A large language model driven aero-engine system modeling and verification method and system

Abstract

This invention relates to the field of aero-engine technology and discloses a method and system for aero-engine system modeling and verification driven by a large language model. By acquiring relevant textual requirements of the aero-engine as a whole, and parsing the natural language requirements using a large language model, standardized requirement items are formed. From these, the overall engine functions are extracted to form a functional architecture. The functional architecture is then mapped to the logical and physical architectures, ultimately completing the scheme design and requirement decomposition. The aero-engine system modeling is completed using the large language model. This invention achieves efficient, accurate, and compliant aero-engine system modeling by constructing a coupling consistency overall score calculation model and performing coupling consistency verification on the logical architecture model. It avoids the shortcomings of existing large language model-based system modeling processes, such as unclear mapping between functional and logical architectures and inconsistencies in the main gas path topology and shaft coupling relationships within the logical architecture.

Description

Technical Field

[0001] This invention relates to the field of aero-engine technology, and discloses a method and system for modeling and verifying aero-engine systems driven by a large language model. Background Technology

[0002] As a complex system comprised of compressors, combustion chambers, turbines, and control systems, aero-engines rely heavily on manual operation when performing layer-by-layer mapping and tracing of "requirement-function-logic-physical (RFLP)" using existing methods. This process is cumbersome and prone to errors. Furthermore, there is a lack of automated consistency verification capabilities for system models. In particular, current specification-driven methods lack sufficient intelligence, making it difficult to perform deep semantic parsing and logical transformation of natural language requirements. This results in significant bottlenecks in overall engine requirements analysis, functional architecture integrity verification, parameter rationality assessment, and differentiated modeling of subsystems supporting multiple solutions, thus hindering the efficiency and quality of engine design. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for modeling and verifying aero-engine systems driven by a large language model, which can avoid the defects that are easy to exist in the existing system modeling process based on large language models, such as unclear mapping between functional architecture and logical architecture, and inconsistent main gas path topology and shaft coupling relationship in logical architecture.

[0004] To achieve the above-mentioned technical effects, the technical solution adopted by the present invention is as follows:

[0005] A large language model-driven method for modeling and validating aero-engine systems includes: Obtain input text for modeling aero-engine systems, including aero-engine overall requirements text, mission profile text, design constraint text, and historical design knowledge text; Using a large language model, component objects, interfaces, parameters, constraints, and traceability information in the input text are identified to construct aero-engine modeling semantic units. The traceability information is used to characterize the source of each corresponding component object, interface, parameter, and constraint. Each aero-engine modeling semantic unit includes at least an aero-engine category, component object name, interface definition, parameter definition, constraint definition, and source identifier. The component objects include one or more combinations of fans, compressors, combustion chambers, turbines, nozzles, sensors, actuators, controllers, and shafting components. The semantic units of the aero-engine modeling are mapped to standardized requirement items. Based on the aero-engine category and the standardized requirement items, a functional architecture model is generated using a large language model to describe the functional blocks of the aero-engine system and the dependencies, input-output relationships, and interface connections between the functional blocks. The functional blocks are the carrying units of the aero-engine system functions, which include intake, air splitting, air compression, combustion heat release, gas expansion and power generation, exhaust, thrust output, sensing and detection, control and regulation, and shaft power transmission, which are formed by decomposing the overall engine requirements. Based on the aforementioned functional architecture model, a logical architecture model for the aero-engine is generated using a large language model; wherein, the logical architecture model includes logical components, ports, interfaces, and connection relationships that carry the functional blocks; A coupling consistency total score calculation model is constructed, and coupling consistency verification is performed on the logical architecture model. If the output value of the coupling consistency total score calculation model is less than or equal to the first preset score threshold, it is determined that there are logical anomalies in the logical architecture model, and the large language model corrects the current logical architecture model until a logical architecture model without logical anomalies is generated. Based on the logical architecture model without logical anomalies, a corresponding physical architecture model is generated using a large language model. The physical architecture model includes physical component instances, port constraints, interface relationships, and parameter relationships formed by mapping the logical components. The physical component instances are the specific physical structures of the logical components in the aero-engine system. Based on the mapping relationship between logical components and physical component instances, the overall requirements are decomposed into physical component instances step by step to generate scheme design results and requirement decomposition results. Based on the scheme design results, requirement decomposition results, logical architecture model and physical architecture model, an aero-engine system model is generated using a large language model.

[0006] Furthermore, the method for performing coupling consistency verification on the logical architecture model includes: Construct a model for calculating the overall coupling consistency score ,in The overall coupling consistency score for the logical architecture model is given. Main gas path topology consistency score. Scoring is given for shaft coupling consistency. To control the consistency score of the feedback chain, For cross-view traceability consistency scoring, , , , This represents the weight coefficient corresponding to each scoring item, and ;in: Main gas path topology consistency score , Scoring of key node coverage in the logical architecture model. , The number of key logical component nodes matched in the generated logical architecture model. The key logical component nodes are logical component nodes that carry the main air path function of the aero-engine. For turbojet engines, the key logical component nodes include at least compressor, combustion chamber, turbine and nozzle. For turbofan engines, the key logical component nodes include at least fan, bypass duct components, compressor, combustion chamber, turbine and nozzle. This represents the total number of critical logic component nodes in the baseline path corresponding to the type of aero-engine. Scoring the sequential consistency of the logical architecture model. , For the first The location of each key logical component node in the generated logical architecture model path. For the first The location of each key logical component node in the baseline path corresponding to the type of aero-engine; Scoring the branching structure of the logical architecture model. , , , This represents the weight coefficient corresponding to each scoring item, and ; Shaft coupling consistency score , For the basic scoring of the shaft system, , This represents the total number of key logical component nodes that must be connected through the same shaft system in the logical architecture model for the current engine type. The number of critical logical component node pairs incorrectly established in the logical architecture model. This represents the total number of key logical component nodes missing in the logical architecture model. , This is the preset penalty coefficient; Scoring is given for cross-axle mis-attachment penalty items. , This represents the total number of all logical components connected to all axis systems in the logical architecture model. The number of logical components in the logical architecture model that are assigned to an axis system inconsistent with their original axis system; , This represents the weight coefficient corresponding to each scoring item, and ; Control feedback chain consistency score , The number of control chains that form a complete closed loop for each component of an aero-engine. This represents the total number of control chains in the logical architecture model. This represents the number of control edges in the control chain that characterize the directed control connections between adjacent nodes. To control the number of edges in the correct direction. , This represents the weight coefficient corresponding to each scoring item, and ; Cross-view traceability consistency score , This refers to the number of standardized requirement items for which a complete correspondence has been established between standardized requirement items and physical logical components. The total number of standardized requirement items generated.

[0007] Furthermore, after generating the corresponding physical architecture model using the large language model, and before decomposing the overall machine requirements into physical component instances to generate solution design results and requirement decomposition results, the process also includes constructing a parameter-topology joint verification and scoring model for the physical architecture model, performing parameter-topology joint verification on the physical architecture model; if the output value of the parameter-topology joint verification and scoring model is less than or equal to a second preset scoring threshold, it is determined that the physical architecture model has parameter distortion or insufficient physical constraints, and the large language model corrects the physical architecture model until there are no defects of parameter distortion or insufficient physical constraints in the physical architecture model.

[0008] Furthermore, the parameter-topology joint verification scoring model of the physical architecture model is as follows:

[0009] in The parameter-topology joint total score of the physical architecture model. Scoring the trend consistency of physical component instances. Scoring is applied to the topology location interface matching of physical component instances. For the coupled scoring of aero-engine parameters, Scoring based on aircraft engine type characteristics. , , , This represents the weight coefficient corresponding to each scoring item, and ;in: Trend consistency score of physical component instances , For the first The trend determination results for individual physical component instances, for compression components, indicate that when both the outlet pressure and inlet pressure are higher than the inlet temperature, the corresponding... Otherwise, the corresponding For expansion-type work components, when the outlet pressure is lower than the inlet pressure and the outlet temperature is lower than the inlet temperature, the corresponding... Otherwise the corresponding , This indicates the total number of physical component instances that need to participate in role trend verification in the baseline path corresponding to the type of aircraft engine. Topology location interface matching score of physical component instances ,in This represents the total number of physical component instances that need to participate in topology location interface verification, preset in the baseline path corresponding to the aircraft engine type. This refers to the number of erroneous path connections in the physical architecture model, such as physical component instances being connected to paths outside the preset ones, lacking corresponding upstream or downstream logical connections, or having inconsistent connection relationships with preset adjacent component objects. This represents the number of erroneous physical component instances in the physical architecture model that are located in positions outside their preset axis system. , This is the preset penalty coefficient; Aero-engine parameter coupling rating , This represents the total pressure ratio coupling error. , This refers to the overall pressure ratio of an aircraft engine. This refers to the fan pressure ratio. For low-pressure compressor pressure ratio, This refers to the pressure ratio of the high-pressure compressor. For bypass ratio error, , The bypass ratio of an aero engine. This represents the bypass flow in the diversion path. Core machine traffic; For traffic allocation error, , This represents the total flow rate at the fan outlet. , , This represents the weighting coefficients corresponding to each error term, and ; Main airflow continuity error , The main air inlet flow rate Main air outlet flow rate The bleed air flow rate is the bleed air volume from the main air passage. For thermal trend error, , This indicates the number of times thermal trend violations occur. The number of times thermal trend violations occur refers to the number of times when the judgment result of each physical component instance participating in the thermal trend check does not meet the preset thermal trend rule according to its component role. This indicates the total number of times the physical component instances that need to participate in the thermal trend check in the baseline path corresponding to the type of aero-engine have performed trend determination; , This represents the weighting coefficients corresponding to each error term, and ; Aircraft engine type feature matching score , Predetermine the number of physical component instances in the physical architecture model that match the key components of the current engine type. This indicates the number of critical components that should be present in the current engine type.

[0010] To achieve the above-mentioned technical effects, the present invention also provides a large language model-driven aero-engine system modeling and verification system for implementing the aforementioned large language model-driven aero-engine system modeling and verification method, comprising: The input text determination module is used to obtain the input text for modeling the aero-engine system. The input text includes the aero-engine overall requirements text, mission profile text, design constraint text, and historical design knowledge text. The semantic unit construction module is used to identify component objects, interfaces, parameters, constraints, and traceability information in the input text using a large language model, and construct aero-engine modeling semantic units. The traceability information is used to characterize the source of the corresponding component objects, interfaces, parameters, and constraints. Each aero-engine modeling semantic unit includes at least an aero-engine category, component object name, interface definition, parameter definition, constraint definition, and source identifier. The component objects include one or more combinations of fans, compressors, combustion chambers, turbines, nozzles, sensors, actuators, controllers, and shafting components. The functional architecture generation module is used to map the aero-engine modeling semantic units into standardized requirement items, and based on the aero-engine category and the standardized requirement items, use a large language model to generate a functional architecture model that describes the functional blocks of the aero-engine system and the dependencies, input-output relationships and interface connection relationships between the functional blocks; the functional blocks are the carrying units of the aero-engine system functions, and the aero-engine system functions include intake, air splitting, air compression, combustion heat release, gas expansion and power generation, exhaust, thrust output, sensing and detection, control and regulation and shaft power transmission formed by the decomposition of the overall engine requirements; A logical architecture generation module is used to generate a logical architecture model of an aero-engine based on the functional architecture model and using a large language model; wherein, the logical architecture model includes logical components, ports, interfaces and connection relationships that carry the functional blocks; The coupling consistency verification module is used to construct a coupling consistency total score calculation model and perform coupling consistency verification on the logical architecture model. If the output value of the coupling consistency total score calculation model is less than or equal to the first preset score threshold, it is determined that there is a logical anomaly in the logical architecture model, and the large language model corrects the current logical architecture model until a logical architecture model without logical anomalies is generated. The physical architecture generation module is used to generate a corresponding physical architecture model based on a logical architecture model that does not contain any logical anomalies, using a large language model. The physical architecture model includes physical component instances, port constraints, interface relationships, and parameter relationships formed by mapping the logical components. The physical component instances are the specific physical structures of the logical components in the aero-engine system. The system model generation module is used to decompose the overall requirements of the machine into physical component instances step by step according to the mapping relationship between logical components and physical component instances, generating scheme design results and requirement decomposition results. Based on the scheme design results, requirement decomposition results, logical architecture model and physical architecture model, the module uses a large language model to generate an aero-engine system model.

[0011] Furthermore, the method for performing coupling consistency verification on the logical architecture model in the coupling consistency verification module includes: Construct a model for calculating the overall coupling consistency score ,in The overall coupling consistency score for the logical architecture model is given. Main gas path topology consistency score. Scoring is given for shaft coupling consistency. To control the consistency score of the feedback chain, For cross-view traceability consistency scoring, , , , This represents the weight coefficient corresponding to each scoring item, and ;in: Main gas path topology consistency score , Scoring of key node coverage in the logical architecture model. , The number of key logical component nodes matched in the generated logical architecture model. The key logical component nodes are logical component nodes that carry the main air path function of the aero-engine. For turbojet engines, the key logical component nodes include at least compressor, combustion chamber, turbine and nozzle. For turbofan engines, the key logical component nodes include at least fan, bypass duct components, compressor, combustion chamber, turbine and nozzle. This represents the total number of critical logic component nodes in the baseline path corresponding to the type of aero-engine. Scoring the sequential consistency of the logical architecture model. , For the first The location of each key logical component node in the generated logical architecture model path. For the first The location of each key logical component node in the baseline path corresponding to the type of aero-engine; Scoring the branching structure of the logical architecture model. , , , This represents the weight coefficient corresponding to each scoring item, and ; Shaft coupling consistency score , For the basic scoring of the shaft system, , This represents the total number of key logical component nodes that must be connected through the same shaft system in the logical architecture model for the current engine type. The number of critical logical component node pairs incorrectly established in the logical architecture model. This represents the total number of key logical component nodes missing in the logical architecture model. , This is the preset penalty coefficient; Scoring is given for cross-axle mis-attachment penalty items. , This represents the total number of all logical components connected to all axis systems in the logical architecture model. The number of logical components in the logical architecture model that are assigned to an axis system inconsistent with their original axis system; , This represents the weight coefficient corresponding to each scoring item, and ; Control feedback chain consistency score , The number of control chains that form a complete closed loop for each component of an aero-engine. This represents the total number of control chains in the logical architecture model. This represents the number of control edges in the control chain that characterize the directed control connections between adjacent nodes. Number of control edges with correct orientation , ; Cross-view traceability consistency score , This refers to the number of standardized requirement items for which a complete correspondence has been established between standardized requirement items and physical logical components. The total number of standardized requirement items generated.

[0012] Furthermore, it also includes a joint verification module, which, after generating the corresponding physical architecture model using the large language model and before decomposing the overall machine requirements to physical component instances to generate the scheme design results and requirement decomposition results, also includes constructing a parameter-topology joint verification scoring model for the physical architecture model, performing parameter-topology joint verification on the physical architecture model; if the output value of the parameter-topology joint verification scoring model is less than or equal to a second preset scoring threshold, it is determined that the physical architecture model has parameter distortion or insufficient physical constraints, and the large language model corrects the physical architecture model until there are no defects of parameter distortion or insufficient physical constraints in the physical architecture model.

[0013] Furthermore, the parameter-topology joint verification scoring model of the physical architecture model of the joint verification module is as follows:

[0014] in The parameter-topology joint total score of the physical architecture model. Scoring the trend consistency of physical component instances. Scoring is applied to the topology location interface matching of physical component instances. For the coupled scoring of aero-engine parameters, Scoring based on aircraft engine type characteristics. , , , This represents the weight coefficient corresponding to each scoring item, and ;in: Trend consistency score of physical component instances , For the first The trend determination results for individual physical component instances, for compression components, indicate that when both the outlet pressure and inlet pressure are higher than the inlet temperature, the corresponding... Otherwise, the corresponding For expansion-type work components, when the outlet pressure is lower than the inlet pressure and the outlet temperature is lower than the inlet temperature, the corresponding... Otherwise the corresponding , This indicates the total number of physical component instances that need to participate in role trend verification in the baseline path corresponding to the type of aircraft engine. Topology location interface matching score of physical component instances ,in This represents the total number of physical component instances that need to participate in topology location interface verification, preset in the baseline path corresponding to the aircraft engine type. This refers to the number of erroneous path connections in the physical architecture model, such as physical component instances being connected to paths outside the preset ones, lacking corresponding upstream or downstream logical connections, or having inconsistent connection relationships with preset adjacent component objects. This represents the number of erroneous physical component instances in the physical architecture model that are located in positions outside their preset axis system. , This is the preset penalty coefficient; Aero-engine parameter coupling rating , This represents the total pressure ratio coupling error. , This refers to the overall pressure ratio of an aircraft engine. This refers to the fan pressure ratio. For low-pressure compressor pressure ratio, This refers to the pressure ratio of the high-pressure compressor. For bypass ratio error, , The bypass ratio of an aero engine. This represents the bypass flow in the diversion path. Core machine traffic; For traffic allocation error, , This represents the total flow rate at the fan outlet. , , This represents the weighting coefficients corresponding to each error term, and ; Main airflow continuity error , The main air inlet flow rate Main air outlet flow rate The bleed air flow rate is the bleed air volume from the main air passage. For thermal trend error, , This indicates the number of times thermal trend violations occur. The number of times thermal trend violations occur refers to the number of times when the judgment result of each physical component instance participating in the thermal trend check does not meet the preset thermal trend rule according to its component role. This indicates the total number of times the physical component instances that need to participate in the thermal trend check in the baseline path corresponding to the type of aero-engine have performed trend determination; , This represents the weighting coefficients corresponding to each error term, and ; Aircraft engine type feature matching score , Predetermine the number of physical component instances in the physical architecture model that match the key components of the current engine type. This indicates the number of critical components that should be present in the current engine type.

[0015] Compared with the prior art, the beneficial effects of this invention are: 1. This invention obtains relevant texts of the overall requirements of an aero-engine, parses the natural language requirements using a large language model, forms standardized requirement items, extracts the overall functions from them and forms a functional architecture; then maps the functional architecture to the logical architecture and physical architecture, and finally completes the scheme design and requirement decomposition, and uses a large language model to complete the system modeling of the aero-engine system.

[0016] 2. After generating the logical architecture model, this invention constructs a coupling consistency total score calculation model and performs coupling consistency verification on the logical architecture model, thereby achieving efficient, accurate and compliant system modeling of aero-engine systems; it avoids the defects that are easy to exist in the existing system modeling process based on large language models, such as non-standard extraction of requirement semantics, unclear mapping between functional architecture and logical architecture, and inconsistent main gas path topology and shaft coupling relationship in the logical architecture. Attached Figure Description

[0017] Figure 1 This is a flowchart of the modeling and verification method for aero-engine systems driven by a large language model, as shown in Example 1 or 2. Detailed Implementation

[0018] 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.

[0019] Example 1 See Figure 1 A large language model-driven method for modeling and validating aero-engine systems includes: Obtain input text for modeling aero-engine systems, including aero-engine overall requirements text, mission profile text, design constraint text, and historical design knowledge text; Using a large language model, component objects, interfaces, parameters, constraints, and traceability information in the input text are identified to construct aero-engine modeling semantic units. The traceability information is used to characterize the source of each corresponding component object, interface, parameter, and constraint. Each aero-engine modeling semantic unit includes at least an aero-engine category, component object name, interface definition, parameter definition, constraint definition, and source identifier. The component objects include one or more combinations of fans, compressors, combustion chambers, turbines, nozzles, sensors, actuators, controllers, and shafting components. The semantic units of the aero-engine modeling are mapped to standardized requirement items. Based on the aero-engine category and the standardized requirement items, a functional architecture model is generated using a large language model to describe the functional blocks of the aero-engine system and the dependencies, input-output relationships, and interface connections between the functional blocks. The functional blocks are the carrying units of the aero-engine system functions, which include intake, air splitting, air compression, combustion heat release, gas expansion and power generation, exhaust, thrust output, sensing and detection, control and regulation, and shaft power transmission, which are formed by decomposing the overall engine requirements. Based on the aforementioned functional architecture model, a logical architecture model for the aero-engine is generated using a large language model; wherein, the logical architecture model includes logical components, ports, interfaces, and connection relationships that carry the functional blocks; A coupling consistency total score calculation model is constructed, and coupling consistency verification is performed on the logical architecture model. If the output value of the coupling consistency total score calculation model is less than or equal to the first preset score threshold, it is determined that there are logical anomalies in the logical architecture model, and the large language model corrects the current logical architecture model until a logical architecture model without logical anomalies is generated. Based on the logical architecture model without logical anomalies, a corresponding physical architecture model is generated using a large language model. The physical architecture model includes physical component instances, port constraints, interface relationships, and parameter relationships formed by mapping the logical components. The physical component instances are the specific physical structures of the logical components in the aero-engine system. Based on the mapping relationship between logical components and physical component instances, the overall requirements are decomposed into physical component instances step by step to generate scheme design results and requirement decomposition results. Based on the scheme design results, requirement decomposition results, logical architecture model and physical architecture model, an aero-engine system model is generated using a large language model.

[0020] In this embodiment, by acquiring relevant textual requirements for the entire aero-engine, and parsing the natural language requirements using a large language model, standardized requirement items are formed. From these, the overall engine functions are extracted to form a functional architecture. The functional architecture is then mapped to the logical and physical architectures, ultimately completing the scheme design and requirement decomposition. The large language model is then used to complete the aero-engine system modeling. Furthermore, after generating the logical architecture model, this embodiment constructs a coupling consistency overall score calculation model and performs coupling consistency verification on the logical architecture model. This achieves efficient, accurate, and compliant aero-engine system modeling, avoiding the shortcomings of existing large language model-based system modeling processes, such as non-standard requirement semantic extraction, unclear mapping between functional and logical architectures, and inconsistent main air path topology and shaft coupling relationships within the logical architecture.

[0021] Based on the same inventive concept, this embodiment also provides a large language model-driven aero-engine system modeling and verification system for implementing the aforementioned large language model-driven aero-engine system modeling and verification method, including: The input text determination module is used to obtain the input text for modeling the aero-engine system. The input text includes the aero-engine overall requirements text, mission profile text, design constraint text, and historical design knowledge text. The semantic unit construction module is used to identify component objects, interfaces, parameters, constraints, and traceability information in the input text using a large language model, and construct aero-engine modeling semantic units. The traceability information is used to characterize the source of the corresponding component objects, interfaces, parameters, and constraints. Each aero-engine modeling semantic unit includes at least an aero-engine category, component object name, interface definition, parameter definition, constraint definition, and source identifier. The component objects include one or more combinations of fans, compressors, combustion chambers, turbines, nozzles, sensors, actuators, controllers, and shafting components. The functional architecture generation module is used to map the aero-engine modeling semantic units into standardized requirement items, and based on the aero-engine category and the standardized requirement items, use a large language model to generate a functional architecture model that describes the functional blocks of the aero-engine system and the dependencies, input-output relationships and interface connection relationships between the functional blocks; the functional blocks are the carrying units of the aero-engine system functions, and the aero-engine system functions include intake, air splitting, air compression, combustion heat release, gas expansion and power generation, exhaust, thrust output, sensing and detection, control and regulation and shaft power transmission formed by the decomposition of the overall engine requirements; A logical architecture generation module is used to generate a logical architecture model of an aero-engine based on the functional architecture model and using a large language model; wherein, the logical architecture model includes logical components, ports, interfaces and connection relationships that carry the functional blocks; The coupling consistency verification module is used to construct a coupling consistency total score calculation model and perform coupling consistency verification on the logical architecture model. If the output value of the coupling consistency total score calculation model is less than or equal to the first preset score threshold, it is determined that there is a logical anomaly in the logical architecture model, and the large language model corrects the current logical architecture model until a logical architecture model without logical anomalies is generated. The physical architecture generation module is used to generate a corresponding physical architecture model based on a logical architecture model that does not contain any logical anomalies, using a large language model. The physical architecture model includes physical component instances, port constraints, interface relationships, and parameter relationships formed by mapping the logical components. The physical component instances are the specific physical structures of the logical components in the aero-engine system. The system model generation module is used to decompose the overall requirements of the machine into physical component instances step by step according to the mapping relationship between logical components and physical component instances, generating scheme design results and requirement decomposition results. Based on the scheme design results, requirement decomposition results, logical architecture model and physical architecture model, the module uses a large language model to generate an aero-engine system model.

[0022] The large language model-driven aero-engine system modeling and verification system in this embodiment also includes a joint verification module. After generating the corresponding physical architecture model using the large language model, and before decomposing the overall requirements into physical component instances to generate scheme design results and requirement decomposition results, it also includes constructing a parameter-topology joint verification scoring model for the physical architecture model and performing parameter-topology joint verification on the physical architecture model. If the output value of the parameter-topology joint verification scoring model is less than or equal to a second preset scoring threshold, it is determined that the physical architecture model has parameter distortion or insufficient physical constraints, and the large language model corrects the physical architecture model until there are no defects of parameter distortion or insufficient physical constraints in the physical architecture model.

[0023] Example 2 See Figure 1 A large language model-driven method for modeling and validating aero-engine systems includes: Step 1: Obtain the input text for modeling the aero-engine system. The input text includes the aero-engine overall requirements text, mission profile text, design constraint text, and historical design knowledge text.

[0024] Step 2: Use a large language model to identify component objects, interfaces, parameters, constraints, and traceability information in the input text to construct aero-engine modeling semantic units; wherein, the traceability information is used to characterize the source of the corresponding component objects, interfaces, parameters, and constraints respectively, and the aero-engine modeling semantic unit includes at least the aero-engine category, component object name, interface definition, parameter definition, constraint definition, and source identifier, and the component objects include one or more combinations of fans, compressors, combustion chambers, turbines, nozzles, sensors, actuators, controllers, and shaft components; In this embodiment, the relevant parameters of the aero-engine include pressure ratio, bypass ratio, speed, flow rate, total temperature, total pressure, nozzle area, fuel flow rate, and thrust parameters; the constraints include main gas path topology constraints, bypass flow path topology constraints, high and low pressure shaft coupling constraints, control closed-loop constraints, and thermodynamic parameter trend constraints.

[0025] Step 3: Map the aero-engine modeling semantic units to standardized requirement items, and based on the aero-engine category and the standardized requirement items, use a large language model to generate a functional architecture model to describe the functional blocks of the aero-engine system and the dependencies, input-output relationships, and interface connection relationships between the functional blocks; the functional blocks are the carrying units of the aero-engine system functions, and the aero-engine system functions include intake, air splitting, air compression, combustion heat release, gas expansion and power generation, exhaust, thrust output, sensing and detection, control and regulation, and shaft power transmission, which are formed by decomposing the overall requirements. In this embodiment, the standardized requirement items include at least requirement identifier, requirement source, component object, functional description, interface definition, parameter definition, constraint definition, and traceability identifier; the component object, interface definition, parameter definition, and constraint definition are standardized in combination with the characteristics of the aero-engine system. Component objects include fans, compressors, combustion chambers, turbines, nozzles, sensors, actuators, controllers, and shaft components; interface definitions include air interfaces, fuel interfaces, gas interfaces, mechanical shaft interfaces, and control signal interfaces; parameter definitions include pressure ratio, bypass ratio, speed, flow rate, total temperature, total pressure, nozzle area, fuel flow rate, and thrust; and constraint definitions include main air path topology constraints, bypass flow path topology constraints, shaft coupling constraints, control closed-loop constraints, and thermodynamic trend constraints.

[0026] Step 4: Based on the functional architecture model, generate the logical architecture model of the aero-engine using a large language model; wherein, the logical architecture model includes logical components, ports, interfaces and connection relationships that carry the functional blocks; In this embodiment, the logical component is an intermediate abstract carrying unit of the functional block before the specific implementation layer. It is used to describe the composition and interaction relationship of the components that undertake the corresponding functions, without limiting the specific physical implementation form.

[0027] Step 5: Construct a coupling consistency total score calculation model and perform coupling consistency verification on the logical architecture model; if the output value of the coupling consistency total score calculation model is less than or equal to the first preset score threshold, it is determined that there are logical anomalies in the logical architecture model, and the large language model corrects the current logical architecture model until a logical architecture model without logical anomalies is generated. In this embodiment, the constructed coupling consistency total score calculation model is as follows: ,in The overall coupling consistency score for the logical architecture model is given. Main gas path topology consistency score. Scoring is given for shaft coupling consistency. To control the consistency score of the feedback chain, For cross-view traceability consistency scoring, , , , This represents the weight coefficient corresponding to each scoring item, and ;in: Main gas path topology consistency score , Scoring of key node coverage in the logical architecture model. , The number of key logical component nodes matched in the generated logical architecture model. The key logical component nodes are logical component nodes that carry the main air path function of the aero-engine. For turbojet engines, the key logical component nodes include at least compressor, combustion chamber, turbine and nozzle. For turbofan engines, the key logical component nodes include at least fan, bypass duct components, compressor, combustion chamber, turbine and nozzle. This represents the total number of critical logic component nodes in the baseline path corresponding to the type of aero-engine. Scoring the sequential consistency of the logical architecture model. , For the first The location of each key logical component node in the generated logical architecture model path. For the first The location of each key logical component node in the baseline path corresponding to the type of aero-engine; Scoring the branching structure of the logical architecture model. , , , This represents the weight coefficient corresponding to each scoring item, and .

[0028] In this embodiment, for a turbojet engine, the reference path is the main air path formed by the compressor, combustion chamber, turbine, and nozzle connected in sequence; for a turbofan engine, the reference path is the core engine path and bypass (outer bypass) path formed by the fan, bypass duct components, compressor, combustion chamber, turbine, and nozzle.

[0029] Shaft coupling consistency score , For the basic scoring of the shaft system, , This represents the total number of key logical component nodes that must be connected through the same shaft system in the logical architecture model for the current engine type. The number of critical logical component node pairs incorrectly established in the logical architecture model. This represents the total number of key logical component nodes missing in the logical architecture model. , This is the preset penalty coefficient; Scoring is given for cross-axle mis-attachment penalty items. , This represents the total number of all logical components connected to all axis systems in the logical architecture model. The number of logical components in the logical architecture model that are assigned to an axis system inconsistent with their original axis system; , This represents the weight coefficient corresponding to each scoring item, and ; Control feedback chain consistency score , The number of control chains that form a complete closed loop for each component of an aero-engine. This represents the total number of control chains in the logical architecture model. This represents the number of control edges in the control chain that characterize the directed control connections between adjacent nodes. To control the number of edges in the correct direction. , This represents the weight coefficient corresponding to each scoring item, and ; Cross-view traceability consistency score , This refers to the number of standardized requirement items for which a complete correspondence has been established between standardized requirement items and physical logical components. The total number of standardized requirement items generated.

[0030] This embodiment can promptly identify problems such as main air path breakage, incorrect connection sequence of key components, shaft coupling error, and non-closed loop control feedback chain during the logical architecture model generation stage, avoiding the system modeling defects caused by the disconnect between logical architecture and engine mechanism in existing methods.

[0031] Step 6: Based on the logical architecture model without logical anomalies, a corresponding physical architecture model is generated using a large language model. The physical architecture model includes physical component instances, port constraints, interface relationships, and parameter relationships formed by mapping the logical components. The physical component instances are the specific physical structures of the logical components in the aero-engine system, used to characterize the specific engine components, assemblies, or devices that implement the functions of the logical components.

[0032] Step 7: Construct a parameter-topology joint verification scoring model for the physical architecture model, and perform parameter-topology joint verification on the physical architecture model; In this embodiment, the parameter-topology joint verification scoring model of the physical architecture model is:

[0033] in The parameter-topology joint total score of the physical architecture model. Scoring the trend consistency of physical component instances. Scoring is applied to the topology location interface matching of physical component instances. For the coupled scoring of aero-engine parameters, Scoring based on aircraft engine type characteristics. , , , This represents the weight coefficient corresponding to each scoring item, and ;in: Trend consistency score of physical component instances , For the first The trend determination results for individual physical component instances, for compression components, indicate that when both the outlet pressure and inlet pressure are higher than the inlet temperature, the corresponding... Otherwise, the corresponding For expansion-type work components, when the outlet pressure is lower than the inlet pressure and the outlet temperature is lower than the inlet temperature, the corresponding... Otherwise the corresponding , This indicates the total number of physical component instances that need to participate in role trend verification in the baseline path corresponding to the type of aircraft engine. Topology location interface matching score of physical component instances ,in This represents the total number of physical component instances that need to participate in topology location interface verification, preset in the baseline path corresponding to the aircraft engine type. This refers to the number of erroneous path connections in the physical architecture model, such as physical component instances being connected to paths outside the preset ones, lacking corresponding upstream or downstream logical connections, or having inconsistent connection relationships with preset adjacent component objects. This represents the number of erroneous physical component instances in the physical architecture model that are located in positions outside their preset axis system. , This is the preset penalty coefficient; Aero-engine parameter coupling rating , This represents the total pressure ratio coupling error. , This refers to the overall pressure ratio of an aircraft engine. This refers to the fan pressure ratio. For low-pressure compressor pressure ratio, This refers to the pressure ratio of the high-pressure compressor. For bypass ratio error, , The bypass ratio of an aero engine. This represents the bypass flow in the diversion path. Core machine traffic; For traffic allocation error, , This represents the total flow rate at the fan outlet. , , This represents the weighting coefficients corresponding to each error term, and ; Main airflow continuity error , The main air inlet flow rate Main air outlet flow rate The bleed air flow rate is the bleed air volume from the main air passage. For thermal trend error, , This indicates the number of times thermal trend violations occur. The number of times thermal trend violations occur refers to the number of times when the judgment result of each physical component instance participating in the thermal trend check does not meet the preset thermal trend rule according to its component role. This indicates the total number of times the physical component instances that need to participate in the thermal trend check in the baseline path corresponding to the type of aero-engine have performed trend determination; , This represents the weighting coefficients corresponding to each error term, and In this embodiment, for compression components, a violation of thermal trend is defined as an instance where the outlet pressure is not higher than the inlet pressure or the outlet temperature is not higher than the inlet temperature; for combustion components, a violation of thermal trend is defined as an instance where the outlet temperature is not higher than the inlet temperature; and for expansion components, a violation of thermal trend is defined as an instance where the outlet pressure is not lower than the inlet pressure or the outlet temperature is not lower than the inlet temperature. The total number of thermal trend checks refers to the total number of times thermal trend determination is performed on each physical component instance participating in the thermal trend check. Each instance of a physical component participating in the thermal trend check performs one thermal trend determination, which is counted as one thermal trend check.

[0034] Aircraft engine type feature matching score , Predetermine the number of physical component instances in the physical architecture model that match the key components of the current engine type. This indicates the number of critical components that should be present in the current engine type.

[0035] If the output value of the parameter-topology joint verification scoring model is less than or equal to the second preset scoring threshold, it is determined that the physical architecture model has abnormal thermal trends, abnormal location interfaces, abnormal parameter coupling, or abnormal type characteristics. The large language model then corrects the physical architecture model until there are no defects such as abnormal thermal trends, abnormal location interfaces, abnormal parameter coupling, or abnormal type characteristics in the physical architecture model.

[0036] This embodiment introduces a parameter-topology joint verification based on parameter values, component roles, topological locations, path affiliation, and engine type after the physical architecture model is generated. It also uses the parameter-topology joint total score to uniformly evaluate the physical architecture. This can identify problems such as unreasonable thermal trends, mismatch between paths and interfaces, and incorrect coupling relationships of engine type-related parameters at the physical implementation layer. It avoids the defects of parameter relationship distortion, unreasonable physical implementation, and insufficient mechanism constraints in the existing system modeling process, thereby ensuring that the aero-engine system modeling language model text that meets the preset modeling constraints and consistency requirements can be generated.

[0037] Step 8: Based on the mapping relationship between the verified logical components and the verified physical component instances, the overall requirements are decomposed into physical component instances level by level to generate scheme design results and requirement decomposition results. Based on the scheme design results, requirement decomposition results, logical architecture model and physical architecture model, the aero-engine system model is generated using a large language model.

[0038] 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 modeling and verifying aero-engine systems driven by a large language model, characterized in that, include: Obtain input text for modeling aero-engine systems, including aero-engine overall requirements text, mission profile text, design constraint text, and historical design knowledge text; Using a large language model, component objects, interfaces, parameters, constraints, and traceability information in the input text are identified to construct aero-engine modeling semantic units. The traceability information is used to characterize the source of each corresponding component object, interface, parameter, and constraint. Each aero-engine modeling semantic unit includes at least an aero-engine category, component object name, interface definition, parameter definition, constraint definition, and source identifier. The component objects include one or more combinations of fans, compressors, combustion chambers, turbines, nozzles, sensors, actuators, controllers, and shafting components. The semantic units of the aero-engine modeling are mapped to standardized requirement items. Based on the aero-engine category and the standardized requirement items, a functional architecture model is generated using a large language model to describe the functional blocks of the aero-engine system and the dependencies, input-output relationships, and interface connections between the functional blocks. The functional blocks are the carrying units of the aero-engine system functions, which include intake, air splitting, air compression, combustion heat release, gas expansion and power generation, exhaust, thrust output, sensing and detection, control and regulation, and shaft power transmission, which are formed by decomposing the overall engine requirements. Based on the aforementioned functional architecture model, a logical architecture model for the aero-engine is generated using a large language model; wherein, the logical architecture model includes logical components, ports, interfaces, and connection relationships that carry the functional blocks; A coupling consistency total score calculation model is constructed, and coupling consistency verification is performed on the logical architecture model. If the output value of the coupling consistency total score calculation model is less than or equal to the first preset score threshold, it is determined that there are logical anomalies in the logical architecture model, and the large language model corrects the current logical architecture model until a logical architecture model without logical anomalies is generated. Based on the logical architecture model without logical anomalies, a corresponding physical architecture model is generated using a large language model. The physical architecture model includes physical component instances, port constraints, interface relationships, and parameter relationships formed by mapping the logical components. The physical component instances are the specific physical structures of the logical components in the aero-engine system. Based on the mapping relationship between logical components and physical component instances, the overall requirements are decomposed into physical component instances step by step to generate scheme design results and requirement decomposition results. Based on the scheme design results, requirement decomposition results, logical architecture model and physical architecture model, an aero-engine system model is generated using a large language model.

2. The method for modeling and verifying aero-engine systems driven by a large language model according to claim 1, characterized in that, The method for performing coupling consistency verification on the logical architecture model includes: Construct a model for calculating the overall coupling consistency score ,in The overall coupling consistency score for the logical architecture model is given. Main gas path topology consistency score. Scoring is given for shaft coupling consistency. To control the consistency score of the feedback chain, For cross-view traceability consistency scoring, , , , This represents the weight coefficient corresponding to each scoring item, and ;in: Main gas path topology consistency score , Scoring of key node coverage in the logical architecture model. , The number of key logical component nodes matched in the generated logical architecture model. The key logical component nodes are logical component nodes that carry the main air path function of the aero-engine. For turbojet engines, the key logical component nodes include at least compressor, combustion chamber, turbine and nozzle. For turbofan engines, the key logical component nodes include at least fan, bypass duct components, compressor, combustion chamber, turbine and nozzle. This represents the total number of critical logic component nodes in the baseline path corresponding to the type of aero-engine. Scoring the sequential consistency of the logical architecture model. , For the first The location of each key logical component node in the generated logical architecture model path. For the first The location of each key logical component node in the baseline path corresponding to the type of aero-engine; Scoring the branching structure of the logical architecture model. , , , This represents the weight coefficient corresponding to each scoring item, and ; Shaft coupling consistency score , For the basic scoring of the shaft system, , This represents the total number of key logical component nodes that must be connected through the same shaft system in the logical architecture model for the current engine type. The number of critical logical component node pairs incorrectly established in the logical architecture model. This represents the total number of key logical component nodes missing in the logical architecture model. , This is the preset penalty coefficient; Scoring is given for cross-axle mis-attachment penalty items. , This represents the total number of all logical components connected to all axis systems in the logical architecture model. The number of logical components in the logical architecture model that are assigned to an axis system inconsistent with their original axis system; , This represents the weight coefficient corresponding to each scoring item, and ; Control feedback chain consistency score , The number of control chains that form a complete closed loop for each component of an aero-engine. This represents the total number of control chains in the logical architecture model. This represents the number of control edges in the control chain that characterize the directed control connections between adjacent nodes. To control the number of edges in the correct direction. , This represents the weight coefficient corresponding to each scoring item, and ; Cross-view traceability consistency score , This refers to the number of standardized requirement items for which a complete correspondence has been established between standardized requirement items and physical logical components. The total number of standardized requirement items generated.

3. The method for modeling and verifying aero-engine systems driven by a large language model according to claim 1, characterized in that, After generating the corresponding physical architecture model using a large language model, and before decomposing the overall machine requirements into physical component instances to generate solution design results and requirement decomposition results, the process includes constructing a parameter-topology joint verification and scoring model for the physical architecture model, performing parameter-topology joint verification on the physical architecture model; if the output value of the parameter-topology joint verification and scoring model is less than or equal to a second preset scoring threshold, it is determined that the physical architecture model has parameter distortion or insufficient physical constraints, and the large language model corrects the physical architecture model until there are no defects of parameter distortion or insufficient physical constraints in the physical architecture model.

4. The method for modeling and verifying aero-engine systems driven by a large language model according to claim 3, characterized in that, The parameters of the physical architecture model—the topology joint verification scoring model—are as follows: in The parameter-topology joint total score of the physical architecture model. Scoring the trend consistency of physical component instances. Scoring is applied to the topology location interface matching of physical component instances. For the coupled scoring of aero-engine parameters, Scoring based on aircraft engine type characteristics. , , , This represents the weight coefficient corresponding to each scoring item, and ;in: Trend consistency score of physical component instances , For the first The trend determination results for individual physical component instances, for compression components, indicate that when both the outlet pressure and inlet pressure are higher than the inlet temperature, the corresponding... Otherwise, the corresponding For expansion-type work components, when the outlet pressure is lower than the inlet pressure and the outlet temperature is lower than the inlet temperature, the corresponding... Otherwise the corresponding , This indicates the total number of physical component instances that need to participate in role trend verification in the baseline path corresponding to the type of aircraft engine. Topology location interface matching score of physical component instances ,in This represents the total number of physical component instances that need to participate in topology location interface verification, preset in the baseline path corresponding to the aircraft engine type. This refers to the number of erroneous path connections in the physical architecture model, such as physical component instances being connected to paths outside the preset ones, lacking corresponding upstream or downstream logical connections, or having inconsistent connection relationships with preset adjacent component objects. This represents the number of erroneous physical component instances in the physical architecture model that are located in positions outside their preset axis system. , This is the preset penalty coefficient; Aero-engine parameter coupling rating , This represents the total pressure ratio coupling error. , This refers to the overall pressure ratio of an aircraft engine. This refers to the fan pressure ratio. For low-pressure compressor pressure ratio, This refers to the pressure ratio of the high-pressure compressor. For bypass ratio error, , The bypass ratio of an aero engine. This represents the bypass flow in the diversion path. Core machine traffic; For traffic allocation error, , This represents the total flow rate at the fan outlet. , , This represents the weighting coefficients corresponding to each error term, and ; Main airflow continuity error , The main air inlet flow rate Main air outlet flow rate The bleed air flow rate is the bleed air volume from the main air passage. For thermal trend error, , This indicates the number of times thermal trend violations occur. The number of times thermal trend violations occur refers to the number of times when the judgment result of each physical component instance participating in the thermal trend check does not meet the preset thermal trend rule according to its component role. This indicates the total number of times the physical component instances that need to participate in the thermal trend check in the baseline path corresponding to the type of aero-engine have performed trend determination; , This represents the weighting coefficients corresponding to each error term, and ; Aircraft engine type feature matching score , Predetermine the number of physical component instances in the physical architecture model that match the key components of the current engine type. This indicates the number of critical components that should be present in the current engine type.

5. A large language model-driven aero-engine system modeling and verification system, used to implement the large language model-driven aero-engine system modeling and verification method of claim 1, characterized in that, include: The input text determination module is used to obtain the input text for modeling the aero-engine system. The input text includes the aero-engine overall requirements text, mission profile text, design constraint text, and historical design knowledge text. The semantic unit construction module is used to identify component objects, interfaces, parameters, constraints, and traceability information in the input text using a large language model, and construct aero-engine modeling semantic units. The traceability information is used to characterize the source of the corresponding component objects, interfaces, parameters, and constraints. Each aero-engine modeling semantic unit includes at least an aero-engine category, component object name, interface definition, parameter definition, constraint definition, and source identifier. The component objects include one or more combinations of fans, compressors, combustion chambers, turbines, nozzles, sensors, actuators, controllers, and shafting components. The functional architecture generation module is used to map the aero-engine modeling semantic units into standardized requirement items, and based on the aero-engine category and the standardized requirement items, use a large language model to generate a functional architecture model that describes the functional blocks of the aero-engine system and the dependencies, input-output relationships and interface connection relationships between the functional blocks; the functional blocks are the carrying units of the aero-engine system functions, and the aero-engine system functions include intake, air splitting, air compression, combustion heat release, gas expansion and power generation, exhaust, thrust output, sensing and detection, control and regulation and shaft power transmission formed by the decomposition of the overall engine requirements; A logical architecture generation module is used to generate a logical architecture model of an aero-engine based on the functional architecture model and using a large language model; wherein, the logical architecture model includes logical components, ports, interfaces and connection relationships that carry the functional blocks; The coupling consistency verification module is used to construct a coupling consistency total score calculation model and perform coupling consistency verification on the logical architecture model. If the output value of the coupling consistency total score calculation model is less than or equal to the first preset score threshold, it is determined that there is a logical anomaly in the logical architecture model, and the large language model corrects the current logical architecture model until a logical architecture model without logical anomalies is generated. The physical architecture generation module is used to generate a corresponding physical architecture model based on a logical architecture model that does not contain any logical anomalies, using a large language model. The physical architecture model includes physical component instances, port constraints, interface relationships, and parameter relationships formed by mapping the logical components. The physical component instances are the specific physical structures of the logical components in the aero-engine system. The system model generation module is used to decompose the overall requirements of the machine into physical component instances step by step according to the mapping relationship between logical components and physical component instances, generating scheme design results and requirement decomposition results. Based on the scheme design results, requirement decomposition results, logical architecture model and physical architecture model, the module uses a large language model to generate an aero-engine system model.

6. The large language model-driven aero-engine system modeling and verification system according to claim 5, characterized in that, The coupling consistency verification module includes a method for performing coupling consistency verification on the logical architecture model, comprising: Construct a model for calculating the overall coupling consistency score ,in The overall coupling consistency score for the logical architecture model is given. Main gas path topology consistency score. Scoring is given for shaft coupling consistency. To control the consistency score of the feedback chain, For cross-view traceability consistency scoring, , , , This represents the weight coefficient corresponding to each scoring item, and ;in: Main gas path topology consistency score , Scoring of key node coverage in the logical architecture model. , The number of key logical component nodes matched in the generated logical architecture model. The key logical component nodes are logical component nodes that carry the main air path function of the aero-engine. For turbojet engines, the key logical component nodes include at least compressor, combustion chamber, turbine and nozzle. For turbofan engines, the key logical component nodes include at least fan, bypass duct components, compressor, combustion chamber, turbine and nozzle. This represents the total number of critical logic component nodes in the baseline path corresponding to the type of aero-engine. Scoring the sequential consistency of the logical architecture model. , For the first The location of each key logical component node in the generated logical architecture model path. For the first The location of each key logical component node in the baseline path corresponding to the type of aero-engine; Scoring the branching structure of the logical architecture model. , , , This represents the weight coefficient corresponding to each scoring item, and ; Shaft coupling consistency score , For the basic scoring of the shaft system, , This represents the total number of key logical component nodes that must be connected through the same shaft system in the logical architecture model for the current engine type. The number of critical logical component node pairs incorrectly established in the logical architecture model. This represents the total number of key logical component nodes missing in the logical architecture model. , This is the preset penalty coefficient; Scoring is given for cross-axle mis-attachment penalty items. , This represents the total number of all logical components connected to all axis systems in the logical architecture model. The number of logical components in the logical architecture model that are assigned to an axis system inconsistent with their original axis system; , This represents the weight coefficient corresponding to each scoring item, and ; Control feedback chain consistency score , The number of control chains that form a complete closed loop for each component of an aero-engine. This represents the total number of control chains in the logical architecture model. This represents the number of control edges in the control chain that characterize the directed control connections between adjacent nodes. Number of control edges with correct orientation , ; Cross-view traceability consistency score , This refers to the number of standardized requirement items for which a complete correspondence has been established between standardized requirement items and physical logical components. The total number of standardized requirement items generated.

7. The large language model-driven aero-engine system modeling and verification system according to claim 5, characterized in that, It also includes a joint verification module, which is used to generate the corresponding physical architecture model after using the large language model, and before decomposing the overall machine requirements into physical component instances to generate the scheme design results and requirement decomposition results. It also includes a parameter-topology joint verification scoring model for constructing the physical architecture model, and performs parameter-topology joint verification on the physical architecture model. If the output value of the parameter-topology joint verification scoring model is less than or equal to the second preset scoring threshold, it is determined that the physical architecture model has parameter distortion or insufficient physical constraints, and the large language model corrects the physical architecture model until there are no defects of parameter distortion or insufficient physical constraints in the physical architecture model.

8. The large language model-driven aero-engine system modeling and verification system according to claim 7, characterized in that, The parameters of the physical architecture model of the joint verification module—the topology joint verification scoring model—are as follows: in The parameter-topology joint total score of the physical architecture model. Scoring the trend consistency of physical component instances. Scoring is applied to the topology location interface matching of physical component instances. For the coupled scoring of aero-engine parameters, Scoring based on aircraft engine type characteristics. , , , This represents the weight coefficient corresponding to each scoring item, and ;in: Trend consistency score of physical component instances , For the first The trend determination results for individual physical component instances, for compression components, indicate that when both the outlet pressure and inlet pressure are higher than the inlet temperature, the corresponding... Otherwise, the corresponding For expansion-type work components, when the outlet pressure is lower than the inlet pressure and the outlet temperature is lower than the inlet temperature, the corresponding... Otherwise the corresponding , This indicates the total number of physical component instances that need to participate in role trend verification in the baseline path corresponding to the type of aircraft engine. Topology location interface matching score of physical component instances ,in This represents the total number of physical component instances that need to participate in topology location interface verification, preset in the baseline path corresponding to the aircraft engine type. This refers to the number of erroneous path connections in the physical architecture model, such as physical component instances being connected to paths outside the preset ones, lacking corresponding upstream or downstream logical connections, or having inconsistent connection relationships with preset adjacent component objects. This represents the number of erroneous physical component instances in the physical architecture model that are located in positions outside their preset axis system. , This is the preset penalty coefficient; Aero-engine parameter coupling rating , This represents the total pressure ratio coupling error. , This refers to the overall pressure ratio of an aircraft engine. This refers to the fan pressure ratio. For low-pressure compressor pressure ratio, This refers to the pressure ratio of the high-pressure compressor. For bypass ratio error, , The bypass ratio of an aero engine. This represents the bypass flow in the diversion path. Core machine traffic; For traffic allocation error, , This represents the total flow rate at the fan outlet. , , This represents the weighting coefficients corresponding to each error term, and ; Main airflow continuity error , The main air inlet flow rate Main air outlet flow rate The bleed air flow rate is the bleed air volume from the main air passage. For thermal trend error, , This indicates the number of times thermal trend violations occur. The number of times thermal trend violations occur refers to the number of times when the judgment result of each physical component instance participating in the thermal trend check does not meet the preset thermal trend rule according to its component role. This indicates the total number of times the physical component instances that need to participate in the thermal trend check in the baseline path corresponding to the type of aero-engine have performed trend determination; , This represents the weighting coefficients corresponding to each error term, and ; Aircraft engine type feature matching score , Predetermine the number of physical component instances in the physical architecture model that match the key components of the current engine type. This indicates the number of critical components that should be present in the current engine type.

Images