Simulation configuration generation method and system based on physical-numerical constraint feasible region

Abstract

The invention provides a simulation configuration generation method and system based on a physical-numerical constraint feasible region. The method comprises the following steps: constructing a unified simulation parameter vector according to a continuous parameter set and a discrete parameter set; calculating a key dimensionless physical quantity according to the continuous parameter set; constructing a physical consistency constraint set and a numerical stability constraint set, and constructing a feasible region according to the physical consistency constraint set and the numerical stability constraint set; generating an initial candidate parameter vector from a natural language or a semi-structured engineering demand input by a user through semantic analysis and parameter extraction; judging whether the initial candidate parameter vector belongs to a feasible region or not; if yes, generating a simulation configuration file; if not, executing a hierarchical projection correction strategy; intelligent generation of simulation configuration is achieved, the success rate of one-time starting is remarkably increased, the number of manual parameter adjustment times is reduced, dependence of simulation analysis on personal experience of engineers is reduced, and efficient and reliable technical support is provided for numerical simulation of complex engineering problems.

Description

Technical Field

[0001] This invention relates to the field of automation and intelligent modeling technology, and in particular to a method and system for generating simulation configurations based on physical-numerical constraint feasible regions. Background Technology

[0002] With the development of computational fluid dynamics and multiphysics coupling simulation technology, simulation software is increasingly widely used in engineering design and scientific research. However, the simulation configuration process involves a large number of coupling parameters, including physical model selection, mesh size setting, time step control, solver type selection, and boundary condition setting. These parameters have complex physical dependencies and numerical stability constraints.

[0003] In existing technologies, simulation configuration typically relies on manual experience or rule enumeration. Engineers need to repeatedly consult theoretical manuals and software documentation, and adjust parameter combinations through multiple trial calculations, depending on the specific problem. This process is time-consuming, labor-intensive, and highly dependent on personal experience. For complex multiphysics coupling problems, the parameter space grows exponentially, and manual configuration can hardly account for all constraints.

[0004] Although there have been attempts in recent years to use artificial intelligence to automatically generate configuration files, existing methods have the following obvious shortcomings: First, most methods only focus on semantic matching or template replacement, mapping user-input natural language requirements to preset configuration templates, without providing a unified model for physical consistency and numerical stability. The generated configuration files may be semantically reasonable, but they diverge in numerical computation due to violations of stability conditions.

[0005] Second, some methods that consider constraints employ a "hard rule check" approach, which directly reports an error or replaces a parameter with a preset fixed value when the parameter does not meet the rules. This approach lacks flexibility and cannot make minimal corrections while maintaining the user's original intent, resulting in a significant deviation between the generated configuration and the user's expectations.

[0006] Third, existing technologies fail to effectively handle joint optimization problems involving a mixture of continuous and discrete parameters, often treating the two types of parameters separately, which cannot guarantee the global optimality of the final configuration. Summary of the Invention

[0007] In view of this, the purpose of this invention is to provide a simulation configuration generation method and system based on the feasible domain of physical-numerical constraints, so as to realize the intelligent generation of simulation configuration, significantly improve the first-time start-up success rate, reduce the number of manual parameter tunings, reduce the dependence of simulation analysis on the personal experience of engineers, and provide efficient and reliable technical support for the numerical simulation of complex engineering problems.

[0008] In a first aspect, embodiments of the present invention provide a simulation configuration generation method based on a physical-numerical constraint feasible region, the method comprising: Obtain the continuous parameter set and the discrete parameter set; A unified simulation parameter vector is constructed based on the continuous parameter set and the discrete parameter set; The key dimensionless physical quantities are calculated based on the continuous parameter set; wherein, the key dimensionless physical quantities include Reynolds number, Mach number, CFL number and wall value; Construct a set of physical consistency constraints, wherein the set of physical consistency constraints includes flow state and model selection constraints, turbulence model parameter constraints, boundary conditions and physical field constraints; Construct a set of numerical stability constraints, which includes time step stability constraints, grid resolution constraints, and iterative convergence constraints. Based on the set of physical consistency constraints and the set of numerical stability constraints, a feasible region is constructed. Obtain user input in natural language or semi-structured engineering requirements; The natural language or semi-structured engineering requirements are semantically parsed and parameters are extracted to generate an initial candidate parameter vector; Determine whether the initial candidate parameter vector belongs to the feasible region; If it is found to be true, a simulation configuration file is generated and imported into mainstream simulation software to perform calculations. If it does not belong to the category, then the hierarchical projection correction strategy is executed.

[0009] Furthermore, the continuous parameter set includes geometric feature length, feature velocity, fluid density, dynamic viscosity, temperature, mesh feature size, time step, and dimensionless wall distance; The set of discrete parameters includes the turbulence model, solver, energy equation options, and boundary condition types.

[0010] Furthermore, the flow state and model selection constraints are implemented in the following ways: When the Reynolds number is greater than or equal to the turbulence transition threshold, the turbulence model is enabled; When the Mach number is greater than or equal to a preset threshold, a compressible solver is selected and the energy equation is enabled; When the Mach number is less than a preset threshold, an incompressible solver is selected.

[0011] Furthermore, the turbulence model parameter constraints are implemented in the following way: Obtain the LES analytical scale and minimum number of grid cells; When the LES model is selected, the mesh feature size is less than or equal to the ratio of the LES analytical scale to the minimum number of meshes; When a RANS model is selected, the wall values ​​satisfy the effective range of the model; The boundary conditions and physical field constraints are implemented in the following ways: Define thermal property parameters when a temperature gradient exists or the energy equation is enabled; When dealing with multiphase flow problems, an interphase force model is defined.

[0012] Furthermore, the time step stability constraint includes explicit and implicit formats; The grid resolution constraint includes both overall resolution and local resolution; The iterative convergence constraint is that the ratio of the norm of the residual vector to the norm of the initial residual vector is less than or equal to the convergence tolerance.

[0013] Furthermore, the hierarchical projection correction strategy is implemented in the following way: For fixed continuous variables, use enumeration or heuristic search algorithms in the discrete variable space to search for feasible discrete combinations that are closest to the original discrete choices, and use the feasible discrete combinations as discrete variables. Based on differentiable constraint functions, a constrained gradient descent method is used to project the parameters back to the boundary of the feasible region after each iteration; alternatively, a Lagrange function is constructed, and the optimal solution is gradually approximated by solving a quadratic programming subproblem. If a feasible solution cannot be found after hierarchical optimization, the restrictions on the modification of the discrete variables are relaxed, and discrete-continuous joint optimization is performed until a feasible solution is found or no solution is output. By performing the above hierarchical correction, the optimal correction parameters that satisfy all constraints are obtained.

[0014] Furthermore, the method also includes: Define a stability scoring function; When the stability scoring function is less than a preset threshold, the system automatically triggers a fine-tuning mechanism. During the fine-tuning process, it continues to iterate until the stability scoring function is greater than or equal to the preset threshold, or the maximum number of iterations is reached.

[0015] Secondly, embodiments of the present invention provide a simulation configuration generation system based on a physical-numerical constraint feasible region, the system comprising: The parameter set acquisition module is used to acquire continuous parameter sets and discrete parameter sets; The simulation parameter vector construction module is used to construct a unified simulation parameter vector based on the continuous parameter set and the discrete parameter set; The calculation module is used to calculate key dimensionless physical quantities based on the continuous parameter set; wherein, the key dimensionless physical quantities include Reynolds number, Mach number, CFL number and wall value; The physical consistency constraint set construction module is used to construct a physical consistency constraint set, wherein the physical consistency constraint set includes flow state and model selection constraints, turbulence model parameter constraints, boundary conditions and physical field constraints; A numerical stability constraint set construction module is used to construct a numerical stability constraint set, which includes time step stability constraints, grid resolution constraints, and iterative convergence constraints. A feasible region construction module is used to construct a feasible region based on the set of physical consistency constraints and the set of numerical stability constraints. The engineering requirements acquisition module is used to acquire natural language or semi-structured engineering requirements input by the user. The generation module is used to generate an initial candidate parameter vector by semantic parsing and parameter extraction of the natural language or semi-structured engineering requirements; The judgment module is used to determine whether the initial candidate parameter vector belongs to the feasible region; if it does, a simulation configuration file is generated and imported into mainstream simulation software for calculation; if it does not, a hierarchical projection correction strategy is executed.

[0016] Thirdly, embodiments of the present invention provide an electronic device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the method described above.

[0017] Fourthly, embodiments of the present invention provide a computer-readable medium having processor-executable non-volatile program code that causes the processor to perform the method described above.

[0018] This invention provides a simulation configuration generation method and system based on a physics-numerical constraint feasible region, including: obtaining a continuous parameter set and a discrete parameter set; constructing a unified simulation parameter vector based on the continuous parameter set and the discrete parameter set; calculating key dimensionless physical quantities based on the continuous parameter set; wherein the key dimensionless physical quantities include Reynolds number, Mach number, CFL number, and wall value; constructing a set of physical consistency constraints, wherein the set of physical consistency constraints includes flow state and model selection constraints, turbulence model parameter constraints, boundary conditions and physical field constraints; and constructing a set of numerical stability constraints, wherein the set of numerical stability constraints includes time step stability constraints, grid resolution constraints, and iterative convergence constraints. The system constructs a feasible region based on the set of physical consistency constraints and the set of numerical stability constraints; it obtains user-inputted natural language or semi-structured engineering requirements; it generates an initial candidate parameter vector by semantic parsing and parameter extraction of the natural language or semi-structured engineering requirements; it determines whether the initial candidate parameter vector belongs to the feasible region; if it does, it generates a simulation configuration file and imports it into mainstream simulation software for calculation; if it does not, it executes a hierarchical projection correction strategy; this system achieves intelligent generation of simulation configurations, significantly improving the first-time startup success rate, reducing the number of manual parameter tunings, and reducing the reliance of simulation analysis on engineers' personal experience, providing efficient and reliable technical support for numerical simulation of complex engineering problems.

[0019] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0021] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0022] Figure 1 The flowchart shows the simulation configuration generation method based on the feasible region of physical-numerical constraints provided in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram of a simulation configuration generation system based on a physical-numerical constraint feasible region provided in Embodiment 1 of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] To facilitate understanding of this embodiment, the embodiments of the present invention will be described in detail below.

[0025] Example 1: Figure 1 The flowchart shows the simulation configuration generation method based on the feasible region of physical-numerical constraints provided in Embodiment 1 of the present invention.

[0026] Reference Figure 1 The method includes the following steps: Step S101: Obtain the continuous parameter set and the discrete parameter set; Step S102: Construct a unified simulation parameter vector based on the continuous parameter set and the discrete parameter set; Step S103: Calculate the key dimensionless physical quantities based on the continuous parameter set; wherein, the key dimensionless physical quantities include Reynolds number, Mach number, CFL number and wall value; Step S104: Construct a set of physical consistency constraints, which includes flow state and model selection constraints, turbulence model parameter constraints, boundary conditions and physical field constraints. Step S105: Construct a set of numerical stability constraints, which includes time step stability constraints, grid resolution constraints, and iterative convergence constraints. Step S106: Construct the feasible region based on the set of physical consistency constraints and the set of numerical stability constraints; Step S107: Obtain the user's input of natural language or semi-structured engineering requirements; Step S108: Generate an initial candidate parameter vector by semantic parsing and parameter extraction of natural language or semi-structured engineering requirements; Step S109: Determine whether the initial candidate parameter vector belongs to the feasible region; Step S110: If it belongs to the category, generate a simulation configuration file and import the simulation configuration file into mainstream simulation software to perform calculations. Step S111: If it does not belong to the category, then execute the hierarchical projection correction strategy.

[0027] This application aims to achieve intelligent generation of simulation configurations, significantly improve the first-time startup success rate, reduce the number of manual parameter adjustments, reduce the reliance of simulation analysis on engineers' personal experience, and provide efficient and reliable technical support for numerical simulation of complex engineering problems.

[0028] Furthermore, the continuous parameter set includes geometric feature length, feature velocity, fluid density, dynamic viscosity, temperature, mesh feature size, time step, and dimensionless wall distance; The set of discrete parameters includes turbulence models, solvers, energy equation options, and boundary condition types.

[0029] Specifically, a unified simulation parameter vector is first constructed, incorporating continuous physical parameters and discrete solution parameters into the same mathematical space. Let the simulation parameter vector be:

[0030] in, Represents a continuous set of parameters. This represents a discrete set of parameters. Taking a typical computational fluid dynamics problem as an example, the parameters can be specifically represented as:

[0031] Where L is the geometric characteristic length, in meters (m); and U is the characteristic flow velocity, in meters per second (m / s). ρ is the fluid density, measured in kilograms per cubic meter (kg / m³); μ is the dynamic viscosity, measured in Pa·s. T represents temperature, measured in Kelvin (K); Δx represents the mesh feature size, measured in meters (m). Δt is the time step, in seconds (s); y+ is the dimensionless wall distance; The type of turbulence model is a discrete variable, and its values ​​can be such as non-turbulent model, standard k-ε model, SST k-ω model, LES model, etc. The solver type is a discrete variable, and its values ​​can be taken as pressure-based solver or density-based solver. These are options for the energy equation, representing discrete variables, with values ​​such as enabled or disabled. It is a boundary condition type, and it consists of discrete variables.

[0032] After obtaining or completing the above parameters based on the input requirements, the system automatically calculates the key dimensionless physical quantities: The Reynolds number Re is defined as:

[0033] The Mach number Ma is defined as: Where c represents the speed of sound in the fluid, for an ideal gas: In the formula, γ is the specific heat ratio and R is the gas constant.

[0034] For explicit time-progression formats, the CFL number is defined as: For implicit formats, appropriate stability criteria can be used, such as Fourier numbers or von Neumann stability analysis results.

[0035] Wall value Estimate: in, Let y be the friction velocity and y be the height of the first layer of mesh.

[0036] Furthermore, the flow state and model selection constraints are implemented in the following ways: When the Reynolds number is greater than or equal to the turbulence transition threshold, the turbulence model is enabled; When the Mach number is greater than or equal to a preset threshold, select the compressible solver and enable the energy equation; When the Mach number is less than a preset threshold, the incompressible solver is selected.

[0037] Furthermore, the turbulence model parameter constraints are achieved in the following ways: Obtain the LES analytical scale and minimum number of grid cells; When selecting the LES model, the mesh feature size is less than or equal to the ratio of the LES analytical scale to the minimum number of meshes; When a RANS model is selected, the wall values ​​meet the effective range of the model. Boundary conditions and physical field constraints are implemented in the following ways: Define thermal property parameters when a temperature gradient exists or the energy equation is enabled; When dealing with multiphase flow problems, an interphase force model is defined.

[0038] Specifically, a physical consistency constraint set P is constructed, which includes model dependency logic constraints and parameter value range constraints, specifically manifested as follows: (1) Flow state and model selection constraints: If Re ≥ Re_turb (turbulence transition threshold, usually taken as 5000), then the turbulence model is enabled, i.e. M_turb ≠None; If Ma ≥ 0.3, then select the compressible solver, i.e., S_solver = density basis solver, and enable the energy equation E_eqn = True; If Ma < 0.3, the incompressible solver can be selected, and the energy equation can be enabled or disabled depending on the temperature change.

[0039] (2) Parameter constraints of the turbulence model: If the LES model is chosen, then the following condition must be met: Δx ≤ L_les / N_les_min, where L_les is the LES analytical scale and N_les_min is the minimum number of grid cells; If the RANS model is selected, the wall value The model needs to meet the effective range, such as the standard wall function requirement of 30 < <300, while low Reynolds number models require ≈ 1.

[0040] (3) Boundary conditions and physical field constraints: If a temperature gradient exists or the energy equation is enabled, then define the thermal properties (such as thermal conductivity and specific heat capacity). If it is a multiphase flow problem, then define an interphase force model; The above constraints are expressed in the form of logical predicates: P(θ) = True indicates that all physical consistency conditions are true.

[0041] Furthermore, time step stability constraints include explicit and implicit formats; Mesh resolution constraints include global resolution and local resolution; The iterative convergence constraint is that the ratio of the norm of the residual vector to the norm of the initial residual vector is less than or equal to the convergence tolerance.

[0042] Specifically, a set of numerical stability constraints N is constructed, which is expressed in the form of inequalities to ensure that the numerical computation process does not diverge: (1) Time step stability constraint: For explicit formats: , Typically, the value is taken as 0.5~1.0; For implicit formats: corresponding stability conditions can be defined based on format characteristics, or precision constraints can be used instead of stability constraints.

[0043] (2) Mesh resolution constraint: Overall resolution: ,in, This represents the minimum number of grid divisions to ensure that basic flow characteristics can be captured. Local resolution: For the boundary layer, The specific value depends on the selected turbulence model.

[0044] (3) Iterative convergence constraint: Where R(θ) is the residual vector, This is for convergence tolerance.

[0045] The above constraints constitute a system of inequalities: ,in, For constraint functions.

[0046] Construct a joint feasible region Ω, which is defined as the set of parameters that simultaneously satisfies both the set of physical consistency constraints and the set of numerical stability constraints:

[0047] Since the parameter space contains both continuous and discrete variables, Ω is a mixed integer feasible region, whose boundary is defined by nonlinear inequalities and discontinuous logical conditions. This is one of the key features that distinguishes this invention from existing technologies.

[0048] The system receives user input in natural language or semi-structured engineering requirements, and generates an initial candidate parameter vector θ_candidate through semantic parsing and parameter extraction. If the user input is incomplete, the system completes it based on physical common sense and default empirical values, such as the default density of air being 1.225 kg / m³ and the default viscosity being 1.8 × 10⁻⁶. -5 Pa·s.

[0049] The system determines whether θ_candidate belongs to the feasible region Ω. If θ_candidate satisfies all constraints, it proceeds directly to step nine to generate the simulation configuration file; otherwise, it proceeds to step eight to perform constraint projection correction.

[0050] Furthermore, the layered projection correction strategy is implemented in the following way: Step S201: Fix the continuous variables, and use an enumeration method or heuristic search algorithm in the discrete variable space to search for feasible discrete combinations that are closest to the original discrete choices, and use the feasible discrete combinations as discrete variables. Step S202: Based on the differentiable constraint function, a constrained gradient descent method is used to project the parameters back to the boundary of the feasible region after each iteration; or, a Lagrange function is constructed, and the optimal solution is gradually approximated by solving the quadratic programming subproblem. Step S203: If a feasible solution still cannot be found after hierarchical optimization, the restriction on the modification of the discrete variables is relaxed, and discrete-continuous joint optimization is performed until a feasible solution is found or no solution information is output. Step S204: By performing the above hierarchical correction, the optimal correction parameters that satisfy all constraints are obtained.

[0051] Specifically, when the candidate parameters do not satisfy the Ω condition, this application proposes a hierarchical projection correction strategy, which maps the candidate parameters to the feasible region with minimal changes by solving the following mixed-variable optimization problem: Objective function:

[0052] Constraints:

[0053] Where ||·|| represents the Euclidean norm, and d(·,·) represents the distance metric of discrete variables (such as Hamming distance). and The weighting coefficients reflect the user's preference for modifying continuous and discrete parameters.

[0054] This optimization problem is solved using a hierarchical iterative strategy: First level (discrete variable correction): fixed continuous variables The algorithm searches the space of discrete variables for discrete combinations that satisfy P(θ) = True. Since the space of discrete variables is usually finite, enumeration or heuristic search (such as genetic algorithms) can be used to find feasible discrete combinations that are closest to the original discrete choices.

[0055] The second layer (continuous variable correction): After the discrete variables are determined, the problem is transformed into constrained optimization of continuous variables. One of the following two methods can be used: Method A (Gradient Projection Method): For differentiable constrained functions, a constrained gradient descent method is used, projecting the parameters back to the feasible region boundary after each iteration. Method B (Sequential Quadratic Programming): Construct a Lagrangian function and gradually approximate the optimal solution by solving quadratic programming subproblems. The third layer (joint optimization): If a feasible solution cannot be found after layer optimization, the restrictions on the modification of discrete variables are relaxed, and discrete-continuous joint optimization is performed until a feasible solution is found or no solution is output.

[0056] Through the above hierarchical correction, the optimal correction parameter θ* that satisfies all constraints is obtained. This parameter preserves the user's original intent as much as possible while ensuring feasibility.

[0057] Furthermore, the method also includes the following steps: Step S301: Define the stability scoring function; Step S302: When the stability scoring function is less than the preset threshold, the system automatically triggers the fine-tuning mechanism. During the fine-tuning process, it continues to iterate until the stability scoring function is greater than or equal to the preset threshold, or the maximum number of iterations is reached.

[0058] Specifically, in stability scoring and adaptive fine-tuning, to further improve simulation robustness, this application defines a stability scoring function. :

[0059] Where α, β, and γ are weighting coefficients, each reflecting the CFL margin, Reynolds number fit margin, and wall resolution margin, respectively. Below the preset threshold At that time, the system automatically triggers the fine-tuning mechanism: First, adjust the time step Δt, with an adjustment factor of 0.5~0.8; if the time step adjustment still does not meet the requirements, then adjust the mesh size Δx; if the mesh adjustment leads to excessive computational cost, then prompt the user to change the turbulence model.

[0060] The fine-tuning process continues iteratively until... ≥ Or it may reach the maximum number of iterations.

[0061] The system automatically generates a simulation configuration file based on the parameter vector θ* that satisfies the Ω condition and achieves the required stability score. This file includes: governing equation settings (e.g., Navier-Stokes equations, energy equations, component transport equations, etc.); mesh parameters (mesh size, boundary layer mesh requirements, mesh type, etc.); time progression parameters (time step, total computation time, time format, etc.); solver parameters (solver type, discretization scheme, relaxation factor, convergence criterion, etc.); physical model parameters (turbulence model, multiphase flow model, combustion model, etc.); boundary conditions (inlet boundary, outlet boundary, wall conditions, etc.); initial conditions (flow field initialization parameters); and output control (monitoring point settings, data output frequency, etc.).

[0062] The generated configuration file can be directly imported into mainstream simulation software (such as ANSYS Fluent, OpenFOAM, CFX, STAR-CCM+, etc.) to perform calculations.

[0063] Compared with the prior art, the present invention has the following significant technical effects: First, it achieves unified modeling of physical model selection and numerical stability conditions. This invention is the first to incorporate physical consistency constraints (logical rules) and numerical stability constraints (inequality systems) into the same mathematical framework, constructing a mixed integer feasible region Ω, thus solving the problem of separate handling of the two types of constraints in traditional methods.

[0064] Second, by constructing a feasible region, we avoid generating parameter combinations that result in physical conflicts or numerical divergences. Any generated simulation configuration must undergo rigorous verification by Ω to ensure that it is physically reasonable and numerically computable, fundamentally eliminating calculation failures caused by inappropriate parameters.

[0065] Third, the method reduces the number of manual parameter adjustments through constrained projection correction. When candidate parameters do not meet the constraints, the layered projection correction strategy of this invention automatically adjusts the parameters with minimal changes, freeing engineers from tedious manual debugging. In typical application scenarios, the number of manual parameter adjustments is reduced by more than 80%.

[0066] Fourth, it significantly improves the first-time startup success rate of simulations. Experiments show that the simulation configuration generated using this method increases the first-time startup success rate in complex turbulent problems from less than 40% with traditional methods to over 85%, greatly improving the efficiency of engineering analysis.

[0067] Fifth, improve numerical convergence stability. By adaptively fine-tuning the stability scoring function S(θ), the residual oscillation amplitude during the calculation process is reduced by 60%, and the number of iterations required to achieve convergence is reduced by 30% to 50%.

[0068] Sixth, excellent scalability. The method framework of this invention is not limited to computational fluid dynamics, but can be extended to various scientific computing fields such as structural mechanics, electromagnetic field simulation, and multiphysics coupling, demonstrating good versatility and promotional value.

[0069] Seventh, it reduces reliance on engineers' personal experience. By transforming domain knowledge into computable constraint functions, this method enables even novice engineers to generate expert-level simulation configurations, promoting the widespread application of simulation technology.

[0070] Taking the aerodynamic analysis of a certain aircraft wing as an example, the specific implementation process of the present invention is explained.

[0071] The input conditions are: Geometric parameters: wingspan 2 meters, chord length 0.5 meters, characteristic length L is taken as chord length 0.5 m; Flow conditions: Incoming flow velocity U = 30 m / s, angle of attack 3°; Medium properties: Air, default density ρ = 1.225 kg / m³, dynamic viscosity μ = 1.8 × 10⁻⁶ -5 Pa·s; User requirement: Analyze the lift coefficient and drag coefficient of the wing.

[0072] Step 1: Parameter Initialization The system constructs an initial parameter vector based on the input: θ_c = {L=0.5, U=30, ρ=1.225, μ=1.8e-5, T=300, Δx=0.01, Δt=0.001, y+=30, ...} θ_d = {M_turb=None, S_solver=pressure base, E_eqn=False, ...} Step 2: Calculation of dimensionless parameters The system automatically calculates the key dimensionless number:

[0073] Speed ​​of sound in air:

[0074] Mach number:

[0075] Step 3: Physical Consistency Check According to Re = 1.02×10 6 If the value is greater than 5000, it is determined to be turbulent flow, and the physical constraints require that the turbulence model be enabled.

[0076] Based on Ma = 0.086 < 0.3, the flow is determined to be incompressible, and the pressure-based solver can be selected. The energy equation can be disabled.

[0077] The system detected that the initial M_turb = None violated the physical constraints and triggered projection correction.

[0078] Step 4: Layered Projection Correction Discrete variable correction layer: M_turb is corrected from None to the turbulence model closest to the original intent. Considering the high Reynolds number and the fact that it is an airborne outflow problem, the system searches in the discrete space and selects the standard k-ε model (M_turb = k-epsilon). This model has good applicability to external flow problems and requires minimal modification from the original intent (no model specified).

[0079] Continuous variable correction layer: After fixing the discrete variables, check the numerical stability constraints.

[0080] Calculate the initial number of CFLs:

[0081] Setting CFL_max = 1.0 and CFL = 3 > 1 violates the stability constraint.

[0082] The system initiates continuous variable optimization, aiming to minimize ||θ_c - θ_c_candidate||², with the constraint CFL ≤ 1. The solution yields that Δt should be adjusted as follows:

[0083] Step 5: Stability Scoring Calculate the stability score S(θ) of the corrected parameters: CFL margin: 1 - 1.0 / 1.0 = 0; Reynolds number adaptation margin: When Re is within the turbulent range, the margin is relatively high; y+ margin: Initially set y+ = 30. For the standard k-ε model combined with the standard wall function, 30 < y+ < 300, meeting the requirements; The comprehensive score is higher than the threshold, and no further fine-tuning is required.

[0084] Step Six: Generation of configuration file The system generates a simulation configuration file based on the corrected parameter vector θ*, and the main contents include: Solver settings: Pressure-based solver, steady-state calculation; Turbulence model: Standard k-ε model, standard wall function; Mesh parameters: Global size 0.01 m, and the height of the first layer of the boundary layer ensures y+ ≈ 30; Boundary conditions: Velocity inlet 30 m / s, pressure outlet, no-slip on the wall; Discretization format: Second-order upwind format; Convergence criterion: Residual below 10 -5 ; Implementation effect: After the generated configuration file is imported into ANSYS Fluent, the calculation starts smoothly. After 800 iterations of convergence, the lift coefficient and drag coefficient are in good agreement with the wind tunnel experimental data, and the relative error is less than 5%. Compared with the traditional manual configuration, the configuration time is shortened from 2 hours to 5 minutes, and there is no need for multiple trial calculations and adjustments.

[0085] Example Two: Thermal protection analysis of hypersonic vehicle Input conditions: Geometry: Blunt body, head radius R = 0.3 m; Inflow: Mach number Ma = 8, altitude 30 km, static pressure p = 1197 Pa, static temperature T = 226.5 K; Requirement: Analyze the stagnation heat flux density Processing procedure: Since Ma = 8 > 0.3, the system automatically selects the density-based solver and enables the energy equation and compressibility effects.

[0086] Calculate the Reynolds number Re ≈ 2.5×10 6 , determine it as turbulent flow, and select the SST k-ω turbulence model applicable to hypersonic flow.

[0087] Stability constraint requirements: For hypersonic flow, shock capturing requires a sufficiently high mesh resolution. The system automatically adjusts the mesh size Δx from the initial 0.05 m to 0.01 m; the time step Δt is adjusted from 10 -5 s to 10 -7s satisfies the CFL condition.

[0088] The generated configuration file successfully captured the bow shock wave, and the calculated value of the stagnation point heat flux density deviated from the theoretical formula (Fay-Riddell formula) by less than 8%.

[0089] Example 3: Simulation of a Multiphase Flow Reactor Input conditions: Geometry: Stirred reactor, 1 m in diameter; Medium: gas-liquid two-phase system, air-water system; Operating conditions: stirring speed 200 rpm, aeration rate 0.5 vvm Requirement: Analyze gas holdup and gas-liquid mass transfer coefficient. Processing procedure: Based on the requirements of multiphase flow, the system automatically activates the Euler-Euler multiphase flow model and adds interphase forces (drag, lift, and turbulent diffusion forces).

[0090] Physical consistency check: After enabling multiphase flow, it is necessary to define the interphase exchange coefficient, and the system will automatically supplement the default Schiller-Naumann drag force model.

[0091] Numerical stability check: The CFL conditions for multiphase flow are more stringent than those for single-phase flow. The system time step was adjusted from the initial 0.01 s to 0.002 s to ensure computational stability.

[0092] The generated configuration file ran successfully for 200 seconds of physical time, and the gas holdup distribution matched the experimental measurements well, providing a reliable basis for reactor optimization design.

[0093] The three application examples above demonstrate that the method of the present invention can adapt to a wide range of simulation scenarios, from low-speed incompressible flow to hypersonic compressible flow, and from single-phase flow to multiphase flow. It can automatically generate high-quality simulation configurations that meet both physical and numerical constraints, significantly improving the efficiency and reliability of engineering simulation analysis.

[0094] Example 2: Figure 2 This is a schematic diagram of a simulation configuration generation system based on a physical-numerical constraint feasible region provided in Embodiment 2 of the present invention.

[0095] Reference Figure 2 The system includes: The parameter set acquisition module is used to acquire continuous parameter sets and discrete parameter sets; The simulation parameter vector construction module is used to construct a unified simulation parameter vector based on the continuous parameter set and the discrete parameter set; The calculation module is used to calculate key dimensionless physical quantities based on a continuous set of parameters; among which, key dimensionless physical quantities include Reynolds number, Mach number, CFL number and wall value; The Physical Consistency Constraint Set Construction Module is used to construct a set of physical consistency constraints, which includes flow state and model selection constraints, turbulence model parameter constraints, boundary conditions and physical field constraints. The numerical stability constraint set construction module is used to construct a set of numerical stability constraints, which includes time step stability constraints, grid resolution constraints, and iterative convergence constraints. The feasible region construction module is used to construct the feasible region based on the set of physical consistency constraints and the set of numerical stability constraints. The engineering requirements acquisition module is used to acquire natural language or semi-structured engineering requirements input by the user. The generation module is used to generate an initial candidate parameter vector from natural language or semi-structured engineering requirements through semantic parsing and parameter extraction; The judgment module is used to determine whether the initial candidate parameter vector belongs to the feasible region; if it does, a simulation configuration file is generated and imported into mainstream simulation software for calculation; if it does not, a hierarchical projection correction strategy is executed.

[0096] This invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the simulation configuration generation method based on the physical-numerical constraint feasible region provided in the above embodiments.

[0097] This invention also provides a computer-readable medium having processor-executable non-volatile program code, on which a computer program is stored. When the computer program is run by a processor, it executes the steps of the simulation configuration generation method based on the physical-numerical constraint feasible region described above.

[0098] The computer program product provided in this embodiment of the invention includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0099] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and apparatus described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0100] Furthermore, in the description of the embodiments of the present invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention based on the specific circumstances.

[0101] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0102] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to needs to have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0103] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A simulation configuration generation method based on physical-numerical constraint feasible region, characterized in that, The method includes: Obtain the continuous parameter set and the discrete parameter set; Based on the continuous parameter set and the discrete parameter set, a unified simulation parameter vector is constructed; The key dimensionless physical quantities are calculated based on the continuous parameter set; wherein, the key dimensionless physical quantities include Reynolds number, Mach number, CFL number and wall value; Construct a set of physical consistency constraints, wherein the set of physical consistency constraints includes flow state and model selection constraints, turbulence model parameter constraints, boundary conditions and physical field constraints; Construct a set of numerical stability constraints, which includes time step stability constraints, grid resolution constraints, and iterative convergence constraints. Based on the set of physical consistency constraints and the set of numerical stability constraints, a feasible region is constructed. Obtain user input in natural language or semi-structured engineering requirements; The natural language or semi-structured engineering requirements are semantically parsed and parameters are extracted to generate an initial candidate parameter vector; Determine whether the initial candidate parameter vector belongs to the feasible region; If it is found to be true, a simulation configuration file is generated and imported into mainstream simulation software to perform calculations. If it does not belong to the category, then the hierarchical projection correction strategy is executed.

2. The simulation configuration generation method based on the physical-numerical constraint feasible region according to claim 1, characterized in that, The continuous parameter set includes geometric feature length, feature velocity, fluid density, dynamic viscosity, temperature, mesh feature size, time step, and dimensionless wall distance; The set of discrete parameters includes the turbulence model, solver, energy equation options, and boundary condition types.

3. The simulation configuration generation method based on the physical-numerical constraint feasible region according to claim 1, characterized in that, The flow state and model selection constraints are implemented in the following ways: When the Reynolds number is greater than or equal to the turbulence transition threshold, the turbulence model is enabled; When the Mach number is greater than or equal to a preset threshold, a compressible solver is selected and the energy equation is enabled; When the Mach number is less than a preset threshold, an incompressible solver is selected.

4. The simulation configuration generation method based on the feasible region of physical-numerical constraints according to claim 1, characterized in that, The turbulence model parameter constraints are achieved in the following ways: Obtain the LES analytical scale and minimum number of grid cells; When the LES model is selected, the mesh feature size is less than or equal to the ratio of the LES analytical scale to the minimum number of meshes; When a RANS model is selected, the wall values ​​satisfy the effective range of the model; The boundary conditions and physical field constraints are implemented in the following ways: Define thermal property parameters when a temperature gradient exists or the energy equation is enabled; When dealing with multiphase flow problems, an interphase force model is defined.

5. The simulation configuration generation method based on the feasible region of physical-numerical constraints according to claim 1, characterized in that, The time step stability constraints include explicit and implicit formats. The grid resolution constraint includes both overall resolution and local resolution; The iterative convergence constraint is that the ratio of the norm of the residual vector to the norm of the initial residual vector is less than or equal to the convergence tolerance.

6. The simulation configuration generation method based on the feasible region of physical-numerical constraints according to claim 1, characterized in that, The layered projection correction strategy is implemented in the following way: With continuous variables fixed, an enumeration method or heuristic search algorithm is used in the discrete variable space to search for feasible discrete combinations that are closest to the original discrete choices, and these feasible discrete combinations are used as discrete variables. Based on differentiable constraint functions, a constrained gradient descent method is used to project the parameters back to the boundary of the feasible region after each iteration; alternatively, a Lagrange function is constructed, and the optimal solution is gradually approximated by solving a quadratic programming subproblem. If a feasible solution cannot be found after hierarchical optimization, the restrictions on the modification of the discrete variables are relaxed, and discrete-continuous joint optimization is performed until a feasible solution is found or no solution is output. By performing the above hierarchical correction, the optimal correction parameters that satisfy all constraints are obtained.

7. The simulation configuration generation method based on the feasible region of physical-numerical constraints according to claim 1, characterized in that, The method further includes: Define a stability scoring function; When the stability scoring function is less than a preset threshold, the system automatically triggers a fine-tuning mechanism. During the fine-tuning process, it continues to iterate until the stability scoring function is greater than or equal to the preset threshold, or the maximum number of iterations is reached.

8. A simulation configuration generation system based on a physics-numerical constraint feasible region, characterized in that, The system includes: The parameter set acquisition module is used to acquire continuous parameter sets and discrete parameter sets; The simulation parameter vector construction module is used to construct a unified simulation parameter vector based on the continuous parameter set and the discrete parameter set. The calculation module is used to calculate key dimensionless physical quantities based on the continuous parameter set; wherein, the key dimensionless physical quantities include Reynolds number, Mach number, CFL number and wall value; The physical consistency constraint set construction module is used to construct a physical consistency constraint set, wherein the physical consistency constraint set includes flow state and model selection constraints, turbulence model parameter constraints, boundary conditions and physical field constraints; A numerical stability constraint set construction module is used to construct a numerical stability constraint set, which includes time step stability constraints, grid resolution constraints, and iterative convergence constraints. A feasible region construction module is used to construct a feasible region based on the set of physical consistency constraints and the set of numerical stability constraints. The engineering requirements acquisition module is used to acquire natural language or semi-structured engineering requirements input by the user. The generation module is used to generate an initial candidate parameter vector by semantic parsing and parameter extraction of the natural language or semi-structured engineering requirements; The judgment module is used to determine whether the initial candidate parameter vector belongs to the feasible region; if it does, a simulation configuration file is generated and imported into mainstream simulation software for calculation; if it does not, a hierarchical projection correction strategy is executed.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor executes the computer program, it implements the method described in any one of claims 1 to 7.

10. A computer-readable medium having processor-executable non-volatile program code, characterized in that, The program code causes the processor to execute the method described in any one of claims 1 to 7.

Images