Intelligent simulation method and device, equipment, storage medium and computer program product

Abstract

The invention provides an intelligent simulation method, device and equipment, a storage medium and a computer program product, and belongs to the technical field of intelligent simulation, the method comprises the following steps: inputting simulation demand information into a first large model to obtain simulation logic information output by the first large model; inputting the simulation logic information into the second large model to obtain a simulation language output by the second large model; constructing a simulation process according to the simulation language; under the condition that the simulation process meets the simulation requirement, user operation information based on the simulation process is input into the third model, and simulation operation prompt information output by the third model is obtained. According to the method, the problems of high learning cost and low simulation efficiency caused by simulation software operation complexity can be solved. By utilizing large model simulation, the software operation process is simplified, the simulation work difficulty is reduced, and the entry threshold of the simulation field is reduced, so that the simulation efficiency and precision are improved, and the innovation process is accelerated.

Description

Technical Field

[0001] The present invention relates to the field of intelligent simulation technology, and in particular to an intelligent simulation method, apparatus, device, storage medium and computer program product. Background Art

[0002] Simulation technology has become an indispensable tool. It not only helps engineers predict product performance and optimize design parameters, but also significantly reduces the manufacturing and testing costs of physical prototypes. However, the current simulation process is complex and time-consuming, requiring users to possess deep expertise and the skills to operate advanced simulation tools. This is a huge challenge for many designers, especially new employees who lack relevant backgrounds. This makes it difficult for engineers to master both design and simulation technology. As a result, the design and simulation departments in enterprises are often independent, and the ratio of designers to simulators is unbalanced. As a result, simulation tasks often accumulate, making it impossible to quickly verify and optimize solutions, which seriously affects the efficiency of product design and innovation. Based on this, how to implement intelligent simulation to improve simulation efficiency has become an urgent problem that needs to be solved. Summary of the Invention

[0003] The present invention provides an intelligent simulation method, apparatus, device, storage medium and computer program product to solve the defect of low simulation efficiency in the prior art, realize intelligent simulation, and improve simulation accuracy and efficiency.

[0004] The present invention provides an intelligent simulation method, comprising the following steps: Inputting simulation requirement information into the first large model to obtain simulation logic information output by the first large model; Inputting the simulation logic information into the second large model to obtain a simulation language output by the second large model; Constructing a simulation process according to the simulation language; In the case where the simulation process meets the simulation requirements, user operation information based on the simulation process is input into the third model to obtain simulation operation prompt information output by the third model; Among them, the first model is trained based on simulation theory information, simulation specification information and historical simulation specification information; the second model is trained based on the programming syntax and code library required for simulation; the third model is trained based on simulation software operation tutorial information.

[0005] According to an intelligent simulation method provided by the present invention, the third model includes a user operation analysis module, a parameter retrieval module and an operation prompt output module; The user operation analysis module is used to analyze the user operation information to determine the next simulation operation; The parameter retrieving module is used to retrieve the empirical parameters required for the next simulation operation from the first large model, and / or retrieve the material parameters required for the next simulation operation from a material library; The operation prompt output module is used to generate the simulation operation prompt information based on the next simulation operation, or to generate the simulation operation prompt information based on the next simulation operation and the parameters retrieved by the parameter retrieval module.

[0006] According to an intelligent simulation method provided by the present invention, constructing a simulation process according to the simulation language includes: Importing the simulation language into the CAE software through a computer-aided engineering (CAE) interface; Functional modules are retrieved based on the CAE software, and the retrieved functional modules are associated to construct the simulation process.

[0007] According to an intelligent simulation method provided by the present invention, the method further includes: Matching simulation logic information from a template library according to the simulation requirement information; When the target simulation logic information is matched, the target simulation logic information is input into the second large model to obtain the simulation language output by the second large model.

[0008] According to an intelligent simulation method provided by the present invention, after constructing the simulation process according to the simulation language, the method further includes: When the simulation process meets the simulation requirements, the simulation logic information output by the first large model and the simulation language output by the second large model are stored in a template library.

[0009] According to an intelligent simulation method provided by the present invention, after constructing the simulation process according to the simulation language, the method further includes: If the simulation process does not meet the simulation requirements, modify the simulation requirement information; The modified simulation requirement information is re-input into the first large model.

[0010] The present invention also provides an intelligent simulation device, comprising the following modules: A first simulation module, configured to input simulation requirement information into a first large model and obtain simulation logic information output by the first large model; A second simulation module, configured to input the simulation logic information into a second large model to obtain a simulation language output by the second large model; A simulation process construction module, used to construct a simulation process according to the simulation language; The third simulation module is used to input user operation information based on the simulation process into the third model when the simulation process meets the simulation requirements, and obtain simulation operation prompt information output by the third model; wherein, the first model is trained based on simulation theory information, simulation specification information and historical simulation specification information; the second model is trained based on the programming syntax and code library required for simulation; the third model is trained based on the simulation software operation tutorial information.

[0011] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any of the above-described intelligent simulation methods when executing the computer program.

[0012] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the intelligent simulation methods described above.

[0013] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned intelligent simulation methods.

[0014] The intelligent simulation method, device, equipment, storage medium and computer program product provided by the present invention are as follows: by inputting simulation requirement information into a first large model, simulation logic information output by the first large model is obtained; the simulation logic information is input into a second large model, simulation language output by the second large model is obtained; according to the simulation language, a simulation process is constructed; when the simulation process meets the simulation requirements, user operation information based on the simulation process is input into a third large model, simulation operation prompt information output by the third large model is obtained; wherein, the first large model is obtained by training based on simulation theory information, simulation specification information and historical simulation specification information; the second large model is obtained by training based on the programming grammar and code library required for simulation; the third large model is obtained by training based on simulation software operation tutorial information. The present invention can solve the problems of high learning cost and low simulation efficiency caused by the complexity of simulation software operation. By utilizing large model simulation, the software operation process is simplified, the difficulty of simulation work is reduced, and the entry threshold of the simulation field is lowered, thereby improving simulation efficiency and accuracy and accelerating the innovation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0016] Figure 1 It is a flow chart of the intelligent simulation method provided by the present invention.

[0017] Figure 2 It is a flow chart of the large model-based intelligent simulation method provided by the present invention.

[0018] Figure 3 It is a structural diagram of the intelligent simulation device provided by the present invention.

[0019] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0020] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0021] The following combination Figure 1-Figure 4 The intelligent simulation method, apparatus, device, storage medium and computer program product of the present invention are described.

[0022] Figure 1 It is a flow chart of the intelligent simulation method provided by the present invention, such as Figure 1 As shown, the method includes the following: Step 101: Input simulation requirement information into a first large model to obtain simulation logic information output by the first large model.

[0023] Simulation requirement information can be input into the first model in a variety of forms, including voice, text, images, and structured data. It should be understood that simulation requirement information refers to all necessary information that designers or simulators need to provide to the simulation tool or system to ensure accurate and efficient simulation execution. This information can include basic properties of the simulation object, boundary conditions, target outputs, and other information.

[0024] Specifically, the simulator receives a simulation task from the design end, which provides detailed information such as the boundary conditions and properties of the simulation object. For example, the simulation object: the physical system or component that needs to be simulated (such as mechanical structure, fluid system, heat conduction system, etc.); boundary conditions: external constraints or input conditions of the simulation object (such as fixed constraints, loads, temperature, pressure, etc.); material properties: material properties of the simulation object (such as density, elastic modulus, thermal conductivity, etc.); geometric information: the geometric shape and size of the simulation object (such as parametric description); simulation target: the problem to be solved or the output result (such as stress distribution, temperature field, flow field, etc.).

[0025] Simulators can input simulation requirement information into the first model in the form of voice or text. For example: Voice input: "This is a simulation task for a mechanical component made of aluminum alloy. The bottom is fixed, and the top is subject to random vibration. I want to know whether the structural strength meets the requirements." After receiving this information, the first model can perform the following steps: 1) Extract key information from the input, such as material properties, boundary conditions, simulation objectives, etc. 2) Understand the physical meaning and engineering background of the simulation task, such as if this is a structural mechanics problem that requires calculating stress distribution; 3) Generate the corresponding simulation logic based on the task requirements. It should be understood that simulation logic can include: determining the simulation method to be used (such as finite element analysis, computational fluid dynamics, etc.); breaking down the simulation task into specific operational steps (such as meshing, boundary condition setting, solver selection, etc.); recommending appropriate simulation parameters based on the input information (such as mesh density, time step, convergence criteria, etc.); and specifying how to output and process simulation results (such as stress contours, data tables, etc.).

[0026] For example, the first model can output the following simulation logic: Random vibration analysis is required for the structure. Before random vibration analysis, modal analysis is required to obtain the structure's mode shapes and frequencies, which are then imported into the random vibration analysis module for random vibration analysis. The modal analysis steps include importing the model, setting material properties, meshing, setting boundary conditions, and performing analysis-related settings. The random vibration analysis steps include setting vibration excitation and analysis-related settings.

[0027] The first model is trained based on simulation theory information, simulation specification information, and historical simulation specification information. The first model can be a natural language processing (NLP) model, a multimodal model, a reinforcement learning model, a simulation-specific model, etc.

[0028] It should be understood that simulation theory information refers to the scientific principles and mathematical models that underpin simulation technology, such as finite element theory, fluid dynamics theory, and heat conduction theory. Training in simulation theory information can include: large-scale models learning the basic concepts, formulas, and principles of simulation theory; and understanding the physical background and mathematical descriptions of different simulation types (such as structural mechanics, fluid dynamics, and heat conduction). The purpose of simulation theory information is to enable large-scale models to understand the physical meaning of simulation tasks and generate simulation processes and parameter settings that conform to theoretical logic.

[0029] It should be understood that simulation specification information refers to the standards and rules that need to be followed during the simulation process. For example, specifications for meshing (such as mesh density and element type), boundary condition setting specifications (such as fixed constraints and load application methods), solver selection specifications (such as static solvers and dynamic solvers), and result processing specifications (such as stress cloud maps and data output formats). Training content for simulation specification information can include: large models learning the operating procedures and best practices in simulation specifications, and understanding the specification requirements of different simulation tasks. The role of simulation specification information is to enable large models to generate simulation logic that complies with specifications and reduce errors and unreasonable settings during the simulation process.

[0030] It should be understood that historical simulation specification information refers to the accumulated simulation task records, experience data, and case libraries. For example, it includes successful simulation cases and their parameter settings, failed simulation cases and their cause analysis, and optimization solutions for different simulation tasks. Training with historical simulation specification information can include: allowing the large model to learn from the experiences and lessons learned from historical simulation tasks, as well as understanding the commonalities and characteristics of different simulation tasks. The purpose of historical simulation specification information is to enable the large model to draw on historical experience, generate more reasonable simulation logic, and improve the efficiency and success rate of simulation tasks.

[0031] Among them, the first large model can be trained in the following way: 1) Data collection: Collect simulation theories (such as finite element analysis and fluid dynamics), organize simulation specifications (such as meshing and boundary condition settings), and obtain historical simulation data (such as success cases and failure analysis). 2) Data preprocessing: Cleaning data, removing noise, and converting text and code into a format that the model can process (such as tokens); 3) Model selection: Choose an appropriate model architecture, such as Transformer or Generative Pre-trained Transformer (GPT). 4) Model training: Train on large-scale text and code datasets to learn general capabilities, and fine-tune based on simulation theory, specifications, and historical data to adapt to simulation tasks; 5) Evaluation: Check whether the generated simulation logic conforms to the theory and specifications, and verify whether the functions are accurate in actual simulation.

[0032] By integrating simulation theory, simulation specifications, and historical simulation specifications for training, the First Large Model is able to: Based on simulation theory, the model understands the physical meaning and objectives of simulation tasks; based on simulation specifications, the model generates standardized simulation processes and parameter settings; and based on historical simulation specifications, the model generates more efficient and reliable simulation logic. This allows the First Large Model to provide scientific, standardized, and efficient solutions for simulation tasks, significantly improving simulation efficiency and reliability.

[0033] Step 102: input the simulation logic information into the second largest model to obtain the simulation language output by the second largest model.

[0034] The second model is primarily used to convert simulation logic information into a language recognizable by simulation software (such as CAE software). Its core function is code generation. The second model is trained based on the programming syntax and code library required for simulation. It can convert simulation logic information into scripts or commands for the simulation software. It should be understood that specific forms of simulation languages can include script files and command line instructions.

[0035] The second largest model can be a code generation model or a translation model, such as a Transformer-based code generation model or a sequence-to-sequence (Seq2Seq) model. The training process of the second largest model includes the following key steps: 1) Data Collection: Collect simulation software scripts or command examples (such as ANSYS APDL scripts, AbaqusPython scripts, OpenFOAM configuration files, etc.); collect simulation logic information corresponding to the simulation scripts (such as natural language descriptions, structured data, etc.); organize the simulation software's programming syntax and common commands to build a code library; 2) Data preprocessing: Remove noise (such as comments and blank lines) from simulation scripts and logic information, convert text and code into tokens that the model can process, align simulation logic information with the corresponding simulation script, and construct input-output pairs; 3) Model training: Pre-training is performed on large-scale code datasets (such as GitHub code) to learn general code generation capabilities. Fine-tuning is performed on datasets of simulation scripts and logic information to adapt the model to simulation tasks. By adjusting the model architecture, hyperparameters, and training strategies, the model's accuracy and efficiency are improved. 4) Evaluation and Verification: Check whether the generated code meets the syntax requirements of the simulation software and whether it can be executed correctly. Verify whether the generated code can achieve the expected simulation function. Further optimize the model based on user feedback.

[0036] After pre-training the second largest model, the simulation logic information is input into the second largest model to obtain the simulation language output by the second largest model. It should be understood that the second largest model parses the input simulation logic information, extracts key steps and parameters, and then generates corresponding scripts or commands based on the syntax and code library of the simulation software. Specifically, the main processing process of the second largest model to convert simulation logic information into simulation language is as follows: Parsing input: If the input is a natural language description, the model will use natural language processing (NLP) technology to parse the text and extract key information (such as simulation steps, parameter settings, etc.); if the input is structured data, the model will directly extract the fields and values.

[0037] Semantic understanding: The model recognizes the simulation tasks in the input (such as "import geometry model", "set material properties", etc.), and extracts task-related parameters (such as file path, material properties, boundary conditions, etc.).

[0038] Code generation: The model converts simulation tasks and parameters into corresponding code snippets according to the grammatical rules of the simulation software, and the model combines all code snippets into a complete simulation script.

[0039] After inputting the simulation logic information into the second largest model, the model will convert it into a language recognizable by the simulation software through steps such as parsing, semantic understanding, code generation and optimization. This process significantly improves the simulation efficiency and automation level.

[0040] Step 103: construct a simulation process according to the simulation language.

[0041] The simulation process refers to the specific steps and operation sequence required to complete a simulation task. It defines the entire execution process from the beginning to the end of the simulation task, including input, processing, calculation and output.

[0042] Specifically, key information (such as geometric model path, material properties, boundary conditions, etc.) is extracted from the simulation language, the simulation language is decomposed into specific operation steps, and then the operation steps are organized into a logically clear process in the order of pre-processing, solving, and post-processing to ensure that the input and output of each step can be correctly connected. Finally, the organized operation steps are converted into a process executable by the simulation software to generate script files, command line instructions or configuration files.

[0043] In one embodiment, step 103 specifically includes: importing the simulation language into the CAE software through the computer-aided engineering (CAE) interface; retrieving functional modules based on the CAE software, and associating the retrieved functional modules to construct a simulation process.

[0044] Computer-Aided Engineering (CAE) refers to the technology of using computer software to simulate, analyze and optimize engineering problems.

[0045] It should be understood that associating modules can form a complete simulation process because the simulation process consists of multiple steps or tasks, which require data transfer, logical dependencies, and parameter sharing. Associating different modules (such as the modal analysis module and the vibration analysis module) through simulation code (simulation language) enables them to work together to form a complete simulation process, ensuring the consistency and correctness of the simulation process. For example, the associated content can include the following aspects: 1) Data Transfer: The output of one module serves as the input to another. For example, the output of a modal analysis module (such as natural frequencies and mode shapes) can serve as the input to a vibration analysis module. Specifically, the modal analysis results can be extracted through code and passed to the vibration analysis module.

[0046] 2) Logical Dependencies: The execution order between modules must conform to the simulation logic. For example, modal analysis must be performed before vibration analysis can be performed based on its results. Specifically, the execution order of modules can be clearly defined in the code to ensure the correctness of the process.

[0047] 3) Parameter Sharing: Certain parameters (such as material properties and geometric models) need to be shared across multiple modules. For example, modal analysis and vibration analysis may use the same material properties and geometric models. Specifically, these parameters can be defined in the code and called upon across different modules.

[0048] By linking modules through code, an efficient and automated simulation process can be built, reducing human intervention and improving simulation efficiency and accuracy.

[0049] Step 104 , when the simulation process meets the simulation requirements, user operation information based on the simulation process is input into the third largest model to obtain simulation operation prompt information output by the third largest model.

[0050] A simulation process that meets simulation requirements means it can fully and accurately achieve the user's goals and output simulation results that meet the requirements. For example, a simulation process that meets simulation requirements must meet the following conditions: the simulation process covers all steps required to achieve the simulation goal (such as pre-processing, solving, and post-processing); the input and output of each step are properly connected to ensure the successful completion of the simulation task; and the parameter settings in the simulation process (such as mesh density, time step, convergence criteria, etc.) conform to simulation theory and specifications.

[0051] If the simulation process meets the simulation requirements, user operation information based on the simulation process is input into the third model, and simulation operation prompt information is output by the third model. For example, user operation information is recorded during the simulation process and input into the third model. The third model analyzes the user operation information, identifies problems or optimization points, generates simulation operation prompt information, and returns it to the user. The user adjusts the operation according to the prompt information to complete the simulation task.

[0052] It should be understood that the third model continuously acquires the user's current simulation operation, predicts the next operation, and generates operation prompts until the simulation task is completed. Through this cyclical process, users can complete simulation tasks according to best practices and obtain high-quality simulation results.

[0053] In one embodiment, the third large model includes a user operation analysis module, a parameter retrieval module and an operation prompt output module; the user operation analysis module is used to analyze user operation information to determine the next simulation operation; the parameter retrieval module is used to retrieve the empirical parameters required for the next simulation operation from the first large model, and / or retrieve the material parameters required for the next simulation operation from the material library; the operation prompt output module is used to generate simulation operation prompt information based on the next simulation operation, or generate simulation operation prompt information based on the next simulation operation and the parameters retrieved by the parameter retrieval module.

[0054] Specifically, the user operation analysis module is primarily used to analyze the user's operation records in the current simulation process (such as geometry model import, material property setting, boundary condition setting, etc.). Based on the user's operation, it determines the progress and status of the current simulation process and then predicts the next simulation operation (such as meshing, solver setting, result output, etc.). For example, if the user has completed geometry model import and material property setting, the module will determine that the next operation is "setting boundary conditions."

[0055] The parameter retrieval module is primarily used to retrieve empirical parameters from the primary model and material parameters from the material library. It should be understood that the primary model provides best-practice parameters (such as mesh density, time step, and convergence criteria) based on simulation theory and historical data. The material library stores characteristic parameters of various materials (such as density, elastic modulus, and thermal conductivity), and the module retrieves the corresponding material parameters based on the simulation object. For example, if the next step is to "Set Material Properties," the module will retrieve the density, elastic modulus, and Poisson's ratio of steel from the material library.

[0056] The Operation Prompt Output Module is primarily used to generate prompts based on the next action determined by the User Operation Analysis Module, such as "Next: Set Boundary Conditions." Alternatively, it can generate more detailed prompts based on the parameters provided by the Parameter Retrieval Module, such as "Set Boundary Conditions: Fixed Bottom, Apply 5000N Force to Top." For example, if the next action is "Generate Mesh," the module will generate the prompt: "Generate Mesh. A medium mesh density (e.g., ESIZE, 5) is recommended."

[0057] It should be understood that the collaboration between the modules is as follows: the user operation analysis module first analyzes the user's operation information and determines the next simulation operation. The parameter retrieval module retrieves the required parameters from the first large model or material library according to the next operation. Finally, the operation prompt output module generates simulation operation prompt information based on the next operation and the retrieved parameters, and returns it to the user.

[0058] Through the collaborative work of the user operation analysis module, parameter retrieval module and operation prompt output module, the user's operation information is analyzed, the required parameters are retrieved, and simulation operation prompt information is generated. This structured design can significantly improve the efficiency, accuracy and user experience of simulation tasks and realize intelligent simulation.

[0059] It should be understood that the third model is trained based on simulation software operation tutorial information. The model's training data primarily comes from simulation software operation tutorials, user manuals, official documentation, and related operation examples. Through this data, the model can learn the simulation software's operation steps, parameter settings, best practices, and other knowledge, thus gaining the ability to analyze and guide user operations.

[0060] Among them, the training process of the third model can include: 1) Data collection: Collect simulation software operation tutorials, example cases, user experience, code scripts, etc. 2) Data preprocessing: Remove noise from the tutorial (such as comments, blank lines, special symbols, etc.), convert text and code into tokens that the model can process, align the operation steps with the corresponding code scripts or parameter settings, construct input-output pairs, and annotate key information (such as operation steps, parameter names, error messages, etc.) for supervised learning; 3) Model selection: You can choose Transformer-based models, sequence-to-sequence (Seq2Seq) models, and domain-specific models (specialized models trained for simulation software operation tutorials). 4) Model Training: Pre-training is performed on large-scale text and code datasets (such as GitHub code and simulation software documentation) to learn general language and code representation capabilities. Self-supervised learning methods (such as masked language modeling and next sentence prediction) are used for training. Fine-tuning is performed on a dataset of simulation software operation tutorials to adapt the model to simulation tasks. Supervised learning methods are used to input text or code from the operation tutorials and output operation steps, parameter settings, or prompts. By adjusting the model architecture, hyperparameters, and training strategies, the accuracy and efficiency of the model are improved, and user feedback data is used to further optimize the model. 5) Evaluation and Verification: Check whether the generated code meets the syntax requirements of the simulation software and whether it can be executed correctly. Verify whether the generated prompt information can guide the user to complete the simulation task.

[0061] It should be understood that the third model trained based on the simulation software operation tutorial information can have the following functions: parsing the user's operation information and identifying the progress and status of the current simulation process; recommending the parameters required for the next operation based on the best practices in the operation tutorial; generating prompt information for the next operation based on the knowledge in the operation tutorial; error detection and correction: identifying errors in user operations and providing correction suggestions.

[0062] The intelligent simulation method provided by the embodiment of the present invention inputs simulation requirement information into the first large model to obtain simulation logic information output by the first large model; inputs simulation logic information into the second large model to obtain simulation language output by the second large model; constructs a simulation process according to the simulation language; when the simulation process meets the simulation requirements, inputs user operation information based on the simulation process into the third large model to obtain simulation operation prompt information output by the third large model; wherein, the first large model is obtained by training based on simulation theory information, simulation specification information and historical simulation specification information; the second large model is obtained by training based on the programming grammar and code library required for simulation; and the third large model is obtained by training based on simulation software operation tutorial information. The present invention can solve the problems of high learning cost and low simulation efficiency caused by the complexity of simulation software operation. By utilizing large model simulation, the software operation process is simplified, the difficulty of simulation work is reduced, and the entry threshold of the simulation field is lowered, thereby improving simulation efficiency and accuracy and accelerating the innovation process.

[0063] Based on the above embodiment, the present invention further includes: Step 210: Match simulation logic information from a template library according to the simulation requirement information; Step 211: When the target simulation logic information is matched, the target simulation logic information is input into the second large model to obtain the simulation language output by the second large model.

[0064] The template library stores standardized simulation logic information for various simulation tasks. Each template may include simulation type, simulation steps, parameter settings, and applicable scenarios.

[0065] Specifically, refer to Figure 2 , converts the simulation requirement information input by the user into a form that the model can understand. For example, it extracts key information (such as simulation object, material properties, boundary conditions, target output, etc.) and converts the information into a feature vector or embedding representation. Then, a matching algorithm is used to find the template that is most similar to the user's requirement from the template library. The matching algorithm may include: 1) Rule-based matching: Filter templates based on preset rules (such as simulation type, material properties, etc.); 2) Similarity-based matching: Calculate the similarity between user requirements and templates (such as cosine similarity, Euclidean distance, etc.); 3) Machine learning-based matching: training classification or ranking models to predict the most appropriate template.

[0066] Finally, the matched simulation logic information is input into the second largest model to obtain the simulation language output by the second largest model.

[0067] By constructing a rich template library and designing a precise matching algorithm, the embodiment of the present invention can quickly generate simulation logic that meets user needs, significantly improving simulation efficiency and reliability. At the same time, it can also adapt to more complex simulation needs by continuously optimizing and expanding the template library.

[0068] Based on the above embodiment, after constructing the simulation process according to the simulation language, the method further includes: Step 310: When the simulation process meets the simulation requirements, the simulation logic information output by the first large model and the simulation language output by the second large model are stored in a template library.

[0069] It should be understood that reference Figure 2 If the simulation process meets the simulation requirements, the simulation logic information output by the first model and the simulation language output by the second model are verified to be correct and capable of guiding the user to complete the simulation task. At this point, the simulation logic information and simulation language are stored in the template library. Only verified correct information needs to be stored in the template library to avoid storing erroneous or incomplete information.

[0070] It should be understood that the information in the template library can only be reused efficiently when it is correct. Storing verified information can reduce the user's trial and error costs during reuse and improve the efficiency of simulation tasks.

[0071] Based on the above embodiment, after constructing the simulation process according to the simulation language, the method further includes: Step 410: if the simulation process does not meet the simulation requirements, modify the simulation requirement information; Step 411: re-input the modified simulation requirement information into the first large model.

[0072] It should be understood that reference Figure 2 ,When the simulation process cannot meet the user ,requirements, the simulation requirement information is modified and the ,modified simulation requirement information is re-entered into the first model. ,This process is an iterative optimization mechanism that ,enables the simulation process to gradually meet the user ,goals by continuously adjusting the simulation requirement information.

[0073] For example, if the simulation process cannot achieve the user's goals, such as the results are not as expected, the accuracy is insufficient, the calculation time is too long, etc. Analyze the reasons why the simulation process does not meet the requirements, which may be due to unclear simulation requirement information, unreasonable parameter settings, or inaccurate target definitions. Further modify the simulation requirement information, where the modifications may include: modifying the simulation goals (such as improving accuracy, optimizing calculation efficiency, etc.), modifying simulation parameters (such as mesh density, time step, material properties, etc.) and redefining the simulation requirement information to make it clearer and more specific. For example, the original requirement: "Calculate stress distribution." The modified requirement: "Calculate stress distribution with an accuracy error of less than 5%, and the calculation time is controlled within 1 hour." The modified simulation requirements are input into the first large model to regenerate the simulation logic. Specifically, the simulation process is re-executed based on the new simulation logic to verify the results. This involves checking whether the new simulation process meets the modified simulation requirements and continuing the iteration process. If the requirements are still not met, the simulation requirements are further modified and re-input into the first large model until the simulation process meets the requirements.

[0074] This embodiment of the present invention modifies the simulation requirement information and re-enters the first large model when the simulation process does not meet the requirements. This is an iterative optimization mechanism. By continuously adjusting the simulation requirement information, the simulation process gradually meets the user's goals, improving the quality and efficiency of the simulation results. This mechanism can significantly increase the flexibility and adaptability of simulation tasks, achieving intelligent simulation.

[0075] In order to further analyze and illustrate the intelligent simulation method proposed in the present invention, refer to the following examples.

[0076] To solve the problems of high learning cost and low simulation efficiency caused by the complexity of simulation software operations. The embodiment of the present invention specifically proposes an intelligent simulation method based on a large model, which can be implemented through an intelligent simulation system. The system allows users to describe design intentions and simulation requirements in simple language. The system automatically parses these instructions with the help of the deep understanding and simulation knowledge learning ability of the large language model, and provides professional simulation suggestions and corresponding simulation templates / processes to designers. During the simulation process, the system gradually guides the user to complete the necessary simulation settings through an intuitive user interface, such as the selection of loading objects and the input of load sizes. In addition, the intelligent simulation platform also has adaptive learning capabilities, and can continuously optimize its recommendation strategies based on historical simulation cases and results, thereby improving simulation accuracy and efficiency.

[0077] By lowering the threshold and complexity of simulation technology, the system will promote the popularization of simulation technology, allowing more designers to make full use of simulation tools like professional simulation engineers, promote the deep integration of design and simulation, achieve faster and higher-quality design-simulation iterations, and accelerate product innovation.

[0078] refer to Figure 2 , the intelligent simulation process based on large models can include: 1. The simulator receives a simulation task from the design end, which provides detailed information such as the boundary conditions and properties of the simulation object; 2. The simulation personnel input the simulation requirements into the large model 1 (i.e., the first large model) in the form of voice or text. The large model 1 interprets and outputs the simulation logic. 3. Large model 2 (i.e., the second largest model) converts the simulation logic output by large model 1 into a language recognizable by the simulation software; 4. Import the language output by large model 2 into CAE software through CAE interface; 5. Automatically call the corresponding modules or functions in CAE software; 6. Tutorial on using the software for Large Model 3 (the third large model). The software provides on-screen operations, including picking and analyzing, checking the next steps, and providing corresponding prompts to guide the user through the necessary input and settings, such as importing simulation objects, selecting load surfaces, and entering load magnitudes. Experience-based parameters such as mesh size can be retrieved from Large Model 1 by Large Model 3, while deterministic parameters such as material properties can be directly retrieved from the material library and then filled in by the user after judgment. 7. The system prompts that the settings are completed and the personnel start the simulation calculation.

[0079] The above involves three major models. The basic idea is to leverage the language understanding capabilities of these models, input relevant professional data, and then perform training and fine-tuning to enable them to acquire corresponding capabilities. The three major models are described in detail below: Large Model 1: The training and fine-tuning data consists of simulation theory, simulation specifications, and historical simulation specifications. Basic simulation rules are learned from these data. For example, modal analysis is required before vibration analysis, thus translating simulation requirements into simulation logic. Furthermore, simulation experience and techniques are learned from these data, which are then used by Large Model 3 to query and retrieve relevant parameters.

[0080] Large Model 2: The training and fine-tuning data is the programming syntax of the CAE background code and the related code library. The simulation logic generated by Large Model 1 is converted into a language (code) that can be recognized by the CAE software. After the code is imported into the software through the corresponding interface, the functional modules (such as modal analysis and vibration analysis modules) can be called up and linked to each other to build a simulation process.

[0081] Large Model 3: The training and fine-tuning data is the CAE software tutorial. During the detailed setup phase, on-screen operations are used to pick and analyze the data, verify the next steps, and provide corresponding prompts to guide the user through the necessary setup inputs. When parameters such as mesh size and material properties are required, they are retrieved from Large Model 1 and the material library.

[0082] It should be understood that in order to better divide the work among the models and thus achieve better results, a step-by-step conversion method is used for multiple large models. Alternatively, these large models can be integrated into one large model to directly generate the final modeling commands. For example, the functions of multiple large models can be integrated into one large model, and modeling commands can be generated directly based on the simulation requirements. Based on this, the step-by-step conversion steps can be reduced, improving efficiency. At the same time, all functions are integrated into one model, making it easier to manage and maintain. Directly generating modeling commands can reduce delays in intermediate links.

[0083] The large-scale model-based intelligent simulation method provided by the embodiments of the present invention simplifies the operational procedures of simulation software by utilizing artificial intelligence large-scale models, automatically constructs simulation processes, and provides prompts to assist personnel in completing specific settings. This reduces the learning cost for new users, allowing beginners to leverage their previous simulation experience and skills, significantly improving simulation efficiency and accuracy. This not only accelerates the innovation process, but also enhances the user experience through interactive prompts and supports the handling of complex simulation tasks, thereby promoting the widespread application of simulation technology in the design field. It has significant practical application value and promotion prospects.

[0084] The intelligent simulation device provided by the present invention is described below. The intelligent simulation device described below and the intelligent simulation method described above can be referenced to each other.

[0085] refer to Figure 3 The intelligent simulation device provided by the present invention includes a first simulation module 301, a second simulation module 302, a simulation process construction module 303 and a third simulation module 304.

[0086] The first simulation module 301 is used to input simulation requirement information into the first large model and obtain simulation logic information output by the first large model; The second simulation module 302 is used to input the simulation logic information into the second large model to obtain the simulation language output by the second large model; A simulation process construction module 303 is used to construct a simulation process according to the simulation language; The third simulation module 304 is used to input user operation information based on the simulation process into the third model when the simulation process meets the simulation requirements, and obtain simulation operation prompt information output by the third model; wherein, the first model is trained based on simulation theory information, simulation specification information and historical simulation specification information; the second model is trained based on the programming syntax and code library required for simulation; and the third model is trained based on simulation software operation tutorial information.

[0087] The intelligent simulation device provided by the embodiment of the present invention inputs simulation requirement information into the first large model to obtain simulation logic information output by the first large model; inputs simulation logic information into the second large model to obtain simulation language output by the second large model; constructs a simulation process according to the simulation language; and when the simulation process meets the simulation requirements, inputs user operation information based on the simulation process into the third large model to obtain simulation operation prompt information output by the third large model; wherein, the first large model is obtained by training based on simulation theory information, simulation specification information and historical simulation specification information; the second large model is obtained by training based on the programming grammar and code library required for simulation; and the third large model is obtained by training based on simulation software operation tutorial information. The present invention can solve the problems of high learning cost and low simulation efficiency caused by the complexity of simulation software operation. By utilizing large model simulation, the software operation process is simplified, the difficulty of simulation work is reduced, and the entry threshold of the simulation field is lowered, thereby improving simulation efficiency and accuracy and accelerating the innovation process.

[0088] In one embodiment, the third large model includes a user operation analysis module, a parameter retrieval module and an operation prompt output module; the user operation analysis module is used to analyze the user operation information to determine the next simulation operation; the parameter retrieval module is used to retrieve the empirical parameters required for the next simulation operation from the first large model, and / or retrieve the material parameters required for the next simulation operation from the material library; the operation prompt output module is used to generate the simulation operation prompt information based on the next simulation operation, or generate the simulation operation prompt information based on the next simulation operation and the parameters retrieved by the parameter retrieval module.

[0089] In one embodiment, the simulation process construction module 303 is specifically used to: import the simulation language into the CAE software through the computer-aided engineering (CAE) interface; call the functional modules based on the CAE software, and associate the called functional modules to construct the simulation process.

[0090] In one embodiment, the first simulation module 301 is further used to: match simulation logic information from a template library according to the simulation requirement information; and when the target simulation logic information is matched, input the target simulation logic information into the second large model to obtain the simulation language output by the second large model.

[0091] In one embodiment, the simulation process construction module 303 is further used to: when the simulation process meets the simulation requirements, store the simulation logic information output by the first large model and the simulation language output by the second large model in a template library.

[0092] In one embodiment, the simulation process construction module 303 is further used to: modify the simulation requirement information when the simulation process does not meet the simulation requirement; and re-input the modified simulation requirement information into the first large model.

[0093] Figure 4 An example of a physical structure diagram of an electronic device is shown below. Figure 4 As shown, the electronic device may include: a processor 410, a communications interface 420, a memory 430, and a communications bus 440, wherein the processor 410, the communications interface 420, and the memory 430 communicate with each other via the communications bus 440. The processor 410 may call logic instructions in the memory 430 to execute an intelligent simulation method, which includes: inputting simulation requirement information into a first large model to obtain simulation logic information output by the first large model; inputting the simulation logic information into a second large model to obtain a simulation language output by the second large model; constructing a simulation process based on the simulation language; and, if the simulation process meets the simulation requirements, inputting user operation information based on the simulation process into a third large model to obtain simulation operation prompt information output by the third large model. The first large model is trained based on simulation theory information, simulation specification information, and historical simulation specification information; the second large model is trained based on programming syntax and code library required for simulation; and the third large model is trained based on simulation software operation tutorial information.

[0094] Furthermore, the logic instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion 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, stored in a storage medium, includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0095] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the intelligent simulation method provided by the above methods, which includes: inputting simulation requirement information into a first large model to obtain simulation logic information output by the first large model; inputting the simulation logic information into a second large model to obtain a simulation language output by the second large model; constructing a simulation process according to the simulation language; when the simulation process meets the simulation requirements, inputting user operation information based on the simulation process into a third large model to obtain simulation operation prompt information output by the third large model; wherein, the first large model is trained based on simulation theory information, simulation specification information and historical simulation specification information; the second large model is trained based on the programming syntax and code library required for simulation; and the third large model is trained based on simulation software operation tutorial information.

[0096] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the intelligent simulation method provided by the above-mentioned methods, the method comprising: inputting simulation requirement information into a first large model to obtain simulation logic information output by the first large model; inputting the simulation logic information into a second large model to obtain a simulation language output by the second large model; constructing a simulation process according to the simulation language; when the simulation process meets the simulation requirements, inputting user operation information based on the simulation process into a third large model to obtain simulation operation prompt information output by the third large model; wherein, the first large model is trained based on simulation theory information, simulation specification information and historical simulation specification information; the second large model is trained based on the programming syntax and code library required for simulation; and the third large model is trained based on simulation software operation tutorial information.

[0097] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0098] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An intelligent simulation method, characterized in that: include: Inputting simulation requirement information into the first large model to obtain simulation logic information output by the first large model; Inputting the simulation logic information into the second large model to obtain a simulation language output by the second large model; Constructing a simulation process according to the simulation language; In the case where the simulation process meets the simulation requirements, user operation information based on the simulation process is input into the third model to obtain simulation operation prompt information output by the third model; Among them, the first model is trained based on simulation theory information, simulation specification information and historical simulation specification information; the second model is trained based on the programming syntax and code library required for simulation; the third model is trained based on simulation software operation tutorial information.

2. The intelligent simulation method according to claim 1, characterized in that: The third model includes a user operation analysis module, a parameter retrieval module and an operation prompt output module; The user operation analysis module is used to analyze the user operation information to determine the next simulation operation; The parameter retrieving module is used to retrieve the empirical parameters required for the next simulation operation from the first large model, and / or retrieve the material parameters required for the next simulation operation from a material library; The operation prompt output module is used to generate the simulation operation prompt information based on the next simulation operation, or to generate the simulation operation prompt information based on the next simulation operation and the parameters retrieved by the parameter retrieval module.

3. The intelligent simulation method according to claim 1, characterized in that: The step of constructing a simulation process according to the simulation language includes: Importing the simulation language into the CAE software through a computer-aided engineering (CAE) interface; Functional modules are retrieved based on the CAE software, and the retrieved functional modules are associated to construct the simulation process.

4. The intelligent simulation method according to claim 1, characterized in that: The method further comprises: Matching simulation logic information from a template library according to the simulation requirement information; When the target simulation logic information is matched, the target simulation logic information is input into the second large model to obtain the simulation language output by the second large model.

5. The intelligent simulation method according to claim 1, characterized in that: After constructing the simulation process according to the simulation language, the method further includes: When the simulation process meets the simulation requirements, the simulation logic information output by the first large model and the simulation language output by the second large model are stored in a template library.

6. The intelligent simulation method according to claim 1, characterized in that: After constructing the simulation process according to the simulation language, the method further includes: If the simulation process does not meet the simulation requirements, modify the simulation requirement information; The modified simulation requirement information is re-input into the first large model.

7. An intelligent simulation device, characterized in that: include: A first simulation module, configured to input simulation requirement information into a first large model and obtain simulation logic information output by the first large model; A second simulation module, configured to input the simulation logic information into a second large model to obtain a simulation language output by the second large model; A simulation process construction module, used to construct a simulation process according to the simulation language; The third simulation module is used to input user operation information based on the simulation process into the third model when the simulation process meets the simulation requirements, and obtain simulation operation prompt information output by the third model; wherein, the first model is trained based on simulation theory information, simulation specification information and historical simulation specification information; the second model is trained based on the programming syntax and code library required for simulation; the third model is trained based on the simulation software operation tutorial information.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the intelligent simulation method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the intelligent simulation method according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the intelligent simulation method according to any one of claims 1 to 6 is implemented.

Images