Aircraft design method and system based on intelligent agent system

Abstract

The invention belongs to the technical field of artificial intelligence, and discloses an aircraft design method and system based on an intelligent agent system.The system comprises a basic platform layer, a unified data layer, a model algorithm layer, a tool application layer and a user interaction layer, and the basic platform layer provides unified management heterogeneous computing power resources and provides a large-model full-life-cycle development environment; the unified data layer supports model training through data management and standardization construction of a data chain, training, deployment and dynamic optimization of open source and domain models are realized in the model algorithm layer, and the user interaction layer is provided with a comprehensive design workbench and a unified portal for users. And the tool application layer intelligently upgrades an aircraft design tool and constructs an automatic process of aircraft design through a Copilot mode and an Agent mode. According to the method, the cross-professional collaboration efficiency in aircraft design is remarkably improved, and the collaboration bottleneck in a traditional design process is solved.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology, and relates to an artificial intelligence system for aircraft design, and particularly to an aircraft design method and system based on an intelligent agent system. Background Technology

[0002] Aircraft design and development is a technology-intensive, multidisciplinary systems engineering project involving complex product integration, highly concurrent engineering, deep professional collaboration, and end-to-end knowledge fusion. Promoting the digital transformation of aircraft design by combining R&D processes with information systems and industrial software is currently one of the main ways to improve design efficiency and shorten iteration cycles. However, existing design methodologies still have significant shortcomings in addressing the demands of multidisciplinary collaboration and intelligentization, mainly manifested in low collaboration efficiency, low knowledge reuse rates, insufficient automation of toolchains, and a lack of data standardization.

[0003] While research has been conducted both domestically and internationally on intelligent aircraft design technologies, traditional methods such as manual coordination and distributed tool applications heavily rely on personnel experience when handling cross-disciplinary tasks, making it difficult to achieve automated optimization and real-time data verification. Different professionals' analyses of the same design problem often yield inconsistent results due to differences in experience, and the lack of a unified knowledge base hinders the systematic identification and resolution of potential design conflicts, further exacerbating the extended development cycle and quality risks.

[0004] To enhance the intelligence level and R&D efficiency of aircraft design, meet the requirements of multidisciplinary collaboration and data closure, ensure the full implementation of cross-disciplinary task coordination, knowledge reuse, and automated process orchestration, and promote the transformation of aircraft design from traditional manual drive to a human-machine collaborative agile and intelligent model, it is urgent to build a full-process automated design method based on intelligent agent system to achieve a comprehensive improvement in design efficiency, quality, and innovation capabilities. Summary of the Invention

[0005] The purpose of this invention is to solve the above problems. This invention provides an aircraft design method and system based on an intelligent agent system. By introducing cutting-edge digital and artificial intelligence technologies, it carries out the construction of a large-scale model in the aviation vertical field that is closely integrated with the aircraft R&D process, and builds a new R&D process of "large-scale model + aircraft design".

[0006] Technical solution: An aircraft design system based on an intelligent agent architecture includes a basic platform layer, a unified data layer, a model algorithm layer, a tool application layer, and a user interaction layer. The basic platform layer provides unified management of heterogeneous computing resources and a large-scale model lifecycle development environment. The unified data layer supports model training by building a data chain through data governance and standardization. The model algorithm layer implements the training, deployment, and dynamic optimization of open-source and domain-specific models. The user interaction layer includes a comprehensive design workbench and a unified portal for users. The tool application layer intelligently upgrades aircraft design tools and builds an automated process for aircraft design through Copilot and Agent modes. Specifically, the Copilot mode embeds intelligent agents as plug-ins into aircraft industrial software to improve human-computer interaction and decision-making efficiency, while the Agent mode uses intelligent agents and workflow orchestration technology to build an end-to-end automated toolchain for aircraft design.

[0007] A vehicle system design method based on an intelligent agent architecture, used to design the aforementioned vehicle design system based on an intelligent agent architecture, includes the following steps: S1 constructs a hierarchical architecture of intelligent agents, including system-level intelligent agents, cluster-level intelligent agents, and professional-level intelligent agents; S2 is the collaborative logic and workflow based on the R&D process settings for aircraft design. S3 deploys functional modules and toolchains related to aircraft design for intelligent agents; S4 provides intelligent agents with a professional knowledge graph and tool interface for aircraft design. S5 establishes a human-machine collaboration platform between intelligent agents and users; S6 collects design process data in real time, regularly evaluates the efficiency indicators of the intelligent agent in aircraft design, and completes data closure and continuous optimization. S7, based on the coverage of the intelligent agent in the stage design of the aircraft and the coverage of aviation industry standards, completes the adequacy check and standardization of the intelligent agent.

[0008] Furthermore, in S1, the system-level intelligent agent coordinates the top-level requirements in the aircraft design and decomposes them into large system design tasks to system-level intelligent agents. The system-level intelligent agents coordinate the large system design tasks and decompose them into subsystem design tasks to cluster-level intelligent agents. The cluster-level intelligent agents manage and execute the subsystem design tasks and call upon professional-level intelligent agents. The professional-level intelligent agents are embedded in the aircraft design tools to execute specific design tasks in the aircraft design and provide real-time optimization suggestions.

[0009] Furthermore, in S2, by analyzing the R&D process of aircraft design, the overall task of aircraft design is decomposed into sub-tasks that can be executed by intelligent agents, and the collaboration logic and workflow rules between intelligent agents are designed. The research and development process of aircraft design is broken down into sub-tasks based on the work breakdown structure, and a corresponding agent level is assigned to each task. The responsibilities of agents at different levels are clarified, including requirement decomposition, data verification, and specific algorithm execution. Then, the execution order of agents, data transmission paths, and verification nodes are defined through visualization tools to ensure collaboration across agent levels.

[0010] Furthermore, in S3, the digital agile design environment is implemented through private cloud deployment, with a unified data chain in the middle; the top-level intelligent agent decomposes the requirements according to the development stage and monitors the quality indicators of the entire process; the middle-level intelligent agent coordinates cross-disciplinary data transfer and refines tasks to the professional domain level; the professional-level intelligent agent is embedded in the design tool software through AI plug-ins; the execution tool layer adopts a human-computer interaction mode to embed AI capabilities into the existing tool software while providing real-time optimization suggestions.

[0011] Furthermore, in S4, historical design data is integrated to construct a knowledge graph to support agents in calling the knowledge base in real time to assist decision-making. The standardized integration of agents and design tool software is completed by uniformly calling the tool API interface through the MCP protocol; and a workflow orchestration tool is developed.

[0012] Furthermore, in S5, the human-machine collaboration platform integrates design software tools and AI plugins, and initiates preset or custom agent processes through conversational interaction.

[0013] Furthermore, in S6, design process data is collected in real time through a closed loop of "data-model-tool-user", system efficiency is evaluated regularly, and knowledge base content, toolchain configuration and workflow logic are iteratively optimized based on user feedback.

[0014] Furthermore, in S7, it verifies whether the design process covers all design stages, ensuring the integrity and consistency of cross-disciplinary interface data transfer.

[0015] The beneficial effects of this invention are as follows.

[0016] 1. This invention significantly improves cross-disciplinary collaboration efficiency and solves the collaboration bottleneck in traditional design processes. In traditional aircraft design processes, multidisciplinary collaboration relies on manual coordination, leading to ambiguous task allocation, data transmission delays, and frequent conflicts. This invention constructs a four-layer intelligent agent hierarchical architecture (system-level, cluster-level, and professional-level intelligent agents), realizing the autonomous decomposition and hierarchical execution of design tasks, fundamentally solving the problem of low efficiency in cross-disciplinary collaboration. First, this invention solves the task layering and transmission mechanism. Through the system-level intelligent agent (AIASoS), top-level requirements are analyzed, and tasks are decomposed step by step into executable sub-tasks at the system-level (AIAS), cluster-level (AIAA), and professional-level (AIA) levels. For example, in the discipline simulation optimization design, in the aerodynamic-load-strength-structure coupled design, AIASoS decomposes the "optimize lift-drag ratio" task into sub-tasks such as aerodynamic shape design (executed by the aerodynamic professional-level intelligent agent) and structural weight constraints (executed by the structural professional-level intelligent agent), and transmits intermediate data (such as airfoil parameters and structural weight) through real-time communication protocols between intelligent agents (such as MCP). Secondly, this invention performs real-time data verification and conflict detection. System-level intelligent agents coordinate cross-disciplinary data transfer, such as the dynamic verification of aerodynamic simulation results (lift coefficient, drag coefficient) and structural simulation results (stress distribution, weight). If a contradiction is found (e.g., aerodynamic optimization leading to excessive structural weight), the system-level intelligent agent triggers a conflict resolution process, automatically adjusting task priorities or reallocating resources. In traditional processes, such conflicts require repeated manual iterations, while this invention, through autonomous negotiation by intelligent agents, significantly shortens conflict resolution time. Furthermore, this invention integrates standardized toolchains. The MCP protocol unifies tool API interfaces (such as CAD, CAE, and PDM systems), supporting seamless integration of heterogeneous tools. For example, when an aerodynamic-level intelligent agent calls ANSYS Fluent for simulation, it can automatically use weight data provided by a structural-level intelligent agent as boundary conditions, avoiding manual configuration errors. Toolchain integration reduces cross-disciplinary data transfer latency. This invention also enables visual workflow orchestration. Through drag-and-drop configuration of the toolchain (e.g., a closed-loop process of "design → simulation → verification"), designers can quickly define the collaborative logic of the intelligent agents.

[0017] 2. This invention significantly improves knowledge reuse rate and design quality, overcoming the limitations of experience-based reliance. Traditional design heavily relies on engineers' experience, leading to implicit knowledge, low reuse rate, and a tendency for "repetitive work." This invention, by constructing a structured professional knowledge base and AI plugins, achieves explicit, systematic, and intelligent access to design experience, effectively improving knowledge reuse rate and design quality. In knowledge base construction and intelligent access, based on RAG (Retrieval Enhanced Generation) technology, historical design data, industry standards, and expert experience bases are transformed into a structured knowledge graph. During tasks, the intelligent agent can access knowledge nodes in the knowledge base, combine them with current task constraints, generate suggestions, and provide optimization suggestions through AI plugins for real-time design optimization.

[0018] 3. This invention achieves an end-to-end closed loop of design, simulation, and verification through agent-driven automated process orchestration and seamless integration of the toolchain, significantly shortening the iteration cycle. Traditional toolchains rely on manual operation, resulting in low automation levels, numerous process breakpoints, and low efficiency. This invention adopts an agent-based model for full-process automation, with professional-grade agents autonomously orchestrating the "design → simulation → verification" closed-loop process. For example, after the aerodynamic agent generates the initial airfoil, it automatically calls the CFD tool for simulation, feeds the results back to the structural agent for strength verification, and then the system-level agent evaluates the overall performance. Simultaneously, it enables adaptive adjustment of simulation parameters. For instance, when the structural CAE simulation detects excessive stress in a certain area, the AI ​​plugin automatically recommends "local stiffener design" and regenerates the geometric model without manual intervention.

[0019] 4. This invention ensures the reliability and scalability of the design process through data standardization and closed-loop optimization mechanisms. Traditional design data lacks unified standards, leading to difficulties in traceability, version confusion, and compliance risks. This invention achieves traceability, compliance, and self-evolution capabilities in the design process through standardized data management and continuous optimization closed loops. The MCP protocol is used to define cross-tool and cross-level data formats and transmission rules, ensuring consistency in unit format, precision, and semantics of input and output across all agents. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 The diagram shows the logical architecture of the intelligent agent system for the aircraft design of this invention.

[0022] Figure 2A schematic diagram of the digital agile design environment architecture that empowers the AI ​​intelligent agent of this invention.

[0023] Figure 3 This is a schematic diagram of the unified portal large model + AI tools of the present invention.

[0024] Figure 4 This is a schematic diagram of the integrated design workbench of the present invention.

[0025] Figure 5 A schematic diagram of intelligent agents and workflow planning for "multidisciplinary simulation optimization design". Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] The features and illustrative embodiments of various aspects of the present invention will now be described in detail. Numerous specific details are set forth in the following detailed description to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention may be practiced without requiring some of these specific details. The following description of embodiments is merely intended to provide a better understanding of the invention by illustrating examples of the invention. The invention is by no means limited to any specific setups and methods set forth below, but covers any improvements, substitutions, and modifications to structures, methods, and devices without departing from the spirit of the invention. Well-known structures and techniques are not shown in the drawings and the following description to avoid unnecessarily obscuring the invention.

[0028] In the description of this invention, it should be noted that the directions or positional relationships indicated by terms such as "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer" are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing and simplifying the invention, and should not be construed as limiting the invention. Furthermore, the use of ordinal numbers (e.g., "first and second," etc.) is for distinguishing objects and is not limited to this order, and should not be construed as indicating or implying relative importance.

[0029] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly, encompassing both direct connection and indirect connection via an intermediate medium. Those skilled in the art can understand the specific meaning of these terms in this invention based on the specific circumstances.

[0030] It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other, and the various embodiments can be referenced and cited in each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0031] Example 1.

[0032] An aircraft design system based on an intelligent agent architecture includes a basic platform layer, a unified data layer, a model algorithm layer, a tool application layer, and a user interaction layer. The basic platform layer provides unified management of heterogeneous computing resources and a large-scale model lifecycle development environment. The unified data layer supports model training by building a data chain through data governance and standardization. The model algorithm layer implements the training, deployment, and dynamic optimization of open-source and domain-specific models. The user interaction layer includes a comprehensive design workbench and a unified portal for users. The tool application layer intelligently upgrades aircraft design tools and builds an automated process for aircraft design through Copilot and Agent modes. Specifically, the Copilot mode embeds intelligent agents as plug-ins into aircraft industrial software to improve human-computer interaction and decision-making efficiency, while the Agent mode uses intelligent agents and workflow orchestration technology to build an end-to-end automated toolchain for aircraft design.

[0033] A vehicle system design method based on an intelligent agent architecture, used to design the aforementioned vehicle design system based on an intelligent agent architecture, includes the following steps: S1 constructs a hierarchical architecture of intelligent agents, including system-level intelligent agents, cluster-level intelligent agents, and professional-level intelligent agents; S2 is the collaborative logic and workflow based on the R&D process settings for aircraft design. S3 deploys functional modules and toolchains related to aircraft design for intelligent agents; S4 provides intelligent agents with a professional knowledge graph and tool interface for aircraft design. S5 establishes a human-machine collaboration platform between intelligent agents and users; S6 collects design process data in real time, regularly evaluates the efficiency indicators of the intelligent agent in aircraft design, and completes data closure and continuous optimization. S7, based on the coverage of the intelligent agent in the stage design of the aircraft and the coverage of aviation industry standards, completes the adequacy check and standardization of the intelligent agent.

[0034] In S1, the system-level intelligent agent coordinates the top-level requirements in the aircraft design and decomposes them into large system design tasks to system-level intelligent agents. The system-level intelligent agent coordinates the large system design tasks and decomposes them into subsystem design tasks to cluster-level intelligent agents. The cluster-level intelligent agent manages and executes the subsystem design tasks and calls on professional-level intelligent agents. The professional-level intelligent agents are embedded in the aircraft design tool to execute specific design tasks in the aircraft design and provide real-time optimization suggestions.

[0035] In S2, by analyzing the R&D process of aircraft design, the overall task of aircraft design is decomposed into sub-tasks that can be executed by intelligent agents, and the collaboration logic and workflow rules between intelligent agents are designed. The research and development process of aircraft design is broken down into sub-tasks based on the work breakdown structure, and a corresponding agent level is assigned to each task. The responsibilities of agents at different levels are clarified, including requirement decomposition, data verification, and specific algorithm execution. Then, the execution order of agents, data transmission paths, and verification nodes are defined through visualization tools to ensure collaboration across agent levels.

[0036] In S3, the digital agile design environment is implemented through private cloud deployment, with a unified data chain in the middle; the top-level intelligent agent decomposes the requirements according to the development stage and monitors the quality indicators of the entire process; the middle-level intelligent agent coordinates cross-disciplinary data transfer and refines tasks to the professional domain level; the professional-level intelligent agent is embedded in the design tool software through AI plug-ins; the execution tool layer adopts a human-computer interaction mode to embed AI capabilities into the existing tool software while providing real-time optimization suggestions.

[0037] In S4, historical design data is integrated to construct a knowledge graph to support agents in calling the knowledge base in real time to assist decision-making. The standardized integration of agents and design tool software is completed by calling the tool API interface in a unified manner through the MCP protocol. Workflow orchestration tools are also developed.

[0038] In S5, the human-machine collaboration platform integrates design software tools and AI plugins, and initiates preset or custom Agent processes through conversational interaction.

[0039] In S6, design process data is collected in real time through a closed loop of "data-model-tool-user", system efficiency is evaluated regularly, and knowledge base content, toolchain configuration and workflow logic are iteratively optimized based on user feedback.

[0040] In S7, verify whether the design process covers all design stages and ensure the integrity and consistency of cross-disciplinary interface data transfer.

[0041] Example 2: This invention addresses the problems of low cross-disciplinary collaboration efficiency, low knowledge reuse rate, insufficient toolchain automation, and lack of data standardization in traditional aircraft design processes. It proposes an aircraft design method and system based on an intelligent agent architecture. This method constructs a four-layer logical architecture comprising system-level, cluster-level, and specialty-level intelligent agents. It integrates the hierarchical task delivery achieved by the intelligent agent architecture, and incorporates key technologies such as knowledge base construction, MCP (Modular Communication Protocol) tool interfaces, and visual workflow orchestration to form an intelligent design closed loop. This solution significantly improves design efficiency, enhances knowledge reuse rate, ensures cross-disciplinary data consistency, and supports flexible expansion to adapt to the parallel development of multiple models, ultimately achieving a comprehensive improvement in aircraft design efficiency, quality, and collaborative capabilities. Through autonomous task decomposition by intelligent agents, real-time optimization by AI plugins (such as CAD parameter recommendation and CAE adaptive simulation), and end-to-end automated processes in an agent-based mode (such as "design → simulation → verification"), it supports flexible expansion to adapt to the parallel development of multiple models.

[0042] An aircraft design method and system based on intelligent agent architecture includes a basic platform layer, a unified data layer, a model algorithm layer, a tool application layer, and a user interaction layer.

[0043] Basic platform layer: Responsible for unified management of heterogeneous computing resources and providing a development environment for the entire lifecycle of large models; Unified Data Layer: Builds a high-quality data chain through data governance and standardization to support model training; Model algorithm layer: Implements the training, deployment, and dynamic optimization of open-source and domain-specific models; Tool application layer: Intelligent upgrade of traditional tools and construction of automated processes through Copilot and Agent modes; Copilot mode: Embedding AI capabilities into traditional industrial software to improve its human-computer interaction and decision-making efficiency; Agent pattern: Building an end-to-end automation toolchain through intelligent agents and workflow orchestration technology; User interaction layer: Enables human-machine collaboration and automated process operation through a comprehensive design workbench and unified portal.

[0044] The present invention then provides an aircraft design method and system based on an intelligent agent system, the method comprising the following steps: The first step is to build a hierarchical architecture for intelligent agents. The second step is to plan intelligent agents and workflows based on the R&D process. The third step is to deploy functional modules and toolchains. The fourth step is to build a professional knowledge base and tool interfaces. Fifth step: Build a human-machine collaboration platform Step 6: Data closed loop and continuous optimization Step 7: Adequacy Check and Standardization The specific principles of the aircraft design method and system based on intelligent agent architecture provided by this invention are as follows: Step 1: Construct an intelligent agent hierarchical architecture Based on the Work Breakdown Structure (WBS) of aircraft, a four-layer intelligent agent logical architecture is defined: the system-level intelligent agent (AIASoS) coordinates top-level requirements and decomposes tasks to system-level intelligent agents (AIAS); the system-level intelligent agents coordinate the design of the large system and allocate tasks to cluster-level intelligent agents (AIAA); the cluster-level intelligent agents manage subsystem designs and call upon professional-level intelligent agents (AIA); and the professional-level intelligent agents execute specific design tasks (such as structural optimization and aerodynamic simulation) through AI algorithms. This architecture achieves autonomous decomposition and execution of design tasks through hierarchical division of labor, ensuring cross-disciplinary collaboration and efficient resource scheduling. The intelligent agent architecture is as follows: Figure 1 As shown.

[0045] This invention, based on the Work Breakdown Structure (WBS) for aircraft, decomposes complex design tasks into manageable hierarchical subtasks and maps them to a four-layer intelligent agent logical architecture. Each layer of intelligent agents undertakes specific responsibilities, forming a top-down task decomposition and a bottom-up data feedback closed loop. The specific hierarchical division is as follows: The core responsibilities of the AIASoS (AI-based Intelligent Agent System) are: to coordinate top-level requirements (such as performance indicators, weight limits, and airworthiness standards), decompose aircraft development goals into system-level tasks, monitor the progress and quality indicators throughout the entire process, and coordinate cross-system resource allocation.

[0046] The core responsibilities of the System-Level Intelligent Agent (AIAS) are: to coordinate the design of large systems (such as aerodynamic-structural coupling and propulsion system matching) and to ensure cross-disciplinary data consistency (such as verifying aerodynamic simulation results against structural weight constraints). It decomposes system-level tasks into cluster-level tasks (e.g., breaking down "wing design" into sub-tasks such as "airfoil optimization," "structural layout," and "skin material selection"). The AIAS interacts with the AIAA cluster via the MCP protocol (Modular Communication Protocol), defining data transmission formats (e.g., aerodynamic simulation results in JSON format containing lift coefficient, drag coefficient, and pressure distribution data).

[0047] The core responsibilities of the Intelligent Agent Cluster (AIAA) are: to manage subsystem design (such as wings, fuselage, and avionics systems), to call upon specialized intelligent agents to perform specific tasks (such as finite element analysis and aerodynamic optimization), to monitor the progress and quality of subsystem design, to ensure compatibility with the overall architecture (such as matching avionics system wiring with structural holes), and to call upon the specialized algorithms and tools of the specialized intelligent agents (AIA) (such as structural topology optimization algorithms and aerodynamic CFD simulation tools) through API interfaces to complete tasks.

[0048] The core responsibility of the professional-grade intelligent agent (AIA) is to execute specific design tasks through AI algorithms and embed traditional tools (such as CAD and CAE) to provide real-time optimization suggestions (such as parameter recommendations and simulation adaptive adjustments).

[0049] Step 2: Design intelligent agents and workflows based on the R&D process By analyzing the R&D process, the overall task is decomposed into sub-tasks that can be executed by intelligent agents, and the collaboration logic and workflow rules between intelligent agents are designed. Specifically, this includes: Task mapping: Based on the work breakdown structure (WBS), the design process is broken down into sub-tasks, and each task is assigned a corresponding agent level (architecture level, system level, cluster level, professional level) to clarify its responsibilities (such as requirement decomposition, data verification, and specific algorithm execution). Agent and workflow planning: Define the execution order of agents, data transmission paths and verification nodes through visualization tools to ensure cross-level collaboration (e.g., a closed-loop process from requirement input to simulation verification).

[0050] Step 3: Deploy functional modules and toolchain The AI-powered digital agile design environment is deployed via a private cloud, enabling centralized deployment and access from underlying computing resources and large model bases to upper-level design workbench and unified portal. A unified and standardized high-quality data chain ensures consistency, security, and traceability of data across the entire chain. Through closed-loop linkage of "data-model-tools-users" in the R&D process, the efficient and secure operation of the AI-powered system is guaranteed.

[0051] The top-level intelligent agent decomposes requirements based on the development stage and monitors quality indicators throughout the entire process; the middle-level intelligent agent coordinates cross-disciplinary data transfer (such as verification of structural and aerodynamic simulation results) and refines tasks to the professional domain level; the professional-level intelligent agent embeds AI plugins into traditional tools (such as CAD parameter recommendation and CAE adaptive mesh generation); the execution tool layer adopts the Copilot mode of human-computer interaction (AI-assisted single-point work, improving human-computer interaction and decision-making efficiency by embedding AI capabilities into existing software tools (such as CAD, CAE, PDM, etc.) to provide real-time optimization suggestions, or orchestrates end-to-end automated processes (such as a "design → simulation → verification" closed loop) through an autonomous agent mode, achieving seamless toolchain integration and automated task execution. The functional architecture is as follows: Figure 2 As shown.

[0052] Step 4: Build a professional knowledge base and tool interfaces By leveraging RAG / Prompt technology (RAG-enhanced retrieval and prompt word engineering) to integrate historical design data, a structured knowledge graph is constructed, supporting agents to call the knowledge base in real time to assist decision-making. The MCP protocol is used to uniformly call tool API interfaces (such as CAD, CAE, and PDM systems) to achieve standardized integration of AI and tools. A visual workflow orchestration tool is developed, supporting drag-and-drop configuration of toolchains (such as the "automatic code generation → testing" process) and dynamically adjusting node logic to improve tool reusability and process flexibility.

[0053] Step 5: Build a human-machine collaboration platform The integrated design workbench combines traditional tools with AI plugins (such as CAD + AI parameter recommendations), providing an immersive collaborative environment that supports real-time multi-user collaboration and version control. The unified portal serves as an automated process scheduling center, initiating preset or custom agent processes through conversational interaction (such as "one-click generation of structural optimization solutions") and visually monitoring task progress (such as remaining simulation time). It integrates result delivery and feedback mechanisms, forming a dual-track collaborative ecosystem of "human-machine collaborative design" and "automated process management." Figure 3 , Figure 4 As shown.

[0054] Step Six: Data Loop and Continuous Optimization By collecting design process data in real time through a closed loop of "data-model-tool-user" (such as simulation results and user operation logs), and continuously injecting thought chains and knowledge to optimize the decision-making logic of the intelligent agent, the system efficiency indicators (such as design cycle shortening rate and error rate reduction rate) are regularly evaluated. Based on user feedback, the knowledge base content, toolchain configuration and workflow logic are iteratively optimized to ensure that the system can continuously adapt to complex design requirements and improve its intelligence level.

[0055] Step 7: Adequacy Check and Standardization Verify that the design process covers all stages (e.g., conceptual design → detailed design → verification), ensure the integrity and consistency of cross-disciplinary interface data transfer (e.g., matching structural weight with aerodynamic simulation input); follow aerospace industry standards, manage the iteration records of models, tools, and configurations through a version control system, achieve standardization, traceability, and compliance of the design process, and ultimately achieve high efficiency, reliability, and scalability of the entire design process.

[0056] Example 3: The following example illustrates the aircraft design method and system based on intelligent agent architecture provided by this invention through a multidisciplinary simulation optimization process in the design of the fuselage structure subsystem of a certain type of aircraft.

[0057] Building an intelligent agent hierarchical architecture For the design of the fuselage structure subsystem of a certain type of aircraft, based on its WBS (Work Breakdown Structure), "Multidisciplinary Simulation Optimization Design" is selected as a typical module to construct a four-layer intelligent agent hierarchical architecture, as shown in the table below.

[0058] Table 1. Hierarchical Architecture of the Intelligent Agent in "Multidisciplinary Simulation Optimization Design"

[0059] Based on R&D process intelligence agents and workflow planning Based on the "Multidisciplinary Simulation Optimization Design" R&D process, intelligent agents and workflows are planned. The "Multidisciplinary Simulation Optimization Design" R&D process can be roughly described as follows: First, establish an integrated optimization design framework for the entire load calculation process, including structural modal calculation, static-elasticity correction, maneuver simulation, and load calculation screening. Then, for the computationally complex and time-consuming numerical analysis process, establish a local response surface model based on a surrogate model to achieve high-efficiency global simulation. Finally, based on the set load and strength optimization objectives, conduct optimization design based on the global simulation process of the local surrogate model, realizing the coupled design of flight control, load, and strength disciplines. The planning of intelligent agents and workflows for the "Multidisciplinary Simulation Optimization Design" is as follows: Figure 5 As shown.

[0060] Deployment functional modules and toolchain In "Multidisciplinary Simulation Optimization Design", the following functional modules and toolchains are deployed: Table 2. Intelligent Agents and Corresponding Functions in "Multidisciplinary Simulation Optimization Design"

[0061] Building a professional knowledge base and tool interfaces Knowledge base construction Reports such as "Aircraft fuselage structure optimization" and "Aircraft fuel tank load calculation report" are extracted from historical cases and a knowledge graph is generated using RAG technology. The knowledge base is accessed through a question-and-answer format (e.g., inputting "material selection under high temperature environment" returns titanium alloy cases and parameters).

[0062] Tool interface standardization The APIs of CAD, CAE, and PDM systems or tools are encapsulated using the MCP protocol, supporting direct calls by intelligent agents (such as "Agent AIA requests CAE tool to perform simulation"); SDK toolkits are developed, and new tools (such as load spectrum construction tools) can be quickly integrated.

[0063] Workflow orchestration Visual drag-and-drop configuration toolchain, for example: Input: aerodynamic load data → Intelligent agent AIA generates cross-sectional parameters → CAE tool simulation → Output strength report.

[0064] Table 3 Examples of Analysis Rules

[0065] Building a human-machine collaboration platform Integrated Design Workbench It integrates CAD+AI plugins, allowing structural engineers and materials experts to edit 3D models simultaneously; and records iteration history through version control (e.g., "V1.0: Aluminum alloy used; V2.0: Titanium alloy used").

[0066] Unified Portal The program features a conversational start-up function (e.g., inputting "Generate structural optimization scheme" will automatically invoke the Agent mode process); it also provides visual monitoring of simulation progress (e.g., "Current progress: 70%, remaining time: 2 hours") and generates a comprehensive report containing simulation results and design documents.

[0067] Data closed loop and continuous optimization The system collects design process data (such as simulation results and user operation logs) in real time through a closed-loop "data-model-tool-user" system, continuously injecting and optimizing the agent's decision-making logic using knowledge from the thought chain. For example, if the cross-sectional dimensions recommended by the agent are repeatedly rejected manually, the system automatically records and optimizes the structural strength calculation logic. Efficiency indicators (such as design cycle shortening rate and strength analysis error rate) are periodically statistically analyzed, and the knowledge base content and workflow logic are iteratively optimized based on user feedback. Experimental data shows: Before optimization: Strength analysis took 120 hours, with a safety factor error rate of 25%; After optimization: Strength analysis took 70 hours, with a safety factor error rate of 8%. Adequacy Check and Standardization Design sufficiency verification: Check whether the entire process covers all stages (such as conceptual design → detailed design → verification) and ensure that the aerodynamic load data and structural simulation inputs are completely matched. Coverage analysis tools were used to confirm that all design tasks (such as node strength verification) had been completed.

[0068] Standardization and Compliance: Follow the requirements of the research and development procedures to ensure that task breakdown and tool usage comply with industry standards; By managing iterative records through baselines, traceability is supported (such as "V2.0 Improvement Measures: Strengthening Material Selection Rules").

[0069] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the embodiments described herein are for illustration and explanation only and are not intended to limit the invention.

[0070] In summary, for the design of the aircraft's fuselage structural subsystem, a multidisciplinary simulation optimization based on an intelligent agent system was implemented. By constructing a four-layer intelligent agent architecture (from system level to professional level) and deploying a knowledge base and toolchain, a total of 236 design optimization points were identified, including 38 structural weight redundancy issues, 157 cross-disciplinary data conflicts, and 41 simulation parameter mismatches. Targeted improvement measures were proposed (such as adding response surface model nodes and optimizing material selection rules). This significantly improved the collaborative efficiency, accuracy, and reliability of the aircraft's structural design, ensuring that key subsystems meet the requirements of multidisciplinary coupled optimization. Finally, it should be noted that the above embodiments are for illustration only and not for limiting the technical solutions of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention without departing from the spirit and scope of the present invention. Any modifications or partial substitutions should be covered within the scope of the claims of the present invention.

[0071] Other supplementary notes This invention is innovative in the following respects: a) Establish an aircraft design architecture and hierarchical model based on an intelligent agent system. By constructing a four-layer intelligent agent hierarchical architecture (system-level, cluster-level, and professional-level), AI technology is deeply integrated with the aircraft R&D process for the first time, forming an autonomous task decomposition and execution capability. This solves problems such as low efficiency of manual coordination and incomplete toolchain connection in traditional design, making the design process intuitive and operable, and significantly improving engineers' understanding and execution efficiency of complex tasks.

[0072] b) Intelligent execution mechanism integrating knowledge base and toolchain It innovatively combines a professional knowledge base (built using RAG technology), MCP protocol tool interface, and visual workflow orchestration technology to achieve a closed-loop design of "knowledge-driven + tool automation," which solves the shortcomings of traditional methods that rely on human experience and are difficult to qualitatively analyze, and has significant engineering practice value.

[0073] c) Standardized, scalable end-to-end design process A standardized process has been established, from requirements decomposition to toolchain orchestration, supporting layered expansion. A data closed-loop mechanism is used to continuously collect design process information and continuously check and verify the completeness of the entire process coverage. At the same time, in accordance with aviation industry standards, a version control system is used to ensure the traceability of the design process, significantly improving the standardization, completeness, and operability of the R&D process, and supporting the parallel development and continuous optimization of multiple models.

[0074] The above detailed embodiments are a description of the present invention. It should not be considered that the specific embodiments of the present invention are limited to these descriptions. For those skilled in the art, several simple deductions and substitutions can be made without departing from the concept of the present invention, and all of these should be considered to fall within the protection scope of the present invention.

Claims

1. An aircraft design system based on an intelligent agent architecture, characterized in that, It comprises a basic platform layer, a unified data layer, a model algorithm layer, a tool application layer, and a user interaction layer. The basic platform layer provides unified management of heterogeneous computing resources and a development environment for the entire lifecycle of large models. The unified data layer supports model training by building a data chain through data governance and standardization. The model algorithm layer enables the training, deployment, and dynamic optimization of open-source and domain-specific models. The user interaction layer features a comprehensive design workbench and a unified portal for users. The tool application layer intelligently upgrades aircraft design tools and builds automated processes for aircraft design through Copilot and Agent modes. Copilot mode embeds intelligent agents as plug-ins into aircraft industrial software to improve human-computer interaction and decision-making efficiency. Agent mode builds an end-to-end automated toolchain for aircraft design through intelligent agents and workflow orchestration technology.

2. A method for designing an aircraft system based on an intelligent agent architecture, used to design an aircraft design system based on an intelligent agent architecture as described in claim 1, characterized in that, Includes the following steps: S1 constructs a hierarchical architecture of intelligent agents, including system-level intelligent agents, cluster-level intelligent agents, and professional-level intelligent agents; S2 is the collaborative logic and workflow based on the R&D process settings for aircraft design. S3 deploys functional modules and toolchains related to aircraft design for intelligent agents; S4 provides intelligent agents with a professional knowledge graph and tool interface for aircraft design. S5 establishes a human-machine collaboration platform between intelligent agents and users; S6 collects design process data in real time, regularly evaluates the efficiency indicators of the intelligent agent in aircraft design, and completes data closure and continuous optimization. S7, based on the coverage of the intelligent agent in the stage design of the aircraft and the coverage of aviation industry standards, completes the adequacy check and standardization of the intelligent agent.

3. The aircraft system design method based on intelligent agent architecture according to claim 2, characterized in that, In S1, the system-level intelligent agent coordinates the top-level requirements in the aircraft design and decomposes them into large system design tasks to system-level intelligent agents. The system-level intelligent agent coordinates the large system design tasks and decomposes them into subsystem design tasks to cluster-level intelligent agents. The cluster-level intelligent agent manages and executes the subsystem design tasks and calls on professional-level intelligent agents. The professional-level intelligent agents are embedded in the aircraft design tool to execute specific design tasks in the aircraft design and provide real-time optimization suggestions.

4. The aircraft system design method based on intelligent agent architecture according to claim 2, characterized in that, In S2, by analyzing the R&D process of aircraft design, the overall task of aircraft design is decomposed into sub-tasks that can be executed by intelligent agents, and the collaboration logic and workflow rules between intelligent agents are designed. The research and development process of aircraft design is broken down into sub-tasks based on the work breakdown structure, and a corresponding agent level is assigned to each task. The responsibilities of agents at different levels are clarified, including requirement decomposition, data verification, and specific algorithm execution. Then, the execution order of agents, data transmission paths, and verification nodes are defined through visualization tools to ensure collaboration across agent levels.

5. The aircraft system design method based on intelligent agent architecture according to claim 2, characterized in that, In S3, the digital agile design environment is implemented through private cloud deployment, with a unified data chain in the middle; the top-level intelligent agent decomposes the requirements according to the development stage and monitors the quality indicators of the entire process; the middle-level intelligent agent coordinates cross-disciplinary data transfer and refines tasks to the professional domain level; the professional-level intelligent agent is embedded in the design tool software through AI plug-ins; the execution tool layer adopts a human-computer interaction mode to embed AI capabilities into the existing tool software while providing real-time optimization suggestions.

6. The aircraft system design method based on intelligent agent architecture according to claim 2, characterized in that, In S4, historical design data is integrated to build a knowledge graph to support agents in calling the knowledge base in real time to assist decision-making. The standardized integration of agents and design tool software is completed by calling the tool API interface in a unified manner through the MCP protocol. Develop workflow orchestration tools.

7. The aircraft system design method based on intelligent agent architecture according to claim 2, characterized in that, In S5, the human-machine collaboration platform integrates design software tools and AI plugins, and initiates preset or custom Agent processes through conversational interaction.

8. The aircraft system design method based on intelligent agent architecture according to claim 2, characterized in that, In S6, design process data is collected in real time through a closed loop of "data-model-tool-user", system efficiency is evaluated regularly, and knowledge base content, toolchain configuration and workflow logic are iteratively optimized based on user feedback.

9. The aircraft system design method based on intelligent agent architecture according to claim 2, characterized in that, In S7, verify whether the design process covers all design stages and ensure the integrity and consistency of cross-disciplinary interface data transfer.

Images