Large-language-model-based intelligent design optimization system for large aluminum alloy castings

Abstract

The invention discloses an intelligent design optimization system for a large aluminum alloy casting based on a large language model, and aims to improve the casting design efficiency, optimize material selection and improve the data sharing safety. The system comprises five modules: an intelligent design analysis module, a material selection and optimization module, a multi-modal data analysis module, a feedback learning module and a block chain data management module. The intelligent design analysis module performs semantic analysis on historical design documents and standards by using a large language model, automatically generates a structure optimization scheme, and reduces stress concentration and defect risks. And the material selection and optimization module recommends an optimal aluminum alloy material formula based on design requirements and performance indexes, and proposes process parameter adjustment suggestions. And the multi-modal data analysis module integrates the text, the simulation data and the material characteristics to generate a design optimization scheme. The feedback learning module collects production feedback data, continuously optimizes the model, and enhances the design adaptability. And the block chain data management module records and protects key data in the design process to ensure the integrity and security of the data. The system can remarkably improve the casting design efficiency and quality, has good industrial application prospects, and is suitable for design and optimization of castings of complex structures such as high-speed rails, aerospace and high-voltage power transmission and transformation.

Description

Technical Field

[0001] The present invention relates to the fields of casting technology and artificial intelligence technology, and in particular to an intelligent design optimization system combined with a large language model, which is suitable for structural design optimization, material selection and performance improvement of large aluminum alloy castings. Background Art

[0002] Large aluminum alloy castings are widely used in high-speed rail, aerospace, and ultra-high voltage power transmission and transformation. They have complex structures and large sizes, and require extremely high design accuracy and performance. However, traditional casting design optimization methods usually rely on finite element analysis (FEM) and expert experience. These methods have the following shortcomings in processing multimodal data, generating design optimization solutions, and adapting to different application scenarios: Low design efficiency: Traditional design optimization relies on multiple simulation analyses and manual adjustments, which is time-consuming and prone to human errors.

[0003] Lack of intelligent material selection and optimization: Existing design methods rely on experience to make decisions on material selection, and it is difficult to comprehensively consider microstructure and performance indicators, resulting in low material utilization efficiency.

[0004] Difficulties in data sharing and knowledge protection: In large-scale project collaboration, there are great challenges in sharing and protecting design data and optimization solutions, especially when multiple parties are involved.

[0005] With the rapid development of large language models (such as Tongyi Qianwen), they have demonstrated strong capabilities in natural language understanding, knowledge reasoning and generative analysis. Applying large language models to the design optimization of large aluminum alloy castings can significantly improve design efficiency and quality. Summary of the invention

[0006] The present invention aims to solve the following technical problems: (1) Improve the design optimization efficiency of large aluminum alloy castings and reduce stress concentration and structural defects; (2) Provide intelligent material selection and optimization solutions to improve casting performance and material utilization; (3) Realize trusted data sharing and protection in the design process, and support multi-party collaboration and intellectual property protection.

[0007] In order to solve the above technical problems, the present invention proposes an intelligent design optimization system for large aluminum alloy castings based on a large language model, and the system includes the following modules: (1) Intelligent design analysis module: Uses a large language model to analyze historical design documents, standard specifications, and case data to automatically generate structural optimization solutions. This module aims to reduce common stress concentration and structural defects in traditional design methods through natural language processing (NLP) and deep learning technology, improve design accuracy, and optimize design efficiency.

[0008] The core functions of the intelligent design analysis module include: Historical design data analysis: Use the big language model to analyze historical design documents (such as design drawings, design reports, historical cases, etc.) to identify potential design defects and structural problems. The big language model can understand and summarize the structural and performance data in these documents, extract valuable information, and help designers understand common problems in past designs and avoid repeating mistakes.

[0009] Semantic understanding and optimization generation: Through semantic analysis of industry standards and design specifications, the intelligent design analysis module can automatically extract and match standard requirements related to the current design, and generate a design solution that meets the specifications based on these requirements. For example, the system will automatically identify the design elements of the casting, such as wall thickness, structural dimensions, reinforcement areas, etc., and combine the stress analysis results to recommend design modification suggestions to reduce stress concentration and avoid excessive stress areas.

[0010] Stress analysis and structural optimization: The intelligent design analysis module can automatically identify possible stress concentration areas such as sharp corners, sharp turns or discontinuous structures by combining stress distribution data, and provide optimization suggestions (such as adding fillets or improving transition areas). By working in conjunction with other modules, the module can effectively reduce stress concentration and improve the structural strength and reliability of castings.

[0011] The workflow of the intelligent design analysis module is as follows: Input data: System input includes historical design documents, casting design parameters, industry standard specifications, design drawings, etc. These data can be structured data (such as CAD drawings, engineering drawings) or unstructured data (such as technical documents, reports). For unstructured data, the system uses a large language model to perform text parsing and data extraction to extract meaningful information.

[0012] Data preprocessing: For different types of input data, the system first performs data cleaning and preprocessing. Text data is processed by word segmentation, denoising, vectorization, etc. Drawing data may need to be converted into a standardized CAD format or other suitable input format. The system will standardize various types of data to ensure data consistency and operability.

[0013] Data analysis and semantic matching: The big language model conducts in-depth analysis of design documents, industry standards and cases, identifies standards, specifications and experience related to the current design, and generates semantic matching results based on specific application scenarios. For example, it analyzes the wall thickness, support structure, and accessory requirements of the current casting, and provides optimization suggestions for designers based on common design solutions in historical cases.

[0014] Design optimization generation: Based on the analysis results, the system will automatically generate multiple optimization design solutions. These solutions will take into account factors such as material properties, mechanical properties, and manufacturing processes, and strive to achieve the best in terms of structural strength, durability, and production efficiency. After the solution is generated, the system will provide an analysis of the advantages and disadvantages of each solution to help designers choose the most appropriate solution.

[0015] Output results: The system outputs the generated design plan as a detailed report or engineering drawing for reference by design engineers. The report not only contains structural optimization suggestions, but also provides explanations and reasoning processes for each optimization result, so that designers can understand the principles behind the optimization suggestions. The report may also include potential risk assessments for design implementation and propose corresponding improvement measures.

[0016] Special Features Real-time feedback and adaptability: The intelligent design analysis module has real-time feedback. Designers can adjust the design plan based on real-time production data, test results, etc. The system will further train and optimize the model based on this feedback information to improve the accuracy of the model in specific applications.

[0017] Collaborative design and multi-scenario adaptability: This module can not only handle the design optimization of a single casting, but also support the collaborative work of multiple design teams. In the multi-person collaborative mode, the system can provide personalized design suggestions for different team members to ensure the overall consistency and efficiency of the design solution.

[0018] Key technical points Reasoning capability of large language models: The intelligent design analysis module relies on the powerful reasoning capability of large language models. It can combine natural language processing, deep learning algorithms and expert knowledge to automatically extract valuable information and optimize the design based on this information.

[0019] Stress and defect detection: By analyzing stress concentrations and defects in historical designs, the intelligent design analysis module can automatically identify potential problems and make structural optimization suggestions to reduce stress concentrations, avoid local failures and other common design defects.

[0020] Optimization algorithm and multi-objective optimization: This module not only focuses on the strength and performance of the structure, but also considers multiple objectives such as the manufacturability of the casting, production cost and material utilization. The system uses a multi-objective optimization algorithm to find the best design solution under multiple constraints.

[0021] (2) Material selection and optimization module: Combined with the large language model and material science database, it analyzes the performance and microstructural characteristics of aluminum alloy materials and recommends the optimal material formula and process parameter adjustment suggestions. The core purpose of this module is to improve the material performance and manufacturing quality of castings, ensure that the material utilization efficiency is maximized and the production cost is reduced while meeting the design requirements.

[0022] The core functions of the material selection and optimization module include: Material performance analysis: This module can analyze a large amount of aluminum alloy material performance data through a large language model, including but not limited to the mechanical properties, thermal properties, corrosion resistance, conductivity, etc. of aluminum alloys. These performance data come from material science databases, literature and production practices. By comparing different material properties and design requirements, the module can automatically recommend the type of material suitable for the current casting design.

[0023] Microstructure Analysis: This module can analyze the microstructure of aluminum alloys in detail, considering the effects of different components and treatment processes on the microstructure of aluminum alloys (such as grain size, precipitation phase, solid solubility, etc.). By combining the large language model with the material science database, the module can accurately identify the microstructure characteristics of the material under different processes and predict its potential impact on the performance of the casting.

[0024] Process optimization suggestions: Based on the analysis of aluminum alloy materials, the material selection and optimization module can automatically generate process adjustment suggestions based on the application scenario, size, performance requirements and other conditions of the casting. For example, the system can recommend the alloying element ratio, heat treatment process, casting temperature and cooling rate of aluminum alloy to improve the strength, corrosion resistance and ductility of the casting.

[0025] Multi-material comparison and screening: This module is not limited to selecting one material. It can compare multiple aluminum alloy materials, combine the design requirements of the casting, conduct comprehensive evaluation from multiple aspects, and finally select the material that best meets the requirements. This process includes a comprehensive trade-off of multiple factors such as material cost, performance, and machinability.

[0026] The workflow of the material selection and optimization module is as follows: Input data: The input data of this module includes basic information of aluminum alloy materials (such as chemical composition, performance indicators), design requirements of castings (such as load-bearing capacity, corrosion resistance, heat treatment requirements, etc.), and historical production data. The data can come from different sources, such as material science databases, casting production records, laboratory test results, etc.

[0027] Material properties and microstructure analysis: The material selection and optimization module first conducts an in-depth analysis of the input aluminum alloy material performance data through a large language model. This process includes matching the material performance data with the microstructural characteristics of the aluminum alloy and evaluating the performance of different aluminum alloys under different production process conditions. For example, by analyzing the effects of alloy composition and heat treatment process on grain size and precipitation phase, its effects on casting strength, plasticity and fatigue properties are predicted.

[0028] Generate recommended solutions: Based on the material performance analysis results and the design requirements of the casting, the module automatically generates optimal material formula recommendations. These recommendations take into account the optimal ratio of alloy elements, optimized heat treatment processes, and other process parameters required in the manufacturing process (such as casting temperature, cooling rate, pouring temperature, etc.). This process involves not only the selection of a single material, but also the comparison between different materials to ensure that the final recommended solution achieves the best balance between performance and cost.

[0029] Performance simulation and verification: While recommending materials, the module also performs simulation analysis, such as mechanical properties simulation, thermal analysis, corrosion evaluation, etc. By working in collaboration with other modules, it simulates the performance of materials under actual process conditions and further optimizes process parameters to improve material performance and casting quality.

[0030] Output results: The module finally outputs the recommended material formula, process adjustment suggestions and performance prediction report. The report includes the chemical composition, performance indicators, microstructure characteristics of the material, and the process parameters required to achieve the best performance. Design engineers can make final decisions based on this report, select appropriate materials, and develop detailed process plans accordingly.

[0031] Key technical points

[0032] Reasoning and analysis capabilities of large language models: The material selection and optimization module relies on the reasoning and analysis capabilities of large language models to extract the most critical material property information by analyzing a large amount of material data, technical literature, and production practices. The large language model can quickly understand the relationship between material performance and design requirements, and automatically generate optimization solutions based on these relationships.

[0033] Correlation analysis between microstructure and performance: The module can deeply analyze the microstructure of aluminum alloys, identify the effects of different alloy compositions and heat treatment processes on the microstructure, and then infer their potential effects on casting performance. For example, the relationship between the grain size of aluminum alloys and their strength and ductility can be analyzed, and appropriate alloy compositions and heat treatment processes can be recommended to achieve optimal casting performance.

[0034] Multivariable optimization: During the process of process parameter optimization, the module can consider the interaction of multiple variables, such as the proportion of alloying elements, temperature and cooling rate of the casting process, etc. Through the multi-objective optimization algorithm of the large language model, the module can automatically find the optimal combination of process parameters to ensure the maximum performance of the casting.

[0035] Cost-effectiveness analysis: During the material selection process, the module also considers the cost-effectiveness of the material, combines the material's performance and market price, and automatically recommends a cost-effective material formula, helping to reduce production costs while meeting performance requirements.

[0036] Special Features

[0037] Real-time feedback and online optimization: The material selection and optimization module can perform online optimization based on feedback data from the actual production process. For example, if the performance of a certain material is found to be not as good as expected during the actual casting process, the system will adjust the model based on real-time feedback data to improve the recommended material and process solutions.

[0038] Collaboration and intelligent decision-making: This module can work with other modules in the system (such as the intelligent design analysis module and the multimodal data analysis module) to share information and provide comprehensive decision support. For example, the material selection module can adjust the recommended material formula based on the stress distribution data generated by the design scheme to ensure the best match between material selection and structural design.

[0039] (3) Multimodal data analysis module: Integrates text data, simulation analysis results and material properties to generate multimodal design optimization solutions. This module can provide more comprehensive design suggestions by comprehensively analyzing various forms of data, improve the overall performance of castings, reduce defects, improve production efficiency, and adapt to various changes under different design and process conditions, ultimately providing the best solution for the intelligent design of castings.

[0040] The core functions of the multimodal data analysis module include: Data integration and multimodal analysis: This module can integrate data from different sources, such as historical design documents, simulation analysis results (e.g., structural stress distribution, thermal flow simulation results), material properties, experimental data, production feedback, etc. Through the large language model, the module can perform semantic analysis and association on these different types of data, identify potential optimization opportunities, and generate targeted design suggestions.

[0041] Intelligent interpretation of simulation analysis results: By cooperating with the simulation analysis module, this module can understand and interpret simulation results such as stress distribution diagrams and heat flow analysis diagrams. These simulation results are generated based on finite element analysis (FEA), fluid dynamics simulation (CFD) and other means, and can describe the performance of castings under different processes and loading conditions. The multimodal data analysis module combines these simulation results with other data (such as material properties, design requirements, etc.) to propose design improvement suggestions, such as optimizing geometry, adjusting casting parameters or selecting different material combinations.

[0042] Comprehensive design optimization: Based on multiple data sources, the module can generate multi-dimensional design optimization solutions. It not only considers material selection and structural design, but also improves the design of castings based on simulation analysis results (such as stress, temperature distribution, etc.). For example, if the simulation analysis shows that there is a large stress concentration in certain areas, the system will automatically recommend increasing the thickness of the area, adjusting the structural shape, or optimizing the cooling system to reduce stress concentration and design defects.

[0043] Automatic design suggestion generation: Based on multimodal data analysis, the module can automatically generate a series of design optimization suggestions and output a complete design improvement plan. These suggestions include geometry optimization, material selection adjustment, process parameter fine-tuning, etc., and will provide a reasonable parameter range. Designers can make decisions quickly and implement optimization based on these suggestions.

[0044] The workflow of the multimodal data analysis module includes the following steps: Data input: The module's input data includes historical design documents, standard design specifications, material properties, simulation analysis results (such as mechanical properties, thermal flow analysis, cooling simulation results, etc.), experimental data, production feedback information, etc. The data source can come from the design team, simulation tools, production lines, testing systems, etc.

[0045] Data preprocessing and analysis: The module first preprocesses the input text data, simulation results, and material properties. This process includes data cleaning, standardization, format conversion, and other operations. Subsequently, the module uses a large language model to perform semantic analysis and feature extraction on these data to identify potential key factors and relationships. For example, the system can analyze common design defects in historical design documents and compare them with simulation results to infer possible causes.

[0046] Simulation results and material property analysis: By combining simulation analysis results, the module can gain in-depth understanding of the performance of castings under specific process conditions. For example, by analyzing the stress distribution diagram, the module can identify areas of stress concentration and compare them with the mechanical properties of the material (such as tensile strength, ductility, etc.) to derive possible optimization solutions. Similarly, the results of thermal flow simulation can reveal the thermal stress and temperature distribution of castings, which affect the consolidation of materials and the quality of castings. The system can automatically adjust design parameters such as cooling rate, pouring temperature, etc. to avoid defects.

[0047] Multi-dimensional optimization solution generation: The module generates multi-dimensional optimization solutions by comprehensively analyzing simulation results, material properties and design requirements. These solutions not only include suggestions for design modifications, but also include optimized process parameters (such as pouring temperature, cooling rate, pouring method, etc.). For example, when the module finds that there is too high stress in certain areas of the casting, it will recommend increasing the thickness of the area or adjusting the layout of the cooling system to reduce the problem of stress concentration.

[0048] Optimization solution output: After completing data analysis and design optimization, the design solution output by the module includes the following: Structural optimization suggestions: suggestions for improvement in stress distribution, heat flow, material properties, etc.; Material optimization recommendation: Based on the multimodal analysis results, the system will recommend appropriate material formula and processing technology; Process parameter adjustment: Based on the simulation results and material properties, the system recommends adjusting process parameters such as cooling rate and casting temperature to improve casting quality.

[0049] Design decision support: The module will eventually provide designers with a detailed optimization report with visual results (such as stress diagrams, temperature distribution diagrams, material distribution diagrams, etc.). Designers can make decisions quickly and optimize casting design based on these reports and suggestions.

[0050] Key technical points

[0051] Multimodal data integration and correlation analysis: One of the key technologies of the multimodal data analysis module is the ability to effectively integrate data from different sources. It can not only understand and analyze simulation results, but also correlate with historical design data, material properties, production feedback, etc., to generate a more comprehensive and accurate design optimization solution.

[0052] Intelligent interpretation and optimization of simulation results: This module can automatically interpret complex simulation analysis results, such as stress distribution diagrams, temperature field diagrams, flow field diagrams, etc., and can combine these results with other design parameters (such as materials, geometric shapes, etc.) to automatically generate optimization solutions.

[0053] Feedback learning and adaptive optimization: In the actual production process, the module can collect production feedback data in real time, including defect information, production efficiency, etc., so as to continuously optimize the analysis model and design optimization plan. Based on the reasoning ability of the large language model, the module can adaptively adjust the design plan to cope with changes in the production environment.

[0054] Interdisciplinary optimization algorithms: The interdisciplinary optimization algorithms used in the multimodal data analysis module can combine knowledge from different disciplines (such as mechanics, thermodynamics, materials science, etc.) to perform comprehensive design optimization. These algorithms can find a balance between different design parameters to ensure the best overall performance of the casting.

[0055] Special Features

[0056] Real-time simulation and optimization: This module can be connected to the online simulation system in real time. When the simulation conditions change, the system can immediately update the design plan and give new optimization suggestions.

[0057] Design data visualization: The module can display complex simulation results and design optimization suggestions in a graphical way, such as stress distribution diagram, temperature field diagram, etc., so that designers can understand and evaluate the design scheme more intuitively.

[0058] Automatic report generation: After completing the design optimization, the module will automatically generate an optimization report and semantically optimize the report content through a large language model, making the report easy to understand and convenient for designers to refer to.

[0059] (4) Feedback learning module: collects production feedback data and performs online learning and optimization based on a large language model. The main purpose of this module is to improve the accuracy of design optimization suggestions, respond to changes in the production environment, and continuously optimize the design process so that the system can adaptively adjust and optimize when facing different production conditions. Through feedback learning, the system can dynamically improve the quality of the design solution, reduce casting defects, improve production efficiency, and ultimately achieve intelligent casting design and manufacturing.

[0060] The core functions of the feedback learning module include: Production feedback data collection: The module can collect key data in the production process in real time by connecting with production lines, quality inspection systems, sensors, etc. Including but not limited to casting defect data (such as pores, cracks, inclusions, etc.), process parameters in the production process (such as temperature, pressure, pouring speed, etc.), inspection reports, operator feedback information, etc. These data provide the system with real-time feedback based on actual production conditions, which is the basis for improving design and optimizing processes.

[0061] Data analysis and diagnosis: Through the large language model, the feedback learning module can conduct in-depth analysis of the collected feedback data, and automatically identify the root cause of the problem by combining previous design data and simulation results. The system can identify potential defects in design, materials, processes, etc., and propose specific optimization measures. For example, if the production data shows that some castings have cracking problems caused by stress concentration, the module can analyze whether it is related to the structural features in the design (such as transition angles, inappropriate geometric shapes, etc.) or material properties (such as mismatched tensile strength), and provide corresponding adjustment suggestions.

[0062] Model-based online learning: The feedback learning module continuously adjusts and optimizes the large language model through the input of real-time feedback data. Whenever the system obtains new production feedback, the model will be adjusted through the online learning mechanism so that it can more accurately identify design problems under different production conditions. The online learning process does not require human intervention. The system can autonomously analyze new data, modify or expand previous design rules, and ensure continuous improvement of the system. This mechanism makes the system highly adaptive and can cope with different production environments, casting types and process conditions.

[0063] Optimization strategy update: Based on the analysis results, the module can update the design optimization strategy. For example, if the feedback data shows that a certain aluminum alloy has excessive expansion at a specific temperature, the system can automatically adjust the temperature control suggestions based on historical data and online learning results to reduce the deformation and defects of the casting. Through continuous feedback learning, the system can not only optimize the specific design plan, but also form a new process strategy to improve the quality of castings and production stability.

[0064] Adaptive adjustment and optimization suggestion generation: The system can adaptively optimize the design scheme, process flow and material formula through feedback learning. Based on production feedback, the system will automatically propose suggestions for adjusting the design or process, such as improving the cooling rate, adjusting the pouring temperature, optimizing the alloy composition ratio, etc. These optimization suggestions are highly practical and targeted, and can effectively solve problems that arise in actual production.

[0065] The workflow of the feedback learning module includes the following steps: Production feedback data collection: This module collects key data in the production process in real time by connecting with production equipment, quality inspection systems, sensors, etc. The data content includes but is not limited to defect statistics (such as pores, cracks, inclusions, etc.), production process parameters (such as temperature, pressure, speed, etc.), operator feedback information, inspection reports, etc. These data provide valuable basis for subsequent analysis.

[0066] Data analysis and problem diagnosis: The system combines the collected feedback data with historical design data, simulation results and material properties, and conducts in-depth analysis and diagnosis through a large language model. The model can identify potential problems in the production process, such as pores on the surface of castings and cracks inside, and determine whether these defects are related to the design structure, material properties or process parameters.

[0067] Online learning and model updating: Based on the feedback data during the production process, the system will conduct online learning. Through the reasoning ability of the GPT-4 model, the system will analyze the difference between the newly input feedback data and the historical data, and automatically adjust the optimization strategy. For example, if certain castings frequently have certain defects in a specific production environment, the system will adjust its diagnostic rules and optimization strategies to cope with different production conditions.

[0068] Design and process optimization suggestions: The feedback learning module will automatically generate design or process optimization suggestions based on the new learning results. These suggestions can be structural adjustments (such as thickening weak areas, adding cooling channels, etc.) or process adjustments (such as modifying the pouring temperature, controlling the cooling rate, etc.). Suggestions can also include material adjustments (such as choosing a more suitable aluminum alloy formula) to adapt to changes in the production environment.

[0069] Optimization suggestion output and decision support: Finally, the system will generate an optimization suggestion report, which contains analysis, optimization suggestions and adjustment plans for defect problems. Designers and process engineers can make decisions quickly based on the report content, implement design or process adjustments, and ensure the improvement of casting production quality.

[0070] Key technical points

[0071] Real-time data collection and dynamic feedback: This module can obtain feedback data from the production line in real time, and make dynamic adjustments based on information such as defects and process parameters in actual production to ensure that the system can always reflect the actual production situation.

[0072] Online learning and adaptive optimization: The large language model has powerful reasoning and learning capabilities, and can automatically learn and optimize through production feedback data without human intervention. This online learning mechanism enables the system to continuously improve and perfect the design optimization plan and enhance its adaptive capabilities.

[0073] Interdisciplinary data fusion: The feedback learning module can not only process data from different fields (such as design, materials, processes, quality inspection, etc.), but also combine these data for comprehensive analysis to provide interdisciplinary comprehensive solutions for design and process optimization.

[0074] Intelligent optimization suggestions: Based on the reasoning ability of the large language model and the support of feedback data, the system can generate specific and targeted optimization suggestions to solve practical problems in casting production. These suggestions are of high value in terms of accuracy and practicality.

[0075] Adaptability to production environment: Through online learning, the feedback learning module can adapt to changes in different production environments. In the event of abnormal situations or new process conditions during the production process, the system can quickly adjust the optimization strategy and provide new design or process adjustment solutions.

[0076] The feedback learning module can be widely used in various foundry production environments, especially in the following scenarios: Large-scale casting production: In the process of mass production of aluminum alloy castings, real-time feedback and optimization can help quickly adjust process parameters and design solutions to improve production efficiency and product quality.

[0077] Diversified product design: For different casting products, production conditions and process parameters may be different. Through the feedback learning module, the system can adaptively adjust the design and process for different product types to meet diverse needs.

[0078] High-demand quality control: In the production of aluminum alloy castings with high quality requirements (such as aerospace, automotive and other fields), real-time feedback and optimization can significantly reduce the defect rate and ensure that the quality of castings meets strict standards.

[0079] (5) Blockchain data management module: Blockchain technology is used to record key data in the design optimization process and realize trusted data storage and sharing. Through the immutability, transparency and distributed storage characteristics of blockchain, this module can ensure the integrity and security of all key design data, support multi-party cooperation and data sharing, and protect intellectual property rights to prevent data from being maliciously tampered with or leaked. The core goal of the blockchain data management module is to achieve data security, traceability, data sharing and cooperation in the casting design process, especially in multi-party cooperation and cross-border projects, to ensure the trust of all parties in the data and enhance the efficiency of collaboration.

[0080] The core functions of the blockchain data management module include: Unalterable record of design data: Through blockchain technology, all key data in the design process (such as design solutions, optimization suggestions, simulation results, material formulas, etc.) are recorded in the form of blocks. Once the data is recorded in the blockchain, it cannot be modified or deleted. This unalterable feature ensures the transparency of the design process and the authenticity of the data. In this way, not only can human tampering of data be avoided, but the source of each design decision can also be traced.

[0081] Data sharing and multi-party collaboration support: The distributed ledger of blockchain technology enables multiple design teams, suppliers, and manufacturers to share design data without mutual trust. All parties can obtain the latest design solutions, optimization suggestions, and other key information by simply accessing the blockchain network. Blockchain technology supports cross-departmental, cross-organizational, and even cross-national collaboration by providing a secure and transparent data sharing platform. Through smart contracts, participants can automatically execute specific agreements, such as data sharing and intellectual property protection during the design optimization process.

[0082] Data integrity assurance: Blockchain ensures the integrity of all data during storage by encrypting and hashing the design data. Each time design optimization data is generated, a unique fingerprint is generated through the hash algorithm. These fingerprints will be recorded in the blockchain. Any tampering will cause the hash value to be inconsistent, which can be detected at the first time, thereby ensuring the security and consistency of the data.

[0083] Automatic execution of smart contracts: Through smart contracts, blockchain modules can automatically execute predetermined tasks, such as data verification, permission control, payment settlement, etc. Smart contracts can ensure that all parties abide by the agreement when participating in design optimization, avoiding contract disputes or data security issues caused by human negligence. For example, when designing data sharing, smart contracts can automatically verify permissions and confirm which data is available for viewing and modification to prevent unauthorized access and tampering.

[0084] Intellectual property protection: The encryption and storage methods of blockchain can effectively protect the intellectual property rights of designers. During the design process, all data upload, modification and sharing records will be permanently stored in the blockchain, forming a traceable historical record. These records can not only prevent external malicious tampering, but also provide strong evidence for intellectual property disputes, ensuring that the intellectual property rights of designers in cooperation are protected.

[0085] The workflow of the blockchain data management module can be divided into the following steps:

[0086] Data recording and encryption: When new data (such as design solutions, simulation results, material analysis, etc.) is generated during the design optimization process, the system will encrypt this data through the blockchain and generate a hash value. These hash values ​​and encrypted data will be added to the blockchain as a block. Each newly generated block will depend on the previous block to ensure that the historical chain of data is coherent and cannot be tampered with.

[0087] Execution of smart contracts: When multiple design teams or suppliers collaborate, smart contracts can automatically execute specific tasks according to pre-set rules. For example, when one party uploads new design data, the smart contract automatically triggers the data sharing and access control mechanism to ensure that the collaborators access and modify the data in accordance with the agreement. If there are design changes or optimization suggestions, smart contracts can ensure that all relevant parties are notified in a timely manner and respond to the data.

[0088] Multi-party data sharing and collaboration: The permissions of each party on the blockchain platform are automatically controlled by smart contracts. Partners, designers, suppliers, etc. can access shared design data according to the agreement. Through the blockchain, the system can ensure that all shared data is authentic and complete, and each design plan and optimization suggestion can be clearly traced back to the original record, avoiding the risk of data leakage and tampering.

[0089] Data query and audit: All parties involved in design optimization can query all data stored on the blockchain at any time and access specific data blocks and their historical records through the blockchain browser. This process can effectively help auditors or project managers verify the authenticity and integrity of data, especially in the context of project handover or intellectual property protection, to provide necessary evidence.

[0090] Data update and tracking: Blockchain technology is not only suitable for data storage, but also for tracking data updates. During the design optimization process, if there are new design improvements or modifications, the system will store these updated data on the blockchain, recording the time, changes and participants of each design optimization. These updates will form a clear chain record, so that every step in the design process can be traced and verified.

[0091] Key technical points

[0092] The immutability of blockchain: Blockchain uses hashing algorithms and encryption technology to ensure that data cannot be tampered with. Once data is recorded, it cannot be modified or deleted. Any tampering with the data will cause the blockchain to break and issue a warning in time. This feature ensures the authenticity and reliability of data during the design process.

[0093] Encryption technology and data privacy protection: Through the encryption mechanism of blockchain, the transmission and storage of design data are guaranteed to be secure. Even when multiple parties share data, blockchain can still protect the privacy and sensitive information of all parties and prevent unauthorized access and data leakage.

[0094] Automation and decentralized management of smart contracts: Smart contracts ensure the automatic management and execution of design data and optimization suggestions. Through smart contracts, the system can customize rules and automatically execute them, such as automatically updating design data after uploading, or controlling data access rights based on protocols, thereby reducing manual intervention and improving collaboration efficiency.

[0095] Distributed storage and data redundancy: Blockchain stores data through distributed ledgers, and all participants can access the data in the blockchain. Redundant storage of data ensures high availability of information and ensures that all parties can access the latest design optimization solutions and related data at any time.

[0096] Applicable scenarios

[0097] Multi-party collaborative design projects: In projects involving multiple design teams, manufacturers, suppliers, etc., the blockchain data management module provides a secure and transparent data sharing platform to ensure that every step in the design process can be traced and verified. Whether it is material selection, structural optimization or process improvement, all design data can be securely shared among different partners.

[0098] International cooperation projects: In cross-border or international projects, design teams from different countries and regions often face the problem of data sharing and privacy protection. Blockchain technology can ensure the safe sharing of design data on a global scale, avoid information leakage and protect the intellectual property rights of all parties.

[0099] Intellectual property protection: Blockchain records every key node in the design optimization process, forming an unalterable historical record, providing designers with strong intellectual property protection. This not only prevents malicious tampering, but also provides strong evidence for patent applications, technology transfer, etc. Beneficial Effects

[0100] The present invention has the following beneficial effects: Intelligent design and optimization: Analyze design data through large language models, automatically generate optimization solutions, and significantly improve design efficiency.

[0101] Intelligent material selection: Combining the material science database with a large language model, the system can recommend the optimal material formula to improve casting performance.

[0102] Multimodal analysis and feedback learning: Through multimodal data analysis and feedback learning, the system can continuously optimize design suggestions and adapt to different application scenarios.

[0103] Trusted data sharing and protection: Use blockchain technology to achieve trusted sharing of design data and knowledge protection, and support multi-party collaboration. BRIEF DESCRIPTION OF THE DRAWINGS

[0104] Figure 1: Schematic diagram of the overall system architecture.

[0105] Figure 2 : Workflow diagram of the intelligent design analysis module.

[0106] Figure 3 : Analysis flow chart of material selection and optimization module.

[0107] Figure 4 : Optimized schematic diagram of the multimodal data analysis module.

[0108] Figure 5 : Model update flowchart of the feedback learning module.

[0109] Figure 6 :Data storage diagram of blockchain data management module. DETAILED DESCRIPTION

[0110] Example 1 like Figure 2 As shown in the figure, the intelligent design analysis module uses a large language model to analyze historical design documents, standard specifications, and case data. Through semantic analysis, the system extracts key information related to stress concentration and design defects, and generates optimization suggestions, including structural improvements and material adjustment suggestions. The module can identify common design problems such as sharp corners and stress concentration in transition areas, and automatically recommends adding fillets or adjusting structural dimensions to reduce stress.

[0111] like Figure 3 As shown in the figure, the material selection and optimization module combines the big language model and the material science database to analyze the chemical composition, microstructure characteristics and performance indicators of aluminum alloys. The system can recommend the optimal material formula according to design requirements and application scenarios, and make corresponding process adjustment suggestions, such as adjusting the proportion of alloy elements to improve tensile strength and corrosion resistance.

[0112] like Figure 4 As shown in the figure, the multimodal data analysis module integrates text data, simulation results (such as stress distribution diagrams, heat flow simulation results) and material property data. The system performs comprehensive analysis based on the large language model and generates optimization solutions, such as adjusting design parameters, optimizing cooling rate and adjusting pouring temperature, to improve the first-time pass rate of castings.

[0113] like Figure 5 As shown in Figure 1, the feedback learning module collects actual feedback data from production (such as defect statistics, inspection reports, and process adjustment records) to update and train the large language model online. The system uses new feedback data to optimize design suggestions and improve the accuracy and adaptability of the design.

[0114] like Figure 6As shown in the figure, the blockchain data management module records the key data nodes in the design optimization process through blockchain technology, and uses smart contracts to automatically store and share data to ensure data integrity and security. This module supports permission control and data traceability to protect the intellectual property rights of the design team.

[0115] The large-scale aluminum alloy casting intelligent design optimization system based on the large language model proposed in this invention uses artificial intelligence models and blockchain technology to achieve intelligent optimization of casting design, scientific decision-making on material selection, and trusted protection of data sharing. In practical applications, the system can significantly improve design efficiency and product quality, and has broad industrial application prospects and significant economic benefits.

Claims

1. An intelligent design optimization system for large aluminum alloy castings based on a large language model, characterized in that: The system comprises: Intelligent design analysis module, which uses a large language model to analyze historical design documents, industry standards and case data, automatically generates the optimal casting structure design solution, and reduces stress concentration and design defects; The material selection and optimization module combines the large language model and material science database to intelligently recommend aluminum alloy material formulas based on design requirements and performance indicators, and puts forward corresponding process adjustment suggestions; Multimodal data analysis module, which is used to comprehensively analyze text data, simulation results and material properties to generate multimodal optimization solutions to improve casting structural performance and manufacturing efficiency; Feedback learning module, which is used to collect actual production feedback data and continuously optimize the large language model to improve the accuracy and practicality of design suggestions; The blockchain data management module is used to record data and analysis results in the design optimization process, realize trusted storage and sharing of data through blockchain technology, and ensure the integrity and security of design data.

2. The system according to claim 1, characterized in that The intelligent design analysis module can combine the global casting design database, perform semantic analysis on the design requirements of different application scenarios, and generate personalized structural optimization solutions.

3. The system according to claim 1, characterized in that The material selection and optimization module can analyze the microstructural characteristics of aluminum alloy materials and generate the optimal material formulation scheme in combination with the reasoning ability of the large language model to improve the strength and corrosion resistance of the castings.

4. The system according to claim 1, characterized in that The multimodal data analysis module can integrate simulation analysis tools, combine simulation results such as stress distribution and thermal flow analysis, optimize design parameters, and output design improvement suggestions.

5. The system according to claim 1, characterized in that The feedback learning module collects inspection data, defect statistics and operator feedback information during the production process, and uses a large language model to perform online learning and model updates to enhance the system's design optimization capabilities and adaptability.

6. The system according to claim 1, characterized in that The blockchain data management module realizes automatic recording and sharing of data through smart contracts, ensures that all data in the design optimization process cannot be tampered with, and supports multi-party collaboration and knowledge protection.

7. The system according to claim 1, characterized in that The intelligent design analysis module can generate a variety of design solutions, perform semantic interpretation and performance evaluation through a large language model, and output a highly readable optimization report for design engineers to refer to and make decisions.

8. The system according to claim 1, characterized in that The multimodal data analysis module can automatically analyze historical simulation data and experimental data, and combine the large language model reasoning results to propose design parameter adjustment suggestions based on performance indicators to improve the first-time pass rate of castings.

9. The system according to claim 1, characterized in that The blockchain data management module can hash and store key data nodes in the design process and support authority control to ensure the security and privacy of design data.

10. The system according to claim 1, characterized in that The system can be deployed via the cloud, supporting remote access and real-time collaboration, enabling multiple design teams to share data and design suggestions and perform collaborative optimization.

Images