A multi-aircraft model migration optimization method, system, computing device and storage medium for large aviation models

Abstract

The present invention relates to the technical field of large aviation model optimization, and provides a multi-aircraft model migration optimization method and system for large aviation models to solve the problem that the existing technology relies on static feature space and is difficult to maintain the stability and consistency of knowledge migration in complex aerodynamic environments. The method of the present invention includes: constructing a flight mechanics feature library based on the flight mechanics parameters and control response data of the source aircraft model and the target aircraft model; performing coupled simulation to generate a compensation parameter set; establishing a model knowledge migration channel, reconstructing the knowledge migration channel according to the compensation parameter set, generating a target migration channel and a migration weight control function; adjusting the target migration channel and monitoring the distribution offset of the aerodynamic characteristic knowledge component, updating the aerodynamic difference characteristics, and iteratively performing coupled simulation to adjust the transmission state register operation until the distribution offset enters the steady-state operation range. The present invention realizes high-precision knowledge migration from wide-body aircraft to narrow-body aircraft, and meets aviation airworthiness safety standards.

Description

Technical Field

[0001] The present invention relates to the technical field of aviation large model optimization, and in particular to a multi-aircraft model migration optimization method and system for aviation large models. Background Art

[0002] In the aviation sector, with the accelerating pace of aircraft model upgrades and the growing demand for coordinated operations across multiple aircraft models, effectively transferring aerodynamic knowledge across models has become a key technical challenge. This is particularly true in scenarios where aerodynamic characteristics differ significantly, such as from wide-body aircraft to Airbus narrow-body aircraft. The significant differences in flight dynamics parameters and control responses make it difficult for traditional model transfer methods to meet the requirements for high-precision, low-error knowledge reuse.

[0003] Existing research has proposed a transfer learning framework based on unified feature space mapping to address the problem of aerodynamic characteristic transfer between different aircraft models. This framework constructs a universal flight data representation space, projects the flight data of the source and target aircraft models into this space, and uses domain adaptation technology to align the distribution differences between the two. A dynamic weighting mechanism is then used to adjust the degree of involvement of the source model knowledge in the training of the target model, thereby optimizing the knowledge transfer from one aircraft model to another. However, existing methods still have significant limitations when migrating aircraft models with large differences in aerodynamic characteristics. Specifically, their reliance on the construction of a static feature space can easily lead to information distortion and offset during the knowledge transfer process. The lack of a real-time monitoring and feedback adjustment mechanism for the knowledge distribution status along the migration path makes it difficult to maintain the stability and consistency of knowledge transfer in complex aerodynamic environments. Summary of the Invention

[0004] The present invention provides a multi-aircraft model migration optimization method and system for large aviation models, which are used to solve the problems in the existing technology that rely on static feature space, easily lead to information distortion and offset during knowledge migration, and are difficult to maintain the stability and consistency of knowledge migration in complex aerodynamic environments.

[0005] In a first aspect, the present invention provides a multi-aircraft migration optimization method for a large aviation model, comprising:

[0006] Based on the acquired flight dynamics parameters and control response data of the source and target aircraft, a flight dynamics feature library containing aerodynamic difference characteristics is constructed;

[0007] performing coupled simulation operations according to the flight mechanics feature library to generate a compensation parameter set;

[0008] Establish a knowledge transfer channel from the pre-trained source model to the target model;

[0009] Reconstructing the transmission buffer of the knowledge transfer channel according to the compensation parameter set to obtain a target transfer channel, and generating a transfer weight control function;

[0010] Using the migration weight control function, adjusting the transmission status register of the target migration channel to obtain an adjusted migration channel;

[0011] Monitor the distribution offset of the aerodynamic characteristic knowledge component in the adjusted migration channel during the migration process, and update the aerodynamic difference characteristics of the flight dynamics feature library according to the distribution offset, so as to iteratively perform the coupling simulation operation to the operation of adjusting the transmission state register until the distribution offset enters the steady-state operation range.

[0012] Optionally, a coupled simulation operation is performed according to the flight dynamics feature library to generate a compensation parameter set, including:

[0013] calculating the dynamic fluid pressure of the source aircraft and the target aircraft under the same flight state parameters based on the aerodynamic difference characteristics and flight state parameters of the flight dynamics feature library, so as to generate a solid deformation boundary constraint based on the dynamic fluid pressure;

[0014] The interaction process of the source aircraft and the target aircraft is synchronously stimulated under the same flight state parameters to obtain the fluid pressure distribution responses corresponding to the source aircraft and the target aircraft respectively;

[0015] The fluid pressure distribution response is transmitted to the effective area of the solid deformation boundary constraint to calculate the response difference value between the source aircraft model and the target aircraft model under the same fluid pressure distribution response, and a compensation parameter set is generated based on the correspondence between the response difference value and the flight state parameter.

[0016] Optionally, the fluid pressure distribution response is transmitted to the effective area of the solid deformation boundary constraint to calculate the response difference between the source aircraft and the target aircraft under the same fluid pressure distribution response, and a compensation parameter set is generated based on the corresponding relationship between the response difference and the flight state parameter, including:

[0017] The fluid pressure distribution response is transmitted to the action area of the solid deformation boundary constraint to obtain the solid deformation response quantities corresponding to the source model and the target model under the same fluid pressure distribution response, so as to calculate the response quantity difference value;

[0018] According to the altitude layer identification, speed range code and flight attitude angle of the flight state parameters, a three-dimensional flight state coordinate is formed;

[0019] Associating the response quantity difference value with a storage location corresponding to the three-dimensional flight state coordinate, and connecting the response quantity difference values of adjacent storage locations to construct a plurality of difference value change trajectories;

[0020] All difference value change trajectories are converted into corresponding executable compensation instruction parameters, and all executable compensation instruction parameters are combined to generate a compensation parameter set.

[0021] Optionally, a knowledge transfer channel is established from a pre-trained source model to a target model, including:

[0022] Based on the hardware resource distribution of the distributed computing node cluster, the pre-trained source aircraft model is allocated computing resources and loaded with the model to obtain an airworthiness certification model instance.

[0023] Initialize the target aircraft model based on its structure definition to obtain the model instance to be migrated.

[0024] According to the airworthiness certification model instance and the model instance to be migrated, a bidirectional data route is configured between the source aircraft model large model and the target aircraft model large model to form a basic migration channel;

[0025] According to the aerodynamic characteristics knowledge migration requirements, a feature extraction operator is implanted into the basic migration channel to generate a feature enhancement migration channel;

[0026] Synchronously loading the encoding rules of the flight state parameters into the feature enhancement migration channel to obtain a synchronized migration channel;

[0027] According to aviation data security specifications, the synchronized migration channel is securely encapsulated to obtain a knowledge migration channel.

[0028] Optionally, based on the compensation parameter set, reconstructing the transmission buffer of the knowledge transfer channel to obtain a target transfer channel, and generating a transfer weight control function, including:

[0029] generating aerodynamic difference compensation instruction parameters and load spectrum shift coefficients according to the compensation parameter set;

[0030] Obtaining an original airfoil eigenvector according to the aerodynamic characteristic knowledge component transmitted in the knowledge transfer channel;

[0031] Adjusting the pressure distribution parameters of the original airfoil eigenvector according to the pressure correction amount of the aerodynamic difference compensation instruction parameter to reconstruct the transmission buffer of the knowledge transfer channel to obtain a reconstructed transfer channel;

[0032] Calculating a change gradient value based on the flight state parameters and calculating a weight factor based on the load spectrum shift coefficient;

[0033] Compiling an initial migration weight control function, and loading the migration weight control function into the reconstructed migration channel to obtain a target migration channel;

[0034] The rudder effect response verifier of the target migration channel is started to obtain a verification result. According to the verification result and the weight factor, the operation boundary value of the initial migration weight control function is set to obtain the migration weight control function.

[0035] In a second aspect, the present invention provides a multi-aircraft migration optimization system for large aviation models, comprising:

[0036] A construction module is used to construct a flight dynamics feature library containing aerodynamic difference features based on the acquired flight dynamics parameters and control response data of the source aircraft and the target aircraft;

[0037] A simulation module, configured to perform coupled simulation operations based on the flight mechanics feature library to generate a set of compensation parameters;

[0038] Establish a module for establishing a knowledge transfer channel from a pre-trained source model to a target model;

[0039] A reconstruction module, configured to reconstruct the transmission buffer of the knowledge migration channel according to the compensation parameter set to obtain a target migration channel and generate a migration weight control function;

[0040] an adjustment module, configured to adjust the transmission status register of the target migration channel by using the migration weight control function to obtain an adjusted migration channel;

[0041] An updating module is used to monitor the distribution offset of the aerodynamic characteristic knowledge component in the adjusted migration channel during the migration process, and update the aerodynamic difference characteristics of the flight mechanics feature library according to the distribution offset, so as to iteratively perform the coupling simulation operation to the operation of adjusting the transmission state register until the distribution offset enters the steady-state operation range.

[0042] In a third aspect, an embodiment of the present invention provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a multi-aircraft migration optimization method for a large aviation model as described in the first aspect above.

[0043] In a fourth aspect, an embodiment of the present invention provides a computer storage medium storing a computer program, which, when executed by a computer, implements a multi-aircraft migration optimization method for a large aviation model as described in the first aspect.

[0044] In the present invention, based on the acquired flight dynamics parameters and control response data of the source aircraft model and the target aircraft model, a flight dynamics feature library containing aerodynamic difference characteristics is constructed; according to the flight dynamics feature library, a coupling simulation operation is performed to generate a compensation parameter set; a knowledge migration channel from a pre-trained source aircraft model large model to a target aircraft model large model is established; according to the compensation parameter set, the transmission buffer of the knowledge migration channel is reconstructed to obtain a target migration channel, and a migration weight control function is generated at the same time; using the migration weight control function, the transmission status register of the target migration channel is adjusted to obtain an adjusted migration channel; the distribution offset of the aerodynamic characteristic knowledge component in the adjusted migration channel during the migration process is monitored, and the aerodynamic difference characteristics of the flight dynamics feature library are updated according to the distribution offset, so as to iteratively perform the coupling simulation operation to the operation of adjusting the transmission status register until the distribution offset enters the steady-state operation range.

[0045] Beneficial effects of the present invention:

[0046] The technical solution provided by the present invention builds a flight dynamics feature library to unify the aerodynamic parameters of heterogeneous aircraft models, solving the feature space fragmentation problem caused by the dispersion of aircraft model data in traditional methods and providing a benchmark data source for cross-aircraft model migration. It also performs fluid and solid coupled simulations to physically quantify aerodynamic differences. Compensation parameter sets are generated based on differences in solid deformation responses, replacing static feature space mapping and eliminating aerodynamic characteristic distortion (such as wing bending and torsion deformation differences). It also achieves privacy-safe knowledge transfer. Through bidirectional routing and secure encapsulation in a distributed architecture, it mitigates the risk of sensitive flight data leakage and meets aviation data compliance requirements. It dynamically corrects feature space offsets and adjusts airfoil pressure distribution in real time based on the compensation parameter set, addressing information distortion caused by static feature space. It also simultaneously generates a safety-constrained migration weight control function to prevent migration process divergence. It achieves adaptive migration strength control by adjusting the transmission state register. It dynamically adjusts the knowledge transfer strength based on the real-time change rate of flight state parameters, addressing the response lag problem of traditional weighted mechanisms and adapting to aerodynamic mutation scenarios in the transonic region. It also triggers feature library updates and simulation iterations by monitoring distribution offsets, overcoming the lack of real-time feedback in existing methods and maintaining migration stability. Furthermore, based on the compensation parameter set, the transmission buffer of the migration channel is dynamically reconstructed. Specifically, the pressure distribution parameters are corrected through aerodynamic difference compensation instructions to reconstruct the channel feature space. The weight factors of the load spectrum offset coefficient and the change gradient value are compiled to compile the migration weight control function, and the operational boundary values are set through the rudder effect safety verifier to achieve dual embedding of physical characteristics and safety constraints. Through dynamic correction of pressure distribution parameters, the information distortion problem of traditional unified feature space mapping in aerodynamic mutation scenarios (such as transonic shock wave offset) is solved, ensuring that narrow-body aircraft accurately inherit the aerodynamic characteristics of wide-body aircraft. The rudder effect response verifier converts flight safety boundaries (such as rudder surface deflection limits) into operational boundary values of the weight control function, solving the problem of insufficient stability of traditional dynamic weighting mechanisms in complex aerodynamic environments and ensuring that the migration process meets airworthiness standards.

[0047] These and other aspects of the present invention will become more readily apparent from the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] Figure 1 A flowchart of a multi-aircraft model migration optimization method for a large aviation model provided by the present invention is shown;

[0050] Figure 2 A schematic structural diagram of a multi-aircraft model migration optimization system for large aviation models provided by the present invention is shown;

[0051] Figure 3 A schematic structural diagram of a computing device provided by the present invention is shown. DETAILED DESCRIPTION

[0052] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0053] In some of the processes described in the specification and claims of the present invention and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent the order of precedence, nor do they limit "first" and "second" to be different types.

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0055] To address the limitations of existing transfer learning methods in migrating aerodynamic characteristics across aircraft models, such as the difficulty of static feature spaces adapting to complex fluid-solid coupling effects and the lack of a dynamic monitoring and control mechanism for distribution offsets during knowledge transfer, the present invention focuses on modeling the aerodynamic differences between the source and target aircraft models. By constructing a flight mechanics feature library containing flight mechanics difference characteristics and combining fluid-solid coupling simulation to generate a set of compensation parameters, the present invention achieves dynamic reconstruction of the knowledge transfer channel and adaptive adjustment of the transmission state. Simultaneously, a migration weight control function and a distribution offset feedback mechanism are introduced to form a closed-loop iterative optimization structure, effectively improving the stability and accuracy of the migration process and breaking through the performance bottleneck of traditional solutions in highly differentiated migration scenarios such as from wide-body aircraft to narrow-body aircraft. Figure 1 The present invention provides a flowchart of a multi-aircraft model migration optimization method for a large aviation model. Figure 1 As shown, the method includes:

[0056] Step 101: constructing a flight dynamics feature library containing aerodynamic difference features based on the acquired flight dynamics parameters and control response data of the source aircraft and the target aircraft;

[0057] In this step, the source aircraft model refers to the aircraft model that provides the basis for knowledge transfer (such as a wide-body passenger aircraft model), and its aerodynamic model serves as the starting point for migration. The target aircraft model refers to the new aircraft model to which the knowledge is transferred (such as a narrow-body passenger aircraft model), which needs to adapt to the aerodynamic characteristics of the source aircraft model. Flight dynamics parameters refer to physical quantities that reflect aerodynamic performance, including airfoil bending and torsional stiffness and Reynolds number sensitivity coefficient. Maneuvering response data refers to data such as control surface deflection efficiency and roll response delay, which are used to quantify flight control characteristics. Aerodynamic difference characteristics refer to quantitative indicators of differences between aircraft models, such as the difference in lift curve slope and the center of pressure offset. The flight dynamics feature library refers to a structured database that stores aerodynamic difference feature matrices according to flight status.

[0058] In an embodiment of the present invention, based on the flight dynamics parameters (including airfoil bending-torsional stiffness and lift coefficient curve) and control response data (such as rudder deflection rate and elevator efficiency) of a source aircraft model (such as a wide-body passenger aircraft model) and a target aircraft model (such as a narrow-body passenger aircraft model), aerodynamic difference features (such as wing center of pressure offset) are extracted through feature fusion technology to construct a structured flight dynamics feature library. The library stores the aerodynamic characteristics differences between aircraft models in matrix form, providing input for subsequent simulations.

[0059] Step 102: performing a coupled simulation operation according to the flight dynamics feature library to generate a compensation parameter set;

[0060] In this step, the compensation parameter set refers to a parameter set of all executable compensation instruction parameters, which is used for channel reconstruction.

[0061] In an embodiment of the present invention, based on the aerodynamic difference characteristics of the flight mechanics feature library, the Navier-Stokes equations are solved using computational fluid dynamics flow field to calculate the dynamic fluid pressure, and a solid deformation boundary constraint is generated based on the dynamic fluid pressure. The interaction process between the source aircraft and the target aircraft is synchronously stimulated under the same flight state parameters to obtain the fluid pressure distribution responses corresponding to the source aircraft and the target aircraft respectively; the fluid pressure distribution response is transmitted to the action area of the solid deformation boundary constraint to calculate the response difference value between the source aircraft and the target aircraft under the same fluid pressure distribution response, and a compensation parameter set is generated based on its correspondence with the flight state parameters.

[0062] Step 103: Establish a knowledge transfer channel from the pre-trained source model to the target model;

[0063] In this step, the source aircraft model refers to a pre-trained neural network model that learns the aerodynamic characteristics of the source aircraft. The target aircraft model refers to the initial model framework to be migrated, inheriting the knowledge structure of the source model. The knowledge migration channel refers to the data transmission path in the distributed architecture, including bidirectional data routing and feature extraction operators.

[0064] In an embodiment of the present invention, computing resources are allocated and model loading is performed on the pre-trained source aircraft model in the hardware resource distribution of the distributed computing node cluster, and the target aircraft model model is initialized. Based on the results of the above steps, a basic channel is established by configuring bidirectional data routing, into which an aerodynamic feature extraction operator (such as an airfoil lift-to-drag ratio capturer) is implanted, and encoding rules for flight state parameters are loaded. Finally, a knowledge transfer channel is generated through secure encapsulation.

[0065] Step 104: reconstructing the transmission buffer of the knowledge transfer channel according to the compensation parameter set to obtain a target transfer channel, and generating a transfer weight control function;

[0066] In this step, the transfer buffer refers to the temporary storage area for aerodynamic characteristic knowledge components in memory. The target migration channel refers to the migration path reconstructed by the buffer. The migration weight control function is a machine code function that dynamically controls the knowledge transfer intensity (such as the bandwidth allocation ratio).

[0067] In an embodiment of the present invention, aerodynamic difference compensation instruction parameters and load spectrum offset coefficients are generated based on a compensation parameter set; the original airfoil characteristic vector is obtained based on the aerodynamic characteristic knowledge component transmitted in the knowledge migration channel; the pressure distribution parameters of the original airfoil characteristic vector are adjusted based on the pressure correction amount of the aerodynamic difference compensation instruction parameters to reconstruct the transmission buffer of the knowledge migration channel to obtain a reconstructed migration channel; the change gradient value is calculated based on the flight state parameters, and a weight factor of the load spectrum offset coefficient and the change gradient value is established to compile an initial migration weight control function, which is loaded into the reconstructed migration channel to obtain a target migration channel; the operating boundary value of the initial migration weight control function is further set to obtain a migration weight control function.

[0068] Step 105: using the migration weight control function, adjusting the transmission status register of the target migration channel to obtain an adjusted migration channel;

[0069] In this step, the transfer status register refers to the hardware register that stores the channel bandwidth and priority configuration. The adjusted migration path refers to the final migration path after weight adjustment, and the status register value has been updated.

[0070] In an embodiment of the present invention, an executable code segment of a migration weight control function is interpreted to separate a transmission intensity control parameter; a boundary constraint value of a rudder effect response verifier is obtained, and a transmission intensity correction value is calculated in combination with the real-time change rate of the flight state parameter and the transmission intensity control parameter; the correction value is input into a transmission scheduler of a target migration channel to trigger a bandwidth reallocation operation of the transmission scheduler; and the transmission status register of the target migration channel is adjusted according to the updated bandwidth configuration to obtain an adjusted migration channel.

[0071] Step 106: monitoring the distribution offset of the aerodynamic characteristic knowledge component in the adjusted migration channel during the migration process, and updating the aerodynamic difference characteristics of the flight dynamics feature library according to the distribution offset, so as to iteratively perform the coupled simulation operation to the operation of adjusting the transmission state register until the distribution offset enters the steady-state operation range;

[0072] In this step, aerodynamic knowledge components refer to core migration data, such as the airfoil pressure distribution vector and the rudder efficiency gain coefficient. The distribution offset refers to the difference (standard deviation) between the distribution of knowledge components and the baseline during the migration process. The steady-state operating range refers to the allowable fluctuation range of the distribution offset (-0.15 to +0.15), which is determined by airworthiness standards.

[0073] In this embodiment of the present invention, the distribution offset of the aerodynamic characteristic knowledge component (e.g., the airfoil pressure distribution vector) in the adjusted migration channel is monitored. The distribution offset is calculated as the current distribution standard deviation minus the baseline distribution standard deviation. When the distribution offset exceeds the steady-state operating range (±0.15), the aerodynamic difference characteristics in the flight dynamics feature library are updated (e.g., the leading edge pressure coefficient threshold is increased by 10%), and the adjustment operation from the fluid-structure interaction simulation to the transfer status register is retriggered until the offset converges.

[0074] The embodiment of the present invention quantifies the aerodynamic differences between aircraft models through fluid-structure coupling simulation to generate dynamic compensation parameters; utilizes buffer reconstruction to correct the characteristic space distortion problem, breaking through the limitations of static mapping; combines the weight control mechanism of rudder effectiveness safety verification to solve the migration instability defect under complex working conditions; and uses a closed-loop iterative mechanism to optimize the migration path in real time through offset monitoring, ultimately achieving high-precision knowledge transfer from wide-body aircraft to narrow-body aircraft, improving the error convergence speed and meeting airworthiness safety standards.

[0075] The present invention provides a specific embodiment, step 102, performing a coupled simulation operation based on the flight dynamics feature library to generate a compensation parameter set, specifically comprising the following steps:

[0076] Step 201: Calculating the dynamic fluid pressure of the source aircraft and the target aircraft under the same flight state parameters based on the aerodynamic difference characteristics and flight state parameters of the flight dynamics feature library, and generating solid deformation boundary constraints based on the dynamic fluid pressure;

[0077] In this step, the flight state parameter refers to the three-dimensional combination of flight altitude (unit: km), Mach number (unit: 1), and angle of attack (unit: degrees), which uniquely identifies a specific flight condition. Dynamic fluid pressure refers to the time-varying pressure value (unit: kPa) exerted by the airflow on the aircraft surface. It is calculated by multiplying the air density by the square of the velocity and then by the pressure coefficient. Solid deformation boundary constraints refer to the structural displacement limits derived from the fluid pressure (e.g., wing deflection ≤ 6mm) and are calculated based on the material flexibility matrix (deformation = pressure × flexibility).

[0078] In an embodiment of the present invention, based on the aerodynamic difference characteristics of the flight mechanics feature library (such as the difference in wing bending and torsional stiffness) and flight state parameters (such as altitude 10km / Mach 0.8), the Navier-Stokes equations are solved by computational fluid dynamics to calculate the dynamic fluid pressure of the source aircraft and the target aircraft under the same operating conditions. Dynamic fluid pressure = air density × velocity squared × pressure coefficient; solid deformation boundary constraints are generated based on the dynamic fluid pressure. Specifically, the fluid pressure value is multiplied by the material flexibility matrix (which reflects the material stiffness characteristics) to obtain a displacement constraint value (such as the maximum displacement of the wing leading edge ≤120mm), which is used as the solid domain simulation boundary constraint.

[0079] Step 202: Synchronously stimulate the interaction process of the source aircraft and the target aircraft under the same flight state parameters to obtain fluid pressure distribution responses corresponding to the source aircraft and the target aircraft respectively;

[0080] In this step, the fluid pressure distribution response refers to the set of pressure values at discrete points on the aircraft surface (e.g., 30 kPa at the leading edge, 50 kPa at the maximum thickness point, and 20 kPa at the trailing edge), which is obtained by solving the computational fluid dynamics flow field.

[0081] In an embodiment of the present invention, under the same flight state parameters (such as an altitude of 10 km and Mach 0.8), the distributed simulation platform synchronously stimulates the fluid-solid interaction process of the source aircraft and the target aircraft, specifically including using the finite volume method to discretize the fluid domain and calculate the surface pressure distribution, such as the surface pressure distribution on the upper surface of the wing = the total inlet pressure - the dynamic fluid pressure × the loss coefficient; mapping it to the solid grid nodes (such as through radial basis function interpolation) to obtain the fluid pressure distribution response of the source aircraft and the target aircraft respectively, where the row vector represents the chord position and the column vector represents the surface pressure distribution.

[0082] Step 203: Transmitting the fluid pressure distribution response to the area of action of the solid deformation boundary constraint to calculate the difference in response between the source aircraft and the target aircraft under the same fluid pressure distribution response, and generating a compensation parameter set based on the correspondence between the difference in response and the flight state parameters;

[0083] In this step, the response difference value refers to the absolute difference in the solid deformation of the two models under the same working conditions.

[0084] In an embodiment of the present invention, the fluid pressure distribution response is transmitted to the action area of the solid deformation boundary constraint (such as the 20% chord length area of the leading edge), and the solid equilibrium equation is solved based on the solid deformation boundary constraint to obtain the solid deformation response; the response difference between the source aircraft model and the target aircraft model under the same pressure distribution is calculated, and the response difference = |source wing shape variable-target wing shape variable|; three-dimensional flight state coordinates are formed according to the altitude layer identifier, speed range code and flight attitude angle of the flight state parameters; the response difference is associated with the storage location corresponding to the three-dimensional flight state coordinate, and the response difference values of adjacent storage locations are connected to construct multiple difference value change trajectories; all difference value change trajectories are converted into corresponding executable compensation instruction parameters, and all executable compensation instruction parameters are combined to generate a compensation parameter set.

[0085] The embodiments of the present invention accurately quantify the differences in aerodynamic-structure coupling between aircraft models through physical simulation. Specifically, the dynamic fluid pressure calculation breaks through the limitations of empirical formulas and reflects the impact of real transonic shock waves. The bidirectional fluid-solid coupling solves the errors of traditional one-way simulation and captures the feedback effect of wing deformation on airflow. The three-dimensional trajectory compensation parameters eliminate the discrete errors of the static lookup table method and achieve continuous adaptation of the entire flight envelope.

[0086] The present invention provides a specific embodiment, step 203, transmitting the fluid pressure distribution response to the effective area of the solid deformation boundary constraint to calculate the response difference between the source aircraft model and the target aircraft model under the same fluid pressure distribution response, and generating a compensation parameter set based on the corresponding relationship between the response difference and the flight state parameter, specifically including the following steps:

[0087] Step 211: transmitting the fluid pressure distribution response to the action area of the solid deformation boundary constraint to obtain the solid deformation response quantities corresponding to the source model and the target model under the same fluid pressure distribution response, so as to calculate the response quantity difference value;

[0088] In this step, the solid deformation response refers to the physical deformation of the aircraft structure under the action of fluid pressure (unit: mm), including wing deflection and fuselage bending.

[0089] In this embodiment of the present invention, a fluid pressure distribution response (e.g., a series of pressure values on the wing's upper surface) is transferred via a finite element coupling interface to the region where solid deformation boundary constraints are applied (e.g., the connection point between the front spar and the skin). Finite element analysis is then used to determine the structural stress distribution. This allows the determination of the solid deformation response under the same pressure for both the source aircraft (a wide-body aircraft) and the target aircraft (a narrow-body aircraft). For example, wing deflection = structural stress distribution × flexibility. The absolute difference between the two values is then calculated: the response difference = |source aircraft wing deflection - target aircraft wing deflection|.

[0090] Step 212: Generating three-dimensional flight state coordinates based on the altitude layer identifier, speed range code, and flight attitude angle of the flight state parameters;

[0091] In this step, the altitude level identifier refers to the digital code of the flight altitude (an integer from 0 to 15), with 0 corresponding to sea level and 15 corresponding to 15 km altitude, according to the international standard flight level hierarchy. The speed range code refers to the discretized Mach number code (an integer from 0 to 12), with 0 corresponding to stationary and 12 corresponding to Mach 1.2, covering the subsonic to supersonic flight envelope. The flight attitude angle refers to the aircraft's angle of attack range (-10° to +40°), with negative values indicating a dive attitude and positive values indicating a climb attitude. The three-dimensional flight state coordinate is an integer triple (e.g., [10, 8, 5]) consisting of the altitude level identifier, speed range code, and flight attitude angle, uniquely identifying a specific flight condition.

[0092] In this embodiment of the present invention, flight state parameters are parsed. Specifically, the altitude level identifier includes values from 0 to 15, corresponding to 0-15 km (e.g., an altitude of 10 km is coded as 10); the speed range code includes values from 0 to 12, corresponding to Mach 0-1.2 (e.g., Mach 0.8 is coded as 8); and the flight attitude angle range is from -10° to +40° (e.g., an angle of attack of 5°). These three parameters are combined to form a three-dimensional flight state coordinate (e.g., [10, 8, 5]), which serves as a spatial positioning reference for aerodynamic differences.

[0093] Step 213: Associating the response difference value with the storage location corresponding to the three-dimensional flight state coordinate, and connecting the response difference values of adjacent storage locations to construct multiple difference value change trajectories;

[0094] In this step, the storage location refers to the physical address of the memory. The difference value change trajectory refers to the interpolated continuous curve of the response quantity difference between adjacent three-dimensional coordinate points, reflecting the gradual change of aerodynamic differences with flight status.

[0095] In an embodiment of the present invention, the response amount difference value (such as 1.7 mm) is associated with the storage location corresponding to the three-dimensional coordinate, such as memory address = base address + height × 13 × 51 + speed × 51 + angle; the adjacent coordinate units are traversed, and the response amount difference values are connected by linear interpolation, where interpolation = previous response amount difference value + (next response amount difference value - previous response amount difference value) × step length ratio, to construct a difference value change trajectory, such as a continuous curve showing that the deflection difference at an attack angle of 5° to 6° increases from 1.7 mm to 1.9 mm.

[0096] Step 214: convert all difference value change trajectories into corresponding executable compensation instruction parameters, combine all executable compensation instruction parameters, and generate a compensation parameter set;

[0097] In this step, the executable compensation instruction parameter refers to a compensation command in a machine code format, which can be directly executed by the flight control computer.

[0098] In an embodiment of the present invention, the difference value change trajectory is converted into machine-executable compensation instruction parameters. Specifically, key points in the difference value change trajectory, such as the response quantity difference value corresponding to an angle of attack of 5° or 6°, are extracted and compiled into an avionics instruction format; all executable compensation instruction parameters are combined to generate a compensation parameter set.

[0099] The embodiments of the present invention resolve traditional empirical estimation errors through finite element analysis of stress-deformation conversion; discretize continuous flight states into addressable storage units, breaking through the limitations of two-dimensional lookup tables; and generate continuous compensation curves through linear interpolation to eliminate the adaptation blind spots of aerodynamic mutations in the transonic region.

[0100] The present invention provides a specific embodiment, step 103, establishing a knowledge transfer channel from a pre-trained source model large model to a target model large model, specifically comprising the following steps:

[0101] Step 301: Based on the hardware resource distribution of the distributed computing node cluster, computing resources are allocated and model loading is performed on the pre-trained source aircraft model to obtain an airworthiness certification model instance;

[0102] In this step, the distributed computing node cluster refers to the integrated modular avionics architecture of the avionics system, consisting of partitioned computers for flight control, navigation, and engine control, interconnected via a backplane bus. Hardware resource distribution refers to the physical allocation of CPU cores, memory blocks, and I / O ports. The airworthiness certification model instance refers to a DO-178C Level A certified neural network instance, including model weights, runtime environment, and digital signature, ready for deployment on avionics hardware.

[0103] In this embodiment of the present invention, based on the hardware resource distribution of a distributed computing node cluster (such as an integrated modular avionics module of an avionics system), for example, CPU cores 0-3 are allocated to the flight control partition. Memory partition management technology is used to allocate resources for the pre-trained source aircraft model: 128MB of memory space is allocated to load the model weight file; and the model instance is compiled using the DO-178C Level A certified tool chain to generate an airworthiness certified model instance.

[0104] Step 302: Initialize the target model large model according to the structure definition of the target model large model to obtain the model instance to be migrated;

[0105] In this step, the model instance to be migrated refers to the initialized entity of the target model neural network, containing an untrained weight matrix and an empty data buffer, waiting for knowledge injection.

[0106] In an embodiment of the present invention, based on the structural definition of the target aircraft model, such as the number of neural network layers and nodes of a narrow-body passenger aircraft model, a model initialization function is called to assign random initial values to the weight matrix, and the assigned data is written to the memory address of the target node to generate a model instance to be migrated.

[0107] Step 303: configuring a bidirectional data route between the source aircraft model and the target aircraft model based on the airworthiness certification model instance and the model instance to be migrated, thereby forming a basic migration channel.

[0108] In this step, bidirectional data routing refers to two independent virtual links that enable data transmission from the source model to the target model (forward) and from the target back to the source (reverse). The basic migration channel refers to the minimum communication path that only includes bidirectional routing and does not embed data processing functions.

[0109] In an embodiment of the present invention, a virtual link is configured through an aviation data exchange network switch based on an airworthiness certification model instance (e.g., address 0x5000) and a model instance to be migrated (e.g., address 0x8000), wherein the virtual link from the aircraft model large model to the target aircraft model large model is numbered 101, with a bandwidth of 64 kilobits per second; the virtual link from the target aircraft model large model to the aircraft model large model is numbered 102, with a bandwidth of 32 kilobits per second, thereby constructing a basic migration channel, and storing the routing table information of the channel at address 0x3000.

[0110] Step 304: Based on the aerodynamic characteristics knowledge migration requirements, a feature extraction operator is implanted into the basic migration channel to generate a feature enhancement migration channel.

[0111] In this step, the aerodynamic knowledge transfer requirement refers to the core aerodynamic parameters that must be retained during the migration process, such as the lift coefficient derivative and the center of pressure position. Feature extraction operators are hardware-accelerated mathematical units, such as convolution kernels implemented using field-programmable gate arrays, used to extract features such as airfoil pressure gradients in real time. The feature-enhanced migration channel is a data path that adds feature extraction capabilities to the basic migration channel, enabling online processing of aerodynamic knowledge components.

[0112] In an embodiment of the present invention, a feature extraction operator is injected into the basic migration channel according to the requirements of aerodynamic characteristic knowledge migration. Specifically, the VL101 data packet processing function is modified, and a convolution kernel (which is a 3×3 Gaussian filter) is added to extract pressure distribution features. The operator machine code (such as 0x90F2) is written into the channel coprocessor to generate a feature enhancement migration channel, thereby reducing the feature extraction delay.

[0113] Step 305: Synchronously loading the encoding rules of the flight state parameters into the feature enhancement migration channel to obtain a synchronized migration channel;

[0114] In this step, the encoding rule refers to the digital mapping rule of the flight status parameters. The synchronized migration channel refers to the channel that supports time-triggered communication, and each node transmits synchronously according to the global clock.

[0115] In an embodiment of the present invention, the encoding rules for flight state parameters (including altitudes 0 to 15 represented by 4-bit binary, the Mach number multiplied by 10 to convert into an 8-bit integer value, and the angle of attack plus 10 as an offset value) are burned into the field programmable gate array registers of the feature enhancement migration channel; the clocks of each node are synchronized via a time-triggered Ethernet bus to obtain a synchronized migration channel with a state synchronization period of 5ms.

[0116] Step 306: Perform security encapsulation on the synchronized migration channel according to aviation data security specifications to obtain a knowledge migration channel;

[0117] In this step, aviation data security specifications refer to data encryption, access control, and integrity verification requirements defined according to standards.

[0118] In this embodiment of the present invention, the following security measures are implemented for the synchronized migration channel in accordance with aviation data security specifications. These include encrypting the payload data of the VL101 and VL102 virtual links using the Advanced Encryption Standard 256-bit algorithm to ensure confidentiality of the transmitted content; establishing a media access control address whitelist (e.g., 00-0C-29-XX) to restrict network access to only authorized devices, achieving effective access control; and adding a 32-bit cyclic redundancy check (CRC) to each frame of data to ensure data transmission integrity. These measures create a knowledge migration channel that meets the security level requirements of avionics systems.

[0119] The embodiments of the present invention prevent key flight control tasks from being interfered with by the migration process by allocating model resources; implanting feature extraction operators to accurately extract the aerodynamic characteristics of the airfoil to avoid knowledge distortion during transmission; using time-triggered Ethernet to achieve clock synchronization to ensure deterministic cross-node data transmission; and preventing unauthorized access to meet the security requirements of the avionics system security level.

[0120] The present invention provides a specific embodiment, step 104, reconstructing the transmission buffer of the knowledge transfer channel according to the compensation parameter set to obtain a target transfer channel, and generating a transfer weight control function, specifically including the following steps:

[0121] Step 401: generating aerodynamic difference compensation instruction parameters and load spectrum shift coefficients according to the compensation parameter set;

[0122] In this step, the aerodynamic difference compensation instruction parameter refers to the physical correction value parsed from the machine instruction in the compensation parameter set, which is used to adjust the airfoil characteristics. The load spectrum offset coefficient refers to the difference in structural load between the source and target aircraft under the same operating conditions, reflecting the difference in aerodynamic stiffness.

[0123] In this embodiment of the present invention, machine instructions in the compensation parameter set are parsed to extract aerodynamic difference compensation instruction parameters, such as a pressure correction of +1.2 kPa and a load spectrum offset coefficient, such as a bending moment difference of 0.8 kN·m. An instruction decoder based on the x86 instruction set is used to separate the opcode from the parameters.

[0124] Step 402: obtaining an original airfoil eigenvector based on the aerodynamic characteristic knowledge component transmitted in the knowledge transfer channel;

[0125] In this step, the aerodynamic knowledge component refers to the core data units transmitted in the migration channel (such as the airfoil pressure coefficient sequence), which contains key information such as lift and drag characteristics. The original airfoil eigenvector refers to the set of uncorrected pressure distribution parameters of the wing profile, stored in the order [leading edge, maximum thickness point, trailing edge].

[0126] In an embodiment of the present invention, a direct memory access controller is used to read aerodynamic characteristic knowledge components in a transmission buffer of a knowledge transfer channel, such as the pressure distribution sequence [0.8, 0.5, 0.3], and extract an unprocessed original airfoil eigenvector (including the pressure coefficients of the leading edge, maximum thickness, and trailing edge). Subsequently, a 3×3 Gaussian kernel is used as a feature selection filter to perform noise reduction processing on the aerodynamic characteristic knowledge components to obtain a purified original airfoil eigenvector.

[0127] Step 403: adjusting the pressure distribution parameters of the original airfoil eigenvector according to the pressure correction amount of the aerodynamic difference compensation instruction parameter to reconstruct the transmission buffer of the knowledge transfer channel to obtain a reconstructed transfer channel;

[0128] In this step, the pressure correction refers to the specific correction value (in kPa) in the aerodynamic difference compensation instruction, which is added to the original parameter through floating-point operations. The pressure distribution parameter refers to the set of pressure coefficient values at discrete points on the airfoil surface, which is the core variable that determines aerodynamic performance. The reconstructed migration path is the migration path that transmits the buffer pressure parameter update and passes the cyclic redundancy check.

[0129] In an embodiment of the present invention, a floating-point addition operation is performed on the pressure distribution parameters in the original airfoil eigenvector based on the pressure correction value ( ) in the aerodynamic difference compensation instruction parameter, such as the adjusted trailing edge pressure value = the original trailing edge pressure value + the correction value. The updated compensation parameter set is rewritten into the transmission buffer, and a reconstructed migration channel is generated using a 32-bit cyclic redundancy check mechanism. At the same time, bit 7 of the status register is set to 1 to activate the channel.

[0130] Step 404: Calculate the change gradient value based on the flight state parameters, and calculate the weight factor in combination with the load spectrum shift coefficient;

[0131] In this step, the change gradient value refers to the rate of change of the flight state parameter over time, calculated by differential calculation (current value - previous value) divided by the time difference. The weight factor is a mathematical mapping between the load spectrum offset coefficient and the change gradient value, which determines the control intensity.

[0132] In this embodiment of the present invention, a change gradient value is calculated based on current flight state parameters (e.g., altitude 10 kilometers, Mach number 0.8, angle of attack 5 degrees). Specifically, change gradient value = altitude gradient + Mach gradient, where altitude gradient = (current altitude - previous altitude) / time interval; Mach gradient = (current Mach number - previous Mach number) × quantization coefficient, where the quantization coefficient is a constant, such as 100. Subsequently, a weighting factor is calculated: the weighting factor = the load spectrum shift coefficient / change gradient value.

[0133] Step 405: compile an initial migration weight control function, and load the migration weight control function into the reconstructed migration channel to obtain a target migration channel;

[0134] In this step, the initial migration weight control function refers to a machine code function without setting a safety boundary, which implements the basic bandwidth allocation logic.

[0135] In this embodiment of the present invention, the LLVM compiler is used to generate the machine code for the initial migration weight control function, such as opcode 0xB2 plus parameter 0x2A. This machine code is then loaded into the instruction queue of the reconstructed migration channel coprocessor, ultimately generating the target migration channel and updating the channel status word to active.

[0136] Step 406: starting the steering effect response verifier of the target migration channel to obtain a verification result, and setting the operation boundary value of the initial migration weight control function according to the verification result and the weight factor to obtain a migration weight control function;

[0137] In this step, the rudder response verifier refers to the flight control system hardware module, which verifies that the control surface deflection efficiency meets the standard through a table lookup method. The verification result refers to the physical safety boundary of the rudder response verification output, such as the maximum rudder load of 3.2 kN·m, expressed in units of force / torque. The operational boundary value refers to the maximum output value allowed by the migration weight control function, such as the bandwidth limit of 2.14, calculated by multiplying the verification result by the weight factor.

[0138] In this embodiment of the present invention, the rudder response verifier of the target migration channel is activated. Verification parameters, such as a rudder deflection limit of ±25 degrees, are input and compared against the rudder response truth table to obtain a verification result, such as a safety margin of 3.2 kN·m. Based on this result, the operational boundary value of the initial migration weight control function is set: operational boundary value = verification result × weight factor. This boundary value is then burned into bits 12 to 15 of the function register, completing the safety margin configuration.

[0139] The embodiment of the present invention solves the problem of static feature space distortion through real-time adjustment of pressure parameters; the load spectrum-state gradient coupling relationship enables weight distribution to match flight state changes; and the rudder effectiveness verification converts the aircraft control limit into a mathematical boundary, eliminating the risk of loss of control during the migration process.

[0140] The present invention provides a specific embodiment, in which step 403, based on the pressure correction amount of the aerodynamic difference compensation instruction parameter, the pressure distribution parameter of the original airfoil eigenvector is adjusted to reconstruct the transmission buffer of the knowledge transfer channel to obtain a reconstructed transfer channel, specifically comprising the following steps:

[0141] Step 411: Locate the pressure distribution parameter storage area to which the original airfoil eigenvector belongs to obtain the pressure distribution parameters;

[0142] In this step, the pressure distribution parameter storage area refers to the physical address space of the original airfoil eigenvector in the memory (such as 0x5000-0x500B), and stores 32-bit floating-point pressure values in the order of leading edge, maximum thickness, and trailing edge.

[0143] In this embodiment of the present invention, the memory management unit locates the pressure distribution parameter storage area in the original airfoil eigenvector. The physical address range of this area is 0x5000 to 0x500B. The leading edge pressure value is stored at address 0x5000, the maximum thickness point pressure value is stored at address 0x5004, and the trailing edge pressure value is stored at address 0x5008. Each parameter occupies 4 bytes. Direct memory access is used to read the current pressure distribution parameters (for example, [0.75, 0.52, 0.28]) and load them into the CPU register for subsequent processing.

[0144] Step 412: Obtain a pressure correction value of the aerodynamic difference compensation instruction parameter, and superimpose the pressure correction value with the pressure distribution parameter to generate an updated pressure distribution parameter set;

[0145] In this step, the updated pressure distribution parameter set refers to a corrected pressure distribution parameter set, which is generated by floating-point addition of the pressure distribution parameter and the pressure correction value and is stored in a temporary buffer (0x7000).

[0146] In this embodiment of the present invention, the opcode in the aerodynamic difference compensation instruction parameter is parsed to extract the corresponding pressure correction, such as a trailing edge pressure increase of +0.1 kPa. A scalar addition operation is performed via the floating-point unit, using the calculation logic of updated pressure distribution parameter = pressure distribution parameter + pressure correction. This ultimately generates an updated pressure distribution parameter set and temporarily stores it in a buffer.

[0147] Step 413: adding the updated pressure distribution parameter set to the pressure distribution parameter storage area to update the field of the original airfoil eigenvector to obtain a reconstructed airfoil eigenvector;

[0148] In this step, the reconstructed airfoil feature vector refers to a complete data structure including updated pressure distribution parameters and a new version identifier.

[0149] In this embodiment of the present invention, the updated set of pressure distribution parameters is written to the original pressure distribution parameter storage area, overwriting the original data. Subsequently, a field update function is called to update the version identifier of the airfoil eigenvector from 0x01 to 0x02, generating a reconstructed airfoil eigenvector. The memory status word is also updated to indicate "updated." Simultaneously, a hardware interrupt signal is triggered to signal the completion of the parameter update.

[0150] Step 414: In response to a signal indicating that the updating of the original airfoil eigenvector is completed, locking the write permission of the transmission buffer of the knowledge transfer channel;

[0151] In this step, the write permission refers to the write enable flag controlled by the memory protection unit, 0 = writable, 1 = locked.

[0152] In an embodiment of the present invention, in response to an interrupt signal indicating that the update is completed, a write protection flag is set on the transmission buffer (bit 0 is set to 1) through a memory protection unit, restricting write operations to the area to only privileged mode access, thereby ensuring data consistency and security during the reconstruction process.

[0153] Step 415: In the locked state, the reconstructed airfoil eigenvector is written into the aerodynamic characteristic data segment of the transmission buffer, so as to activate the data consistency check mechanism to check the transmission buffer. Based on the check-passed signal, the serial number identifier of the transmission buffer is reset, and a reconstructed migration channel is generated.

[0154] In this step, the aerodynamic characteristics data segment refers to a 12-byte partition in the transmission buffer dedicated to storing airfoil pressure parameters. The data consistency check mechanism involves a hardware process in which dual cores lockstep to calculate a 32-bit cyclic redundancy checksum. Cores A and B independently calculate and compare the results, with an error tolerance of ≤ 1 clock cycle. The serial number identifier refers to the version counter in the transmission buffer, which increments with each reconstruction to prevent data version confusion.

[0155] In this embodiment of the present invention, with write protection enabled (i.e., locked), the reconstructed airfoil characteristics are copied to the aerodynamic characteristics data segment in the transmission buffer. At this point, a data consistency check mechanism is activated, employing a dual-core lockstep approach, with two processing cores synchronously calculating a 32-bit cyclic redundancy check sum. If the two cores' calculation results are consistent, a checksum pass signal is generated, and the write protection flag is cleared (bit 0 is restored to 0). Upon receiving the checksum pass signal, the sequence number in the transmission buffer is incremented to mark this reconstruction as a new version. Simultaneously, the reconstruction complete bit in the transmission status register is updated, ultimately generating a migration channel indicating that the reconstruction is complete.

[0156] The embodiment of the present invention eliminates the risk of data competition that may occur during the reconstruction process through the mechanism of write protection locking, writing and unlocking, ensuring the security and consistency of data operations; at the same time, it introduces a dual redundancy check mechanism and uses dual cores to execute cyclic redundancy check in parallel to provide doubly guaranteed integrity of pressure parameters, meeting the requirements for key data protection in airworthiness standards.

[0157] The present invention provides a specific embodiment, step 105, using the migration weight control function to adjust the transmission status register of the target migration channel to obtain the adjusted migration channel, specifically includes the following steps:

[0158] Step 501: Decoding the executable code segment of the migration weight control function to extract the transmission strength control parameter;

[0159] In this step, the executable code segment refers to the machine instruction sequence compiled from the migration weight control function, stored in memory addresses 0x9000-0x90FF, and directly executed by the CPU. The transmission intensity control parameter refers to the scaling coefficient (a floating-point number between 0.0 and 1.0) within the control function, which determines the basic bandwidth allocation weight and is extracted through operand decoding.

[0160] In this embodiment of the present invention, the CPU instruction decoder parses the executable code segment of the migration weight control function (machine code, such as 0xB2 0x18), extracts the opcode (e.g., 0xB2) and operand (e.g., 0x18), and separates the transmission strength control parameter (e.g., a bandwidth allocation coefficient of 0.75). This is achieved by loading the machine code into the instruction pipeline, separating the opcode and operand in the decoder, and storing the extracted transmission strength control parameter in register R1.

[0161] Step 502: Obtain the boundary constraint value of the rudder effect response verifier, and calculate the transmission intensity correction value in combination with the real-time change rate of the flight state parameter and the transmission intensity control parameter;

[0162] In this step, the boundary constraint refers to the physical safety threshold (in kN·m) output by the rudder effectiveness verifier, reflecting the ultimate load capacity of the aircraft's control surfaces. The transmission strength correction refers to the final bandwidth adjustment coefficient, used to dynamically adapt to changes in flight conditions.

[0163] In this embodiment of the present invention, the boundary constraint value output by the rudder response verifier, for example, a maximum rudder load of 3.2 kN·m, is read. This is combined with the real-time rate of change of the flight state parameter, for example, a climb rate of 2.5 meters per second, and the acquired transmission strength control parameter, such as 0.75. A transmission strength correction is calculated using a floating-point arithmetic unit: Transmission strength correction = Transmission strength control parameter × boundary constraint value ÷ real-time rate of change. For example, 0.75 × 3.2 ÷ 2.5 = 0.96. The calculated transmission strength correction (e.g., 0.96) is then output and stored in a temporary storage area.

[0164] Step 503: inputting the transmission strength correction value into the transmission scheduler of the target migration channel to trigger a bandwidth reallocation operation of the transmission scheduler to obtain an updated bandwidth configuration;

[0165] In this step, the transmission scheduler refers to the hardware control module of the avionics network switch, which implements virtual link bandwidth allocation. The updated bandwidth configuration refers to the new bandwidth value (in Kbps) that takes effect after the reallocation operation and is written to the switch configuration register.

[0166] In an embodiment of the present invention, an aviation full-duplex switched Ethernet switch chip is used to input a transmission strength correction value into the transmission scheduler of the target migration channel. A bandwidth reallocation operation is triggered by sending a write command to the switch configuration register address. The updated bandwidth = the original bandwidth × the transmission strength correction value. The updated bandwidth configuration is then updated to the corresponding field in the virtual link configuration table, such as the bandwidth configuration item of VL101, to generate an updated bandwidth configuration.

[0167] Step 504: adjusting the transmission status register of the target migration channel according to the updated bandwidth configuration to obtain an adjusted migration channel;

[0168] In this step, the transfer status register refers to a hardware register that stores the real-time configuration of the channel.

[0169] In this embodiment of the present invention, based on the updated bandwidth configuration, the bandwidth control bits (bits 0 to 7) of the transmission status register of the target migration channel are written to a new value (e.g., the converted hexadecimal value 0x3D), and the parity bit (e.g., bit 8) is updated. After the write is complete, the register latch signal is activated, generating the adjusted migration channel.

[0170] The embodiment of the present invention dynamically corrects the transmission strength through the real-time change rate of flight status parameters, solving the response lag problem of the traditional fixed weight mechanism under variable speed conditions such as climbing or diving; the boundary constraint value converts the physical limit of the aircraft into a mathematical boundary, eliminating the risk of system overload caused by bandwidth over-allocation; atomic write of registers ensures the consistency of configuration switching and meets the millisecond-level real-time requirements of the avionics system.

[0171] Figure 2 The present invention provides a schematic diagram of the structure of a multi-aircraft migration optimization system for a large aviation model. Figure 2 As shown, the system includes:

[0172] A construction module 21 is used to construct a flight dynamics feature library containing aerodynamic difference features based on the acquired flight dynamics parameters and control response data of the source aircraft model and the target aircraft model;

[0173] A simulation module 22 is configured to perform a coupled simulation operation based on the flight mechanics feature library to generate a compensation parameter set;

[0174] Establishing module 23, for establishing a knowledge transfer channel from the pre-trained source model large model to the target model large model;

[0175] A reconstruction module 24 is configured to reconstruct the transmission buffer of the knowledge transfer channel according to the compensation parameter set to obtain a target transfer channel and generate a transfer weight control function;

[0176] An adjustment module 25 is configured to adjust the transmission status register of the target migration channel using the migration weight control function to obtain an adjusted migration channel;

[0177] The updating module 26 is used to monitor the distribution offset of the aerodynamic characteristic knowledge component in the adjusted migration channel during the migration process, and update the aerodynamic difference characteristics of the flight mechanics feature library according to the distribution offset, so as to iteratively perform the coupling simulation operation to the operation of adjusting the transmission state register until the distribution offset enters the steady-state operation range.

[0178] Figure 2 The multi-model migration optimization system of a large aviation model can perform Figure 1 The implementation principles and technical effects of the multi-model migration optimization method for a large aviation model described in the illustrated embodiment will not be elaborated on here. The specific manner in which each module and unit performs operations in the multi-model migration optimization system for a large aviation model described in the above embodiment has been described in detail in the embodiments of the method and will not be elaborated on here.

[0179] In one possible design, Figure 2 The multi-aircraft migration optimization system of a large aviation model of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0180] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0181] The processing component 32 is used for the above Figure 1 The embodiment provides a multi-aircraft model migration optimization method for a large aviation model.

[0182] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0183] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0184] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0185] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0186] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0187] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0188] The embodiment of the present invention further provides a computer storage medium storing a computer program, which can achieve the above-mentioned Figure 1 The embodiment shown is a multi-aircraft migration optimization method for a large aviation model.

[0189] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0190] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0191] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0192] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A multi-aircraft model migration optimization method for a large aviation model, characterized in that: include: Based on the acquired flight dynamics parameters and control response data of the source and target aircraft, a flight dynamics feature library containing aerodynamic difference characteristics is constructed; performing coupled simulation operations according to the flight mechanics feature library to generate a compensation parameter set; Establish a knowledge transfer channel from the pre-trained source model to the target model; Reconstructing the transmission buffer of the knowledge transfer channel according to the compensation parameter set to obtain a target transfer channel and generate a transfer weight control function; Using the migration weight control function, adjusting the transmission status register of the target migration channel to obtain an adjusted migration channel; monitoring a distribution offset of the aerodynamic characteristic knowledge component in the adjusted migration channel during the migration process, updating the aerodynamic difference characteristics of the flight dynamics feature library according to the distribution offset, and iteratively performing a coupled simulation operation to an operation of adjusting the transmission state register until the distribution offset enters a steady-state operation range; According to the compensation parameter set, the transmission buffer of the knowledge transfer channel is reconstructed to obtain the target transfer channel, and a transfer weight control function is generated, including: generating aerodynamic difference compensation instruction parameters and load spectrum shift coefficients according to the compensation parameter set; Obtaining an original airfoil eigenvector according to the aerodynamic characteristic knowledge component transmitted in the knowledge transfer channel; Adjusting the pressure distribution parameters of the original airfoil eigenvector according to the pressure correction amount of the aerodynamic difference compensation instruction parameter, reconstructing the transmission buffer of the knowledge transfer channel, and obtaining a reconstructed transfer channel; Calculating a change gradient value based on the flight state parameters and calculating a weight factor based on the load spectrum shift coefficient; Compiling an initial migration weight control function, and loading the migration weight control function into the reconstructed migration channel to obtain a target migration channel; The rudder effect response verifier of the target migration channel is started to obtain a verification result. According to the verification result and the weight factor, the operation boundary value of the initial migration weight control function is set to obtain the migration weight control function.

2. The method according to claim 1, characterized in that According to the flight mechanics feature library, a coupled simulation operation is performed to generate a compensation parameter set, including: calculating the dynamic fluid pressure of the source aircraft and the target aircraft under the same flight state parameters based on the aerodynamic difference characteristics and flight state parameters of the flight mechanics feature library, and generating solid deformation boundary constraints based on the dynamic fluid pressure; The interaction process between the source aircraft and the target aircraft is synchronously stimulated under the same flight state parameters to obtain the fluid pressure distribution responses corresponding to the source aircraft and the target aircraft respectively; The fluid pressure distribution response is transmitted to the effective area of the solid deformation boundary constraint, the response difference value between the source aircraft model and the target aircraft model under the same fluid pressure distribution response is calculated, and a compensation parameter set is generated based on the correspondence between the response difference value and the flight state parameter.

3. The method according to claim 2, characterized in that The fluid pressure distribution response is transmitted to the effective area of the solid deformation boundary constraint, the response difference between the source aircraft and the target aircraft under the same fluid pressure distribution response is calculated, and a compensation parameter set is generated based on the corresponding relationship between the response difference and the flight state parameter, including: The fluid pressure distribution response is transmitted to the action area of the solid deformation boundary constraint, the solid deformation response quantities corresponding to the source model and the target model under the same fluid pressure distribution response are obtained, and the difference value of the response quantities is calculated; According to the altitude layer identification, speed range code and flight attitude angle of the flight state parameters, a three-dimensional flight state coordinate is formed; Associating the response difference value with a storage location corresponding to the three-dimensional flight state coordinate, and connecting the response difference values of adjacent storage locations to construct multiple difference value change trajectories; All difference value change trajectories are converted into corresponding executable compensation instruction parameters, and all executable compensation instruction parameters are combined to generate a compensation parameter set.

4. The method according to claim 1, wherein Establish a knowledge transfer channel from the pre-trained source model to the target model, including: Based on the hardware resource distribution of the distributed computing node cluster, the pre-trained source aircraft model is allocated computing resources and loaded with the model to obtain an airworthiness certification model instance. Initialize the target aircraft model based on its structure definition to obtain the model instance to be migrated. According to the airworthiness certification model instance and the model instance to be migrated, a bidirectional data route is configured between the source aircraft model large model and the target aircraft model large model to form a basic migration channel; According to the aerodynamic characteristics knowledge migration requirements, a feature extraction operator is implanted into the basic migration channel to generate a feature enhancement migration channel; Synchronously loading the encoding rules of the flight state parameters into the feature enhancement migration channel to obtain a synchronized migration channel; According to aviation data security specifications, the synchronized migration channel is securely encapsulated to obtain a knowledge migration channel.

5. The method according to claim 1, wherein According to the pressure correction amount of the aerodynamic difference compensation instruction parameter, the pressure distribution parameter of the original airfoil eigenvector is adjusted, and the transmission buffer of the knowledge transfer channel is reconstructed to obtain a reconstructed transfer channel, including: Locating a pressure distribution parameter storage area to which the original airfoil eigenvector belongs, and obtaining the pressure distribution parameters; Obtaining a pressure correction value of the aerodynamic difference compensation instruction parameter, and superimposing the pressure correction value with the pressure distribution parameter to generate an updated pressure distribution parameter set; Adding the updated pressure distribution parameter set to the pressure distribution parameter storage area, updating the field of the original airfoil eigenvector, and obtaining a reconstructed airfoil eigenvector; In response to a signal indicating that the updating of the original airfoil eigenvector is completed, locking a write permission of a transmission buffer of the knowledge transfer channel; In the locked state, the reconstructed airfoil feature vector is written into the aerodynamic characteristic data segment of the transmission buffer, the data consistency verification mechanism is activated to verify the transmission buffer, and the serial number identifier of the transmission buffer is reset according to the signal that passes the verification, and the reconstructed migration channel is generated.

6. The method according to claim 1, characterized in that Using the migration weight control function, adjusting the transmission status register of the target migration channel to obtain an adjusted migration channel includes: Decoding the executable code segment of the migration weight control function to extract the transmission strength control parameter; Obtaining a boundary constraint value of a rudder effect response verifier, and calculating a transmission intensity correction value based on the real-time change rate of the flight state parameter and the transmission intensity control parameter; Inputting the transmission strength correction amount into the transmission scheduler of the target migration channel to trigger a bandwidth reallocation operation of the transmission scheduler to obtain an updated bandwidth configuration; According to the updated bandwidth configuration, the transmission status register of the target migration channel is adjusted to obtain an adjusted migration channel.

7. A multi-aircraft migration optimization system for large aviation models, characterized by: include: A construction module is used to construct a flight dynamics feature library containing aerodynamic difference features based on the acquired flight dynamics parameters and control response data of the source aircraft and the target aircraft; A simulation module, configured to perform coupled simulation operations based on the flight mechanics feature library to generate a set of compensation parameters; Establish a module for establishing a knowledge transfer channel from a pre-trained source model to a target model; A reconstruction module, configured to reconstruct the transmission buffer of the knowledge migration channel according to the compensation parameter set, obtain a target migration channel, and generate a migration weight control function; an adjustment module, configured to adjust the transmission status register of the target migration channel by using the migration weight control function to obtain an adjusted migration channel; an updating module, configured to monitor a distribution offset of the aerodynamic characteristic knowledge component in the adjusted migration channel during the migration process, update the aerodynamic difference characteristics of the flight dynamics feature library according to the distribution offset, and iteratively perform a coupled simulation operation to an operation of adjusting the transmission state register until the distribution offset enters a steady-state operation range; According to the compensation parameter set, the transmission buffer of the knowledge transfer channel is reconstructed to obtain the target transfer channel, and a transfer weight control function is generated, including: generating aerodynamic difference compensation instruction parameters and load spectrum shift coefficients according to the compensation parameter set; Obtaining an original airfoil eigenvector according to the aerodynamic characteristic knowledge component transmitted in the knowledge transfer channel; Adjusting the pressure distribution parameters of the original airfoil eigenvector according to the pressure correction amount of the aerodynamic difference compensation instruction parameter, reconstructing the transmission buffer of the knowledge transfer channel, and obtaining a reconstructed transfer channel; Calculating a change gradient value based on the flight state parameters and calculating a weight factor based on the load spectrum shift coefficient; Compiling an initial migration weight control function, and loading the migration weight control function into the reconstructed migration channel to obtain a target migration channel; The rudder effect response verifier of the target migration channel is started to obtain a verification result. According to the verification result and the weight factor, the operation boundary value of the initial migration weight control function is set to obtain the migration weight control function.

8. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a multi-aircraft migration optimization method for a large aviation model as described in any one of claims 1 to 6.

9. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, the multi-aircraft model migration optimization method of the aviation large model according to any one of claims 1 to 6 is implemented.

Images