Semantic-driven aircraft function system optimization design method and system

Abstract

The invention belongs to the technical field of aircraft design, particularly relates to a semantic-driven aircraft function system optimization design method and system, and aims to solve the problem of function system optimization which can decouple design logic and consider efficiency and accuracy. The method comprises the following steps: constructing a quinary semantic modeling set of an aircraft function system, and generating a structured design instruction according to the quinary semantic modeling set; assembling static parameters, dynamically inputting and structuring a design instruction by using an Assemble module to obtain a complete design instruction; and by taking the complete design instruction as input, performing optimization by using LLM until an optimization termination condition is met, and obtaining an optimization result. According to the semantic-driven aircraft function system optimization design method and system provided by the invention, an efficient and reliable scheme is provided for rapid optimization design of the aircraft function system through semantic driving.

Description

Technical Field

[0001] This invention belongs to the field of aircraft design technology, specifically relating to a semantic-driven aircraft functional system optimization design method and system, as well as an electronic device. Background Technology

[0002] Aircraft functional systems serve as a crucial bridge connecting overall design and component implementation, and their architecture directly determines the performance, quality, maintainability, and safety of the entire aircraft. With the increasing integration of modern aircraft, the coupling between functional systems such as electromechanical, hydraulic, fuel, thermal management, and avionics is becoming increasingly tight, leading to strong nonlinearity and cross-level coupling characteristics in functional system design.

[0003] Traditional aircraft functional system design often relies on a cyclical process of "human experience - simulation verification - manual correction," which is essentially an exploratory design driven by implicit experience and lacks formal, analytical solution mechanisms. This design pattern not only has a lengthy design cycle and high R&D costs, but also makes it difficult to ensure the consistency and repeatability of design solutions in complex engineering systems, failing to meet the dual requirements of modern aircraft for design accuracy and efficiency.

[0004] With the development of multi-objective solution technology, analytical / gradient optimization, heuristic algorithms, and data-driven methods based on surrogate models have made some progress in parameter optimization and multi-objective trade-offs, providing new technical ideas for the optimization design of aircraft functional systems. However, when applied to the optimization of aircraft functional systems, they still face significant limitations: 1) Gradient optimization methods rely on continuously differentiable objective functions and constraint equations, while aircraft functional systems have multi-physics coupling effects, which easily form black-box simulation scenarios, making gradient optimization methods difficult to apply; 2) Swarm intelligence and evolutionary algorithms require evaluation through a large number of samples to ensure the sufficiency of the search. Their computational cost increases exponentially with the increase of the design space dimension, and it is difficult to effectively embed complex constraint logic and physical mechanisms, resulting in insufficient adaptability; 3) Surrogate models are prone to accuracy degradation in complex system design, especially in characterizing the discontinuous characteristics of discrete variables, which cannot guarantee the reliability of the design scheme; 4) Knowledge engineering and model-based systems engineering (MBSE) methods mostly remain at the logical structure level, lacking deep integration with dynamic simulation and multi-objective trade-offs, making it difficult to support the optimization design of complex functional system architectures.

[0005] In summary, existing technologies are insufficient to effectively address the multidisciplinary collaboration, strong coupling, and high-dimensional design requirements of modern aircraft functional systems. There is an urgent need for a functional system optimization scheme that can decouple design logic while balancing efficiency and accuracy. Summary of the Invention

[0006] To address the aforementioned technical problems in the prior art, namely how to achieve functional system optimization that can decouple design logic while balancing efficiency and accuracy, this application provides a semantic-driven aircraft functional system optimization design method and system.

[0007] In a first aspect of this application, a semantic-driven method for optimizing the design of aircraft functional systems is provided, comprising:

[0008] Constructing a set of five-element semantic models for the functional systems of an aircraft , For design issues, For the design goals, For design space, As a key principle, To formalize the solution, structured design instructions are generated based on the five-element semantic modeling set. ;

[0009] The Assemble module is used to assemble static parameters, dynamic inputs, and the structured design instructions to obtain complete design instructions. The static parameters are the static prior parameters input during initialization, and the dynamic inputs are the current design baseline and the current optimization trajectory.

[0010] Using the complete design instructions as input, optimization is performed using a Large Language Model (LLM) until the optimization termination condition is met, resulting in the optimization results of the aircraft functional system. The optimization results include component parameter tables and control logic descriptions.

[0011] Optionally, the step of using the complete design instructions as input and optimizing with a Large Language Model (LLM) until the optimization termination condition is met includes:

[0012] The output of the large language model LLM in the nth round of optimization iteration is obtained based on the complete design instructions. , Let n represent the semantic reasoning of a large language model (LLM), where n is the number of iterations, n≥1 and n is an integer. Based on the current design benchmark, Optimize the current trajectory;

[0013] The output of the Large Language Model (LLM) is processed using semantic decoding operators. Convert to executable physical parameters;

[0014] The executable physical parameters are input into the simulation model of the aircraft functional system to perform simulation and obtain simulation results.

[0015] The evaluation result of the nth round of optimization iteration is obtained based on the simulation results. ;

[0016] Update and optimize the trajectory to obtain a new optimized trajectory. , The time series operator is used, and the nth violation degree is obtained based on the simulation results. If the nth violation degree is less than the historical optimal solution violation degree, the design baseline is updated to obtain a new design baseline, and the historical optimal solution violation degree is updated to the nth violation degree. If the nth violation degree is greater than or equal to the historical optimal solution violation degree, the current design baseline is used as the new design baseline, and the nth violation degree is the violation degree of the nth round of optimization iteration.

[0017] Determine whether the optimization termination condition is met;

[0018] If the optimization termination condition is not met, increment the iteration number n by 1, take the new optimization trajectory as the current optimization trajectory, take the new design benchmark as the current design benchmark, take the current optimization trajectory and the current design benchmark as the dynamic input, and return to the step of using the Assemble module to assemble static parameters, dynamic input and the structured design instructions to obtain complete design instructions;

[0019] If the optimization termination condition is met, the optimization iteration process is terminated, and the optimization result of the aircraft functional system is obtained based on the output of the large language model LLM in the nth round of optimization iteration.

[0020] Optionally, the evaluation result of the nth round of optimization iteration is obtained based on the simulation results. ,include:

[0021] Key performance indicators are extracted based on the simulation results, and key performance feature vectors are obtained based on the key performance indicators.

[0022] Calculate the evaluation result ,in, This represents the key performance feature vector for the nth round of optimization iteration. Let be the refueling time for the i-th fuel tank in the n-th round of optimization iteration. The maximum fuel flow rate in the j-th segment of the pipeline during the nth round of optimization iterations. The maximum fuel flow rate of the k-th valve in the nth round of optimization iteration. The threshold value for the refueling time of the fuel tank. This is the constraint threshold for the maximum fuel flow rate in the pipeline. This is the constraint threshold for the maximum fuel flow rate of the valve. These refer to the number of oil tanks, pipes, and valves, respectively. This is an indicator function; if the condition inside the parentheses is true, then... Takes the value 1, otherwise The value is 0.

[0023] Optionally, satisfying the optimization termination condition includes:

[0024] The semantic reasoning of the Large Language Model (LLM) converges, or the number of iterations is greater than or equal to a preset iteration threshold.

[0025] Optionally, the current optimized trajectory is defined as a semantic record sequence. .

[0026] Optional, initial design baseline As a blank scheme, the initial optimized trajectory This is a sequence of blank semantic records.

[0027] Among them, the design problem The design objects and functional objectives used to define the aircraft's functional systems are specification objects with unique identifiers instantiated using the ReqIF standard; the design objectives The quantitative performance indicators used to define the functional systems of the aircraft are attributes of the specification object; the design space The key principles are used to define the design boundaries of the aircraft's functional systems and are formally described by the GOPPRRE meta-model ontology specification. Used to define high-level constraints based on mechanistic principles and design experience; the formal expression of the solution. This is used to describe the mapping structure of the design scheme in semantic space and physical space. The design scheme is constructed as a topological graph composed of objects, relations and roles based on the meta-meta-model ontology specification GOPPRRE.

[0028] Optionally, the semantic-driven aircraft functional system optimization design method uses a meta-meta-modeling method based on the M0-M3 modeling framework to construct the five-element semantic modeling set.

[0029] Optionally, the semantic-driven aircraft functional system optimization design method uses a task assembly function to generate structured design instructions based on the five-element semantic modeling set.

[0030] In a second aspect of this application, a semantically driven aircraft functional system optimization design system is provided, the system comprising:

[0031] The semantic transformation module is used to construct a five-element semantic modeling set for the functional systems of an aircraft. , For design issues, For the design goals, For design space, As a key principle, To formalize the solution, structured design instructions are generated based on the five-element semantic modeling set. ;

[0032] The Assemble module is used to assemble static parameters, dynamic inputs, and the structured design instructions to obtain complete design instructions. The static parameters are the static prior parameters input during initialization, and the dynamic inputs are the current design baseline and the current optimization trajectory.

[0033] The LLM optimization module is used to optimize the complete design instructions using the Large Language Model (LLM) until the optimization termination condition is met, and obtain the optimization results of the aircraft functional system. The optimization results include component parameter tables and control logic descriptions.

[0034] Optionally, the LLM optimization module includes:

[0035] The semantic reasoning submodule is used to obtain the output of the large language model LLM in the nth round of optimization iteration based on the complete design instructions. , Let n represent the semantic reasoning of a large language model (LLM), where n is the number of iterations, n≥1 and n is an integer. Based on the current design benchmark, Optimize the current trajectory;

[0036] The semantic decoding submodule is used to decode the output of the large language model LLM using semantic decoding operators. Convert to executable physical parameters;

[0037] The simulation submodule is used to input the executable physical parameters into the simulation model of the aircraft functional system for simulation and to obtain simulation results.

[0038] The evaluation submodule is used to obtain the evaluation result of the nth round of optimization iteration based on the simulation results. ;

[0039] The dynamic update submodule is used to update the optimized trajectory and obtain the new optimized trajectory. , The time series operator is used, and the nth violation degree is obtained based on the simulation results. If the nth violation degree is less than the historical optimal solution violation degree, the design baseline is updated to obtain a new design baseline, and the historical optimal solution violation degree is updated to the nth violation degree. If the nth violation degree is greater than or equal to the historical optimal solution violation degree, the current design baseline is used as the new design baseline, and the nth violation degree is the violation degree of the nth round of optimization iteration.

[0040] The termination judgment submodule is used to determine whether the optimization termination condition is met;

[0041] The iteration submodule is used to increment the iteration number n by 1 if the optimization termination condition is not met, take the new optimization trajectory as the current optimization trajectory, take the new design benchmark as the current design benchmark, take the current optimization trajectory and the current design benchmark as the dynamic input, and return to the Assemble module;

[0042] The result output submodule is used to terminate the optimization iteration process if the optimization termination condition is met, and to obtain the optimization result of the aircraft functional system based on the output of the large language model LLM in the nth round of optimization iteration.

[0043] Optionally, the evaluation submodule is specifically used for:

[0044] Key performance indicators are extracted based on the simulation results, and key performance feature vectors are obtained based on the key performance indicators.

[0045] Calculate the evaluation result ,in, This represents the key performance feature vector for the nth round of optimization iteration. Let be the refueling time for the i-th fuel tank in the n-th round of optimization iteration. The maximum fuel flow rate in the j-th segment of the pipeline during the nth round of optimization iterations. The maximum fuel flow rate of the k-th valve in the nth round of optimization iteration. The threshold value for the refueling time of the fuel tank. This is the constraint threshold for the maximum fuel flow rate in the pipeline. This is the constraint threshold for the maximum fuel flow rate of the valve. These refer to the number of oil tanks, pipes, and valves, respectively. This is an indicator function; if the condition inside the parentheses is true, then... Takes the value 1, otherwise The value is 0.

[0046] Optionally, satisfying the optimization termination condition includes:

[0047] The semantic reasoning of the Large Language Model (LLM) converges, or the number of iterations is greater than or equal to a preset iteration threshold.

[0048] Optionally, the current optimized trajectory is defined as a semantic record sequence. .

[0049] Optional, initial design baseline As a blank scheme, the initial optimized trajectory This is a sequence of blank semantic records.

[0050] In the semantically driven aircraft functional system optimization design system, the design problem The design objects and functional objectives used to define the aircraft's functional systems are specification objects with unique identifiers instantiated using the ReqIF standard; the design objectives The quantitative performance indicators used to define the functional systems of the aircraft are attributes of the specification object; the design space The key principles are used to define the design boundaries of the aircraft's functional systems and are formally described by the GOPPRRE meta-model ontology specification. Used to define high-level constraints based on mechanistic principles and design experience; the formal expression of the solution. This is used to describe the mapping structure of the design scheme in semantic space and physical space. The design scheme is constructed as a topological graph composed of objects, relations and roles based on the meta-meta-model ontology specification GOPPRRE.

[0051] Optionally, the semantic transformation module uses a meta-meta-modeling method based on the M0-M3 modeling framework to construct the five-element semantic modeling set.

[0052] Optionally, the semantic transformation module uses a task assembly function to generate structured design instructions based on the five-element semantic modeling set.

[0053] In a third aspect of this application, an electronic device is provided, comprising:

[0054] At least one processor; and,

[0055] A memory communicatively connected to at least one of the processors; wherein,

[0056] The memory stores instructions that can be executed by the processor to implement the semantically driven aircraft functional system optimization design method described above.

[0057] In a fourth aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for execution by the computer to implement the above-described semantically driven aircraft functional system optimization design method.

[0058] In a fifth aspect of this application, a computer program product containing instructions is provided, which, when executed by a computer device, causes the computer device to perform the semantically driven aircraft functional system optimization design method described above.

[0059] The beneficial effects of this application are as follows:

[0060] The video semantic-driven optimization design method and system for aircraft functional systems provided in this application effectively overcome the application bottlenecks of traditional numerical optimization methods in the optimization design of complex aircraft functional systems through a semantic-driven intelligent optimization process and multi-module collaborative work. It can quickly locate feasible design domains in a short time, significantly improving optimization efficiency. Natural language semantic modeling and cross-modal mapping mechanisms decouple design logic, flexibly adapting to the optimization design needs of various aircraft functional systems such as fuel, hydraulics, thermal management, and avionics. When facing scenarios such as topological changes and constraint logic expansion, there is no need to refactor the underlying code; seamless migration of optimization tasks can be achieved solely through natural language interaction. Furthermore, combined with modular design, it can flexibly adapt to various needs. The system offers optimization tasks of varying scales and complexities, demonstrating strong versatility and scalability. It effectively ensures the consistency and integrity of design data, significantly reducing design cycles, manpower costs, and computing power consumption. In particular, it can accurately match the optimization design needs of complex systems with high-dimensional constraints, strong coupling, and multi-objective conflicts, such as fuel systems, hydraulic systems, thermal management systems, and avionics systems. It is especially suitable for scenarios that require rapid generation and evaluation of multiple design schemes, providing reliable support for the rapid architecture design and optimization of complex aircraft functional systems. Moreover, it can be quickly implemented without making significant adjustments to the existing R&D system, providing an efficient and reliable optimization solution for aircraft development and effectively improving the overall efficiency of aircraft development. Attached Figure Description

[0061] Figure 1 A flowchart illustrating one implementation of the semantic-driven aircraft functional system optimization design method of this application;

[0062] Figure 2 This is a flowchart illustrating one implementation of step S103 in the semantic-driven optimization of the aircraft functional system of this application.

[0063] Figure 3 This is a structural block diagram of one implementation of the semantic-driven aircraft functional system optimization design system of this application. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description, in conjunction with the accompanying drawings and examples, further clarifies this application. It should be understood that the specific examples described herein are merely illustrative and not intended to limit the scope of this application. Furthermore, the technical features involved in the various embodiments of this application described below can be combined with each other as long as they do not conflict with each other.

[0065] The present application will now be described in detail with reference to the accompanying drawings. A first aspect of this application provides a semantically driven method for optimizing the design of aircraft functional systems. Figure 1The flowchart illustrating one implementation of the semantic-driven aircraft functional system optimization design method of this application is shown, as follows: Figure 1 As shown, the semantic-driven aircraft functional system optimization design method of the first embodiment of this application includes:

[0066] Step S101: Construct a five-element semantic model set for the aircraft functional system. , For design issues, For the design goals, For design space, As a key principle, To formalize the solution, structured design instructions are generated based on the five-element semantic modeling set. ;

[0067] Step S102: Use the Assemble module to assemble static parameters, dynamic inputs, and the structured design instructions to obtain complete design instructions. The static parameters are the static prior parameters input during initialization, including system performance indicators such as the fuel capacity requirements of each tank, maximum full-fill time, fuel flow rate in pipelines, and fuel flow rate in valves, as well as parameter value constraints such as the selectable range of pipeline diameters and the available product library for valve models. The dynamic inputs are the current design baseline and the current optimization trajectory.

[0068] Step S103: Using the complete design instructions as input, optimize using Large Language Model (LLM) until the optimization termination condition is met, and obtain the optimization result of the aircraft functional system. The optimization result includes component parameter table and control logic description.

[0069] The semantic-driven aircraft functional system optimization design method provided in this application is particularly suitable for the early architecture optimization design stage of aircraft functional systems, and is applicable to the design of complex systems with high-dimensional constraints, strong coupling and multi-objective conflicts, such as fuel systems, hydraulic systems, thermal management systems and avionics systems.

[0070] Specifically, in step S101, a five-element semantic modeling set can be constructed using the meta-meta-modeling method based on the M0-M3 modeling framework. M0-M3 is a four-layer meta-modeling architecture of OMG MOF (Meta-Object Facility, a meta-modeling standard developed by the Object Management Group), which abstracts from bottom to top and instantiates from top to bottom, used to unify model semantics and interoperability. Specifically, the five-element semantic modeling set... middle, For design problems, the ReqIF standard can be used to define design objects and functional objectives based on the original requirements in the aircraft functional system mission statement. This allows for the instantiation of design objects and functional objectives into specification objects with unique identifiers, enabling precise semantic parsing of unstructured text and ensuring the atomicity and traceability of requirement descriptions at the semantic level. For the design goals, quantitative performance metrics are defined, obtained from task instantiation, and mapped to the properties of the specification object; For the design space, which defines the design boundaries of the functional systems of the aircraft, the complex nonlinear characteristics and combinational logic can be formally described using the GOPPRRE (Graph-Object-Point-Property-Relationship-Role, a meta-meta-model ontology specification for Model Base System Engineering (MBSE)) method. For example, "objects" can be used to define component entities, and "extensions" can be used to define the constraint rules of the design boundaries. These are key principles used to define high-level constraints based on self-mechanism and design experience, so that the Large Language Model (LLM) can perform effective logical deduction. For a formal representation of the design scheme, this describes the mapping structure of the design scheme in semantic and physical spaces. The design scheme can be constructed based on the GOPPRRE ontology as a topological graph composed of objects, relationships, and roles. This not only includes physical entities but also explicitly encodes the connection logic and interaction interfaces between entities. Taking the aircraft ground-based refueling system as an example, the aircraft ground-based refueling system needs to safely and efficiently refuel all fuel tanks within a specified time window. An example of the code form for the five-element semantic modeling set is as follows: System content: { “Design Issues”:{ System Scenario: Ground Pressure Refueling System "Functional Requirements":[ {“Number”: “REQ-01”, “Description”: “Simultaneously refuel 4 fuel tanks (T1-T4)”}, {“ID”: “REQ-02”, “Description”: “Flow control via valve”}]}, "Design Goals":{ "Constraints":{ "Full tank time":{"Symbol": "<", "Value": 300, "Unit": "s"}, "Pipe velocity": {"Symbol": "<", "Value": 9, "Unit": "m / s"}, "Valve Flow Rate":{"Symbol": "<", "Value": 3, "Unit": "m / s"}}, "Optimization Goal":{ "Consistency":{"Objective":"Maximize", "Threshold":">0.95"}}}, “Design Space” "Pipe assembly":{"Extended properties": "Diameter", "Range":[40,88], "Unit":"mm"} "Valve Assembly":{"Extended Type: "Model Selection", "Options": "Optional Valve Models"}}, "Key Principles":{ Rule 01: "If the constraint check passes, then assess consistency." Rule 02: "To maximize consistency, prioritize traffic balancing." "Formal expression of the scheme":{ "Graph Structure": "Fueling Network Graph" "Objects": ["Fuel Tanks (T1-T4)", "Pipelines (P1-P7)", "Valve (V1-V4)"], "relation":[ {“Association Type”: “Main Route”, “Components”: [“P1”, “P2”, “P3”], “Function”: “Trunk”}, “Description” Description: "Both ends are connected to the filler neck". {"Association Type": "Branch", "Source": "Main Road", "Direction": ["T1", "T2", "T3", "T4"], “Pathway”: [“P4”, “P5”, “P6”, “P7”]], “Function”: “Assignment”}, "Output format constraints": "JSON object {suggested changes, next solution, design basis}"}}.

[0071] The design problem is clearly defined: the design object is the aircraft ground pressure refueling subsystem, which includes four fuel tanks and is refueled by pressure refueling valves, with the flow rate controlled by the valves. The design objectives set strict physical performance indicators, including a maximum refueling time of no more than 300 seconds, a pipe flow velocity of no more than 9 m / s, a valve flow velocity of no more than 3 m / s, and a refueling time consistency coefficient of no less than 0.95. The design space limits the continuous range of pipe diameter to [40, 88] mm, and valve models must be selected from a predefined standard product library. The key design principles treat refueling time and flow velocity limits as hard feasibility constraints, and refueling time consistency as an optimization objective, thereby guiding the model to maximize performance within the feasible region. The formal expression of the design scheme defines the system's topology, including four fuel tanks, seven pipe sections, and four valves, and defines the output format as a unified JSON structure, including proposed_changes, next_scheme, and rationale.

[0072] Specifically, in step S101, the task assembly function can be used. Structured design instructions are generated based on the quinary semantic modeling set. This transforms the aircraft functional system optimization task into a semantic input that can be understood and manipulated by a large language model. The task assembly function utilizes the RDF triple (Resource Description Framework Triple, a semantic web foundational data model defined by the W3C (World Wide Web Consortium)) structure to establish the design problem. Formal expression of the scheme The logical connections between them demonstrate that the generated structured design instructions not only encode explicit task objectives and constraints, but also implicitly embed semantic elements of mechanistic knowledge and historical experience, constituting the boundary conditions for LLM semantic reasoning. The task assembly function can be implemented using the Assemble module.

[0073] Specifically, in step S102, the Assemble module is used to assemble static parameters, dynamic inputs, and structured design instructions. The complete design instructions are obtained. The Assemble module is a widely used functional component in software across multiple domains. Its core function is to combine multiple independent parts into a whole. In the semantic-driven aircraft functional system optimization design method of this application, a large language model, such as Qwen-Plus, can be selected first, and then the Assemble module of the large language model can be used to complete the assembly. Here, static parameters are the static priors input during initialization (first optimization iteration), and are invariant parameters. Dynamic inputs are the current design baseline and the current optimization trajectory, which is constructed based on historical inputs and outputs. In one possible implementation, the optimization trajectory is defined as a semantic record sequence; for example, the current (nth optimization iteration) optimization trajectory... ,in, This is the output of the Large Language Model (LLM). As a design benchmark, This represents the evaluation result of the i-th optimization iteration. For example, after assembly using the Assemble module, the current complete design instructions can be represented as: .

[0074] In one possible implementation, the initial design baseline is a blank scheme, and the initial optimization trajectory is an empty semantic record sequence. The Assemble module assembles the structured design instructions, the current design baseline, and the optimization trajectory into complete input information for LLM, transforming the optimization problem described in natural language into a conditional generation task for LLM.

[0075] Specifically, in step S103, LLM is used for optimization with complete design instructions as input. It should be noted that LLM optimization is an iterative process; the generation of complete design instructions and the LLM model optimization process are combined. In each iteration, complete design instructions are first generated based on the current dynamic input, i.e., the current design baseline and the current optimization trajectory. LLM then optimizes with these complete design instructions as input. Next, the optimization trajectory and design baseline are updated based on the current iteration result, i.e., the current LLM output. If the iteration process has not ended, i.e., the iteration termination condition has not been met, the next iteration begins based on the new design baseline and optimization trajectory. This process continues until the optimization termination condition is met, resulting in the optimized aircraft function. The optimized results may include component parameter tables and control logic descriptions, as well as performance verification reports (e.g., evaluation of iteration results). Finally, the optimized results can be output in the form of topology diagrams, component parameter tables, control logic descriptions, and performance verification reports.

[0076] Specifically, Figure 2 A flowchart illustrating one embodiment of step S103 is shown, as follows: Figure 2 As shown, step S103 may include:

[0077] Step S1031: Obtain the output of the LLM (Large Language Model) for the nth round of optimization iteration based on the complete design instructions. , Let n represent the semantic reasoning of a large language model (LLM), where n is the number of iterations, n≥1 and n is an integer. Based on the current design benchmark, Optimize the current trajectory;

[0078] Step S1032: Use the semantic decoding operator to process the output of the large language model LLM. The parameters are converted into executable physical parameters. A cross-modal mapping between semantic modeling and physical implementation is achieved through semantic decoding and semantic encoding in subsequent processes. The semantic decoding operator can be implemented by inputting a valid dictionary of design parameters generated by LLM into the solver.

[0079] Step S1033: Input the executable physical parameters into the simulation model of the aircraft functional system to perform simulation and obtain simulation results. For example, the simulation model can be pre-built using AMESim. Based on the specific aircraft functional system (e.g., fuel, hydraulic, thermal management system, etc.), the simulation model is built using professional simulation tools such as AMESim to complete the evaluation of LLM optimization output.

[0080] Step S1034: Obtain the evaluation result of the nth round of optimization iteration based on the simulation results. ;

[0081] Step S1035, based on the evaluation results Update and optimize the trajectory to obtain a new optimized trajectory. , The time series operator is used, and the nth violation degree is obtained based on the simulation results. If the nth violation degree is greater than the (n-1)th violation degree, the design baseline is updated to obtain a new design baseline. If the nth violation degree is less than or equal to the (n-1)th violation degree, the current design baseline is used as the new design baseline. The nth violation degree is the violation degree of the nth round of optimization iteration, and the (n-1)th violation degree is the violation degree of the (n-1)th round of optimization iteration.

[0082] Step S1036: Determine whether the optimization termination condition is met;

[0083] Step S1037: If the optimization termination condition is not met, increment the iteration number n by 1, take the new optimization trajectory as the current optimization trajectory, take the new design benchmark as the current design benchmark, take the current optimization trajectory and the current design benchmark as the dynamic input, and return to the step of using the Assemble module to assemble static parameters, dynamic input and the structured design instructions to obtain complete design instructions;

[0084] Step S1038: If the optimization termination condition is met, terminate the optimization iteration process and obtain the optimization result of the aircraft functional system based on the output of the large language model LLM in the nth round of optimization iteration.

[0085] Specifically, in step S1034, the simulation results can be multimodal results. Depending on the specific optimization design problem and solution method, they may simultaneously include multiple data modalities such as time series, spatial field distribution, images, and semantic labels. Specifically, step S1034 may include: extracting key performance indicators from the simulation results; obtaining key performance feature vectors from the key performance indicators; and calculating the evaluation results based on the key performance feature vectors. Specifically, a semantic encoding operator can also be used to convert the key performance feature vectors into a semantic summary sequence. ,in, To design the input vector, This refers to the key performance feature vector. Let L be the output space of the semantic summary sequence, L be the set of all semantic summary sequences, L be the dimension of the semantic summary sequence, and Y() be the semantic encoding operator used to transform the design input vector and the key performance feature vector into a structured semantic summary sequence that can be understood by a large language model.

[0086] Specifically, in step S1034, a feature extraction operator can be used to extract key performance indicators from the multimodal results. These key performance indicators can include the main indicators reflecting the performance of the aircraft's functional systems. Taking a ground refueling system as an example, key performance indicators can include the maximum refueling time for each fuel tank, the maximum fuel flow rate for each fuel pipeline segment, and the maximum fuel flow rate for each valve, etc., determined according to actual optimization needs. The feature extraction operator can be implemented by extracting detailed indicators such as fuel tank refueling time and fuel flow rate in pipelines / valve segments from the simulation result file. The process of obtaining the key performance feature vector based on key performance indicators can be achieved by assembling the system feature values ​​of the key performance indicators into a key performance feature vector. Taking a ground refueling system with 4 fuel tanks, 7 pipeline sections, and 4 valves as an example, this would result in an array containing 4 + 7 + 4 = 15 elements, where each row of the array represents a key performance feature vector. For example, the array `feature_vector = [280.0, 290.0, 275.0, 285.0; 8.5, 8.2, 8.8, 7.9, 8.1, 8.3, 8.0; 3.0, 3.1, 2.9, ...]` would be obtained. [3.05], the first row contains the maximum refueling time for each fuel tank (e.g., the maximum refueling time for tank T1 is 280.0s, for tank T2 it is 290.0s, and for tank T3 it is 275.0s); the second row contains the maximum fuel flow rate for each pipeline segment (e.g., the maximum fuel flow rate for pipeline P1 is 8.5m / s, for pipeline P2 it is 8.2m / s, for pipeline P3 it is 8.8m / s, and for pipeline P3 it is 8.8m / s). The fuel flow rate is 7.9 m / s; the maximum fuel flow rate in pipe P5 is 8.1 m / s; the maximum fuel flow rate in pipe P6 is 8.3 m / s; and the maximum fuel flow rate in pipe P7 is 8.0 m / s. The third row of elements represents the maximum fuel flow rate for each valve (e.g., the maximum fuel flow rate for valve V1 is 3.0 m / s; the maximum fuel flow rate for valve V2 is 3.1 m / s; the maximum fuel flow rate for valve V3 is 2.9 m / s; and the maximum fuel flow rate for valve V4 is 3.05 m / s).

[0087] Specifically, in step S1034, after obtaining the key performance feature vector, the evaluation result is calculated. ,in, This represents the key performance feature vector for the nth round of optimization iteration. Let be the refueling time for the i-th fuel tank in the n-th round of optimization iteration. The maximum fuel flow rate in the j-th segment of the pipeline during the nth round of optimization iterations. The maximum fuel flow rate of the k-th valve in the nth round of optimization iteration. The threshold value for the refueling time of the fuel tank. This is the constraint threshold for the maximum fuel flow rate in the pipeline. This is the constraint threshold for the maximum fuel flow rate of the valve. These refer to the number of oil tanks, pipes, and valves, respectively. This is an indicator function; if the condition inside the parentheses is true, then... Takes the value 1, otherwise The value is 0. Specifically, semantic encoding operators can also be used to convert key performance feature vectors into semantic summary sequences that LLM can understand, achieving the conversion from simulation results to a language-understandable form. Among them, designing the input vector For the parameters of each physical component of the aircraft's functional system, taking a ground refueling system (e.g., including 4 fuel tanks, 7 pipe sections, and 4 valves) as an example, the input vector is designed to include the diameter of each pipe and the model of each valve. Semantic encoding operators can be implemented using LLM serialization functions to... and The sequence is serialized into structured text to obtain a semantic summary sequence.

[0088] Specifically, in step S1035, after the evaluation of the results of the nth round of optimization iteration is completed, the time-series operator... The design baseline, LLM output, and evaluation results of this round are integrated to update the optimized trajectory, resulting in the new optimized trajectory. The temporal operator is used to write content into the trajectory dictionary in a temporal sequence. It should be noted that the optimization trajectory is not merely a historical record used for numerical updates in the traditional sense, but rather an intrinsic semantic memory in the optimization process. LLM can adaptively adjust its semantic reasoning strategy by observing changes in the design problem, the occurrence of constraint conflicts, and the patterns of simulation feedback in the optimization trajectory, demonstrating semantic-level reasoning characteristics that are different from traditional numerical optimization. The optimization trajectory constitutes a key structure in semantic optimization that carries contextual consistency and policy transfer capabilities. On the other hand, the LLM optimization iteration process can reversely construct a semantic intermediate layer, namely a semantic optimization trajectory, to explicitly organize the design problem and feedback relationship across iterations in the optimization process. Specifically, in step S1035, the nth violation degree is obtained based on the simulation results. If the nth violation degree is less than the historical optimal solution violation degree, that is, the violation degree of the current nth iteration optimization is less than the historical optimal solution violation degree, the design baseline is updated to obtain a new design baseline, and the historical optimal solution violation degree is updated to the nth violation degree, that is, the violation degree of the current nth iteration optimization is taken as the historical optimal solution violation degree; if the nth violation degree is greater than or equal to the historical optimal solution violation degree, the design baseline is not updated, and the current design baseline is taken as the new design baseline, where the nth violation degree is the violation degree of the nth optimization iteration. The violation degree is calculated based on the constraints of the aircraft functional system. It can be obtained by iterating through the parameter values ​​corresponding to each constraint in the simulation results, comparing the parameter value with the threshold corresponding to the constraint, and if the parameter value does not meet the constraint (does not meet the threshold requirement), the violation degree is incremented by 1 (the initial value of the violation degree is 0). The number of times the constraint is not met is counted, and this number is the final violation degree value. Taking a ground refueling system as an example, the violation degree is calculated based on three constraints: ① Iterate through the refueling time of all fuel tanks from the simulation results and compare it with the time threshold in the set static parameters. If the refueling time of a certain fuel tank is greater than the time threshold, the violation degree is increased by 1; ② Iterate through the maximum fuel flow rate of all pipelines from the simulation results and compare it with the pipeline flow rate threshold obtained from the set static parameters. If the maximum fuel flow rate of a certain pipeline segment is greater than the pipeline flow rate threshold, the violation degree is increased by 1; ③ Iterate through the maximum fuel flow rate of all valves from the simulation results and compare it with the valve flow rate threshold in the set static parameters. If the maximum fuel flow rate of a certain valve is greater than the valve flow rate threshold, the violation degree is increased by 1. Thus, the final violation degree value is 3. The design benchmark for the nth round of optimization iteration is the system parameter value that minimizes the degree of violation under the current iteration round. Taking a ground refueling system as an example, if the ground refueling system includes 4 fuel tanks, 7 pipelines and 4 valves, then the design benchmark includes the system parameter values ​​that minimize the degree of violation: the diameter of each pipeline in the 7 pipelines, the model of each valve in the 4 valves, the maximum refueling time of all fuel tanks, the maximum fuel flow rate of all pipelines, the maximum fuel flow rate of all valves, the refueling time of each fuel tank, the maximum flow rate of each pipeline, and the maximum flow rate of each valve, etc.If the violation degree in the nth optimization iteration is greater than that in the previous iteration, the design baseline is recalculated to obtain a new design baseline. This new design baseline can be generated by a large language model based on the provided static parameters (including parameter value constraints and system performance index constraints) and feedback from historical optimization iterations. The parameters of the new design baseline are determined according to the parameter value constraints in the static parameters. For example, based on the input semantic information, combined with aerospace engineering mechanisms and feedback from historical optimization iterations, the large language model analyzes the gap between the performance index and the optimization objective in the current design baseline, as well as the reasons for constraint violations. Within the allowable design space, a new design baseline that balances feasibility and optimization is generated. Taking a ground refueling system as an example, the new design baseline includes pipe diameter and valve type, etc. The valve type, etc., can be further adjusted based on experience. The new design baseline is used as the design baseline for the next optimization iteration (the current design baseline), completing the design baseline update. If the violation degree in the nth optimization iteration is less than or equal to that in the previous iteration, the design baseline is maintained, and the current design baseline optimized in the nth iteration is used as the new design baseline for the next iteration.

[0089] Specifically, in step S1036, satisfying the optimization termination condition may include the convergence of the semantic reasoning of the LLM, or the iteration count n being greater than or equal to a preset iteration threshold. Step S1036 may be to determine whether the semantic reasoning of the LLM has converged, or to determine whether the iteration count n is greater than or equal to the iteration threshold. In one possible implementation, step S1036 may also first determine whether the semantic reasoning of the LLM has converged. If it has converged, then step S1038 is executed. If it has not converged, then it further determines whether the iteration count n is greater than or equal to the iteration threshold. If n is less than the iteration threshold, then step S1037 is executed. If n is greater than or equal to the iteration threshold, then step S1038 is executed.

[0090] Specifically, in step S1037, if the optimization iteration has not terminated, the iteration number n is incremented by 1, the updated optimization trajectory and design benchmark are used as the current optimization trajectory and current design benchmark, and the step of assembling and generating complete design instructions using the Assemble module is returned, that is, step S102 is returned to enter the next round of iteration.

[0091] Specifically, in step S1038, after the optimization iteration terminates, the optimization result of the aircraft functional system is obtained based on the output of the LLM in the nth round of optimization iteration, that is, the output of the LLM at the time of termination. Specifically, the output of the LLM can be converted into physical parameters using a semantic decoding operator, and the LLM iteration process data, inference chain, optimization adjustment logic and historical optimization trajectory can be obtained, thereby obtaining the optimization result.

[0092] In one possible implementation, the semantic-driven aircraft functional system optimization design method provided in this application may further include acquiring optimization data, iterative optimization records, and simulation evaluation results through a data integration interface, such as feasible solution acquisition rate, convergence speed, and computational overhead. The feasible solution acquisition rate is used to evaluate whether an optimization result that satisfies the constraints can be found within a limited computational budget. The convergence speed is used to measure the agility of the algorithm in locating the feasible region. The computational overhead can be measured by the cumulative number of simulation evaluations. The key performance indicators may also include average refueling time and consistency coefficient, which reflect the optimization quality. By comprehensively utilizing components such as curves, pie charts, and tables, the optimization process and evaluation results are displayed, clearly presenting the LLM inference chain and parameter adjustment logic, and supporting the viewing of historical iteration trajectories, forming a visual result. This facilitates secondary decision-making and optimization based on the visual result, improving the traceability and interactivity of the optimization process.

[0093] The semantic-driven optimization design method for aircraft functional systems provided in this application breaks through the bottlenecks of traditional numerical optimization through semantic-driven and multi-module collaboration. It can quickly locate feasible design domains and improve optimization efficiency. Natural language semantic modeling and cross-modal mapping eliminate the need to refactor code when topology / constraints change. Task migration can be achieved through natural language interaction. Modular design adapts to tasks of different scales and complexities, with strong versatility and scalability. It helps ensure the consistency and integrity of optimization design data, shortens the design cycle, and reduces manpower and computing costs. It accurately matches the design requirements of complex systems with high-dimensional constraints, strong coupling, and multi-objective conflicts, and is compatible with existing R&D systems. It provides an efficient and reliable solution for rapid architecture design optimization of aircraft functional systems, improving development efficiency.

[0094] A second aspect of this application provides a semantically driven aircraft functional system optimization system. Figure 3 This paper presents a structural block diagram of one embodiment of the semantic-driven aircraft function system optimization system of this application, as shown in the figure. Figure 3 As shown, the semantically driven aircraft function system optimization system of the third embodiment of this application includes:

[0095] Semantic transformation module 301 is used to construct a five-element semantic modeling set for the functional systems of an aircraft. , For design issues, For the design goals, For design space, As a key principle, To formalize the solution, structured design instructions are generated based on the five-element semantic modeling set. ;

[0096] Assemble module 302 is used to assemble static parameters, dynamic inputs and the structured design instructions to obtain complete design instructions. The static parameters are the static prior parameters input during initialization, and the dynamic inputs are the current design baseline and the current optimization trajectory.

[0097] LLM optimization module 303 is used to optimize the aircraft functional system using the complete design instructions as input, until the optimization termination condition is met, and obtain the optimization result of the aircraft functional system. The optimization result includes component parameter table and control logic description.

[0098] Among them, the design problem The design objects and functional objectives used to define the aircraft's functional systems are specification objects with unique identifiers instantiated using the ReqIF standard; the design objectives The quantitative performance indicators used to define the functional systems of the aircraft are attributes of the specification object; the design space The key principles are used to define the design boundaries of the aircraft's functional systems and are formally described by the GOPPRRE meta-model ontology specification. Used to define high-level constraints based on mechanistic principles and design experience; the formal expression of the solution. This describes the mapping structure of a design scheme in semantic and physical space. The design scheme is constructed as a topological graph composed of objects, relationships, and roles based on the GOPPRRE meta-model ontology specification. In one possible implementation, the semantic transformation module constructs the five-element semantic modeling set using a meta-modeling method based on the M0-M3 modeling framework, and uses a task assembly function to generate structured design instructions based on the five-element semantic modeling set.

[0099] Specifically, the LLM optimization module 303 may include:

[0100] Semantic reasoning submodule 3031 is used to obtain the output of the large language model LLM in the nth round of optimization iteration based on the complete design instructions. , Let n represent the semantic reasoning of a large language model (LLM), where n is the number of iterations, n≥1 and n is an integer. Based on the current design benchmark, Optimize the current trajectory;

[0101] Semantic decoding submodule 3032 is used to decode the output of the large language model LLM using semantic decoding operators. Convert to executable physical parameters;

[0102] Simulation submodule 3033 is used to input the executable physical parameters into the simulation model of the aircraft functional system for simulation and to obtain simulation results.

[0103] Evaluation submodule 3034 is used to obtain the evaluation result of the nth round of optimization iteration based on the simulation results. ;

[0104] The dynamic update submodule 3035 is used to update the optimized trajectory and obtain the new optimized trajectory. , The time series operator is used, and the nth violation degree is obtained based on the simulation results. If the nth violation degree is less than the historical optimal solution violation degree, the design baseline is updated to obtain a new design baseline, and the historical optimal solution violation degree is updated to the nth violation degree. If the nth violation degree is greater than or equal to the historical optimal solution violation degree, the current design baseline is used as the new design baseline. The nth violation degree is the violation degree of the nth round of optimization iteration, and the (n-1)th violation degree is the violation degree of the (n-1)th round of optimization iteration.

[0105] Termination judgment submodule 3036 is used to determine whether the optimization termination condition is met;

[0106] The iteration submodule 3037 is used to increment the iteration number n by 1 if the optimization termination condition is not met, take the new optimization trajectory as the current optimization trajectory, take the new design benchmark as the current design benchmark, take the current optimization trajectory and the current design benchmark as the dynamic input, and return to the Assemble module, that is, return to the step of using the Assemble module to assemble static parameters, dynamic input and the structured design instructions to obtain complete design instructions;

[0107] The result output submodule 3038 is used to terminate the optimization iteration process if the optimization termination condition is met, and to obtain the optimization result of the aircraft functional system based on the output of the large language model LLM in the nth round of optimization iteration.

[0108] Specifically, the evaluation submodule is used for:

[0109] Key performance indicators are extracted from the simulation results, and key performance feature vectors are obtained based on these indicators. The evaluation results are then calculated based on the key performance feature vectors. ,in, This represents the key performance feature vector for the nth round of optimization iteration. Let be the refueling time for the i-th fuel tank in the n-th round of optimization iteration. The maximum fuel flow rate in the j-th segment of the pipeline during the nth round of optimization iterations. The maximum fuel flow rate of the k-th valve in the nth round of optimization iteration. The threshold value for the refueling time of the fuel tank. This is the constraint threshold for the maximum fuel flow rate in the pipeline. This is the constraint threshold for the maximum fuel flow rate of the valve. These refer to the number of oil tanks, pipes, and valves, respectively. This is an indicator function; if the condition inside the parentheses is true, then... Takes the value 1, otherwise The value is 0.

[0110] The evaluation submodule can also be used to convert the key performance feature vector into a semantic summary sequence using a semantic encoding operator. , To design the input vector, This refers to the key performance feature vector. Let L be the output space of the semantic summary sequence, L be the set of all semantic summary sequences, L be the dimension of the semantic summary sequence, and Y() be the semantic encoding operator.

[0111] The condition for satisfying the optimization termination condition may include: the semantic reasoning of the large language model LLM converges, or the number of iterations is greater than or equal to a preset iteration threshold.

[0112] In one possible implementation, the current optimized trajectory is defined as a semantic record sequence. In one possible implementation, the initial design baseline This can be a blank scheme, an initial optimized trajectory. Sequences can be recorded using semantic markup.

[0113] In one possible implementation, the semantic-driven aircraft functional system optimization system provided in this application may further include a result output and interaction module. The result output and interaction module is used to obtain optimization data, iterative optimization records and simulation evaluation results through a data integration interface, and comprehensively utilize components such as curves, pie charts, and tables to display the optimization process and evaluation results, clearly present the LLM inference chain and parameter adjustment logic, and support viewing historical iteration trajectories to form visualized results. This facilitates secondary decision-making and optimization based on the visualized results, and improves the traceability and interactivity of the optimization process.

[0114] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the system described above can be referred to the corresponding processes in the foregoing method embodiments, and therefore will not be repeated here.

[0115] The semantic-driven aircraft functional system optimization system provided in this application breaks through the bottleneck of traditional numerical optimization by semantic driving, quickly locates feasible design domains, and improves optimization efficiency; natural language semantic modeling and cross-modal mapping allow topology / constraint changes to be carried out without code refactoring, and tasks can be transferred through natural language interaction; modular design adapts to tasks of varying complexity, has strong versatility and scalability, ensures the consistency and integrity of design data, shortens the cycle, reduces manpower and computing costs, and is compatible with existing R&D systems, providing an efficient and reliable solution for rapid architecture design optimization of aircraft functional systems.

[0116] It should be noted that the semantic-driven aircraft function system optimization system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the modules or steps in the embodiments of this application can be further decomposed or combined. For example, the modules in the above embodiments can be merged into one module, or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of this application are only for distinguishing the various modules or steps and are not considered as an improper limitation of this application.

[0117] In a third aspect of this application, an electronic device is also provided, comprising: at least one processor; and a memory communicatively connected to at least one of the processors; wherein the memory stores instructions executable by the processor, the instructions being executed by the processor to implement the semantically driven aircraft functional system optimization design method described above.

[0118] In a fourth aspect of this application, a computer-readable storage medium is also provided, the computer-readable storage medium storing computer instructions for execution by the computer to implement the above-described semantically driven aircraft functional system optimization design method.

[0119] In a fifth aspect of this application, a computer program product containing instructions is also provided, which, when executed by a computer device, cause the computer device to perform the semantically driven aircraft functional system optimization design method described above.

[0120] Those skilled in the art will recognize that the modules and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. The programs corresponding to the software modules and method steps can be placed in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art. To clearly illustrate the interchangeability of electronic hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in electronic hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0121] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0122] The terms “first”, “second”, etc., are used to distinguish similar objects, not to describe or indicate a specific order or sequence.

[0123] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to those processes, methods, articles, or apparatus / systems.

[0124] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A semantically driven optimization design method for aircraft functional systems, characterized in that, include: Constructing a set of five-element semantic models for the functional systems of an aircraft , For design issues, For the design goals, For design space, As a key principle, To formalize the solution, structured design instructions are generated based on the five-element semantic modeling set. ; The Assemble module is used to assemble static parameters, dynamic inputs, and the structured design instructions to obtain complete design instructions. The static parameters are the static prior parameters input during initialization, and the dynamic inputs are the current design baseline and the current optimization trajectory. Using the complete design instructions as input, optimization is performed using a Large Language Model (LLM) until the optimization termination condition is met, resulting in the optimization results of the aircraft functional system. The optimization results include component parameter tables and control logic descriptions.

2. The semantic-driven aircraft functional system optimization design method as described in claim 1, characterized in that, The process of using the complete design instructions as input and optimizing with a Large Language Model (LLM) until the optimization termination condition is met includes: The output of the large language model LLM in the nth round of optimization iteration is obtained based on the complete design instructions. , Let n represent the semantic reasoning of a large language model (LLM), where n is the number of iterations, n≥1 and n is an integer. Based on the current design benchmark, Optimize the current trajectory; The output of the Large Language Model (LLM) is processed using semantic decoding operators. Convert to executable physical parameters; The executable physical parameters are input into the simulation model of the aircraft functional system to perform simulation and obtain simulation results. The evaluation result of the nth round of optimization iteration is obtained based on the simulation results. ; Update and optimize the trajectory to obtain a new optimized trajectory. , The time series operator is used, and the nth violation degree is obtained based on the simulation results. If the nth violation degree is less than the historical optimal solution violation degree, the design baseline is updated to obtain a new design baseline, and the historical optimal solution violation degree is updated to the nth violation degree. If the nth violation degree is greater than or equal to the historical optimal solution violation degree, the current design baseline is used as the new design baseline, and the nth violation degree is the violation degree of the nth round of optimization iteration. Determine whether the optimization termination condition is met; If the optimization termination condition is not met, increment the iteration number n by 1, take the new optimization trajectory as the current optimization trajectory, take the new design benchmark as the current design benchmark, take the current optimization trajectory and the current design benchmark as the dynamic input, and return to the step of using the Assemble module to assemble static parameters, dynamic input and the structured design instructions to obtain complete design instructions; If the optimization termination condition is met, the optimization iteration process is terminated, and the optimization result of the aircraft functional system is obtained based on the output of the large language model LLM in the nth round of optimization iteration.

3. The semantic-driven aircraft functional system optimization design method as described in claim 2, characterized in that, The evaluation result of the nth round of optimization iteration is obtained based on the simulation results. ,include: Key performance indicators are extracted based on the simulation results, and key performance feature vectors are obtained based on the key performance indicators. Calculate the evaluation result ,in, This represents the key performance feature vector for the nth round of optimization iteration. Let be the refueling time for the i-th fuel tank in the n-th round of optimization iteration. The maximum fuel flow rate in the j-th segment of the pipeline during the nth round of optimization iterations. The maximum fuel flow rate of the k-th valve in the nth round of optimization iteration. The threshold value for the refueling time of the fuel tank. This is the constraint threshold for the maximum fuel flow rate in the pipeline. This is the constraint threshold for the maximum fuel flow rate of the valve. These refer to the number of oil tanks, pipes, and valves, respectively. This is an indicator function; if the condition inside the parentheses is true, then... Takes the value 1, otherwise The value is 0.

4. The semantic-driven aircraft functional system optimization design method as described in claim 2, characterized in that, The conditions for satisfying the optimization termination include: The semantic reasoning of the Large Language Model (LLM) converges, or the number of iterations is greater than or equal to a preset iteration threshold.

5. The aircraft functional system optimization method as described in claim 2, characterized in that, The current optimized trajectory is defined as a semantic record sequence. Initial design baseline As a blank scheme, the initial optimized trajectory This is a sequence of blank semantic records.

6. The semantic-driven aircraft functional system optimization design method as described in claim 1, characterized in that, The design problem The design objects and functional objectives used to define the aircraft's functional systems are specification objects with unique identifiers instantiated using the ReqIF standard; the design objectives The quantitative performance indicators used to define the functional systems of the aircraft are attributes of the specification object; the design space The key principles are used to define the design boundaries of the aircraft's functional systems and are formally described by the GOPPRRE meta-model ontology specification. Used to define high-level constraints based on mechanistic principles and design experience; the formal expression of the solution. This is used to describe the mapping structure of the design scheme in semantic space and physical space. The design scheme is constructed as a topological graph composed of objects, relations and roles based on the meta-meta-model ontology specification GOPPRRE.

7. A semantically driven aircraft functional system optimization design system, characterized in that, include: The semantic transformation module is used to construct a five-element semantic modeling set for the functional systems of an aircraft. , For design issues, For the design goals, For design space, As a key principle, To formalize the solution, structured design instructions are generated based on the five-element semantic modeling set. ; The Assemble module is used to assemble static parameters, dynamic inputs, and the structured design instructions to obtain complete design instructions. The static parameters are the static prior parameters input during initialization, and the dynamic inputs are the current design baseline and the current optimization trajectory. The LLM optimization module is used to optimize the complete design instructions using the Large Language Model (LLM) until the optimization termination condition is met, and obtain the optimization results of the aircraft functional system. The optimization results include component parameter tables and control logic descriptions.

8. The semantic-driven aircraft functional system optimization design system as described in claim 7, characterized in that, The LLM optimization module includes: The semantic reasoning submodule is used to obtain the output of the large language model LLM in the nth round of optimization iteration based on the complete design instructions. , Let n represent the semantic reasoning of a large language model (LLM), where n is the number of iterations, n≥1 and n is an integer. Based on the current design benchmark, Optimize the current trajectory; The semantic decoding submodule is used to decode the output of the large language model LLM using semantic decoding operators. Convert to executable physical parameters; The simulation submodule is used to input the executable physical parameters into the simulation model of the aircraft functional system for simulation and to obtain simulation results. The evaluation submodule is used to obtain the evaluation result of the nth round of optimization iteration based on the simulation results. ; The dynamic update submodule is used to update the optimized trajectory and obtain the new optimized trajectory. , The time series operator is used, and the nth violation degree is obtained based on the simulation results. If the nth violation degree is less than the historical optimal solution violation degree, the design baseline is updated to obtain a new design baseline, and the historical optimal solution violation degree is updated to the nth violation degree. If the nth violation degree is greater than or equal to the historical optimal solution violation degree, the current design baseline is used as the new design baseline, and the nth violation degree is the violation degree of the nth round of optimization iteration. The termination judgment submodule is used to determine whether the optimization termination condition is met; The iteration submodule is used to increment the iteration number n by 1 if the optimization termination condition is not met, take the new optimization trajectory as the current optimization trajectory, take the new design benchmark as the current design benchmark, take the current optimization trajectory and the current design benchmark as the dynamic input, and return to the Assemble module; The result output submodule is used to terminate the optimization iteration process if the optimization termination condition is met, and to obtain the optimization result of the aircraft functional system based on the output of the large language model LLM in the nth round of optimization iteration.

9. The semantic-driven aircraft functional system optimization design system as described in claim 8, characterized in that, The evaluation submodule is specifically used for: Key performance indicators are extracted based on the simulation results, and key performance feature vectors are obtained based on the key performance indicators. Calculate the evaluation result ,in, This represents the key performance feature vector for the nth round of optimization iteration. Let be the refueling time for the i-th fuel tank in the n-th round of optimization iteration. The maximum fuel flow rate in the j-th segment of the pipeline during the nth round of optimization iterations. The maximum fuel flow rate of the k-th valve in the nth round of optimization iteration. The threshold value for the refueling time of the fuel tank. This is the constraint threshold for the maximum fuel flow rate in the pipeline. This is the constraint threshold for the maximum fuel flow rate of the valve. These refer to the number of oil tanks, pipes, and valves, respectively. This is an indicator function; if the condition inside the parentheses is true, then... Takes the value 1, otherwise The value is 0.

10. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to at least one of the processors; wherein, The memory stores instructions that can be executed by the processor to implement the semantically driven aircraft functional system optimization design method as described in any one of claims 1-6.

Images