is a collection of various feature information in the mold design and manufacturing process. The feature recognition network is a model based on deep learning, which is used to automatically identify and extract key features in the mold. The topological relationship describes the connection and relationship between the features in the model. The root node is the starting node in the tree structure (such as feature tree, octree), representing the highest level of the entire model or data set. The leaf node is the end node in the tree structure, representing the most basic feature or data unit. Each leaf node corresponds to a sub-area. The anisotropic subdivision strategy is a grid division method that divides the mesh according to the geometry in different directions. The mesh is subdivided non-uniformly according to the characteristics and physical requirements of the mold or part. The mating surface refers to the surface of the mold or part that needs to be in close contact and fit during the assembly process. The contact boundary condition is the boundary condition used to describe the contact behavior between different parts in finite element analysis. The penalty function is an optimization method that limits or guides the value range of the variable by introducing a penalty term in the objective function. The augmented Lagrange multiplier method is an algorithm for solving constrained optimization problems. By introducing Lagrange multipliers and penalty terms, the constraint conditions are converted into part of the optimization target. The bidirectional coupling refers to the relationship of mutual influence and feedback between two or more physical fields.

[0069] Obtain the 2D drawings, 3D CAD models and assembly data of the mold assembly to be analyzed, combine them as heterogeneous design data, perform digital processing and format conversion on the heterogeneous design data, convert them uniformly into a common data format, extract the geometric information (such as points, lines, surfaces, etc.), topological information (such as connectivity, adjacency, etc.) and attributes (such as materials, dimensions, tolerances, etc.) of the mold parts, adopt a bottom-up assembly modeling strategy based on the structural hierarchy and assembly relationship of the mold assembly, assemble the mold parts through parametric constraints (such as parallelism, verticality, coaxiality, etc.) and assembly interface definitions (such as mating surfaces, bolt connections, etc.), until all parts are assembled and a complete 3D digital model of the mold is obtained;

[0070] Construct a mold feature dataset, including a CAD model of typical features (such as bosses, grooves, threads, etc.), and manually annotate it. Construct a three-dimensional convolutional neural network through intelligent feature recognition technology based on deep learning. Convert the annotated mold feature dataset into a voxel representation and add it to the three-dimensional convolutional neural network for training. Obtain the feature category probability distribution corresponding to each voxel through forward propagation and back propagation. Repeat the training and adjust the hyperparameters of the three-dimensional convolutional neural network (such as convolution kernel size, pooling layer parameters, etc.) to obtain a trained feature recognition network. Use the trained feature recognition network to extract features from the three-dimensional digital model obtained in step 2 to obtain advanced features, such as convex hull, symmetry, etc.;

[0071] Based on the extracted high-level features, by determining the topological relationship (such as adjacent, contained, etc.) and semantic attributes (such as function, manufacturing process, etc.) between different high-level features, a hierarchical semantic feature tree is constructed with the mold part as the root node and the high-level features as the child nodes. Based on the semantic feature tree, the three-dimensional digital model is divided into multiple sub-regions according to geometric attributes (such as size, shape, etc.) and topological relationships (such as connectivity, adjacency, etc.), and a parameterized mapping table between the sub-region and the three-dimensional digital model is established to record the position, size and association relationship of the sub-region. Based on the improved octree adaptive meshing algorithm, the three-dimensional digital model is spatially divided. At each leaf node of the octree, the characteristic curvature and stress gradient of the sub-region corresponding to the current node are calculated. If the characteristic curvature exceeds the preset curvature threshold, the current node is subdivided according to the direction and size of the characteristic curvature combined with the anisotropic subdivision strategy until the error requirement is met. If the stress gradient is greater than the preset gradient threshold, the current node is subdivided in combination with the tetrahedral unit, the size and type of the mesh unit are adjusted, and the optimization is repeated until the mesh quality requirement is met.

[0072] Based on the mesh units obtained by division, combined with material properties (such as elastic modulus, Poisson's ratio, etc.) and boundary conditions (such as constraints, loads, etc.), an initial finite element model is generated. According to the mold assembly drawing and tolerance standards, the mating surfaces and gap areas between the parts are identified, and face-to-face contact boundary conditions are applied on the mating surfaces. The contact boundary conditions are added to the initial finite element model through the penalty function method. The contact problem is solved in combination with the augmented Lagrange multiplier method to obtain the contact pressure and gap size distribution on the mating surfaces. The contact pressure is applied to the initial finite element model as a load condition, and a two-way coupling solution is performed to obtain the final mold finite element model.

[0073] In this embodiment, the automatic assembly of mold parts is achieved through parametric constraints and assembly interface definitions, which avoids the tediousness and errors of manual modeling and improves the construction efficiency and accuracy of three-dimensional digital models. The three-dimensional convolutional neural network based on deep learning is used to realize automatic recognition and extraction of high-level features of the mold, avoiding the limitations of traditional feature recognition methods. By constructing a hierarchical semantic feature tree, the semantic description and association of high-level features are realized, providing richer and more semantic feature information for subsequent analysis. By considering the contact relationship and boundary conditions between parts, a finite element model that is closer to reality is generated, which improves the analysis accuracy. In summary, this embodiment realizes the automatic conversion and analysis from heterogeneous design data to high-fidelity analysis models, improves the efficiency and quality of mold design and analysis, and has good application prospects and promotion value.

[0074] In an optional embodiment,

[0075] The contact boundary condition is added to the initial finite element model through a penalty function as shown in the following formula:

[0076] ;

[0077] in, P c represents the penalty function value, G c represents the contact boundary, e N represents the normal penalty factor, g N represents the normal gap, l N represents the normal Lagrange multiplier, e T represents the tangential penalty factor, g T represents the tangential relative slip, l T represents the tangential Lagrange multiplier, dS Represents the area differential used to describe the range of integration.

[0078] In this embodiment, the contact problem is converted into a variational inequality problem by introducing normal and tangential penalty functions, and the penalty functions are added to the total potential energy functional of the finite element model in the form of penalty functions, thereby realizing the mathematical description and computational solution of the contact boundary conditions and avoiding cumbersome contact units and contact algorithms. The penalty function method converts the contact problem into a stiffness problem by introducing a virtual spring on the contact surface, thereby avoiding the iterative solution and master-slave surface search in the traditional contact algorithm and improving the computational efficiency of the contact analysis. By considering the contact relationship between parts, especially the nonlinear contact behavior, the finite element model is closer to the real physical situation and the authenticity and reliability of the model are improved. In summary, this embodiment realizes the mathematical description and solution of the contact problem, improves the efficiency and accuracy of the contact analysis, makes the finite element model closer to reality, and provides important support for the analysis and optimization of the mold structure.

[0079] S2. Based on the nonlinear mechanical behavior of the mold material, the material microstructure evolution model is constructed in combination with the crystal plasticity finite element method. The mold material behavior is simulated by characterizing the grain orientation, dislocation density and slip system evolution, and the equivalent mechanical performance parameters of the mold material are extracted. A cross-scale mechanical performance parameter database is constructed. A multi-condition load spectrum is generated according to the mold working conditions, and the multi-physical fields in the coupled injection and die-casting process are determined. The multi-condition load spectrum and the boundary conditions of the multi-physical fields are applied to the mold finite element model. The deformation of the cavity wall is tracked and a stress-strain analysis model is established in combination with the improved adaptive dynamic mesh technology. The stress-strain analysis model is solved in combination with the multi-grid method to obtain the full-field dynamic response and the stress-strain evolution history of the sub-region of the model and the key stress distribution characteristic parameters are combined to construct a sub-region stress characteristic parameter database for failure modes;

[0080] The nonlinear mechanical behavior refers to the phenomenon that a material or structure exhibits a nonlinear response under the action of an external force. The crystal plasticity finite element method is a numerical method used to simulate the plastic deformation behavior of metal materials at a microscopic scale. The microstructure evolution model is used to describe the microstructural changes of materials during processing and service, including processes such as grain growth, phase transformation, precipitation, and recovery recrystallization. The characterization of grain orientation refers to determining the orientation distribution of each grain in a polycrystalline material. The dislocation density is a measure of the number of dislocations (linear defects in a crystal) in a material, usually expressed as a dislocation length per unit area. The slip coefficient is the main parameter in crystallography that describes material deformation, and is defined as the crystal orientation of the slip direction on the slip plane. The equivalent mechanical performance parameter simplifies the complex micromechanical behavior into Indicators of macroscopic mechanical properties, such as equivalent stress, equivalent strain, equivalent elastic modulus and equivalent hardening parameter, the multi-condition load spectrum is a comprehensive indicator that describes the various load conditions experienced by the material or structure during service, the coupled injection and die-casting molding refers to considering the interaction between multiple physical fields (such as fluid, heat, stress, etc.) during the injection and die-casting process, the cavity wall refers to the inner surface in the injection and die-casting mold that is in direct contact with the molten plastic aluminum alloy, the adaptive dynamic grid tracking technology is a numerical simulation method for dealing with large deformation and moving boundary problems, the multi-grid method is a numerical solution technology that accelerates iterative convergence by performing calculations on grids of different resolutions, and the failure mode is a specific way of describing the destruction of a material or structure under extreme conditions.

[0081] In an optional embodiment,

[0082] Based on the nonlinear mechanical behavior of mold materials, a material microstructure evolution model is constructed in combination with the crystal plasticity finite element method. The mold material behavior is simulated by characterizing the grain orientation, dislocation density and slip system evolution, and the equivalent mechanical performance parameters of the mold material are extracted. A cross-scale mechanical performance parameter database is constructed. A multi-condition load spectrum is generated according to the mold working conditions, and the multi-physical field in the coupled injection and die-casting process is determined. The multi-condition load spectrum and the boundary conditions of the multi-physical field are applied to the mold finite element model. The deformation of the cavity wall is tracked in combination with the improved adaptive dynamic mesh technology and a stress-strain analysis model is established. The stress-strain analysis model is solved in combination with the multi-grid method to obtain the full-field dynamic response and the stress-strain evolution history and key stress distribution characteristic parameters of the sub-region of the model are combined to construct a sub-region stress characteristic parameter database for failure modes, including:

[0083] Based on the nonlinear mechanical behavior of the mold material and in combination with the crystal plasticity finite element method, the velocity gradient tensor in the crystal with multiple slip systems is decomposed into an elastic part and a plastic part. For the elastic part, the slip rate is calculated by the flow law of the thermal activation effect and the dislocation density is determined in combination with the critical shear stress. The dislocation density evolution process is determined in combination with the kinetic equation and dislocation multiplication and annihilation. Based on the dislocation density evolution process, the material microstructure evolution model is constructed.

[0084] Based on the material microstructure evolution model, volume units are randomly selected and periodic boundary conditions are applied to simulate the microstructure evolution process of the material under different conditions, and the yield strength, hardening modulus and plastic strain of the polycrystalline polymerization model are extracted as equivalent mechanical performance parameters by numerical homogenization method to construct the cross-scale mechanical performance parameter database;

[0085] Based on the cross-scale mechanical performance parameter database, combined with the load data of the mold under actual working conditions collected by the sensor, the load characteristic parameters under different working conditions are extracted through statistical analysis, and a probability distribution model of the load-time series is established based on the random process theory. A random load sequence is sampled from the probability distribution model through the Monte Carlo method, and a multi-condition load spectrum is synthesized by combining the frequency and duration of different working conditions.

[0086] Establish a three-dimensional injection die-casting model corresponding to the injection die-casting molding process, wherein the three-dimensional injection die-casting model includes a mold cavity, a pouring system and a cooling system, establish a fluid control equation based on the constitutive model of the plastic aluminum alloy material and the three-dimensional injection die-casting model, establish an energy conservation equation in the solidification cooling stage in combination with heat transfer theory, discretize the fluid control equation and the energy conservation equation through a finite volume method, combine multi-physical field coupling conditions, solve the flow field, temperature field and stress field through a single coupling solution strategy, and extract the boundary conditions of each physical field respectively;

[0087] The multi-condition load spectrum and the boundary conditions of each physical field are applied to the mold finite element model, the deformation of the cavity wall is tracked and the grid coordinates are updated through the improved adaptive dynamic mesh technology, based on the updated grid coordinates, the material microstructure evolution model is embedded into the mold finite element model in combination with the crystal plasticity finite element method, a stress-strain analysis model is established and a nonlinear stress-strain equation is constructed, the nonlinear stress-strain equation is solved by an incremental iterative method and accelerated by a multi-grid method, the full-field dynamic response of the mold under the load is obtained, the stress-strain evolution history of the mold in different sub-regions and the key stress distribution characteristic parameters including the maximum principal stress and stress triaxiality are extracted, and a sub-region stress characteristic parameter database oriented to failure modes is constructed.

[0088] The elastic part refers to the recoverable deformation of the material under the action of external force. Elastic deformation follows Hooke's law, that is, stress is proportional to strain. The plastic part refers to the irreversible deformation of the material after exceeding the yield strength. The flow law of the thermal activation effect describes the flow behavior of the material under high temperature or thermal excitation conditions. The critical shear stress refers to the minimum stress required for the slip system to start slipping. The dislocation proliferation refers to the phenomenon that the number of dislocations (linear defects in the crystal structure) inside the material increases during the plastic deformation process. The annihilation refers to the process in which dislocations cancel each other out inside the material. The yield strength is the stress required for the material to begin plastic deformation, and is an indicator of the material's ability to resist plastic deformation. The hardening modulus describes the change in the material's ability to resist deformation during plastic deformation, and is usually defined as the change after the yield point in the stress-strain curve. The slope of the curve is as follows: the plastic strain is the permanent deformation of the material during the plastic deformation stage; the load characteristic parameter describes the load characteristics of the material or structure during service, including the load type, size, direction and time history, etc. The Monte Carlo method is a numerical method for solving mathematical and physical problems through random sampling and statistical analysis; the heat transfer theory studies the heat transfer process between objects, including three basic heat transfer modes: heat conduction, heat convection and heat radiation; the finite volume method is a numerical calculation method for solving partial differential equations; the incremental iteration is a numerical method for solving nonlinear equations, and the solution is obtained by stepwise approximation; the stress triaxiality describes the three-dimensional characteristics of the stress state, which is defined as the ratio of the sum of the three-way normal stress to the equivalent stress.

[0089] Based on the nonlinear mechanical behavior of mold materials and combined with the crystal plasticity finite element method, the velocity gradient tensor in the crystal is decomposed into an elastic part and a plastic part. For the elastic part, the slip rate is calculated by the flow law of the thermal activation effect, and the dislocation density is determined by combining the critical shear stress. The dynamic equation is used to simulate the change of dislocation density in combination with the dislocation proliferation and annihilation mechanism. Based on the evolution of dislocation density, a microstructure evolution model of the material is constructed. Volume units are randomly selected and periodic boundary conditions are applied to simulate the microstructure evolution of the material under different load conditions. Through the numerical homogenization method, effective mechanical performance parameters such as yield strength, hardening modulus and plastic strain are extracted from the polycrystalline aggregation model, and the extracted parameters are constructed into a cross-scale mechanical performance parameter database.

[0090] Sensors are used to collect the load data of the mold under actual working conditions. The load characteristic parameters under different working conditions are extracted through statistical analysis. A probability distribution model of load-time series is established based on random process theory. Random load sequences are generated by sampling from the probability distribution model. The frequency and duration of different working conditions are combined to synthesize multi-condition load spectra. A three-dimensional injection die-casting model including mold cavity, pouring system and cooling system is established. The fluid control equation is established based on the constitutive model of plastic aluminum alloy material. The energy conservation equation of the solidification cooling stage is established in combination with heat transfer theory. The fluid control equation and energy conservation equation are discretized by finite volume method. Combine the coupling conditions of multiple physical fields, solve the flow field, temperature field and stress field respectively through the single coupling solution strategy, extract the boundary conditions of each physical field, apply the multi-condition load spectrum and the boundary conditions of each physical field to the mold finite element model, use the improved adaptive dynamic mesh technology to track the deformation of the cavity wall, and update the mesh coordinates. Combined with the crystal plasticity finite element method, the material microstructure evolution model is embedded in the mold finite element model, and a stress-strain analysis model is established. The nonlinear stress-strain equation is constructed, solved by the incremental iterative method, and combined with the multi-grid method for accelerated solution, the full-field dynamic response of the mold under load is obtained;

[0091] The stress-strain evolution process of the mold in different sub-areas is extracted, and the key stress distribution characteristic parameters including the maximum principal stress and stress triaxiality are extracted. Based on the above characteristic parameters, a sub-area stress characteristic parameter database oriented to failure modes is constructed.

[0092] In this embodiment, by introducing the crystal plasticity finite element method and considering the evolution of material microstructure, the nonlinear mechanical behavior of the mold material is accurately described, the limitations of the traditional constitutive model are overcome, and the accuracy and reliability of the prediction of the material mechanical properties are improved. Through the material microstructure evolution model and the numerical homogenization method, a cross-scale mechanical performance parameter database is established, and the correlation between the material microstructure and the macroscopic mechanical properties is realized, providing more real and reliable material parameters for mold structure analysis. Through sensor data acquisition and random process theory, a multi-condition load spectrum of the mold under actual working conditions is generated, which overcomes the limitations of traditional single condition analysis and improves the authenticity and applicability of mold structure analysis. Through adaptive dynamic mesh technology and crystal plasticity finite element method, the dynamic response analysis of the mold under load is realized, the stress and strain characteristic parameters of different sub-regions of the mold are extracted, the mold failure mechanism is revealed, and an important basis is provided for mold life prediction and optimization. In summary, this embodiment realizes cross-scale modeling and analysis of the mold structure, reveals the mold failure mechanism, and provides important support for mold reliability design.

[0093] In an optional embodiment,

[0094] The slip rate is calculated by the flow law of thermal activation effect as shown in the following formula:

[0095] ;

[0096] in, is the slip rate, represents the reference slip rate, represents the equivalent strain rate, represents the reference strain rate, k B represents the Boltzmann constant, T represents absolute temperature, Q represents the activation energy, t α Indicates α The shear stress on the slip system is represents the critical shear stress, represents the reference stress, m represents the strain rate sensitive parameter, p represents the stress index, q represents the shape factor, sgn Used to determine the sign of the shear stress.

[0097] In this embodiment, by considering the absolute temperature and the Boltzmann constant, the influence of the thermal activation effect on the slip rate can be accurately reflected, so that the model can still be accurate under different temperature conditions. Through the accurate calculation of the slip rate, the evolution process of the internal microstructure of the material can be better simulated. Combined with the multi-condition load spectrum generated by the statistical analysis and the Monte Carlo method, the mechanical response of the mold can be accurately simulated under different working conditions. In summary, this embodiment significantly improves the accuracy and reliability of the material microstructure evolution model, and provides a solid theoretical foundation and technical support for mold design, optimization and performance prediction.

[0098] S3. Construct a mold structure strength assessment model based on the mold failure mechanism analysis, associate the failure mode with the influencing factors and generate a mold structure strength knowledge graph, add the parameterized mapping table and the sub-region stress characteristic parameter database to the strength assessment model for multi-index fusion calculation, obtain the mold strength score and judge whether the mold strength meets the use requirements according to the pre-set strength score threshold. If not, perform a global sensitivity analysis based on the parameterized mapping table to determine the key sub-regions and the corresponding weak links, and generate a topology optimization plan and a size optimization plan in combination with the pre-built optimization design knowledge base and intelligent reasoning technology. Map the topology optimization plan and the size optimization plan to the corresponding sub-regions in the three-dimensional digital model, perform association modification through the parameterized mapping table, repeat iterations until the mold structure strength meets the use requirements, and output the three-dimensional digital model.

[0099] The mold failure mechanism analysis is to study how the mold fails under actual working conditions and the causes and processes of failure. The multi-index fusion calculation refers to the optimization calculation based on comprehensive consideration of multiple performance indicators during mold design and optimization. The global sensitivity analysis is to identify the key factors that have the greatest impact on system performance by calculating the influence of each input variable on the output result. The weak joint refers to the area in the mold structure that is most prone to failure. The intelligent reasoning technology is to apply artificial intelligence and machine learning technology in mold design and failure analysis. By learning from a large amount of historical data and experimental results, a prediction model is established to achieve intelligent prediction and optimization of mold performance and failure behavior. The topology optimization optimizes the performance of the mold by changing the material distribution inside the mold. The size optimization optimizes the performance of the mold by adjusting the geometric size of the mold.

[0100] In an optional embodiment,

[0101] Based on the analysis of the mold failure mechanism, a mold structure strength assessment model is constructed, the failure mode is associated with the influencing factors and a mold structure strength knowledge map is generated, the parameterized mapping table and the sub-region stress characteristic parameter database are added to the strength assessment model for multi-index fusion calculation, the mold strength score is obtained, and the mold strength is judged whether it meets the use requirements according to the preset strength score threshold. If not, a global sensitivity analysis is performed based on the parameterized mapping table to determine the key sub-regions and the corresponding weak links, and a topology optimization scheme and a size optimization scheme are generated in combination with the pre-built optimization design knowledge base and intelligent reasoning technology. The topology optimization scheme and the size optimization scheme are mapped to the corresponding sub-regions in the three-dimensional digital model, and the association modification is performed through the parameterized mapping table. It is repeated until the mold structure strength meets the use requirements, and the three-dimensional digital model output includes:

[0102] Obtain and analyze mold failure cases, determine the failure mode and failure mechanism corresponding to the mold, build a mold structure strength assessment model, determine the key factors affecting the mold structure strength based on the failure mechanism, wherein the key factors include the mechanical properties of the material, the structural design of the mold, and the injection molding and die-casting process parameters, associate the failure mode with the key factors, take the failure mode, material properties, structural parameters, and process parameters as core knowledge elements, define the categories, attributes, and associations of the core knowledge elements through ontology modeling language, build an ontology model, extract knowledge element instances from the knowledge in this field and perform semantic annotations according to the ontology model, determine the semantic associations and build link relationships through natural language processing technology, and obtain the mold structure strength knowledge graph;

[0103] The parameterized mapping table and the sub-region stress characteristic parameter database are added to the strength evaluation model, and multi-index calculation is performed through weighted fusion to obtain a mold strength score corresponding to the overall structure of the mold and compare it with a preset strength score threshold. If the mold strength score is greater than the preset strength score threshold, it is considered that the current mold meets the use requirements, otherwise, it is considered that the current mold does not meet the use requirements;

[0104] If it does not meet the use requirements, then based on the parameterized mapping table, the influence of each sub-region and failure index on the mold strength score is calculated by a global sensitivity analysis method. According to the sensitivity analysis result, the sub-region with the greatest influence on the mold strength score is taken as the key sub-region, and the failure index with the greatest influence on the mold strength score is taken as the weak link. The weak link is matched with the failure mode in the mold structure strength knowledge graph, and the failure cause is determined in combination with intelligent reasoning technology. Based on the failure cause, a topology optimization plan and a size optimization plan are generated;

[0105] For the key sub-area, based on the topology optimization scheme, the variable density method is used to optimize and solve it to obtain the density distribution result. For the weak link, based on the size optimization scheme, the local structure size is optimized through parametric modeling to obtain the local structure size change. The density distribution result and the local structure size change are applied to the corresponding area of ​​the three-dimensional digital model and are associated and modified through a parametric mapping table. The optimization is repeated until the preset maximum number of iterations is reached, and the three-dimensional digital model corresponding to the last iteration and the mold strength score corresponding to the mold are output.

[0106] The mechanical properties refer to the mechanical behavior and characteristics of a material or structure under the action of external forces. The injection die-casting process parameters refer to the various parameters that need to be controlled and adjusted during the injection die-casting molding process to ensure molding quality and efficiency. The ontology modeling language is a formal language used to construct and describe domain knowledge structures, and is used to define concepts, attributes and their relationships. The failure index is a parameter used to evaluate the failure risk and reliability of a material or structure under specific conditions. The variable density method is a topology optimization method that optimizes structural performance by adjusting the distribution density of the material in the design domain. The parametric modeling is a method of constructing and controlling a model by defining parameters and their relationships.

[0107] Obtain mold failure cases and analyze them, determine the corresponding failure modes and failure mechanisms of the molds, build a mold structure strength assessment model, and based on the failure mechanisms, determine the key factors that affect the mold structure strength, including the mechanical properties of the materials, the structural design of the molds, and the injection molding and die-casting process parameters, and associate the failure modes with the key factors;

[0108] Taking failure modes, material properties, structural parameters and process parameters as core knowledge elements, the categories, attributes and associations of core knowledge elements are defined through ontology modeling language, and an ontology model is constructed. Knowledge element instances are extracted from the knowledge in this field, and semantic annotations are performed according to the ontology model. The semantic associations are determined and link relationships are constructed through natural language processing technology to obtain the knowledge graph of mold structure strength.

[0109] The parameterized mapping table and the sub-region stress characteristic parameter database are added to the strength assessment model, and multi-index calculation is performed through weighted fusion to obtain the mold strength score corresponding to the overall structure of the mold. The mold strength score is compared with the preset strength score threshold. If the mold strength score is greater than the preset strength score threshold, it is considered that the current mold meets the use requirements, otherwise, it is considered that the current mold does not meet the use requirements;

[0110] If it does not meet the use requirements, the influence of each sub-region and failure index on the mold strength score is calculated based on the parameterized mapping table through the global sensitivity analysis method. According to the sensitivity analysis results, the sub-region with the greatest influence on the mold strength score is taken as the key sub-region, and the failure index with the greatest influence on the mold strength score is taken as the weak link. The weak link is matched with the failure mode in the mold structure strength knowledge map, and the failure cause is determined by combining intelligent reasoning technology. Based on the failure cause, the topology optimization plan and size optimization plan are generated;

[0111] For key sub-areas, based on the topology optimization solution, the variable density method is used to optimize and solve, and the density distribution results are obtained. For weak links, based on the size optimization solution, the local structure size is optimized through parametric modeling, and the local structure size change is obtained. The density distribution results and the local structure size change are applied to the corresponding areas of the three-dimensional digital model, and the associated modifications are made through the parametric mapping table. The optimization is repeated until the preset maximum number of iterations is reached, and the three-dimensional digital model corresponding to the last iteration is output, and the mold strength score corresponding to the mold is output.

[0112] In this embodiment, by collecting and analyzing actual mold failure cases, the common failure modes and failure mechanisms of the mold during use can be determined, and the main factors affecting the life and performance of the mold can be systematically identified, providing a reliable basis for subsequent optimization. By associating failure modes with key factors and using the ontology modeling language to define the categories, attributes and associations of core knowledge elements, domain knowledge can be systematically and structuredly managed and utilized. Topological optimization and size optimization methods are applied to optimize key sub-areas and weak links respectively, thereby comprehensively improving the structural strength and performance of the mold. In summary, this embodiment realizes the reliability analysis and optimization of the mold structure, greatly improves the efficiency and success rate of mold development, provides reference and guidance for subsequent similar mold designs, and has good prospects for promotion and application.

[0113] In an optional embodiment,

[0114] The variable density method is used to optimize and solve the density distribution, and the density distribution result is shown in the following formula:

[0115] ;

[0116] in, r Represents the relative density of the unit, tanh represents the hyperbolic tangent function, β represents the magnification factor, or e represents a continuous slack variable, H represents the penalty factor, Represents the solid element stiffness matrix.

[0117] In this embodiment, by introducing continuous relaxation variables and hyperbolic tangent functions, the continuity and manufacturing feasibility of the optimization results are improved. By adjusting the amplification factor, the optimization convergence speed can be accelerated and the number of optimization iterations can be reduced. By establishing a stiffness matrix interpolation model, repeated calculation of the unit stiffness matrix is ​​avoided, the optimization efficiency is improved, and the calculation cost is reduced. In summary, this embodiment uses an improved variable density method to perform topological optimization on key sub-areas, and through density interpolation, continuous relaxation, penalty factor and other technologies, a clear and smooth material density distribution result is obtained, which provides strong support for lightweight and high-performance design of molds.

[0118] Figure 2 FIG. 1 is a schematic diagram of a mold structure strength calculation system based on digital design according to an embodiment of the present invention. Figure 2 As shown, the system comprises:

[0119] The first unit is used to obtain the mold assembly to be analyzed and convert it into a three-dimensional digital model, adaptively segment the three-dimensional digital model through intelligent feature recognition technology based on deep learning, extract high-level features of the mold structure and generate a semantic feature tree, dynamically divide the model sub-regions based on the semantic feature tree and set a parameterized mapping table, discretize the model sub-regions through an improved octree adaptive grid, adjust the grid unit size and type in combination with the model characteristic curvature and stress gradient, generate a sub-region finite element model, and construct a mold finite element model in combination with the mold assembly tolerance and assembly gap nonlinear contact algorithm;

[0120] The second unit is used to construct a material microstructure evolution model based on the nonlinear mechanical behavior of the mold material in combination with the crystal plasticity finite element method, simulate the mold material behavior by characterizing the grain orientation, dislocation density and slip system evolution, extract the equivalent mechanical performance parameters of the mold material, and construct a cross-scale mechanical performance parameter database. A multi-condition load spectrum is generated according to the mold working conditions, and the multi-physical fields in the coupled injection and die-casting process are determined. The multi-condition load spectrum and the boundary conditions of the multi-physical fields are applied to the mold finite element model, and the deformation of the cavity wall is tracked and a stress-strain analysis model is established in combination with the improved adaptive dynamic mesh technology. The stress-strain analysis model is solved in combination with the multi-grid method to obtain the full-field dynamic response and construct a sub-region stress characteristic parameter database for failure modes in combination with the stress-strain evolution history and key stress distribution characteristic parameters of the sub-region of the model;

[0121] The third unit is used to build a mold structure strength evaluation model based on mold failure mechanism analysis, associate failure modes with influencing factors and generate a mold structure strength knowledge map, add the parameterized mapping table and the sub-region stress characteristic parameter database to the strength evaluation model for multi-index fusion calculation, obtain the mold strength score, and judge whether the mold strength meets the use requirements according to a preset strength score threshold. If not, a global sensitivity analysis is performed based on the parameterized mapping table to determine the key sub-regions and corresponding weak links, and a topology optimization scheme and a size optimization scheme are generated in combination with a pre-built optimization design knowledge base and intelligent reasoning technology. The topology optimization scheme and the size optimization scheme are mapped to the corresponding sub-regions in the three-dimensional digital model, and the associated modifications are performed through the parameterized mapping table. The iteration is repeated until the mold structure strength meets the use requirements, and the three-dimensional digital model is output.

[0122] According to a third aspect of the embodiments of the present invention,

[0123] An electronic device is provided, comprising:

[0124] processor;

[0125] a memory for storing processor-executable instructions;

[0126] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0127] According to a fourth aspect of the embodiments of the present invention,

[0128] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0129] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for calculating mold structure strength based on data-based design, characterized in that: include: The mold assembly to be analyzed is obtained and converted into a three-dimensional digital model, the three-dimensional digital model is adaptively segmented by intelligent feature recognition technology based on deep learning, the high-level features of the mold structure are extracted and a semantic feature tree is generated, the model sub-regions are dynamically divided based on the semantic feature tree and a parameterized mapping table is set, the model sub-regions are discretized by an improved octree adaptive grid, the grid unit size and type are adjusted in combination with the model characteristic curvature and stress gradient, a sub-region finite element model is generated, and a mold finite element model is constructed in combination with a mold assembly tolerance and an assembly gap nonlinear contact algorithm; Based on the nonlinear mechanical behavior of the mold material, the material microstructure evolution model is constructed in combination with the crystal plasticity finite element method. The mold material behavior is simulated by characterizing the grain orientation, dislocation density and slip system evolution, and the equivalent mechanical performance parameters of the mold material are extracted to construct a cross-scale mechanical performance parameter database. A multi-condition load spectrum is generated according to the mold working conditions, and the multi-physical field in the coupled injection and die-casting molding process is determined. The multi-condition load spectrum and the boundary conditions of the multi-physical field are applied to the mold finite element model. The deformation of the cavity wall is tracked in combination with the improved adaptive dynamic mesh technology and a stress-strain analysis model is established. The stress-strain analysis model is solved in combination with the multi-grid method to obtain the full-field dynamic response and to construct a sub-region stress characteristic parameter database for failure modes in combination with the stress-strain evolution history and key stress distribution characteristic parameters of the sub-region of the model. Based on the analysis of the mold failure mechanism, a mold structure strength assessment model is constructed, the failure mode is associated with the influencing factors and a mold structure strength knowledge map is generated, the parameterized mapping table and the sub-region stress characteristic parameter database are added to the strength assessment model for multi-index fusion calculation, the mold strength score is obtained, and whether the mold strength meets the use requirements is judged according to a preset strength score threshold. If not, a global sensitivity analysis is performed based on the parameterized mapping table to determine the key sub-regions and the corresponding weak links, and a topology optimization scheme and a size optimization scheme are generated in combination with a pre-built optimization design knowledge base and intelligent reasoning technology. The topology optimization scheme and the size optimization scheme are mapped to the corresponding sub-regions in the three-dimensional digital model, and the associated modification is performed through the parameterized mapping table. The iteration is repeated until the mold structure strength meets the use requirements, and the three-dimensional digital model is output.

2. The method according to claim 1, characterized in that The mold assembly to be analyzed is obtained and converted into a three-dimensional digital model, the three-dimensional digital model is adaptively segmented by intelligent feature recognition technology based on deep learning, the high-level features of the mold structure are extracted and a semantic feature tree is generated, the model sub-regions are dynamically divided based on the semantic feature tree and a parameterized mapping table is set, the model sub-regions are discretized by an improved octree adaptive grid, the grid unit size and type are adjusted in combination with the model characteristic curvature and stress gradient, a sub-region finite element model is generated, and a mold finite element model is constructed in combination with the mold assembly tolerance and the assembly gap nonlinear contact algorithm, including: Acquire the two-dimensional drawings, three-dimensional CAD models and assembly data of the mold assembly to be analyzed, combine them as heterogeneous design data, perform digital processing and format conversion on the heterogeneous design data, extract the geometric information, topological information and attributes of the mold parts, assemble the mold parts according to the structural hierarchy and assembly relationship of the mold assembly through a bottom-up assembly modeling strategy, and obtain the three-dimensional digital model through parametric constraints and assembly interface definitions; For the three-dimensional digital model, a mold feature data set is constructed and manually annotated, a three-dimensional convolutional neural network is constructed by using an intelligent feature recognition technology based on deep learning, the mold feature data set with annotations is converted into a voxel representation and added to the three-dimensional convolutional neural network, and a feature category probability distribution corresponding to each voxel is obtained, the training is repeated and the hyperparameters of the three-dimensional convolutional neural network are adjusted to obtain a trained feature recognition network, and feature extraction is performed on the three-dimensional digital model through the feature recognition network to obtain the high-level features, and based on the high-level features, a hierarchical semantic feature tree is constructed with the mold part as a root node and the high-level features as child nodes by determining the topological relationship and semantic attributes between different high-level features; Based on the hierarchical semantic feature tree, the three-dimensional digital model is divided into a plurality of sub-regions according to geometric attributes and topological relationships, and a parameterized mapping table between the sub-regions and the three-dimensional digital model is established; Based on the improved octree adaptive meshing algorithm, the characteristic curvature and stress gradient of the sub-region corresponding to the current node are calculated at each leaf node of the octree. If the characteristic curvature exceeds the preset curvature threshold, the current node is subdivided according to the direction and size of the characteristic curvature in combination with the anisotropic subdivision strategy until the error requirement is met. If the stress gradient is greater than the preset gradient threshold, the current node is subdivided in combination with the tetrahedral unit, and the size and type of the mesh unit are adjusted. Based on the mesh unit, an initial finite element model is generated in combination with material properties and boundary conditions. The mating surfaces and gap areas between parts are identified according to the mold assembly drawing and tolerance standards. Face-to-face contact boundary conditions are applied on the mating surfaces. The contact boundary conditions are added to the initial finite element model through a penalty function. The contact problem is solved in combination with the augmented Lagrange multiplier method to obtain the contact pressure and gap size distribution on the mating surfaces. The contact pressure is applied as a load condition to the initial finite element model for bidirectional coupling to obtain the mold finite element model.

3. The method according to claim 2, characterized in that The contact boundary condition is added to the initial finite element model through a penalty function as shown in the following formula: ； in, П c represents the penalty function value, G c represents the contact boundary, ε N represents the normal penalty factor, g N represents the normal gap, λ N represents the normal Lagrange multiplier, ε T represents the tangential penalty factor, g T represents the tangential relative slip, λ T represents the tangential Lagrange multiplier, dS Represents the area differential used to describe the range of integration.

4. The method according to claim 1, characterized in that: Based on the nonlinear mechanical behavior of mold materials, a material microstructure evolution model is constructed in combination with the crystal plasticity finite element method. The mold material behavior is simulated by characterizing the grain orientation, dislocation density and slip system evolution, and the equivalent mechanical performance parameters of the mold material are extracted. A cross-scale mechanical performance parameter database is constructed. A multi-condition load spectrum is generated according to the mold working conditions, and the multi-physical field in the coupled injection and die-casting process is determined. The multi-condition load spectrum and the boundary conditions of the multi-physical field are applied to the mold finite element model. The deformation of the cavity wall is tracked in combination with the improved adaptive dynamic mesh technology and a stress-strain analysis model is established. The stress-strain analysis model is solved in combination with the multi-grid method to obtain the full-field dynamic response and the stress-strain evolution history and key stress distribution characteristic parameters of the sub-region of the model are combined to construct a sub-region stress characteristic parameter database for failure modes, including: Based on the nonlinear mechanical behavior of the mold material and in combination with the crystal plasticity finite element method, the velocity gradient tensor in the crystal with multiple slip systems is decomposed into an elastic part and a plastic part. For the elastic part, the slip rate is calculated by the flow law of the thermal activation effect and the dislocation density is determined in combination with the critical shear stress. The dislocation density evolution process is determined in combination with the kinetic equation and dislocation multiplication and annihilation. Based on the dislocation density evolution process, the material microstructure evolution model is constructed. Based on the material microstructure evolution model, volume units are randomly selected and periodic boundary conditions are applied to simulate the microstructure evolution process of the material under different load conditions. The yield strength, hardening modulus and plastic strain of the polycrystalline polymerization model are extracted as equivalent mechanical performance parameters through a numerical homogenization method to construct the cross-scale mechanical performance parameter database; Based on the cross-scale mechanical performance parameter database, combined with the load data of the mold under actual working conditions collected by the sensor, the load characteristic parameters under different working conditions are extracted through statistical analysis, and a probability distribution model of the load-time series is established based on the random process theory. A random load sequence is sampled from the probability distribution model through the Monte Carlo method, and a multi-condition load spectrum is synthesized by combining the frequency and duration of different working conditions. Establish a three-dimensional injection die-casting model corresponding to the injection die-casting molding process, wherein the three-dimensional injection die-casting model includes a mold cavity, a pouring system, and a cooling system, establish a fluid control equation based on the constitutive model of the plastic aluminum alloy material and the three-dimensional injection die-casting model, establish an energy conservation equation in the solidification cooling stage in combination with heat transfer theory, discretize the fluid control equation and the energy conservation equation through a finite volume method, combine multi-physical field coupling conditions, solve the flow field, temperature field, and stress field through a single coupling solution strategy, and extract the boundary conditions of each physical field respectively; The multi-condition load spectrum and the boundary conditions of each physical field are applied to the mold finite element model, the deformation of the cavity wall is tracked and the grid coordinates are updated through the improved adaptive dynamic mesh technology, based on the updated grid coordinates, the material microstructure evolution model is embedded into the mold finite element model in combination with the crystal plasticity finite element method, a stress-strain analysis model is established and a nonlinear stress-strain equation is constructed, the nonlinear stress-strain equation is solved by an incremental iterative method and accelerated by a multi-grid method, the full-field dynamic response of the mold under the load is obtained, the stress-strain evolution history of the mold in different sub-regions and the key stress distribution characteristic parameters including the maximum principal stress and stress triaxiality are extracted, and a sub-region stress characteristic parameter database oriented to failure modes is constructed.

5. The method according to claim 4, characterized in that The slip rate is calculated by the flow law of thermal activation effect as shown in the following formula: ； in, represents the slip rate, represents the reference slip rate, represents the equivalent strain rate, represents the reference strain rate, k B represents the Boltzmann constant, T represents absolute temperature, Q represents the activation energy, τ α Indicates α The shear stress on the slip system is represents the critical shear stress, represents the reference stress, m represents the strain rate sensitive parameter, p represents the stress index, q represents the shape factor, sgn Used to determine the sign of the shear stress.

6. The method according to claim 1, characterized in that Based on the analysis of the mold failure mechanism, a mold structure strength assessment model is constructed, the failure mode is associated with the influencing factors and a mold structure strength knowledge map is generated, the parameterized mapping table and the sub-region stress characteristic parameter database are added to the strength assessment model for multi-index fusion calculation, the mold strength score is obtained, and the mold strength is judged whether it meets the use requirements according to the preset strength score threshold. If not, a global sensitivity analysis is performed based on the parameterized mapping table to determine the key sub-regions and the corresponding weak links, and a topology optimization scheme and a size optimization scheme are generated in combination with the pre-built optimization design knowledge base and intelligent reasoning technology. The topology optimization scheme and the size optimization scheme are mapped to the corresponding sub-regions in the three-dimensional digital model, and the association modification is performed through the parameterized mapping table. It is repeated until the mold structure strength meets the use requirements, and the three-dimensional digital model output includes: Obtain and analyze mold failure cases, determine the failure mode and failure mechanism corresponding to the mold, build a mold structure strength assessment model, determine the key factors affecting the mold structure strength based on the failure mechanism, wherein the key factors include the mechanical properties of the material, the structural design of the mold, and the injection molding and die-casting process parameters, associate the failure mode with the key factors, take the failure mode, material properties, structural parameters, and process parameters as core knowledge elements, define the categories, attributes, and associations of the core knowledge elements through ontology modeling language, build an ontology model, extract knowledge element instances from the knowledge in this field and perform semantic annotations according to the ontology model, determine the semantic associations and build link relationships through natural language processing technology, and obtain the mold structure strength knowledge graph; The parameterized mapping table and the sub-region stress characteristic parameter database are added to the strength evaluation model, and multi-index calculation is performed through weighted fusion to obtain a mold strength score corresponding to the overall structure of the mold and compare it with a preset strength score threshold. If the mold strength score is greater than the preset strength score threshold, it is considered that the current mold meets the use requirements, otherwise, it is considered that the current mold does not meet the use requirements; If it does not meet the use requirements, then based on the parameterized mapping table, the influence of each sub-region and failure index on the mold strength score is calculated by a global sensitivity analysis method. According to the sensitivity analysis result, the sub-region with the greatest influence on the mold strength score is taken as the key sub-region, and the failure index with the greatest influence on the mold strength score is taken as the weak link. The weak link is matched with the failure mode in the mold structure strength knowledge graph, and the failure cause is determined in combination with intelligent reasoning technology. Based on the failure cause, a topology optimization plan and a size optimization plan are generated; For the key sub-area, based on the topology optimization scheme, the variable density method is used to optimize and solve it to obtain the density distribution result. For the weak link, based on the size optimization scheme, the local structure size is optimized through parametric modeling to obtain the local structure size change. The density distribution result and the local structure size change are applied to the corresponding area of ​​the three-dimensional digital model and are associated and modified through a parametric mapping table. The optimization is repeated until the preset maximum number of iterations is reached, and the three-dimensional digital model corresponding to the last iteration and the mold strength score corresponding to the mold are output.

7. The method according to claim 6, characterized in that The variable density method is used to optimize and solve the density distribution, and the density distribution result is shown in the following formula: ； in, ρ Represents the relative density of the unit, tanh represents the hyperbolic tangent function, β represents the magnification factor, η e represents a continuous slack variable, H represents the penalty factor, Represents the solid element stiffness matrix.

8. A mold structure strength calculation system based on digital design, used to implement the method described in any one of claims 1 to 7, characterized in that: include: The first unit is used to obtain the mold assembly to be analyzed and convert it into a three-dimensional digital model, adaptively segment the three-dimensional digital model through intelligent feature recognition technology based on deep learning, extract high-level features of the mold structure and generate a semantic feature tree, dynamically divide the model sub-regions based on the semantic feature tree and set a parameterized mapping table, discretize the model sub-regions through an improved octree adaptive grid, adjust the grid unit size and type in combination with the model characteristic curvature and stress gradient, generate a sub-region finite element model, and construct a mold finite element model in combination with the mold assembly tolerance and assembly gap nonlinear contact algorithm; The second unit is used to construct a material microstructure evolution model based on the nonlinear mechanical behavior of the mold material in combination with the crystal plasticity finite element method, simulate the mold material behavior by characterizing the grain orientation, dislocation density and slip system evolution, extract the equivalent mechanical performance parameters of the mold material, and construct a cross-scale mechanical performance parameter database. A multi-condition load spectrum is generated according to the mold working conditions, and the multi-physical fields in the coupled injection and die-casting process are determined. The multi-condition load spectrum and the boundary conditions of the multi-physical fields are applied to the mold finite element model, and the deformation of the cavity wall is tracked and a stress-strain analysis model is established in combination with the improved adaptive dynamic mesh technology. The stress-strain analysis model is solved in combination with the multi-grid method to obtain the full-field dynamic response and construct a sub-region stress characteristic parameter database for failure modes in combination with the stress-strain evolution history and key stress distribution characteristic parameters of the sub-region of the model; The third unit is used to build a mold structure strength evaluation model based on mold failure mechanism analysis, associate failure modes with influencing factors and generate a mold structure strength knowledge map, add the parameterized mapping table and the sub-region stress characteristic parameter database to the strength evaluation model for multi-index fusion calculation, obtain the mold strength score, and judge whether the mold strength meets the use requirements according to a preset strength score threshold. If not, a global sensitivity analysis is performed based on the parameterized mapping table to determine the key sub-regions and corresponding weak links, and a topology optimization scheme and a size optimization scheme are generated in combination with a pre-built optimization design knowledge base and intelligent reasoning technology. The topology optimization scheme and the size optimization scheme are mapped to the corresponding sub-regions in the three-dimensional digital model, and the associated modifications are performed through the parameterized mapping table. The iteration is repeated until the mold structure strength meets the use requirements, and the three-dimensional digital model is output.

9. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method described in any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images