LLM-based autonomous computational 3D multi-physics analysis system and method

Abstract

The present disclosure provides a system for automated structural analysis comprising a computer-aided design [CAD] tool and a simulation tool configured to integrate with the CAD tool. The system further includes a large language model-based system and a geometry reasoning engine. The system is configured to receive a technical design specification, analyze the specification using the LLM-based system, configure the simulation tool based on the analysis, perform a structural analysis using the simulation tool, and generate a report of structural analysis results. The system may fetch additional data from a knowledge base or external database, validate assumptions with a designer, analyze results against the specification, and generate design recommendations if needed. Multiple iterations of analysis and design improvement may be performed until predefined criteria are met.

Description

LLM-BASED AUTONOMOUS COMPUTATIONAL 3D MULTI-PHYSICS ANALYSIS SYSTEM AND METHODTECHNICAL FIELD

[0001] The present disclosure relates to computer-aided design and simulation systems, and more particularly to an LLM-based autonomous 3D computational 3D multi-physics analysis system that integrates CAD tools, simulation tools, and large language models to automate the simulation analysis process.BACKGROUND

[0002] Computer-aided design [CAD] and simulation tools have become useful in modern engineering and product development processes. These tools allow designers and engineers to create, analyze, and optimize complex 3D models before physical prototyping. However, the effective use of these tools often requires expertise and manual effort, limiting their accessibility to a broader range of users.

[0003] Recent advancements in artificial intelligence, particularly in the field of large language models [LLMs], have opened new possibilities for enhancing and automating various aspects of the design and simulation process. These Al-powered systems have demonstrated remarkable capabilities in understanding and generating human-like text, as well as interpreting and manipulating complex data structures.

[0004] The integration of CAD tools, simulation software, and LLMs presents an opportunity to streamline and democratize the computational 3D multi-physics analysis process. By combining these technologies, it may be possible to create more intuitive and automated workflows that can benefit both experienced engineers and novice users.

[0005] WO 2024 / 043934 Al discloses a method for generating instructions using 3D model component relationship extraction. This document relates to automatically generating and providing instructions for tasks in an industrial environment by analyzing CAD files and other information formats. However, while this prior art focuses on generating instructions for human or machine execution of industrial tasks, it does not specifically address the automation of the computational 3D multi-physics analysis process itself or the integration of LLMs for real-time design optimization.

[0006] Despite the potential benefits of integrating CAD, simulation, and Al technologies, several challenges remain. These include ensuring the accuracy and reliability of Al-generated analyses, maintaining compatibility between different software components,and developing user-friendly interfaces that can effectively leverage the power of these combined technologies.

[0007] As the complexity of engineering projects continues to grow, there is an increasing demand for more efficient and accessible tools that can assist in the design and analysis process. Improved methods for automating computational 3D multi-physics analysis, providing real-time feedback, and generating design recommendations could enhance productivity and innovation in various industries, including aerospace, automotive, and construction.SUMMARY

[0008] According to an aspect of the present disclosure, a system for automated computational 3D multi-physics analysis is provided. The system includes a computer-aided design tool, a simulation tool allowing CAD integration, a large language model-based system, and a geometry reasoning engine. The system is configured to receive a technical design specification, analyze the technical design specification using the LLM-based system, configure the simulation tool based on the analysis, perform a computational 3D multiphysics analysis using the simulation tool, and generate a report of the simulation analysis results.

[0009] According to other aspects of the present disclosure, the system may include one or more of the following features. The system may be configured to fetch additional data from external databases or knowledge sources. The system may be configured to validate assumptions with a designer before performing the computational 3D multi-physics analysis. The system may be configured to analyze the simulation analysis results and check whether they meet the technical design specification. The system may be configured to generate design improvement recommendations if the simulation analysis results do not meet the technical design specification.

[0010] In certain cases, the system may validate assumptions with a designer before performing the computational 3D multi-physics analysis. This validation step may involve presenting the designer with a summary of key parameters and assumptions derived from the technical design specification. For example, the system may generate a prompt for the designer that includes: (1) material properties assumed for the analysis, such as Young's modulus, Poisson's ratio, and yield strength; (2) boundary conditions inferred from the geometry and specification, such as fixed supports or applied loads; (3) mesh settings proposed for the finite element analysis, including element type and size; (4) anysimplifications or idealizations made to the model, such as neglecting small features or assuming symmetry.

[0011] The designer may then review these assumptions and provide feedback or corrections if necessary. For instance, the designer might adjust the material properties if a different grade of material is intended or modify the boundary conditions to better reflect the actual usage scenario of the part. This interaction may ensure that the subsequent computational 3D multi-physics analysis is based on accurate and relevant inputs, potentially improving the reliability and applicability of the results.

[0012] According to another aspect of the present disclosure, a method for automated computational 3D multi-physics analysis is provided. The method includes receiving a technical design specification, analyzing the technical design specification using a large language model-based system, configuring a simulation tool based on the analysis, performing a computational 3D multi-physics analysis using the simulation tool, and generating a report of the simulation analysis results.

[0013] According to other aspects of the present disclosure, the method may include one or more of the following features. The method may include fetching additional data from external databases or knowledge sources. The method may include validating assumptions with a designer before performing the computational 3D multi-physics analysis. The method may include analyzing the simulation analysis results and checking whether they meet the technical design specification. The method may include generating design improvement recommendations if the simulation analysis results do not meet the technical design specification. The method may include performing multiple iterations of the computational 3D multi-physics analysis and design improvement process until predefined design criteria are met.

[0014] The disclosure describes an iterative design process for optimizing structural components. This process involves multiple rounds of computational 3D multi-physics analysis simulations and design improvements based on LLM-generated recommendations.

[0015] Based on this iterative design process, improved technical components are generated. These components incorporate optimized geometries, material selections, and load distributions resulting from the design iterations, leading to enhanced performance and reliability. The final design may be obtained after a predefined number of loops or after meeting a predefined design criterion, such as a positive result of the computational 3Dmulti-physics analysis indicating that the analyzed component is able to cope with a predefined load case or boundary conditions.

[0016] The system for automated computational 3D multi-physics analysis comprises several important components that work together to streamline the simulation analysis process. FIG. 1 illustrates a block diagram of the system, showing the interconnections between these components.

[0017] The system includes a computer aided design-tool. The computer aided designtool may be a software application that allows designers to create, modify, and analyze 3D models of parts or assemblies. In certain cases, the computer aided design-tool may be Siemens NX or a similar CAD software package.

[0018] A simulation tool is configured to integrate with the computer aided design-tool. The simulation tool may perform computational 3D multi-physics analysis on the 3D models created in the computer aided design-tool. In certain cases, the simulation tool may be the NX Performance Predictor or a similar integrated simulation software. The simulation tool may be capable of analyzing various aspects of a design, such as stress distribution, deformation, and fatigue life.

[0019] The system further comprises a large language model-based system. The LLM-system may be a sophisticated artificial intelligence system capable of understanding and processing natural language inputs. In certain cases, the LLM-system may be based on GPT-4.0 or a similar advanced language model. The LLM-system may analyze technical specifications, interpret design requirements, and assist in configuring the simulation tool.

[0020] A geometry reasoning engine is also included in the system. The geometry reasoning engine may analyze the geometric features of the 3D model and suggest appropriate boundary conditions and loads for the simulation. In certain cases, the geometry reasoning engine may be referred to as a BC / Load Recommender, as shown in FIG. 1.

[0021] The geometry reasoning engine may incorporate advanced algorithms and machine learning techniques to analyze the geometric features of 3D models. In certain cases, the geometry reasoning engine may utilize computer vision algorithms to identify key important structural elements, such as beams, plates, or joints, within the 3D model. This analysis may help in automatically determining appropriate boundary conditions and load distributions for the simulation.

[0022] The geometry reasoning engine may also leverage historical data and design patterns to make informed suggestions about boundary conditions and loads. In certainimplementations, the engine may maintain a database of previous analyses and their corresponding geometric features, allowing it to draw parallels between new designs and past successful simulations.

[0023] In certain examples, the geometry reasoning engine may employ topological analysis techniques to identify important areas of the 3D model that are likely to experience high stress concentrations. This information may be used to automatically refine mesh density in these areas during the finite element analysis setup, potentially improving the accuracy of the simulation results.

[0024] The geometry reasoning engine may also include functionality to detect symmetry in the 3D model. In certain cases, this capability may be used to simplify the analysis by applying appropriate symmetry boundary conditions, potentially reducing computational requirements and simulation time.

[0025] Implementation examples of the geometry reasoning engine may include: (1) A neural network-based system trained on a large dataset of CAD models and their corresponding suitable boundary conditions and load cases. This system may be able to generalize its learning to new, unseen designs; (2) A rule-based expert system that applies heuristics derived from engineering principles and practices to suggest appropriate simulation setups based on geometric features; (3) A hybrid system that combines machine learning techniques with traditional geometric analysis algorithms, leveraging the strengths of both approaches to provide robust and accurate recommendations; (4) A cloud-based service that utilizes distributed computing resources to perform complex geometric analyses and simulations, allowing for real-time feedback and iterative design optimization.

[0026] In certain implementations, the geometry reasoning engine may provide visual feedback to the designer, highlighting areas of the model where it has identified potential issues or made specific recommendations. This visual feedback may be integrated directly into the CAD interface, allowing for intuitive interaction and rapid design iterations.

[0027] The system may configure the simulation tool based on the analysis performed by the LLM-based system in various ways. In certain cases, the LLM-based system may interpret the technical specifications and generate a configuration file that can be directly imported into the simulation tool. This configuration file may contain parameters such as material properties, mesh settings, boundary conditions, and load cases.

[0028] In other implementations, the LLM-based system may utilize an API provided by the simulation tool to programmatically set up the analysis parameters. The LLM-basedsystem may translate the natural language specifications into a series of API calls that configure the simulation environment, define the analysis type, and specify the output requirements.

[0029] The process of translating the LLM analysis into simulation tool settings may involve several steps. Initially, the LLM-based system may extract key information from the technical specifications, such as material types, expected load conditions, and performance requirements. It may then map this information to specific simulation parameters using a predefined ontology or knowledge base that relates engineering concepts to simulation settings.

[0030] In certain cases, the LLM-based system may employ a rule-based engine to translate high-level design requirements into detailed simulation parameters. This engine may contain a set of if-then rules that determine appropriate simulation settings based on the analyzed specifications.Example 1:

[0031] In a computational 3D multi-physics analysis of an automotive component, the LLM-based system may analyze a specification stating, "The part must withstand a 50 kN load in the vertical direction with a safety factor of 1.5." The system may translate this into simulation tool settings by: (1) Setting the load case to apply a 50 kN force in the negative Y-direction (assuming Y is the vertical axis); (2) Configuring the post-processing to calculate and report the safety factor; (3) Setting up a results filter to highlight areas where the safety factor falls below 1.5; (4) Adjusting the solver settings to ensure sufficient accuracy for safety factor calculations.Example 2:

[0032] For a thermal analysis of an electronic enclosure, the specification may state "The maximum temperature inside the enclosure should not exceed 70°C under normal operating conditions." The LLM-based system may configure the simulation tool by: (1) Setting up a steady-state thermal analysis; (2) Defining heat sources based on the power dissipation of internal components (which may be extracted from additional documentation);(3) Configuring material properties for the enclosure, including thermal conductivity and specific heat capacity; (4) Setting up convection boundary conditions to model heat dissipation to the environment; (5) Creating a temperature probe or result measure tospecifically track the maximum internal temperature; (6) Configuring the output to generate a warning if the maximum temperature exceeds 70°C.

[0033] In both examples, the LLM-based system may also suggest additional analyses or sensitivity studies based on its interpretation of the design requirements and best practices in the relevant engineering domain.

[0034] The system also includes a co-pilot, which may act as a central interface between the various components. The co-pilot may coordinate the interactions between the computer aided design-tool, the simulation tool, the LLM-system, and the geometry reasoning engine. In certain cases, the co-pilot may be referred to as the NX Co-pilot, as illustrated in FIG. 1.

[0035] In certain cases, the system may be configured to work with various simulation tools, including Computational Fluid Dynamics (CFD) software, thermal analysis tools, or multi-physics simulation platforms. This flexibility allows the system to address a wide range of engineering analysis requirements beyond computational 3D multi-physics analysis.

[0036] While the system described in this disclosure is capable of performing computational 3D multi-physics analysis across various domains, it offers particular advantages in the field of structural analysis. Structural analysis is a critical component of engineering design, especially in industries such as aerospace, automotive, and civil engineering, where the integrity and performance of structures are paramount.

[0037] The LLM-based system excels in interpreting complex structural requirements from natural language specifications. It can understand and translate concepts such as loadbearing capacity, stress distribution, fatigue life, and safety factors into precise simulation parameters. This capability is especially valuable in structural analysis, where accurate interpretation of design specifications is crucial for ensuring the safety and reliability of engineered structures.

[0038] In the context of structural analysis, the geometry reasoning engine plays a vital role. It can automatically identify key structural elements such as beams, columns, joints, and load-bearing surfaces within a 3D model. This automated recognition allows for more intelligent application of boundary conditions and loads, which is fundamental to accurate structural simulation.

[0039] The system's ability to suggest mesh refinements in areas of high stress concentration is particularly beneficial for structural analysis. It can automatically identify regions where finer mesh elements are needed, such as around holes, fillets, or other geometric features that typically experience stress concentrations. This adaptive meshingapproach can significantly improve the accuracy of stress calculations without unnecessarily increasing computational overhead.

[0040] For iterative design optimization, the system's capabilities are especially valuable in structural engineering. It can suggest design modifications that specifically target structural performance metrics, such as reducing weight while maintaining strength, optimizing the distribution of material to minimize stress concentrations, or adjusting geometries to improve buckling resistance.

[0041] The system's integration with material databases is particularly relevant for structural analysis. It can automatically select appropriate material models and properties, considering factors such as anisotropy, non-linearity, and temperature dependence, which are critical in advanced structural simulations.

[0042] Furthermore, the system's ability to handle multi-physics simulations is especially beneficial in structural analysis scenarios that involve coupled physics. For example, it can seamlessly integrate thermal effects into a structural analysis, accounting for thermal stresses and deformations, which is crucial in applications such as aerospace structures or power generation equipment.

[0043] By automating many of the complex setup and interpretation tasks involved in structural analysis, this system not only accelerates the design process but also helps to prevent errors that could arise from manual configuration. This is particularly important in structural engineering, where simulation errors could lead to catastrophic failures in real-world applications.

[0044] In summary, while the system offers broad capabilities in computational 3D multi-physics analysis, its features are particularly well-suited to enhancing the efficiency, accuracy, and sophistication of structural analysis tasks, making it an invaluable tool for structural engineers and designers.BRIEF DESCRIPTION OF THE DRAWINGS

[0045] FIG. 1 illustrates a block diagram of an LLM-based autonomous 3D computational 3D multi-physics analysis system, a 3D structural analysis system, according to aspects of the present disclosure.

[0046] FIG. 2 depicts a linear workflow diagram of a computational 3D multi-physics analysis process, according to an embodiment.DETAILED DESCRIPTION

[0047] The present disclosure relates to an automated system for computational 3D multi-physics analysis in computer-aided design environments. This system integrates advanced language processing capabilities with simulation tools to streamline and enhance the simulation analysis process. By leveraging large language models and sophisticated reasoning engines, the system aims to automate many of the complex tasks involved in setting up, running, and interpreting simulations.

[0048] The system described herein may be implemented as a non-transitory computer-readable medium storing instructions that, when executed by a processor, perform operations for automated computational 3D multi-physics analysis. This implementation allows for flexible deployment across various computing environments while ensuring consistent performance and functionality.

[0049] At its core, the system combines several key components to create a comprehensive solution for computational 3D multi-physics analysis. These components work in concert to interpret design specifications, configure simulation parameters, execute analyses, and generate meaningful reports. The integration of these elements enables a more efficient and accessible approach to simulation analysis, potentially reducing the time and expertise required to perform complex simulations.

[0050] By automating various aspects of the computational 3D multi-physics analysis workflow, the system aims to make advanced simulation capabilities more accessible to a broader range of users. This may include designers and engineers who may not have specialized expertise in simulation setup and interpretation. The system's ability to handle technical specifications, retrieve relevant data, and provide design recommendations represents a significant step towards democratizing advanced simulation analysis tools.

[0051] FIG. 1 illustrates a block diagram of an LLM-based autonomous 3D computational 3D multi-physics analysis system. The system comprises several interconnected components that work together to automate the simulation analysis process.

[0052] The system includes a computer aided design-tool CAD. The computer aided design-tool CAD may be a software application that allows designers to create, modify, and analyze 3D models of parts or assemblies. In certain cases, the computer aided design-tool CAD may be Siemens NX or a similar CAD software package.

[0053] A simulation tool SMT is configured to integrate with the computer aided designtool CAD. The simulation tool SMT may perform computational 3D multi-physics analysison the 3D models created in the computer aided design-tool CAD. In certain cases, the simulation tool SMT may be the NX Performance Predictor or a similar integrated simulation software. The simulation tool SMT may be capable of analyzing various aspects of a design, such as stress distribution, deformation, and fatigue life.

[0054] The system further comprises a large language model-based system LMS. The LLM-system LMS may be a sophisticated artificial intelligence system capable of understanding and processing natural language inputs. In certain cases, the LLM-system LMS may be based on GPT-4.0 or a similar advanced language model. The LLM-system LMS may analyze technical specifications, interpret design requirements, and assist in configuring the simulation tool SMT.

[0055] A geometry reasoning engine GRE is also included in the system. The geometry reasoning engine GRE may analyze the geometric features of the 3D model and suggest appropriate boundary conditions and loads for the simulation. In certain cases, the geometry reasoning engine GRE may be referred to as a BC / Load Recommender, as shown in FIG. 1.

[0056] The system also includes a co-pilot COP, which may act as a central interface between the various components. The co-pilot COP may coordinate the interactions between the computer aided design-tool CAD, the simulation tool SMT, the LLM-system LMS, and the geometry reasoning engine GRE. In certain cases, the co-pilot COP may be referred to as the NX Co-pilot, as illustrated in FIG. 1.

[0057] In certain cases, the system may be configured to work with various simulation tools, including Computational Fluid Dynamics [CFD] software, thermal analysis tools, or multi-physics simulation platforms. This flexibility allows the system to address a wide range of engineering analysis needs beyond computational 3D multi-physics analysis.

[0058] Referring to FIG. 1, the system for automated computational 3D multi -physics analysis may integrate with various data sources to enhance the accuracy and comprehensiveness of the simulation analysis process. The system may include a database DBS, which may be an external database containing relevant information for the analysis. Additionally, the system may incorporate a knowledge KWL component, which may serve as a repository of engineering principles, simulation best practices, and historical data.

[0059] In certain cases, the large language model-based system LMS may fetch additional data from the database DBS or the knowledge KWL component. This data retrieval process may be crucial for obtaining comprehensive information required for the computational 3D multi-physics analysis. For example, the LLM-based system LMS mayidentify and retrieve relevant material parameters MPR from a specific material database within the database DBS.

[0060] The material parameters MPR may be essential for conducting accurate structural analyses. These parameters may include properties such as Young's modulus, Poisson's ratio, yield strength, and thermal expansion coefficients. By accessing a dedicated material database, the system may ensure that the most up-to-date and accurate material properties are used in the simulation.

[0061] In certain cases, the system may pull data from a wider range of sources beyond the database DBS and knowledge KWL component. For instance, the system may integrate with supplier databases to automatically fetch component specifications. This integration may allow for more accurate modeling of complex assemblies that incorporate parts from various manufacturers.

[0062] Additionally, the system may connect to Internet of Things [loT] devices to incorporate real-world performance data into simulations. This capability may enable the system to validate and refine simulation models based on actual operational data, potentially improving the accuracy of future analyses.

[0063] The LLM-based system LMS may play a crucial role in interpreting and synthesizing data from these diverse sources. In certain cases, the LLM-based system LMS may analyze unstructured data from technical documents, research papers, or industry standards to extract relevant information for the computational 3D multi-physics analysis. This may allow the system to incorporate the latest industry knowledge and best practices into the analysis process.

[0064] By leveraging these various data sources and integration capabilities, the system may provide a more comprehensive and accurate computational 3D multi-physics analysis. The ability to access and utilize a wide range of relevant data may contribute to more informed decision-making in the design process and potentially lead to improved product performance and reliability.

[0065] FIG. 1 illustrates a system for automated computational 3D multi-physics analysis, preferably structural analysis, which includes various components for user interaction and interface. The system comprises a computer aided design-tool CAD that allows a designer DSG to input and manipulate design data.

[0066] In certain cases, a designer DSG may provide a specification SPC to the computer aided design-tool CAD. The specification SPC may include technical requirements, materialproperties, or other design parameters relevant to the computational 3D multi-physics analysis.

[0067] The system includes a co-pilot COP that facilitates interaction between the designer DSG and other components of the system. The co-pilot COP may interpret the specification SPC and coordinate the analysis process using various system components.

[0068] In certain cases, the co-pilot COP may validate assumptions with the designer DSG before performing the structural analysis. This validation step may ensure that the system's interpretation of the specification SPC aligns with the designer's DSG intent.

[0069] The system may generate a report RPT based on the structural analysis results. In certain cases, the report RPT may be customized based on the recipient. For example, the system may generate technical reports for engineers and simplified visual summaries for nontechnical stakeholders.

[0070] If the structural analysis results do not meet the requirements specified in the specification SPC, the system may generate a design recommendation DSR. The design recommendation DSR may suggest modifications to improve the design's performance or meet specific criteria.

[0071] In certain cases, the user interface of the system may be expanded beyond the traditional computer aided design-tool CAD environment. For example, the system may incorporate a voice-activated interface, allowing the designer DSG to verbally input specifications or request analyses.

[0072] Additionally, the system may include virtual reality [VR] or augmented reality [AR] interfaces. These interfaces may enable the designer DSG to visualize and interact with 3D models in a more immersive environment. For instance, the designer DSG may use VR to explore the structural analysis results or manipulate design elements in a virtual space.

[0073] The co-pilot COP may play a crucial role in facilitating these advanced interfaces, interpreting voice commands, or translating user interactions in VR / AR environments into actionable inputs for the structural analysis system.

[0074] The structural analysis process may be performed using the system as illustrated in FIG. 2. The process may begin with step 1 SOI, where a technical design specification SPC is received. In certain cases, the technical design specification SPC may be provided to a computer-aided design CAD co-pilot COP, such as an NX Co-pilot.

[0075] In step 2 S02, the system may analyze the technical design specification SPC using a large language model -based system LMS. The LLM-based system LMS mayinterpret the technical specifications and prepare configuration data for a simulation tool SMT.

[0076] Step 3 S03 may involve fetching additional data for the simulation. In certain cases, the LLM-based system LMS may retrieve relevant material parameters MPR from external databases or knowledge KWL bases.

[0077] Once the information has been gathered, the system may proceed to step 4 S04. In this step, the system may configure the simulation tool SMT based on the analysis performed by the LLM-based system LMS. The simulation tool SMT may then be initiated with the collected data to perform a structural analysis.

[0078] In step 5 S05, the simulation tool SMT may complete the structural analysis and generate simulation results SMR. These simulation results SMR may be returned to the LLM-based system LMS for further analysis.

[0079] Step 6 S06 may involve direct interaction between the LLM-based system LMS and the simulation tool SMT. In certain cases, the LLM-based system LMS may use API calls to the simulation tool SMT to identify specific data points, such as maximum stress STM values.

[0080] The process may continue with step 7 S07, where the LLM-based system LMS generates a report RPT of the structural analysis results SMR. This report RPT may be based on a predefined template and may be provided to the designer DSG or stored in a product lifecycle management PLM system.

[0081] In certain cases, the process may include step 8 S08, where the system analyzes whether the design DSN criteria have been met based on the simulation results SMR. If the criteria are not met, the system may generate design DSN recommendations for improving the design DSN. This may lead to an iterative process, where the improved design DSN specifications are fed back into step 1 SOI for further analysis and refinement.

[0082] In certain cases, the system may generate design recommendations DSR based on the structural analysis results SMR. The design recommendations DSR may be generated using a combination of rule-based algorithms and machine learning techniques. These recommendations may suggest modifications to various aspects of the design, such as geometry, material selection, or load distribution, to improve performance or meet specific criteria.

[0083] The system may analyze the stress distribution, deformation patterns, and safety factors obtained from the structural analysis to identify areas that require improvement. Incertain implementations, the system may use optimization algorithms to suggest specific changes that could enhance the design's performance while minimizing material usage or manufacturing complexity.

[0084] For example, in analyzing a bracket design for an aerospace application, the system may generate a design recommendation DSRthat suggests: (1) Increasing the fillet radius at high-stress comers to reduce stress concentrations; (2) Adding ribs in specific locations to improve stiffness without significantly increasing weight; (3) Changing the material from aluminum to a titanium alloy in critical areas to enhance strength-to-weight ratio; (4) Adjusting the hole placement to minimize stress around fastener locations.

[0085] The design recommendations DSR may be presented to the designer DSG through the co-pilot COP interface, potentially including visual representations or annotations on the 3D model to illustrate the suggested changes. In certain cases, the system may also provide quantitative estimates of the expected performance improvements for each recommendation, allowing the designer DSG to make informed decisions about which modifications to implement.

[0086] In certain cases, the system may analyze the structural analysis results SMR and check whether the structural analysis results SMR meet the technical design specification SPC. The system may compare various parameters from the structural analysis results SMR, such as maximum stress, deformation, or safety factors, against the requirements specified in the technical design specification SPC.

[0087] If the structural analysis results SMR do not meet the technical design specification SPC, the system may generate a design recommendation DSR. A design recommendation DSR may include suggestions for modifying the design DSN to improve its performance under the specified conditions. For example, a design recommendation DSR may suggest increasing the thickness of certain areas, changing the material, or modifying the geometry to reduce stress concentrations.

[0088] FIG. 2 illustrates an iterative process for design optimization. In step S08, if the design criteria have not been met, the system may provide recommendations on how to improve the design DSN. This process may involve multiple iterations of design specification input, structural analysis, and design change recommendations.

[0089] The system may perform multiple iterations of the structural analysis and design improvement process until predefined design criteria are met. These predefined designcriteria may include specific performance targets, such as maximum allowable stress or minimum safety factors, as defined in the technical design specification SPC.

[0090] In certain cases, the iterative process may continue for a predetermined number of cycles. Alternatively, the process may continue until a satisfactory design DSN is achieved that meets all specified criteria. For example, the final design DSN may be obtained after the structural analysis indicates that the analyzed component is able to cope with a predefined load case or boundary conditions.

[0091] The system may automatically implement the design recommendations DSR and initiate subsequent analysis cycles. In certain cases, the system may present the design recommendations DSR to a designer DSG for review and approval before proceeding with the next iteration.

[0092] By employing this iterative approach, the system may systematically refine and optimize the design DSN to meet the required specifications while considering various constraints and performance criteria.

[0093] The system integration and workflow of the automated structural analysis system combines various components to create a cohesive and efficient process. FIG. 1 illustrates a block diagram of the system, showing the interconnections between key components.

[0094] At the heart of the system is a co-pilot COP, which acts as a central hub for coordinating the various processes. The co-pilot COP interfaces with a computer aided design-tool CAD, allowing a designer DSG to input a specification SPC for the structural analysis.

[0095] The co-pilot COP communicates with an LLM-system LMS, which analyzes the specification SPC and configures a simulation tool SMT based on the analysis. In certain cases, the LLM-system LMS may fetch additional data from a knowledge base KWL or an external database DBS. This additional data may comprise material parameters MPR for use in the structural analysis.

[0096] A geometry reasoning engine GRE works in conjunction with the simulation tool SMT to determine appropriate boundary conditions and loads for the analysis. The simulation tool SMT then performs the structural analysis, generating simulation results SMR and determining maximum stress STM values.

[0097] In certain cases, the system may validate assumptions with the designer DSG before performing the structural analysis. This validation step helps ensure the accuracy and relevance of the analysis parameters.

[0098] Once the simulation is complete, the LLM-system LMS analyzes the simulation results SMR and checks whether they meet the original specification SPC. Based on this analysis, the system may generate a report RPT summarizing the findings.

[0099] FIG. 2 illustrates the workflow of the system in a series of steps S01-S08. The workflow begins with the input of the technical specification and progresses through the analysis, simulation, and reporting stages.

[0100] In certain cases, if the simulation results SMR do not meet the specification SPC, the system may generate a design recommendation DSR. This recommendation may be used to initiate an iterative process of design improvement and re-analysis until predefined design criteria are met.

[0101] The report RPT generated by the LLM-system LMS may be stored directly in a Product Lifecycle Management PLM system, facilitating seamless integration with existing design and engineering processes.

[0102] In certain cases, the report generation may be enhanced with advanced visualization techniques. These techniques may include interactive 3D models or animated simulations, providing a more comprehensive and intuitive representation of the structural analysis results.

[0103] By integrating these various components and processes, the system provides a streamlined and efficient workflow for autonomous structural analysis, reducing the need for manual intervention and potentially improving the speed and accuracy of the design process.

Claims

CLAIMS1. A system for automated computational three-dimensional (3D) multi-physics analysis, the system comprising:a computer-aided design (CAD) tool (CAD);a simulation tool (SMT) configured to integrate with the CAD tool; characterized in that the system further comprises:a large language model (LLM)-based system (LMS); anda geometry reasoning engine (GRE);wherein the system is configured to:receive a technical design specification (SPC);analyze the technical design specification using the LLM-based system; configure the simulation tool based on the analysis;perform a computational 3D multi-physics analysis using the simulation tool; andgenerate a report (RPT) of computational 3D multi-physics analysis results (SMR).

2. The system of claim 1, wherein the system is further configured to:fetch additional data from at least one of a knowledge base (KWL) and an external database (DBS).

3. The system of claim 2, wherein the additional data comprises material parameters (MPR) for use in the computational 3D multi-physics analysis.

4. The system of claim 1, wherein the system is further configured to:validate assumptions with a designer (DSG) before performing the computational 3D multi-physics analysis.

5. The system of claim 1, wherein the system is further configured to:analyze the computational 3D multi-physics analysis results (SMR) and check whether the computational 3D multi-physics analysis results (SMR) meet the technical design specification (SPC).

6. The system of claim 5, wherein the system is further configured to:generate design recommendations (DSR) when the computational 3D multi-physics analysis results (SMR) do not meet the technical design specification (SPC).

7. The system of claim 6, wherein the system is further configured to:perform multiple iterations of the computational 3D multi-physics analysis and design improvement process until predefined design criteria are met.

8. A method for automated computational three-dimensional (3D) multi-physics analysis, the method comprising:receiving a technical design specification (SPC);analyzing the technical design specification using a large language model (LLM)-based system (LMS);configuring a simulation tool (SMT) based on the analysis;performing a computational 3D multi-physics analysis using the simulation tool; and generating a report (RPT) of computational 3D multi-physics analysis results (SMR).

9. The method of claim 8, further comprising:fetching additional data from at least one of a knowledge base (KWL) and an external database (DBS).

10. The method of claim 9, wherein the additional data comprises material parameters (MPR) for use in the computational 3D multi-physics analysis.

11. The method of claim 8, further comprising:validating assumptions with a designer (DSG) before performing the computational 3D multi-physics analysis.

12. The method of claim 8, further comprising:analyzing the computational 3D multi-physics analysis results (SMR) and checking whether the computational 3D multi-physics analysis results (SMR) meet the technical design specification (SPC).

13. The method of claim 12, further comprising:generating design recommendations (DSR) when the computational 3D multi-physics analysis results (SMR) do not meet the technical design specification (SPC).

14. The method of claim 13, further comprising:performing multiple iterations of the computational 3D multi-physics analysis and design improvement process until predefined design criteria are met.

15. A non-transitory computer-readable medium storing instructions that, when executed by a processor, cause the processor to:receive a technical design specification (SPC);analyze the technical design specification using a large language model (LLM)-based system (LMS);configure a simulation tool (SMT) based on the analysis;perform a computational three-dimensional (3D) multi-physics analysis using the simulation tool; andgenerate a report (RPT) of computational 3D multi-physics analysis results (SMR).

16. The non-transitory computer-readable medium of claim 15, wherein the processor is further configured to:fetch additional data from at least one of a knowledge base (KWL) and an external database (DBS).

17. The non-transitory computer-readable medium of claim 16, wherein the additional data comprises material parameters (MPR) for use in the computational 3D multi-physics analysis.

18. The non-transitory computer-readable medium of claim 15, wherein the processor is further configured to:validate assumptions with a designer (DSG) before performing the computational 3D multi-physics analysis.

19. The non-transitory computer-readable medium of claim 15, wherein the processor is further configured to:analyze the computational 3D multi-physics analysis results (SMR) and checking whether they meet the technical design specification (SPC).

20. The non-transitory computer-readable medium of claim 19, wherein the processor is further configured to:generate design recommendations (DSR) when the computational 3D multi-physics analysis results (SMR) do not meet the technical design specification (SPC); and perform multiple iterations of the computational 3D multi-physics analysis and design improvement process until predefined design criteria are met.

Images