A multi-physics finite element intelligent prompting system and method

Abstract

This invention relates to the field of computer-aided engineering simulation technology, specifically to a multi-physics finite element intelligent prompting system and method. It achieves adaptive coordinate system matching through weighted comprehensive evaluation of geometric multi-indicators; establishes a unified multi-physics evaluation framework to comprehensively assess structural, fluid, thermal, electromagnetic, acoustic, seepage single-field, and coupled working conditions; adaptively selects and matches volume elements based on multi-constraint joint scoring, and matches boundary elements of the same order through upstream and downstream forced linkage constraints; combines multi-dimensional matrix features with multi-objective optimization to select the solver; adaptively maps and matches elastoplastic yield criteria based on material mechanical properties; and finally integrates end-to-end linkage evaluation information to output standardized intelligent prompt text. This scheme employs multi-feature fusion evaluation, upstream and downstream parameter linkage constraints, and adaptive hierarchical optimization of working conditions to achieve intelligent parameter recommendation throughout the finite element modeling process, improving the convergence and computational accuracy of multi-physics coupled simulation.

Description

Technical Field

[0001] This invention relates to the field of computer-aided engineering simulation technology, specifically to a multiphysics finite element intelligent prompting system and method. Background Technology

[0002] A complete finite element simulation process typically involves multiple stages, including geometric modeling, physical field definition, mesh generation, boundary condition application, solver configuration, and post-processing. However, with the increasing complexity of engineering systems, the pre-processing stage has become increasingly demanding, posing a key bottleneck to improving analysis efficiency and accuracy.

[0003] Existing finite element analysis software requires operators to possess strong mechanics and mathematics knowledge to manually determine the topological characteristics of the geometric model to select a suitable coordinate system and to decide whether to use tetrahedral or hexahedral elements based on geometric regularity when performing mesh generation and element selection. For inexperienced users, it is often difficult to find a balance between computational accuracy and computational cost, and improper selection of element order can easily lead to shear locking or wasted computational resources.

[0004] When dealing with multi-physics coupled scenarios involving structures, fluids, and electromagnetic fields, users must manually set material constitutive models one by one, such as the Mises criterion for metals or the Drucker-Prag criterion for soil and rock, as well as boundary conditions. Due to the lack of a mandatory linkage constraint mechanism, problems such as order mismatch between volume elements and boundary elements and incorrect load application methods often occur. In addition, in the selection of solvers, users often find it difficult to adaptively select direct or iterative methods based on the symmetry, sparsity, and ill-conditioned nature of the system matrix, leading to difficulties in convergence or even divergence in the solution process.

[0005] Existing software lacks a comprehensive intelligent prompting and error correction mechanism. From geometry import to solution setup, data at each stage is isolated, failing to form a closed-loop feedback loop. Once an initial setting deviates, it is often only discovered after a calculation failure, significantly reducing development efficiency. Summary of the Invention

[0006] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a multiphysics finite element intelligent prompting system and method.

[0007] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: a multiphysics finite element intelligent prompting method, comprising the following steps: S1: Import the geometric model to be analyzed, extract multi-dimensional topological, surface and symmetric geometric features, use a multi-index weighted comprehensive scoring method to judge geometric attributes, and adaptively match the optimal coordinate system in one-dimensional, two-dimensional and three-dimensional coordinate systems based on geometric shape and symmetry features, and generate coordinate system selection prompts. S2: By using a unified multiphysics analysis framework, the simulation problem type is determined by combining the physical field type and analysis attributes. The physical fields include structural, fluid, thermal, electromagnetic, acoustic, and seepage fields. The analysis attributes include static / dynamic, linear / nonlinear, and steady-state / transient, and corresponding analysis type recommendations are output. S3: Based on a comprehensive score considering multiple constraints such as geometric regularity, model dimension, nonlinearity strength, and user accuracy requirements, the system adaptively selects the corresponding volume element and first-order or second-order element order, and generates volume element selection suggestions. S4: Establish an upstream and downstream forced linkage constraint mechanism. Based on the selected volume element type and order, match the corresponding order of point, spring, line, and surface boundary elements, adapt to distributed load, displacement constraint, heat flow, and convection boundary conditions, and generate boundary element matching prompts. S5: Taking multiple constraints such as linear or nonlinear properties of the problem, symmetry of the system matrix, model size, and ill-conditioned matrix, and comprehensively balancing multiple objectives such as convergence speed, memory usage, and computational accuracy, it adaptively selects either a direct method or an iterative method solver and generates solver selection prompts. S6: Adaptively map and match the corresponding elastoplastic yield criterion according to the material type, and generate hardening or softening parameter setting prompts according to the material strain softening characteristics; S7: Integrate all the above-mentioned linkage analysis and prompt information, and output standardized intelligent prompt text in a unified format.

[0008] Further preferred, the geometric attribute assessment in step S1 specifically includes four weighted evaluation dimensions: Dimensional analysis: Based on the topological relationship characteristics of the vertices, edges, faces, and volumes of the geometric object, determine whether it is a one-dimensional linear model, a two-dimensional shell model, or a three-dimensional solid model; Axisymmetry fusion judgment: The structure is judged to be axisymmetric when it meets the requirements of rotational coincidence and circumferential property uniformity by a comprehensive evaluation of multiple characteristics including geometric generatrix shape, cross-sectional uniformity, boundary load distribution, and material distribution. Geometric regularity weighted distinction: The regularity score is calculated by weighting multiple indicators such as aspect ratio, surface curvature consistency, and side angle deviation. If the deviation is less than the preset threshold, it is judged as regular geometry; otherwise, it is irregular geometry. Symmetry feature recognition: Detect the symmetry plane / axis corresponding to the planar mirror symmetry and rotational symmetry of the model, as a prerequisite constraint for coordinate system optimization; The coordinate systems include a one-dimensional x-cartesian coordinate system, a r-polar coordinate system, and a spherical coordinate system; a two-dimensional xy-plane cartesian coordinate system, an xyd-axis symmetric plane coordinate system, an rz-axis symmetric cylindrical coordinate system, a ro-polar coordinate system, a rs-spherical coordinate system, and a so-spherical coordinate system; and a three-dimensional xyz-cartesian coordinate system, a rzo-cylindrical coordinate system, a roz-polar-cylindrical hybrid coordinate system, and a rso-spherical coordinate system, which are automatically selected and matched based on geometric comprehensive characteristics.

[0009] More preferably, the physical field types mentioned in step S2 include structural fields, fluid fields, thermal fields, electromagnetic fields, acoustic fields, and seepage fields; The analytical attributes are further subdivided by physical field as follows: The structural field is divided into elastic, elastoplastic, viscoelastic, viscoelastic-plastic, rigid-plastic, contact, and large deformation scenarios, as well as steady-state and dynamic scenarios of beams, plates, trusses, and beam-plate combinations. Fluid fields are distinguished between steady-state and transient flows, single-phase and multiphase flows, and free surface scenarios; Electromagnetic fields can be categorized into electrostatic, magnetostatic, time-harmonic, and transient scenarios. Thermal fields distinguish between steady-state and transient scenarios; The seepage field is distinguished into saturated and unsaturated, steady-state and transient seepage scenarios; The type of analysis problem is determined by the combination of physical field type and subdivided analysis attributes.

[0010] Further preferred, the adaptive hierarchical selection rule for volume units in step S3 is as follows: Basic matching logic: For regular geometry, quadrilateral elements are used in 2D and hexahedral elements are used in 3D; for irregular geometry, triangular elements are used in 2D and tetrahedral elements are used in 3D. Adaptive upgrade rules: When the system is identified as nonlinear elastoplastic, involves contact action, has a stress concentration sensitive area, or the user presets a high accuracy level, the system will automatically upgrade and recommend second-order elements. In other scenarios, first-order elements will be recommended by default.

[0011] In a further preferred embodiment, the boundary elements and volume elements in step S4 employ an order-forced matching mechanism: When the selected volume element is a first-order element, it matches first-order boundary elements, including point, spring, line, and surface boundary elements. When the selected volume element is a second-order element, match the second-order boundary element; Boundary elements are adapted to the application requirements of distributed loads, displacement constraints, heat flow, and convection boundary conditions.

[0012] Further preferred, the matching decision logic of the solver in step S5 is as follows: For linear problems with symmetric positive definite system matrices, a symmetric direct solver or a preconditional conjugate gradient iterative solver is recommended. For scenarios where the system matrix is ​​asymmetric, we recommend the asymmetric direct solver, the multi-wavefront solver, or the GMRES iterative solver. For nonlinear or large-scale ill-conditioned problem scenarios, it is recommended to use a solver that combines the Newton-Raphson method with preconditioned LU decomposition, over-relaxation iteration, or Gauss-Seidel iteration.

[0013] Further preferred, the matching rule for the elastoplastic yield criterion in step S6 is as follows: Metallic materials should conform to the Mises yield criterion; Geotechnical materials are matched with the Drucker-Plager yield criterion and the Mohr-Coulomb yield criterion; Concrete material matching: concrete damage-plastic model or Drucker-Prager model; At the same time, depending on whether the material contains strain softening characteristics, corresponding setting prompts for hardening parameters or softening parameters will be generated.

[0014] Preferably, a multiphysics finite element intelligent prompting system includes a geometry and coordinate system comprehensive judgment module, a multiphysics problem unified judgment module, a volume element adaptive hierarchical selection module, a boundary element order linkage matching module, a solver multi-objective adaptive selection module, an elastic-plastic yield criterion adaptive mapping module, and a prompting information generation and output module. The geometry and coordinate system integrated judgment module outputs geometric features and coordinate system results to the multiphysics problem unified judgment module and the volume element adaptive hierarchical selection module. The multiphysics problem unified judgment module outputs problem type labels to the volume element adaptive hierarchical selection module, the boundary element order linkage matching module, and the solver multi-objective adaptive selection module. The volume element adaptive hierarchical optimization module outputs the volume element selection result to the boundary element order linkage matching module; The outputs of the boundary element order linkage matching module, the solver multi-objective adaptive selection module, and the elastic-plastic yield criterion adaptive mapping module are all connected to the prompt information generation and output module. The prompt information generation and output module integrates all the analysis results and outputs standardized intelligent prompt text.

[0015] Further preferably, it also includes a hierarchical constraint matching library, which includes a coordinate system weighted matching rule table, a physical field analysis attribute mapping table, a geometric regularity-element hierarchical mapping table, a volume-boundary element order linkage constraint table, a solver multi-objective optimization evaluation table, and a material yield criterion mapping table.

[0016] (III) Beneficial Effects Compared with the prior art, the present invention provides a multiphysics finite element intelligent prompting system and method, which has the following beneficial effects: This solution overcomes the limitations of traditional software that relies solely on manual coordinate system selection. It automatically extracts the topological, surface, and symmetry features of the model and accurately identifies one-dimensional, two-dimensional, three-dimensional, and axisymmetric geometric attributes through a multi-dimensional weighted scoring mechanism. The system can automatically select the optimal coordinate system from Cartesian, polar, cylindrical, and spherical coordinates based on the geometric shape, significantly reducing preprocessing complexity and providing the most suitable mathematical description for subsequent calculations, effectively improving the convergence and accuracy of numerical computation.

[0017] This scheme establishes a forced linkage constraint mechanism between upstream and downstream components. The system adaptively selects the type and order of volume elements based on geometric regularity and nonlinearity, and forcibly matches point, line, and surface boundary elements of the corresponding order accordingly. This linkage mechanism completely eliminates node inconsistencies or calculation errors caused by the mixed use of boundary element and volume element orders, ensuring the rigor and integrity of the finite element model in terms of topological structure.

[0018] This approach comprehensively considers the linear / nonlinear properties, matrix symmetry, and ill-conditioned scale of the problem, breaking the limitations of single solvers. The system automatically balances convergence speed, memory usage, and computational accuracy, making the optimal recommendation between direct and iterative methods. This not only avoids computational divergence or resource waste caused by inappropriate solver selection but also significantly improves the solution efficiency for complex multiphysics coupled problems.

[0019] The system incorporates a comprehensive material yield criterion mapping logic, which can automatically match Mises, Drucker-Prag, or concrete damage-plasticity models based on material type, and intelligently identify strain softening characteristics to generate parameter setting prompts. This function transforms complex mechanical constitutive theory into intuitive automated prompts, greatly lowering the barrier for non-experienced users to perform complex nonlinear analyses and reducing simulation distortion caused by incorrect constitutive model selection.

[0020] By integrating information from the entire chain of geometry, physical fields, mesh, boundaries, and solvers, this solution outputs standardized intelligent prompt text in a unified format, providing navigational assistance for simulation analysis. This effectively avoids human omissions and misoperations, significantly improving the standardization and reliability of finite element analysis results. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the system architecture for the present invention; Figure 2 This is a schematic diagram of the method flow architecture of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figure 1-2 The present invention provides a multiphysics finite element intelligent prompting method, comprising the following steps: S1: Import the geometric model to be analyzed, extract multi-dimensional topological, surface and symmetric geometric features, use a multi-index weighted comprehensive scoring method to judge geometric attributes, and adaptively match the optimal coordinate system in one-dimensional, two-dimensional and three-dimensional coordinate systems based on geometric shape and symmetry features, and generate coordinate system selection prompts. S2: By using a unified multiphysics analysis framework, the simulation problem type is determined by combining the physical field type and analysis attributes. The physical fields include structural, fluid, thermal, electromagnetic, acoustic, and seepage fields. The analysis attributes include static / dynamic, linear / nonlinear, and steady-state / transient, and corresponding analysis type recommendations are output. S3: Based on a comprehensive score considering multiple constraints such as geometric regularity, model dimension, nonlinearity strength, and user accuracy requirements, the system adaptively selects the corresponding volume element and first-order or second-order element order, and generates volume element selection suggestions. S4: Establish an upstream and downstream forced linkage constraint mechanism. Based on the selected volume element type and order, match the corresponding order of point, spring, line, and surface boundary elements, adapt to distributed load, displacement constraint, heat flow, and convection boundary conditions, and generate boundary element matching prompts. S5: Taking multiple constraints such as linear or nonlinear properties of the problem, symmetry of the system matrix, model size, and ill-conditioned matrix, and comprehensively balancing multiple objectives such as convergence speed, memory usage, and computational accuracy, it adaptively selects either a direct method or an iterative method solver and generates solver selection prompts. S6: Adaptively map and match the corresponding elastoplastic yield criterion according to the material type, and generate hardening or softening parameter setting prompts according to the material strain softening characteristics; S7: Integrate all the above-mentioned linkage analysis and prompt information, and output standardized intelligent prompt text in a unified format.

[0024] This solution uses a hierarchical constraint matching library as the underlying rule support. It consists of seven functional modules that operate in a series of coordinated steps with a fixed data flow: a geometry and coordinate system comprehensive judgment module, a multi-physics problem unified judgment module, a volume element adaptive hierarchical selection module, a boundary element order linkage matching module, a solver multi-objective adaptive selection module, an elastic-plastic yield criterion adaptive mapping module, and a prompt information generation and output module. Through the seven-step linkage process from step S1 to step S7, it realizes intelligent parameter recommendation for the entire finite element modeling process.

[0025] The core operating logic is as follows: replacing the traditional single-threshold static decision tree mode, it adopts multi-index weighted comprehensive judgment to complete the matching of geometric attributes and coordinate system, builds a unified framework covering six types of physical fields (structure, fluid, thermal, electromagnetic, acoustic field, seepage) and various coupled working conditions, ensures the consistency of order between volume elements and boundary elements through forced linkage constraints of upstream and downstream parameters, relies on a multi-objective trade-off adaptive optimization solver, combines material property mapping with elastoplastic yield criteria, and finally integrates and outputs standardized prompt text. The entire solution can be realized through secondary development based on existing CAE simulation platforms. All rule logic, geometric analysis, and module interaction are conventional and mature technologies in the fields of computer software and CAE secondary development, and technical personnel in the relevant fields can completely reproduce it without creative labor.

[0026] Geometric multi-feature weighted analysis and coordinate system adaptive matching, corresponding to step S1: Abandoning the rigid judgment logic of traditional single size threshold, it uses a weighted scoring system based on four dimensions: dimensional analysis, axisymmetric fusion discrimination, geometric regularity weighted differentiation, and symmetry feature recognition. This system comprehensively evaluates geometric attributes and then matches the optimal coordinate system from a set of one-dimensional, two-dimensional, and three-dimensional coordinate systems based on the comprehensive score. This avoids coordinate system selection errors caused by misjudgment of a single feature and adapts to complex irregular models.

[0027] Dimensional analysis: By analyzing the topological relationships between the vertices, edges, faces, and volumes of a geometric object, determine whether the model is a one-dimensional rod, a two-dimensional shell, or a three-dimensional solid, and determine the basic dimensions of the coordinate system; Axisymmetry fusion discrimination: Simultaneously verify four characteristics: geometric generatrix shape, cross-sectional uniformity, boundary load distribution, and material distribution. Only when all four meet the circumferential uniformity are they judged as axisymmetric structures, thus preventing misjudgments that are only geometrically symmetrical but have load / material asymmetry. Geometric regularity weighted differentiation: The score is calculated by weighting three indicators: aspect ratio, surface curvature consistency, and side angle deviation. If the deviation is below the threshold, it is judged as regular geometry; otherwise, it is irregular geometry. This provides a preliminary basis for subsequent unit selection. Symmetry feature recognition: Detects plane mirror symmetry and rotational symmetry planes / axis of symmetry, and assists in coordinate system selection and subsequent mesh generation optimization.

[0028] The matched coordinate systems include: one-dimensional x-rectangular coordinate system, r-polar coordinate system, s-spherical coordinate system; two-dimensional xy-plane rectangular coordinate system, xyd-axis symmetric plane coordinate system, rz-axis symmetric cylindrical coordinate system, ro-polar coordinate system, rs-spherical coordinate system, so-spherical coordinate system; three-dimensional xyz-rectangular coordinate system, rzo-cylindrical coordinate system, roz-polar-cylindrical hybrid coordinate system, rso-spherical coordinate system, all of which are standard coordinate systems in the finite element field.

[0029] Geometric feature extraction can directly call commercial geometry kernels, such as ACIS and Parasolid, or the topology query interface of the open-source geometry engine OpenCASCADE to count the number and relationships of vertices, edges, faces, and volumes. Axisymmetry detection, curvature calculation, and symmetry recognition are all mature and common functions in the field of geometry algorithms, and can be implemented by directly calling the built-in functions of the geometry kernel. The weight coefficients of each dimension and the coordinate system mapping rules are pre-stored in the coordinate system weighted matching rule table. The table structure can be stored using a JSON configuration file or a MySQL relational table. The module calls the query through the interface, which is a conventional software development logic that CAE secondary development engineers in this field can directly implement.

[0030] A unified multiphysics analysis framework, corresponding to step S2: Establish a unified judgment framework covering all categories of physical fields, integrate physical field types and subdivided analysis attributes, and combine judgments on simulation problem types of single-field or multi-field coupling, replacing the traditional fragmented mode of independent judgment of a single scene, and adapting to multi-physics field coupling conditions.

[0031] The attributes of each physical field are analyzed in detail below: Structural field: distinguishing between elastic, elastoplastic, viscoelastic, viscoelastic-plastic, rigid-plastic, contact, and large deformation scenarios, as well as steady-state / dynamic scenarios of beams, plates, trusses, and beam-plate combinations; Fluid field: Distinguish between steady-state / transient, single-phase / multiphase flow, and free surface effect scenarios; Electromagnetic fields: distinguish between electrostatic, magnetic, time-harmonic, and transient scenarios; Thermal field: Distinguish between steady-state and transient heat transfer scenarios; Seepage field: distinguish between saturated / unsaturated and steady-state / transient seepage scenarios; Sound field: Adapted for frequency domain / time domain and steady-state / transient acoustic analysis.

[0032] After the user selects the physical field and sub-attributes, the system automatically generates standardized problem type labels and passes them to downstream modules.

[0033] The six physical fields and their sub-attributes are standard classifications for mainstream CAE software such as ANSYS and COMSOL. The system only needs to organize the classification rules into a physical field analysis attribute mapping table and complete the type combination determination through condition judgment logic. Multi-field coupling scenarios can be achieved through multi-attribute superposition matching, which is a conventional back-end data processing logic. No complex algorithm development is required, and it can be directly developed on the basis of the existing simulation platform.

[0034] Multi-constraint adaptive hierarchical selection of solid elements, corresponding to step S3: It adopts a two-level selection logic of basic matching and adaptive upgrading. Based on multiple constraints such as geometric regularity, model dimension, nonlinearity strength and user accuracy requirements, it comprehensively scores and matches volume units of corresponding types and orders, taking into account both computational accuracy and computational efficiency.

[0035] Basic matching logic: For regular geometry, quadrilateral elements are used in 2D and hexahedral elements are used in 3D, resulting in structured meshes and high computational efficiency; for irregular geometry, triangular elements are used in 2D and tetrahedral elements are used in 3D, resulting in unstructured meshes that are suitable for complex curved surfaces. Adaptive upgrade rules: When the comprehensive analysis identifies the behavior as nonlinear elastoplastic, involving contact action, or having a stress concentration sensitive area, or when the user presets a high accuracy level, the system automatically upgrades and recommends second-order elements to improve interpolation accuracy; for conventional linear scenarios, first-order elements are used by default to control computational costs.

[0036] Quadrilaterals, triangles, hexahedrons, tetrahedrons, and first / second-order elements are all built-in elements in the finite element standard element library; selection rules and upgrade thresholds are pre-existing in the geometric regularity-element classification mapping table. After the module reads the geometric regularity score, nonlinearity label, and accuracy requirement parameters, it can look up the table to output the selection results. The branch judgment logic is standard software development content and can be directly connected to the existing CAE element library call interface.

[0037] Forced linkage constraint of volume-boundary element order, corresponding to step S4: A mandatory linkage mechanism for upstream and downstream parameters is established, and the order of boundary elements and volume elements are strictly bound: first-order volume elements are uniformly matched with first-order boundary elements, including four types: points, springs, lines, and surfaces; second-order volume elements are uniformly matched with second-order boundary elements. This ensures that the interpolation accuracy of field variables such as displacement, load, and heat flow is consistent, avoiding numerical oscillations and non-convergence problems caused by order mismatch due to manual selection. It also adapts to the application requirements of various boundary conditions such as distributed loads, displacement constraints, heat flow, and convection.

[0038] The order correspondence between volume elements and boundary elements is pre-existing in the volume-boundary element order linkage constraint table. The boundary element matching module directly receives the order parameters output by the upstream volume element module and can automatically match the corresponding boundary element by looking up the table, without the need for manual secondary configuration. All boundary elements are standard built-in elements of the finite element software, and only the order mapping relationship needs to be established. The technology is mature and the development workload is minimal.

[0039] The multi-objective solver adaptively selects the best solution, corresponding to step S5: With multiple constraints including the linear / nonlinear properties of the problem, the symmetry of the system matrix, the model size, and the ill-conditioned nature of the matrix, and by comprehensively balancing the three objectives of convergence speed, memory usage, and computational accuracy, a classification and optimal matching of direct or iterative solvers is performed. For linear problems with symmetric positive definite matrices: we recommend a symmetric direct solver or a preconditional conjugate gradient (PCG) iterative solver, which balances efficiency and stability. For matrix asymmetric scenarios: We recommend asymmetric direct solvers, multi-wavefront solvers, or GMRES iterative solvers, which are suitable for asymmetric mechanics and electromagnetic equations. For nonlinear, large-scale ill-conditioned problems, we recommend a combination of the Newton-Raphson method with preconditional LU decomposition, over-relaxation iteration, or Gauss-Seidel iteration to address the difficulties in convergence of nonlinear iterations and computational bottlenecks in large models.

[0040] Symmetric direct solvers, GMRES, and the Newton-Raphson method are all commercial / open-source standard solution modules in the finite element field. The solver selection decision logic is organized into a multi-objective optimization evaluation table for solvers. After reading the problem type and matrix characteristic parameters, the module looks up the table and outputs the recommended result. It is a pure branch judgment logic, which does not require complex algorithm development and can be directly connected to the existing solver call interface.

[0041] Material-yield criterion adaptive mapping, corresponding to step S6: The system incorporates a corresponding constraint system for engineering materials and elastoplastic yield criteria. It automatically matches the appropriate yield criterion based on the material category and simultaneously outputs configuration prompts for hardening or softening parameters, taking into account the material's strain softening characteristics. Metallic materials should conform to the Mises yield criterion; Geotechnical materials are matched with the Drucker-Plager yield criterion and the Mohr-Coulomb yield criterion; Concrete material is matched with either the concrete damage-plastic model or the Drucker-Prager model.

[0042] Various yield criteria are classic constitutive models for structural and geotechnical simulations and have been integrated into the material libraries of mainstream CAE software. The correspondence between materials and criteria, as well as parameter prompt text, are stored in the material yield criterion mapping table. The module reads the material type input by the user and then looks up the matching result in the table. This is a static data configuration logic with extremely low development workload.

[0043] The seven modules operate collaboratively in a serial data stream, with upstream and downstream parameters automatically transmitted to form a closed-loop recommendation chain: Geometric and Coordinate System Integrated Judgment Module: The system entry module, which imports geometric models to complete full-dimensional feature recognition and outputs geometric features and coordinate system results. It also includes a unified judgment module for multi-physics problems and a volume element adaptive hierarchical selection module. Multiphysics Problem Unified Judgment Module: Receives physics configuration and coordinate system data, determines the analysis type, and outputs problem type labels to the volume element adaptive hierarchical selection module, the boundary element order linkage matching module, and the solver multi-objective adaptive selection module; Adaptive hierarchical selection module for volume elements: It matches volume elements by combining geometric regularity, dimension and problem type, and outputs the selection results to the boundary element order linkage matching module. Boundary Unit Order Linkage Matching Module: Executes the order-forced linkage rule, matches boundary units of the same order, and outputs the result to the prompt message generation and output module; The solver's multi-objective adaptive optimization module selects solvers based on the characteristics of the problem matrix and the model size, and outputs recommended results. Elastic-plastic yield criterion adaptive mapping module: Matches yield criterion and parameter suggestions according to material type and outputs recommended results; Prompt message generation and output module: Summarizes the output results of all modules, encapsulates them into standardized text, and pushes them to the interactive interface.

[0044] The modules are divided into standard software industry functional breakdowns, and data interaction between modules can be achieved through function calls, interface parameter passing, or message queues, resulting in a clear architecture. All modules can be implemented based on existing CAE software secondary development interfaces, eliminating the need for independent development of the underlying simulation kernel and lowering the deployment threshold.

[0045] Hierarchical constraint matching library underlying architecture: The hierarchical constraint matching library serves as the core basis for end-to-end intelligent recommendation. Unlike traditional static decision trees with a single threshold, it employs a flexible matching logic involving multi-index weighted scoring and the linkage of upstream and downstream parameters. It includes six types of rule tables: coordinate system weighted matching rule table, physical field analysis attribute mapping table, geometric regularity-element hierarchical mapping table, volume-boundary element order linkage constraint table, solver multi-objective optimization evaluation table, and material yield criterion mapping table. Each module calls the corresponding rule table to complete the matching based on its own function, eliminating the need for machine learning model training, making it lightweight and flexibly extensible.

[0046] All rule tables can be stored in Excel configuration files, JSON files, or relational database tables, supporting on-demand editing and updating of rules without modifying the core code. The table structure and query logic are standard database development content, which can be implemented by ordinary software developers in this field, and can be quickly embedded into various self-developed and commercial finite element simulation platforms.

[0047] Detailed Workflow Process 1: System Initialization Phase Start the industrial finite element simulation terminal equipped with this intelligent prompting system, and the system completes a self-test; After loading the built-in hierarchical constraint matching library and reading all six types of rule tables, the seven functional modules enter standby mode. The operator opens the CAE simulation interface and prepares to import the geometric model to be analyzed.

[0048] Process 2: Geometric import and coordinate system analysis and matching, corresponding to step S1: The operator imports a one-dimensional / two-dimensional / three-dimensional geometric model, and the geometry and coordinate system comprehensive analysis module calls the geometry kernel interface to start the analysis. The four detection steps are completed sequentially: dimension assessment, axisymmetric fusion discrimination, regularity weighted calculation, and symmetry feature recognition. The module calls the coordinate system weighted matching rule table, automatically selects the optimal coordinate system based on the comprehensive score, and generates temporary selection prompts. The module synchronously sends geometric feature data and coordinate system names to the multiphysics problem unified judgment module and the volume element adaptive hierarchical selection module.

[0049] Process 3: Physics field configuration and problem type determination, corresponding to step S2: Operators select the physical field type in the interactive interface: single field or multi-field coupling, and set subdivided attributes such as steady state / transient, linear / nonlinear; The unified judgment module for multiphysics problems receives configuration data and coordinate system information, calls the physics analysis attribute mapping table, determines the single-field or coupled analysis type, and generates corresponding prompts. The module distributes standardized problem type labels to the volume element adaptive hierarchical selection module, the boundary element order linkage matching module, and the solver multi-objective adaptive selection module.

[0050] Process 4: Selection of volume element and boundary element linkage, corresponding to steps S3-S4: The adaptive hierarchical selection module for solid elements receives data on geometric regularity, model dimension, and problem nonlinearity. It then calls the geometric regularity-element hierarchical mapping table to perform a comprehensive score and selects the solid element type and order. If the problem is identified as nonlinear, contact, stress concentration, or requires high precision, it automatically upgrades to a second-order element and generates selection suggestions. The module sends the volume element order data to the boundary element order linkage matching module; The boundary element order linkage matching module calls the volume-boundary element order linkage constraint table to automatically match point, spring, line, and surface boundary elements of the same order and generate boundary matching prompts.

[0051] Process 5: Solver and yield criterion matching, corresponding to steps S5-S6: The solver's multi-objective adaptive optimization module reads the problem type, model size, and matrix symmetry attribute, calls the solver's multi-objective optimization evaluation table, filters the optimal solver, and generates selection prompts. The operator enters the engineering materials used in the model, the elastic-plastic yield criterion adaptive mapping module calls the material yield criterion mapping table, matches the corresponding yield criterion, and outputs hardening / softening parameter setting prompts based on the material strain characteristics.

[0052] Process 6: Prompt for integrated output, corresponding to step S7: The prompt information generation and output module summarizes all six types of prompt data: coordinate system, analysis type, volume element, boundary element, solver, and yield criterion. The text is packaged into standardized intelligent prompts according to a preset uniform format and displayed in real time on the simulation interactive interface. Operators can configure finite element modeling parameters with one click or manually, following the prompts.

[0053] Process 7: Simulation Execution and Rule Base Maintenance After the parameters are configured, the finite element solution is started, and the system continuously monitors the status throughout the modeling process. After this simulation is completed, a new model can be imported to repeat the entire process described above. During long-term operation, the rules of the hierarchical constraint matching library can be edited and updated as needed to adapt to more simulation scenarios.

[0054] Typical scenario implementation example: Example 1: Static Elastoplastic Analysis of Metal Components Geometric analysis identifies regular 3D solids and adaptively matches them to the xyz 3D Cartesian coordinate system. It comprehensively identifies metallic materials, static loads, and elastoplastic nonlinear conditions, recommending nonlinear static analysis types. For regular 3D solids without high-precision requirements, first-order hexahedral C8 elements are preferentially recommended. For first-order solid elements, Q4 quadrilateral surface elements are used for linkage matching. For asymmetric stiffness matrices, an adaptive selection of an asymmetric direct solver or a Newton-Raphson method combined with a preconditioned LU decomposition iterative solver is used. For metallic materials, the Mises yield criterion is adaptively mapped.

[0055] System output prompt: Intelligent modeling hints: Coordinate system: xyz three-dimensional rectangular coordinate system (recommended for comprehensive analysis of multiple geometric features) Analysis type: static_nonlinear (nonlinear static analysis, multiphysics comprehensive judgment) Solid element: C8 (first-order hexahedron, multi-constraint adaptive optimization) Boundary element: Q4 quadrilateral surface element (order-linked constraint matching) Solver: Asymmetric direct solver (multi-objective adaptive optimization) Yield Criterion: Mises (Adaptive Mapping Matching of Material Properties) Example 2: Thermal-structural coupling analysis of axisymmetric thick-walled cylinder: Geometric analysis identifies axisymmetric structures rotating around the Z-axis, and adaptively selects the rz two-dimensional axisymmetric coordinate system; comprehensive assessment determines the coupled working conditions of steady-state heat conduction and linear elastic structure, recommending linear static and steady-state thermal coupling problem types; for regular two-dimensional regions, Q4 first-order quadrilateral elements are selected as the optimal choice, and L2 first-order linear elements are used in conjunction to apply temperature and convection boundary conditions; the system's total stiffness matrix is ​​symmetric and positive definite, and an adaptive selection of a symmetric direct solver or a PCG iterative solver is made; linear elastic analysis has no yield criterion.

[0056] Example 3: Contact Analysis of Large Deformation of Soil and Rock Slopes Geometric analysis identifies irregular 3D soil and rock masses and matches them to a 3D rectangular coordinate system (xyz). It comprehensively identifies soil and rock materials, contact conditions, and large deformation nonlinear working conditions, recommending nonlinear static analysis types. For irregular geometries with stress concentration, it adaptively upgrades and recommends W10 second-order tetrahedral elements, linked to T6 triangular surface elements. For ill-conditioned stiffness matrices, it adaptively selects either a multi-wavefront direct solver or a GMRES iterative solver. It adaptively maps the Mohr-Coulomb yield criterion for soil and rock materials and prompts the use of strain softening parameters.

[0057] Example 4: Steady-state seepage-stress coupling analysis of earth-rock dam: Geometric analysis identifies two-dimensional irregular cross-sections, using an xy-plane rectangular coordinate system. A comprehensive assessment of the coupled steady-state seepage and linear elastic stress conditions in saturated soil is conducted, recommending a coupled steady-state seepage and linear static analysis type. For the two-dimensional irregular region, the optimal Q4 element with pore pressure degree of freedom is selected, and pore pressure and displacement constraints are applied using a linked matching L2 linear element. The matrix is ​​symmetric and positive definite, and an adaptively optimized symmetric direct solver is used. Given that the material is saturated soil, the permeability coefficient and Biot modulus are configured, and the seepage-stress coupling option is activated.

[0058] Example 5: Analysis of Incompressible Steady-State Fluid Flow in a Pipeline: Geometric analysis identifies 3D regular pipes, using an xyz 3D Cartesian coordinate system; comprehensive judgment is made for steady-state, single-phase, and incompressible fluid without free surfaces, with steady-state fluid analysis type recommended; high fluid simulation accuracy is required, adaptive upgrade recommends C27 second-order hexahedral elements, and boundary linkage is matched with Q9 second-order surface elements; the solution matrix is ​​symmetric positive definite, and the PCG iterative solver is adaptively selected, with synchronous prompts to set the turbulence model and velocity-pressure coupling algorithm.

[0059] Example 6: Analysis of Electromagnetic Induction Heating: Geometric analysis identifies two-dimensional axisymmetric induction coil workpieces, using the rz two-dimensional axisymmetric coordinate system. A comprehensive assessment of the time-harmonic electromagnetic field and transient thermal nonlinear coupling condition is conducted, recommending analysis of the time-harmonic electromagnetic and transient thermal coupling. For regular geometry, Q9 second-order quadrilateral elements are selected, with boundary linkage matching using L2 line elements to apply electromagnetic excitation and convective boundaries. The matrix is ​​complex symmetric and non-positive definite; an adaptive PARDISO direct method solver is used, with simultaneous prompts for skin depth, material conductivity, and thermal conductivity parameter configurations.

[0060] Example 7: Sound Field-Structure Coupling Analysis: Geometric analysis identifies three-dimensional regular fluid cavities and elastic walls, using a three-dimensional rectangular coordinate system (xyz). A comprehensive assessment of the frequency domain steady-state linear pressure acoustics and structural linear elastic coupling conditions is conducted, recommending a frequency domain acoustic-structure interaction analysis type. The fluid domain adaptively selects a C27 ​​second-order hexahedron, and the structural domain selects a C8 first-order hexahedron. The coupling interface is matched with Q9 second-order surface elements to achieve sound pressure and displacement coupling. Due to matrix asymmetry, the UMFPACK asymmetric direct solver is adaptively selected.

[0061] The above embodiments cover structural, thermal, seepage, fluid, electromagnetic, acoustic field, and multi-field coupling scenarios, fully verifying the versatility and adaptability of the system's multi-feature comprehensive judgment, adaptive optimization, and parameter linkage matching. Those skilled in the art can expand the rule base content according to different engineering scenarios, all of which fall within the protection scope of this invention.

[0062] Although embodiments of the invention have been shown, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multiphysics finite element intelligent prompting method, characterized in that, Includes the following steps: S1: Import the geometric model to be analyzed, extract multi-dimensional topological, surface and symmetric geometric features, use a multi-index weighted comprehensive scoring method to judge geometric attributes, and adaptively match the optimal coordinate system in one-dimensional, two-dimensional and three-dimensional coordinate systems based on geometric shape and symmetry features, and generate coordinate system selection prompts. S2: By using a unified multiphysics analysis framework, the simulation problem type is determined by combining the physical field type and analysis attributes. The physical fields include structural, fluid, thermal, electromagnetic, acoustic, and seepage fields. The analysis attributes include static / dynamic, linear / nonlinear, and steady-state / transient, and corresponding analysis type recommendations are output. S3: Based on a comprehensive score considering multiple constraints such as geometric regularity, model dimension, nonlinearity strength, and user accuracy requirements, the system adaptively selects the corresponding volume element and first-order or second-order element order, and generates volume element selection suggestions. S4: Establish an upstream and downstream forced linkage constraint mechanism. Based on the selected volume element type and order, match the corresponding order of point, spring, line, and surface boundary elements, adapt to distributed load, displacement constraint, heat flow, and convection boundary conditions, and generate boundary element matching prompts. S5: Taking multiple constraints such as linear or nonlinear properties of the problem, symmetry of the system matrix, model size, and ill-conditioned matrix, and comprehensively balancing multiple objectives such as convergence speed, memory usage, and computational accuracy, it adaptively selects either a direct method or an iterative method solver and generates solver selection prompts. S6: Adaptively map and match the corresponding elastoplastic yield criterion according to the material type, and generate hardening or softening parameter setting prompts according to the material strain softening characteristics; S7: Integrate all the above-mentioned linkage analysis and prompt information, and output standardized intelligent prompt text in a unified format.

2. The multiphysics finite element intelligent prompting method according to claim 1, characterized in that, Step S1, the geometric attribute assessment, specifically includes four weighted evaluation dimensions: Dimensional analysis: Based on the topological relationship characteristics of the vertices, edges, faces, and volumes of the geometric object, determine whether it is a one-dimensional linear model, a two-dimensional shell model, or a three-dimensional solid model; Axisymmetry fusion judgment: The structure is judged to be axisymmetric when it meets the requirements of rotational coincidence and circumferential property uniformity by a comprehensive evaluation of multiple characteristics including geometric generatrix shape, cross-sectional uniformity, boundary load distribution, and material distribution. Geometric regularity weighted distinction: The regularity score is calculated by weighting multiple indicators such as aspect ratio, surface curvature consistency, and side angle deviation. If the deviation is less than the preset threshold, it is judged as regular geometry; otherwise, it is irregular geometry. Symmetry feature recognition: Detect the symmetry plane / axis corresponding to the planar mirror symmetry and rotational symmetry of the model, as a prerequisite constraint for coordinate system optimization; The coordinate systems include a one-dimensional x-cartesian coordinate system, a r-polar coordinate system, and a spherical coordinate system; a two-dimensional xy-plane cartesian coordinate system, an xyd-axis symmetric plane coordinate system, an rz-axis symmetric cylindrical coordinate system, a ro-polar coordinate system, a rs-spherical coordinate system, and a so-spherical coordinate system; and a three-dimensional xyz-cartesian coordinate system, a rzo-cylindrical coordinate system, a roz-polar-cylindrical hybrid coordinate system, and a rso-spherical coordinate system, which are automatically selected and matched based on geometric comprehensive characteristics.

3. The multiphysics finite element intelligent prompting method according to claim 1, characterized in that, The physical field types mentioned in step S2 include structural field, fluid field, thermal field, electromagnetic field, acoustic field, and seepage field; The analytical attributes are further subdivided by physical field as follows: The structural field is divided into elastic, elastoplastic, viscoelastic, viscoelastic-plastic, rigid-plastic, contact, and large deformation scenarios, as well as steady-state and dynamic scenarios of beams, plates, trusses, and beam-plate combinations. Fluid fields are distinguished between steady-state and transient flows, single-phase and multiphase flows, and free surface scenarios; Electromagnetic fields can be categorized into electrostatic, magnetostatic, time-harmonic, and transient scenarios. Thermal fields distinguish between steady-state and transient scenarios; The seepage field is distinguished into saturated and unsaturated, steady-state and transient seepage scenarios; The type of analysis problem is determined by the combination of physical field type and subdivided analysis attributes.

4. The multiphysics finite element intelligent prompting method according to claim 1, characterized in that, The adaptive hierarchical selection rule for volume elements in step S3 is as follows: Basic matching logic: For regular geometry, quadrilateral elements are used in 2D and hexahedral elements are used in 3D; for irregular geometry, triangular elements are used in 2D and tetrahedral elements are used in 3D. Adaptive upgrade rules: When the system is identified as nonlinear elastoplastic, involves contact action, has a stress concentration sensitive area, or the user presets a high accuracy level, the system will automatically upgrade and recommend second-order elements. In other scenarios, first-order elements will be recommended by default.

5. The multiphysics finite element intelligent prompting method according to claim 1, characterized in that, In step S4, the boundary elements and volume elements employ an order-forced matching mechanism: When the selected volume element is a first-order element, it matches first-order boundary elements, including point, spring, line, and surface boundary elements. When the selected volume element is a second-order element, match the second-order boundary element; Boundary elements are adapted to the application requirements of distributed loads, displacement constraints, heat flow, and convection boundary conditions.

6. The multiphysics finite element intelligent prompting method according to claim 1, characterized in that, The matching decision logic of the solver in step S5 is as follows: For linear problems with symmetric positive definite system matrices, a symmetric direct solver or a preconditional conjugate gradient iterative solver is recommended. For scenarios where the system matrix is ​​asymmetric, we recommend the asymmetric direct solver, the multi-wavefront solver, or the GMRES iterative solver. For nonlinear or large-scale ill-conditioned problem scenarios, it is recommended to use a solver that combines the Newton-Raphson method with preconditioned LU decomposition, over-relaxation iteration, or Gauss-Seidel iteration.

7. The multiphysics finite element intelligent prompting method according to claim 1, characterized in that, The matching rules for the elastoplastic yield criterion in step S6 are as follows: Metallic materials should conform to the Mises yield criterion; Geotechnical materials are matched with the Drucker-Plager yield criterion and the Mohr-Coulomb yield criterion; Concrete material matching: concrete damage-plastic model or Drucker-Prager model; At the same time, depending on whether the material contains strain softening characteristics, corresponding setting prompts for hardening parameters or softening parameters will be generated.

8. A multiphysics finite element intelligent prompting system applying the multiphysics finite element intelligent prompting method according to any one of claims 1-7, characterized in that, It includes a comprehensive judgment module for geometry and coordinate systems, a unified judgment module for multi-physics problems, an adaptive hierarchical selection module for volume elements, a boundary element order linkage matching module, a multi-objective adaptive selection module for solvers, an adaptive mapping module for elastoplastic yield criteria, and a prompt information generation and output module. The geometry and coordinate system integrated judgment module outputs geometric features and coordinate system results to the multiphysics problem unified judgment module and the volume element adaptive hierarchical selection module. The multiphysics problem unified judgment module outputs problem type labels to the volume element adaptive hierarchical selection module, the boundary element order linkage matching module, and the solver multi-objective adaptive selection module. The volume element adaptive hierarchical optimization module outputs the volume element selection result to the boundary element order linkage matching module; The outputs of the boundary element order linkage matching module, the solver multi-objective adaptive selection module, and the elastic-plastic yield criterion adaptive mapping module are all connected to the prompt information generation and output module. The prompt information generation and output module integrates all the analysis results and outputs standardized intelligent prompt text.

9. The multiphysics finite element intelligent prompting system and method according to claim 8, characterized in that, It also includes a hierarchical constraint matching library, which includes a coordinate system weighted matching rule table, a physical field analysis attribute mapping table, a geometric regularity-element hierarchical mapping table, a volume-boundary element order linkage constraint table, a solver multi-objective optimization evaluation table, and a material yield criterion mapping table.

Images