An ai-physical model coupling driven engine demand analysis method and system
By employing an AI-physical model coupled approach, this method utilizes RAG knowledge retrieval and large language models to analyze aero-engine requirements. Combined with multidimensional constraint analysis and task analysis, it solves the problem of integrating unstructured knowledge with structured physical models in existing technologies, achieving intelligent handling of multiple requirement conflicts and improving design efficiency and accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- AECC SICHUAN GAS TURBINE RES INST
- Filing Date
- 2026-03-04
- Publication Date
- 2026-05-15
AI Technical Summary
Existing aero-engine requirements analysis methods are difficult to effectively integrate unstructured knowledge with structured physical models, and cannot intelligently handle multiple conflicting requirements. As a result, the design process often has to be scrapped and restarted because it cannot meet airworthiness regulations or structural size limitations. Moreover, existing AI-assisted design is inefficient and prone to errors.
By adopting an AI-physical model coupled approach, user needs are analyzed through RAG knowledge retrieval and large language models, combined with multidimensional constraint analysis and task analysis, and optimized using a utility function model, an intelligent closed-loop design is achieved from fuzzy text to precise requirements specifications.
It achieves intelligent closed-loop design from fuzzy text to precise requirements specifications, ensuring that multi-dimensional requirements such as point performance constraints, physical boundary constraints and task objective requirements are met, improving design efficiency and accuracy, and reducing manual intervention.
Smart Images
Figure CN121787019B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of aero-engine technology, and discloses an AI-physics model coupled-driven method and system for engine requirement analysis. Background Technology
[0002] During the conceptual design phase of aero-engines, engineers face the challenge of translating vague, multi-source top-level requirements into precise engineering parameters.
[0003] 1. Limitations of traditional requirements analysis methods
[0004] Traditional aero-engine requirements analysis primarily relies on classical methods such as constraint analysis and carpet plots. While these methods are based on physical modeling and analysis, they suffer from the following problems in practical applications:
[0005] Traditional constraint analysis primarily focuses on aerodynamic and propulsion performance, without simultaneously considering multi-source constraints such as structural dimensions and airworthiness regulations (e.g., noise and emissions). This often leads to the need for a complete overhaul when detailed design is introduced because it cannot meet airworthiness regulations or structural size limitations.
[0006] Carpet diagrams typically only visually represent the trade-off between two design variables and performance indicators. As the number of engine design variables (such as bypass ratio, pressure ratio, turbine inlet temperature, etc.) and performance indicators (such as multi-condition performance, structural constraints, etc.) continues to increase, traditional carpet diagrams struggle to handle high-dimensional design spaces.
[0007] 2. Shortcomings of existing AI-assisted design
[0008] With the development of artificial intelligence technology, although some research has begun to attempt to apply neural networks or large language models (LLM) to engineering design, the current technical approach still has significant shortcomings:
[0009] Coupling unstructured text with physical models is difficult: Airworthiness regulations and user requirements are often in the form of unstructured text in natural language, while engineering simulations rely on rigorous mathematical models that require structured mathematical boundaries. Currently, there is a lack of an efficient mechanism to automatically and accurately transform "textual constraints" into "mathematical boundaries," and there is still a large reliance on manual interpretation and parameter input, which is inefficient and prone to errors.
[0010] The lack of a multi-source requirement conflict resolution mechanism: When faced with typical multi-objective conflicts such as long range, high performance, and structural size limitations, existing requirement analysis methods lack intelligent decision-making mechanisms. They cannot negotiate and compromise like multi-disciplinary human experts, and often can only mechanically output a series of requirement items, still requiring complex manual trade-offs and decisions. They are unable to automatically identify requirement items that comply with airworthiness regulations, user requirements, and other multi-source inputs and are engineering-feasible.
[0011] In summary, the current field of aero-engine design urgently needs a requirement analysis method that can integrate unstructured knowledge with structured physical models and intelligently handle multiple conflicting requirements. Summary of the Invention
[0012] The purpose of this invention is to provide an AI-physical model coupled-driven engine requirement analysis method and system, which can intelligently analyze requirements into multi-dimensional requirements including point performance constraints, physical boundary constraints, and task target requirements. Based on the dual physical model of constraint analysis and task analysis, the feasibility of the design space is verified, realizing an intelligent closed loop from fuzzy text to precise requirement specification design.
[0013] To achieve the above-mentioned technical effects, the technical solution adopted by the present invention is as follows:
[0014] An AI-physics model coupled-driven method for analyzing engine requirements includes:
[0015] Access the user's natural language description of aero-engine design requirements and an industry standard library, which includes airworthiness regulations; use the RAG knowledge retrieval method to parse the industry standard library and obtain standard clause slices that match the design requirements;
[0016] The design requirements are analyzed into multi-dimensional requirements including point performance constraints, physical boundary constraints, and mission objective requirements. The point performance constraints include takeoff distance, climb rate, and landing distance. The physical boundary constraints include engine diameter and weight. The mission objective requirements include range and endurance.
[0017] Based on point performance constraints, key constraint curves are plotted to form feasible domain constraint diagrams for thrust-to-weight ratio and wing loading. The key constraint curves include takeoff distance constraints, climb rate constraints, and landing distance constraints.
[0018] A flight profile mission analysis simulation model is established. The point with the minimum thrust-to-weight ratio within the feasible region of the feasible region constraint diagram is used as the input for flight profile mission analysis. The simulation model is used to simulate and obtain engine performance indicators for each stage of the flight mission profile, including ground taxiing, takeoff, climb, cruise, descent, and landing. The performance indicators include thrust and fuel consumption rate.
[0019] Based on the performance indicators, the predicted value of the mission objective under the minimum thrust-to-weight ratio is obtained through analysis. If the predicted value of the mission objective meets the mission objective requirements, a requirement item containing engine performance indicators is output; otherwise, the point performance constraints or mission objective requirements are adjusted until the predicted value of the mission objective meets the mission objective requirements, and a requirement item containing engine performance indicators is output.
[0020] Furthermore, when the predicted value of the task target does not meet the task target requirements or when there are conflicts between the performance constraints of the multidimensional requirements and the task target requirements, a utility function model is constructed based on the achievement rate of each constraint or requirement index in the multidimensional requirements. The optimization objective is to ensure that the output value of the utility function model meets the preset conditions. Optimization analysis is then performed within the feasible domain boundary to obtain engine performance indicators where the predicted value of the task target meets the task target requirements.
[0021] Furthermore, a utility function model is constructed using Nash game theory, with the expression being: ,in Output values for the utility function model. , This is the required range value. This is the predicted flight distance. , This is the predicted takeoff distance. This is the takeoff distance limit. , Given the exponential weights, the preset condition that the output value of the utility function model satisfies is: .
[0022] Furthermore, after performing optimization analysis within the feasible domain boundary a preset number of times, the engine performance indicators and requirement items obtained from the optimization analysis corresponding to historical design requirements are used as the experience knowledge base for RAG knowledge retrieval. After accessing the industry standard library and design requirements described in the user's natural language, the requirement items containing engine performance indicators are directly output using the RAG knowledge retrieval method.
[0023] Furthermore, methods for directly outputting requirement entries containing engine performance indicators using the RAG knowledge retrieval method include:
[0024] Compare the multidimensional requirements of the current design requirements with the multidimensional requirements corresponding to the historical design requirements in the experience knowledge base. If the similarity is greater than or equal to the preset similarity threshold, output the performance index that the similarity meets the preset similarity condition, and output the corresponding requirement item.
[0025] To achieve the above technical effects, the present invention also provides an AI-physics model coupled-driven engine demand analysis system, comprising:
[0026] The input parsing module is used to access the user's natural language description of the aircraft engine design requirements and the industry standard library, which includes airworthiness regulations; the RAG knowledge retrieval method is used to parse the industry standard library to obtain standard clause slices that match the design requirements;
[0027] The requirement analysis module is used to analyze the design requirements into multi-dimensional requirements including point performance constraints, physical boundary constraints, and mission objective requirements. The point performance constraints include takeoff distance, climb rate, and landing distance. The physical boundary constraints include engine diameter and weight. The mission objective requirements include range and endurance.
[0028] The constraint analysis module is used to draw key constraint curves based on point performance constraints, and to form a feasible domain constraint diagram for thrust-weight ratio and wing loading. The key constraint curves include takeoff distance constraints, climb rate constraints, and landing distance constraints.
[0029] The simulation analysis module is used to establish a flight profile mission analysis simulation model. It finds the point with the minimum thrust-to-weight ratio within the feasible region of the feasible region constraint diagram as the input for flight profile mission analysis. The simulation model is used to simulate and obtain engine performance indicators for each stage of the flight mission profile, including ground taxiing, takeoff, climb, cruise, descent, and landing. The performance indicators include thrust and fuel consumption rate.
[0030] The discrimination output module is used to analyze and obtain the predicted value of the mission target under the minimum thrust-to-weight ratio based on the performance index. If the predicted value of the mission target meets the mission target requirements, it outputs a requirement item containing engine performance indexes; otherwise, it adjusts the point performance constraints or mission target requirements until the predicted value of the mission target meets the mission target requirements, and outputs a requirement item containing engine performance indexes.
[0031] Furthermore, it also includes an optimization analysis module, which is used to construct a utility function model based on the achievement rate of each constraint or requirement index in the multidimensional requirements when the predicted value of the task target does not meet the requirements of the task target. The module takes the output value of the utility function model meeting the preset conditions as the optimization objective and performs optimization analysis within the feasible domain boundary to obtain engine performance indicators where the predicted value of the task target meets the requirements of the task target.
[0032] Furthermore, in the optimization analysis module, a utility function model is constructed using Nash game theory, with the expression being: ,in Output values for the utility function model. , This is the required range value. This is the predicted flight distance. , This is the predicted takeoff distance. This is the takeoff distance limit. , Given the exponential weights, the preset condition that the output value of the utility function model satisfies is: .
[0033] Furthermore, it also includes a knowledge base module, which stores the engine performance indicators and requirement items obtained from optimization analysis corresponding to historical design requirements to form an experience knowledge base for RAG knowledge retrieval. After performing optimization analysis within the feasible domain boundary to a preset number of times, it accesses the industry standard library and design requirements described in the user's natural language, and directly outputs requirement items containing engine performance indicators using the RAG knowledge retrieval method.
[0034] Compared with the prior art, the beneficial effects of this invention are as follows: This invention, through the deep coupling of AI semantic parsing and physical model, intelligently parses the data into multi-dimensional requirements including point performance constraints, physical boundary constraints, and task target requirements. Based on the dual physical model of constraint analysis and task analysis, the feasibility of the design space is verified, ensuring that both point performance constraints and task target requirements are met, thus realizing an intelligent closed loop from fuzzy text to precise requirement specification design. Attached Figure Description
[0035] Figure 1 This is a flowchart of the AI-physics model coupled-driven engine requirement analysis method in Example 1;
[0036] Figure 2 This is a block diagram of the engine requirement analysis system driven by AI-physics model coupling in Example 1;
[0037] Figure 3 This is a flowchart of the AI-physics model coupled-driven engine requirement analysis method in Example 2;
[0038] The module consists of: 1. Input parsing module; 2. Requirement parsing module; 3. Constraint analysis module; 4. Simulation analysis module; 5. Discriminant output module; 6. Optimization analysis module; and 7. Knowledge base module. Detailed Implementation
[0039] The present invention will now be described in further detail with reference to the embodiments and accompanying drawings. However, this should not be construed as limiting the scope of the above-described subject matter of the present invention to the following embodiments; all technologies implemented based on the content of the present invention fall within the scope of the present invention.
[0040] Example 1
[0041] See Figure 1 and Figure 2 An AI-physics model coupled-driven method for analyzing engine requirements includes:
[0042] Access the user's natural language description of aero-engine design requirements and an industry standard library, which includes airworthiness regulations; use the RAG knowledge retrieval method to parse the industry standard library and obtain standard clause slices that match the design requirements;
[0043] The design requirements are analyzed into multi-dimensional requirements including point performance constraints, physical boundary constraints, and mission objective requirements. The point performance constraints include takeoff distance, climb rate, and landing distance. The physical boundary constraints include engine diameter and weight. The mission objective requirements include range and endurance.
[0044] Based on point performance constraints, key constraint curves are plotted to form feasible domain constraint diagrams for thrust-to-weight ratio and wing loading. The key constraint curves include takeoff distance constraints, climb rate constraints, and landing distance constraints.
[0045] A flight profile mission analysis simulation model is established. The point with the minimum thrust-to-weight ratio within the feasible region of the feasible region constraint diagram is used as the input for flight profile mission analysis. The simulation model is used to simulate and obtain engine performance indicators for each stage of the flight mission profile, including ground taxiing, takeoff, climb, cruise, descent, and landing. The performance indicators include thrust and fuel consumption rate.
[0046] Based on the performance indicators, the predicted value of the mission objective under the minimum thrust-to-weight ratio is obtained through analysis. If the predicted value of the mission objective meets the mission objective requirements, a requirement item containing engine performance indicators is output; otherwise, the point performance constraints or mission objective requirements are adjusted until the predicted value of the mission objective meets the mission objective requirements, and a requirement item containing engine performance indicators is output.
[0047] In this embodiment, RAG (Retrieval-augmented Generation) technology is used to integrate natural language requirements and industry standards with a large language model. Through the deep coupling of AI semantic parsing and physical model, the system intelligently parses the requirements into multi-dimensional requirements, including point performance constraints, physical boundary constraints, and task objective requirements. The feasibility of the design space is verified based on the dual physical model of constraint analysis and task analysis, ensuring that the requirements of "performance-physics-task" are met. This achieves an intelligent closed loop from fuzzy text to precise requirement specification design.
[0048] In some other embodiments, when the predicted value of the task target does not meet the task target requirements or when there are conflicts between the performance constraints of the multidimensional requirements and the task target requirements, a utility function model is constructed based on the achievement rate of each constraint or requirement index in the multidimensional requirements. The optimization objective is to ensure that the output value of the utility function model meets the preset conditions. Optimization analysis is performed within the feasible domain boundary to obtain engine performance indicators where the predicted value of the task target meets the task target requirements.
[0049] The utility function model can be a linear weighted summation model or an ideal point distance model. It can also be constructed using Nash game theory, with the expression: ,in Output values for the utility function model. , This is the required range value. This is the predicted flight distance. , This is the predicted takeoff distance. This is the takeoff distance limit. , Given the exponential weights, the preset condition that the output value of the utility function model satisfies is: When a conflict between point performance constraints and task objective requirements is detected, a Nash game is initiated. Using a utility function negotiation model based on Nash games, a Pareto-optimal performance index trade-off is automatically calculated.
[0050] In some other embodiments, after a preset number of optimization analyses are performed within the feasible domain boundary, the engine performance indicators and requirement items obtained from the optimization analyses corresponding to historical design requirements are used as the empirical knowledge base for RAG knowledge retrieval. After accessing the industry standard library and design requirements described in user natural language, the requirement items containing engine performance indicators are directly output using the RAG knowledge retrieval method. For example, the similarity between the multidimensional requirements of the current design requirements and the multidimensional requirements corresponding to historical design requirements in the empirical knowledge base can be compared. If the similarity is greater than or equal to a preset similarity threshold, the performance indicators whose similarity meets the preset similarity condition are output, and the corresponding requirement items are also output.
[0051] Based on the same inventive concept, this embodiment also provides an AI-physics model coupled-driven engine requirement analysis system, including:
[0052] Input parsing module 1 is used to access the aero-engine design requirements described by the user in natural language and the industry standard library, which includes airworthiness regulations; the RAG knowledge retrieval method is used to parse the industry standard library to obtain standard clause slices that match the design requirements;
[0053] The requirement analysis module 2 is used to analyze the design requirements into multi-dimensional requirements including point performance constraints, physical boundary constraints, and mission objective requirements. The point performance constraints include takeoff distance, climb rate, and landing distance. The physical boundary constraints include engine diameter and weight. The mission objective requirements include range and endurance.
[0054] Constraint analysis module 3 is used to draw key constraint curves based on point performance constraints, and to form a feasible domain constraint diagram for thrust-to-weight ratio and wing loading. The key constraint curves include takeoff distance constraints, climb rate constraints, and landing distance constraints.
[0055] Simulation analysis module 4 is used to establish a flight profile mission analysis simulation model. It finds the point with the minimum thrust-to-weight ratio within the feasible region of the feasible region constraint diagram as the input for flight profile mission analysis. The simulation obtains the engine performance indicators for each stage of the flight mission profile, which includes ground taxiing, takeoff, climb, cruise, descent, and landing. The performance indicators include thrust and fuel consumption rate.
[0056] The discrimination output module 5 is used to analyze and obtain the predicted value of the mission target under the minimum thrust-to-weight ratio based on the performance index. If the predicted value of the mission target meets the mission target requirements, it outputs a requirement item containing engine performance indexes; otherwise, it adjusts the point performance constraints or mission target requirements until the predicted value of the mission target meets the mission target requirements, and outputs a requirement item containing engine performance indexes.
[0057] The engine requirement analysis system in this embodiment also includes an optimization analysis module 6 and a knowledge base module 7, wherein:
[0058] The optimization analysis module 6 is used to construct a utility function model based on the achievement rate of each constraint or requirement index in the multidimensional requirements when the predicted value of the task target does not meet the requirements of the task target. The module takes the output value of the utility function model meeting the preset conditions as the optimization objective and performs optimization analysis within the feasible domain boundary to obtain engine performance indicators where the predicted value of the task target meets the requirements of the task target.
[0059] The knowledge base module 7 is used to store the engine performance indicators and requirement items obtained from optimization analysis corresponding to historical design requirements, forming an experience knowledge base for RAG knowledge retrieval. After performing optimization analysis within the feasible domain boundary to a preset number of times, it connects to the industry standard library and design requirements described in the user's natural language, and directly outputs requirement items containing engine performance indicators using the RAG knowledge retrieval method.
[0060] Example 2
[0061] See Figure 3 This embodiment takes the engine power selection of a certain type of civil airliner as an example to describe in detail the process of the AI-physics model coupled-driven engine requirement analysis method of the present invention. The specific process is as follows:
[0062] Step 1: Access the user's natural language description of the aircraft engine design requirements and the industry standard library, which includes airworthiness regulations; use the RAG knowledge retrieval method to parse the industry standard library and obtain standard clause slices that match the design requirements;
[0063] In this embodiment, Sentence-BERT (a sentence vector model based on BERT networks) is used to semantically slice long standard documents such as airworthiness regulations / rules, constructing a hierarchical vector index library with metadata tags such as "applicable aircraft type" and "clause type". During parsing, a large language model is used to extract key entities and implicit intents from user input, generating query embedding vectors, and the most relevant standard clause slices are obtained through an adaptive retrieval strategy. Subsequently, the retrieval results are injected into the large language model as system prompts, automatically identifying implicit airworthiness requirements and completing default mandatory constraints to form a structured context. For example, regarding the selection of engine power for narrow-body passenger aircraft on mainline civil aviation routes, the user inputs: "We need to select an engine for a single-aisle narrow-body passenger aircraft, requiring a range of 5500km, meeting the takeoff requirements of high-altitude airports, and ensuring that noise and emissions meet the latest ICAO standards."
[0064] By using the RAG knowledge retrieval method to search relevant industry standard libraries and historical data, it was found that "high bypass ratio engines, when meeting ICAO noise standards, are often accompanied by an increase in fan diameter, which can easily lead to physical conflicts with the wing-to-ground clearance." Therefore, the AI agent may identify "the latest ICAO standard" through RAG parsing -> search the regulatory library -> map it to "noise margin ≥ standard limit".
[0065] Step 2: Analyze the design requirements into multi-dimensional requirements including point performance constraints, physical boundary constraints, and mission objective requirements. The point performance constraints include takeoff distance, climb rate, and landing distance. The physical boundary constraints include engine diameter and weight. The mission objective requirements include range and endurance.
[0066] In this embodiment, the takeoff runway length is set by pre-identifying the takeoff requirements at high altitudes and then retrieving the design manual. (For example: Takeoff distance restrictions at an airport with an altitude of 3000m) (2500m). Based on RAG analysis and design requirement analysis, the multi-dimensional requirements are as follows:
[0067] Point performance constraints : such as takeoff distance Climb rate Equal mandatory constraints; among which:
[0068] Takeoff distance The constraint calculation is performed using an approximate solution formula based on the ground gliding physical model (considering variables such as thrust-to-weight ratio, wing loading, air density, and instantaneous weight fraction).
[0069] Climb rate The energy maneuver master equation is used for constraint mapping, which maps the transient climb requirement to the sea-level design thrust-to-weight ratio.
[0070] Physical boundary constraints: such as engine diameter, weight, and maximum permissible fan diameter. wait;
[0071] Mission objectives and requirements: such as designed range, endurance, etc.
[0072] Step 3: Based on point performance constraints, draw key constraint curves to form a feasible domain constraint diagram for thrust-to-weight ratio and wing loading. The key constraint curves include takeoff distance constraints, climb rate constraints, and landing distance constraints.
[0073] In this embodiment, a thrust-to-weight ratio-wing load feasible region constraint diagram is drawn, and each constraint boundary is calculated based on the requirements for high-altitude takeoff and single-engine climb.
[0074] Step 4: Establish a flight profile mission analysis simulation model. Find the point with the minimum thrust-to-weight ratio within the feasible region of the feasible region constraint diagram as the input for flight profile mission analysis. Use the simulation model to simulate and obtain engine performance indicators for each stage of the flight mission profile. Each stage of the flight mission profile includes ground taxiing, takeoff, climb, cruise, descent, and landing. The performance indicators include thrust and fuel consumption rate.
[0075] In this embodiment, the minimum value within the feasible region (the point with the smallest thrust-to-weight ratio) is taken. As a preliminary design point, input As input to the flight profile mission analysis simulation model, the simulation obtains engine thrust and fuel consumption rate at each stage of the flight mission profile, and calculates the cruise range. ,in, For cruising flight speed, For fuel consumption rate, Indicates the overall lift-to-drag ratio. It is the ratio of the initial weight to the final weight of the cruise segment.
[0076] Calculations show that The inability to meet the range requirement indicates that the chosen thrust-to-weight ratio was too high to meet the stringent takeoff distance requirements, leading to increased fuel consumption during the cruise phase and reduced fuel load. This results in the inability to achieve the intended range target, demonstrating a conflict between performance constraints and mission objective requirements. The two sides in this game:
[0077] Party A (Mission Objectives and Requirements): The objective is to meet the required flight range. They tend to reduce the thrust-to-weight ratio to reduce fuel consumption and increase fuel capacity.
[0078] Party B (Point Performance Constraints): The objective is to strictly adhere to takeoff distance requirements, which is mandatory. .
[0079] Step 5: Construct a utility function model based on the achievement rate of each constraint or requirement index in the multidimensional requirements. Take the output value of the utility function model satisfying the preset conditions as the optimization objective, and perform optimization analysis within the feasible region boundary to obtain the engine performance index that satisfies the task objective requirements.
[0080] In this embodiment, a utility function model is constructed using Nash game theory, and its expression is: ,in Output values for the utility function model. , This is the required range value. This is the predicted flight distance. , This is the predicted takeoff distance. This is the takeoff distance limit. , Given the exponential weights, the preset condition that the output value of the utility function model satisfies is: .
[0081] Based on the utility function model constructed using Nash game theory, the system searches the Pareto front to find a compromise solution:
[0082] Optimization results: The system found that appropriately relaxing the takeoff distance requirement can reduce the thrust-to-weight ratio and significantly increase the range to meet the requirements. .
[0083] Step 6: Output the requirement items containing engine performance indicators, and output the final optimized requirement specification.
[0084] Demand output: Generate specific demand items, such as demand-01: Ground takeoff thrust ≥ optimized ground takeoff thrust value, demand-02: Cruise fuel consumption rate ≤ optimized cruise fuel consumption rate value.
[0085] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for analyzing engine requirements driven by AI-physics model coupling, characterized in that, include: Access the user's natural language description of aero-engine design requirements and industry standard library, which includes airworthiness regulations; The RAG knowledge retrieval method is used to parse the industry standard library to obtain standard clause slices that match the design requirements. The design requirements are analyzed into multi-dimensional requirements including point performance constraints, physical boundary constraints, and mission objective requirements. The point performance constraints include takeoff distance, climb rate, and landing distance. The physical boundary constraints include engine diameter and weight. The mission objective requirements include range and endurance. Based on point performance constraints, key constraint curves are plotted to form feasible domain constraint diagrams for thrust-to-weight ratio and wing loading. The key constraint curves include takeoff distance constraints, climb rate constraints, and landing distance constraints. A flight profile mission analysis simulation model is established. The point with the minimum thrust-to-weight ratio within the feasible region of the feasible region constraint diagram is used as the input for flight profile mission analysis. The simulation model is used to simulate and obtain engine performance indicators for each stage of the flight mission profile, including ground taxiing, takeoff, climb, cruise, descent, and landing. The performance indicators include thrust and fuel consumption rate. Based on the performance indicators, the predicted value of the mission target under the minimum thrust-to-weight ratio is obtained. If the predicted value of the mission target meets the mission target requirements, a requirement item containing engine performance indicators is output. Otherwise, the point performance constraints or mission target requirements are adjusted until the predicted value of the mission target meets the mission target requirements, and a requirement item containing engine performance indicators is output. When the predicted value of the task objective does not meet the task objective requirements, or when there are conflicts between the performance constraints of the multidimensional requirements or between the task objective requirements, a utility function model is constructed based on the achievement rate of each constraint or requirement index in the multidimensional requirements. The optimization objective is to ensure that the output value of the utility function model meets preset conditions. Optimization analysis is performed within the feasible region boundary to obtain engine performance indicators where the predicted value of the task objective meets the task objective requirements. The utility function model is constructed using Nash game theory, and its expression is: ,in Output values for the utility function model. , This is the required range value. This is the predicted flight distance. , This is the predicted takeoff distance. This is the takeoff distance limit. , Given the exponential weights, the preset condition that the output value of the utility function model satisfies is: .
2. The engine demand analysis method according to claim 1, characterized in that, After the optimization analysis within the feasible domain boundary reaches a preset number of times, the engine performance indicators and requirement items obtained from the optimization analysis corresponding to the historical design requirements are used as the experience knowledge base for RAG knowledge retrieval. After accessing the industry standard library and design requirements described in the user's natural language, the RAG knowledge retrieval method is used to directly output the requirement items containing engine performance indicators.
3. The engine demand analysis method according to claim 2, characterized in that, Methods for directly outputting requirement items containing engine performance indicators using RAG knowledge retrieval methods include: Compare the multidimensional requirements of the current design requirements with the multidimensional requirements corresponding to the historical design requirements in the experience knowledge base. If the similarity is greater than or equal to the preset similarity threshold, output the performance index that the similarity meets the preset similarity condition, and output the corresponding requirement item.
4. An engine requirement analysis system driven by AI-physics model coupling, characterized in that, include: The input parsing module is used to access the aircraft engine design requirements described by the user in natural language and the industry standard library, which includes airworthiness regulations. The RAG knowledge retrieval method is used to parse the industry standard library to obtain standard clause slices that match the design requirements. The requirement analysis module is used to analyze the design requirements into multi-dimensional requirements including point performance constraints, physical boundary constraints, and mission objective requirements. The point performance constraints include takeoff distance, climb rate, and landing distance. The physical boundary constraints include engine diameter and weight. The mission objective requirements include range and endurance. The constraint analysis module is used to draw key constraint curves based on point performance constraints, and to form a feasible domain constraint diagram for thrust-weight ratio and wing loading. The key constraint curves include takeoff distance constraints, climb rate constraints, and landing distance constraints. The simulation analysis module is used to establish a flight profile mission analysis simulation model. It finds the point with the minimum thrust-to-weight ratio within the feasible region of the feasible region constraint diagram as the input for flight profile mission analysis. The simulation model is used to simulate and obtain engine performance indicators for each stage of the flight mission profile, including ground taxiing, takeoff, climb, cruise, descent, and landing. The performance indicators include thrust and fuel consumption rate. The discrimination output module is used to analyze and obtain the predicted value of the mission target under the minimum thrust-weight ratio based on the performance index. If the predicted value of the mission target meets the mission target requirements, it outputs a requirement item containing engine performance index; otherwise, it adjusts the point performance constraint or mission target requirements until the predicted value of the mission target meets the mission target requirements, and outputs a requirement item containing engine performance index. It also includes an optimization analysis module, used to construct a utility function model based on the achievement rate of each constraint or requirement index in the multidimensional requirements when the predicted value of the task target does not meet the task target requirements. The optimization objective is to satisfy a preset condition in the output value of the utility function model, and to perform optimization analysis within the feasible region boundary to obtain engine performance indicators where the predicted value of the task target meets the task target requirements. In the optimization analysis module, a Nash game is used to construct the utility function model, with the expression: ,in Output values for the utility function model. , This is the required range value. This is the predicted flight distance. , This is the predicted takeoff distance. This is the takeoff distance limit. , Given the exponential weights, the preset condition that the output value of the utility function model satisfies is: .
5. The engine demand analysis system according to claim 4, characterized in that, It also includes a knowledge base module, which stores the engine performance indicators and requirement items obtained from optimization analysis corresponding to historical design requirements to form an experience knowledge base for RAG knowledge retrieval. After performing optimization analysis within the feasible domain boundary to a preset number of times, it connects to the industry standard library and design requirements described in the user's natural language, and uses the RAG knowledge retrieval method to directly output requirement items containing engine performance indicators.