Aircraft aerodynamic performance analysis method

Abstract

The invention discloses an aircraft aerodynamic performance analysis method, which comprises the following steps: constructing a multi-dimensional parameterized model integrating aerodynamic / structure / thermal control, and outputting a global design space; based on a global design space, fusing sparse fluid dynamics and engineering data to train an agent model, and verifying the cross-speed domain reliability; decoupling the centroid / lift-drag ratio / thermal load conflict by using the proxy model, and generating a speed domain adaptive configuration library; executing fluid-structure-thermal field strong coupling simulation by utilizing a shock wave-boundary layer self-adaptive grid technology aiming at the speed domain self-adaptive configuration library; based on a fluid-structure-thermal field strong coupling mechanism, utilizing a dynamic sensing technology to predict the aerodynamic performance of the aircraft; and integrating dynamic prediction data and digital simulation, calibrating model parameters, and outputting an aerodynamic performance report and an optimization scheme. According to the method, the problems of insufficient multi-physical field integration, low cross-speed domain reliability, poor configuration self-adaption and low coupling simulation precision in aerodynamic performance analysis of a traditional aircraft are solved.

Description

Technical Field

[0001] This invention relates to the field of aerospace technology, and in particular to a method for analyzing the aerodynamic performance of aircraft. Background Technology

[0002] Currently, traditional aerodynamic performance analysis methods have significant limitations in complex flight environments and multi-physics coupling scenarios: existing methods mostly use parameterized models of a single physics field, which makes it difficult to achieve multi-dimensional integration of aerodynamics, structure and thermal control, resulting in incomplete design space coverage and failure to reflect the comprehensive impact of strong fluid-structure-thermal field coupling on aerodynamic performance, thus limiting the exploration of full-domain design solutions.

[0003] Traditional surrogate models rely on a single data source for training. Under conditions spanning subsonic, supersonic, and hypersonic speeds, their generalization ability is insufficient, reliability verification lacks systematicity, and they struggle to support accurate prediction of aerodynamic performance across the entire speed range. Furthermore, existing configuration libraries are mostly designed for fixed speed ranges, lacking adaptive adjustment mechanisms and failing to dynamically match optimal configurations based on real-time flow field changes. Simultaneously, in fluid-structure-thermal coupling simulations, traditional mesh techniques have limited accuracy in capturing complex flow field structures such as shock waves and boundary layers, leading to significant deviations in the quantification results of strongly coupled dynamic responses. Traditional methods suffer from a disconnect between sensor deployment and the learning framework, making it difficult to construct a real-time perception closed loop between configuration and environment. Moreover, model parameter calibration largely relies on static data, failing to integrate dynamic measured information for dynamic correction. This results in optimization schemes exhibiting substandard performance in multi-condition verification, making it difficult to meet the requirements of next-generation aircraft for high-reliability aerodynamic performance across the entire speed range. This solution addresses the problems of insufficient multiphysics integration, low cross-speed-range reliability, poor configuration adaptability, and low accuracy in coupled simulations inherent in traditional aircraft aerodynamic performance analysis. Summary of the Invention

[0004] This invention provides a method for analyzing the aerodynamic performance of an aircraft, comprising: Construct a multi-dimensional parametric model integrating aerodynamics, structure, and thermal control, and output a global design space; Based on the global design space, a proxy model is trained by integrating sparse fluid dynamics and engineering data to verify the reliability across velocity domains. By using a proxy model to decouple the conflicts between the center of mass, lift-to-drag ratio, and thermal load, a velocity-domain adaptive configuration library is generated. For the velocity domain adaptive configuration library, shock wave-boundary layer adaptive mesh technology is used to perform strong coupling simulation of fluid-structure-thermal field; Based on the strong coupling mechanism of fluid-structure-thermal field, dynamic sensing technology is used to predict the aerodynamic performance of aircraft. It integrates dynamic prediction data with digital simulation, calibrates model parameters, and outputs aerodynamic performance reports and optimization schemes.

[0005] The above-described method for analyzing the aerodynamic performance of aircraft involves constructing a multi-dimensional parametric model integrating aerodynamics, structure, and thermal control, and outputting a global design space, including: Establish a correlation model of aerodynamic / structural / thermal control parameters, and perform multi-dimensional parametric modeling with multi-physics coupling; A global design space is generated based on an integrated model, and the breadth and accuracy of coverage are verified through multiple constraints.

[0006] The above-described method for analyzing the aerodynamic performance of an aircraft, which integrates sparse fluid dynamics and engineering data to train a surrogate model based on a global design space, verifies cross-velocity reliability, includes: By integrating sparse fluid dynamics data with engineering data, a hybrid training set is constructed to train the proxy model. Cross-speed domain operating condition verification was carried out, and the reliability of the surrogate model was verified through dynamic error correction.

[0007] The above-described aerodynamic performance analysis method for aircraft, wherein a surrogate model is used to decouple the center of mass / lift-to-drag ratio / thermal load conflict, generating a velocity-domain adaptive configuration library, includes: Based on the surrogate model, the coupling mechanism of centroid / lift-drag ratio / thermal load is analyzed, and a multi-objective decoupling algorithm is established. Based on the decoupling results, a velocity-domain segmented configuration is generated, and an adaptive configuration library is constructed.

[0008] The above-described aerodynamic performance analysis method for aircraft, wherein, for a velocity-domain adaptive configuration library, shock-boundary layer adaptive mesh technology is used to perform strongly coupled fluid-structure-thermal field simulations, including: Construct an adaptive mesh for dynamic capture of shock waves and boundary layers to improve the resolution of complex flow fields; The simulation of strong coupling between fluid, structure, and thermal field is performed based on adaptive mesh, and the dynamic response of strong coupling between fluid, structure, and thermal field is quantified.

[0009] The above-described method for analyzing the aerodynamic performance of an aircraft, which uses dynamic sensing technology to predict the aerodynamic performance of the aircraft based on a strong fluid-structure-thermal field coupling mechanism, includes: Collect real-time flow field sensor data, correct prediction parameters, and adjust model boundary conditions; By combining the strong coupling mechanism of fluid-structure-thermal field, a trend map of aerodynamic performance across the entire velocity domain is generated.

[0010] A computer-readable storage medium comprising one or more program instructions, the one or more program instructions being executed by a processor as described in any one of claims 1-6, a method for analyzing the aerodynamic performance of an aircraft.

[0011] The beneficial effects achieved by this invention are as follows: This application expands the design space through multi-dimensional integrated modeling, enhances the reliability of the proxy model across velocity domains by integrating multi-source data, improves the accuracy of coupled simulation by using adaptive configuration library and dynamic mesh technology, and enhances the real-time performance and adaptability of the scheme through real-time perception optimization and dynamic calibration, thus efficiently supporting the optimization of aircraft aerodynamic performance. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0013] Figure 1 This is a flowchart of an aerodynamic performance analysis method for an aircraft provided in Embodiment 1 of this application. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Example 1 like Figure 1 As shown, Embodiment 1 of this application provides a method for analyzing the aerodynamic performance of an aircraft, including: S110: Construct a multi-dimensional parametric model integrating aerodynamics, structure, and thermal control, and output the global design space; Aerodynamic, structural, and thermal control parameters are collected, and an integrated model is constructed using a multi-field coupling algorithm to output a global design space. A regression algorithm is then used to establish a parameter correlation model among the three parameters, clarifying the relationships between them. Finally, the design space is validated based on constraints.

[0016] Among them, the global design space is a comprehensive analysis framework that integrates all key design variables, operating parameters and performance constraints of a multi-field coupled system of flow field, structure and thermal field.

[0017] Constructing a multi-dimensional parametric model integrating aerodynamics, structure, and thermal control, and outputting a global design space, includes the following sub-steps: S111: Establish a correlation model of aerodynamic / structural / thermal control parameters and perform multi-dimensional parametric modeling with multi-physics coupling; A multi-field parameter mapping module is constructed to standardize aerodynamic, structural, and thermal control parameters, forming a feature vector with a unified dimension. A physically inspired neural network architecture is used to establish the correlation model: the input layer receives three types of parameter vectors, and the hidden layer consists of five fully connected layers (each containing 100 neurons), using the ReLU activation function to capture nonlinear relationships. A physical constraint layer is introduced, embedding the Navier-Stokes equations, heat conduction equations, and structural mechanics equations.

[0018] A multi-objective optimization strategy was adopted during the training phase: minimizing the mean squared error between the predicted values ​​and the CFD / FEA simulation values ​​was the primary objective. The Adam optimizer was used with an initial learning rate of 0.001, dynamically adjusted using a cosine annealing strategy. The training dataset was divided into training, validation, and test sets in a 3:1:1 ratio, and training continued until the loss function converged.

[0019] An attention mechanism layer is added before the output layer to automatically assign the weights of each parameter's impact on aerodynamic performance. The final output parameter correlation matrix clarifies the influence coefficient of each Mach number change on heat flux density, as well as the coupling relationship between structural deformation and lift coefficient.

[0020] S112: Generates a global design space based on an integrated model, and verifies the breadth and accuracy of coverage through multiple constraints.

[0021] Based on the parameter correlation model, the Latin hypercube sampling algorithm is used to generate initial sample points within the design space, covering the range of key parameters. Multiphysics coupled simulations are performed on each sample point: aerodynamic analysis uses the RANS equation solver, structural analysis uses finite element software, and thermal control analysis combines the thermal network method with Monte Carlo radiation calculations.

[0022] A multi-constraint verification module was constructed, setting three types of constraints: aerodynamic performance constraints, structural strength constraints, and thermal control constraints. A Kriging surrogate model was used for initial screening of sample points, eliminating points that did not meet the constraints and retaining the valid sample groups. Space Filling Index (SFI) and Minimum Distance Index (MDI) were introduced: SFI calculates the fill rate of the design space by the sample points; MDI determines that the Euclidean distance between any two sample points is greater than or equal to 5% of the diagonal length of the design space, and an adaptive sampling strategy is adopted for sparse regions.

[0023] The final output is a full-domain design space dataset, including three types of indicators: aerodynamic performance, structural response, and thermal control, forming a design matrix. Cross-validation confirms that the breadth and accuracy requirements under multiple constraints are met.

[0024] S120: Based on the global design space, a proxy model is trained by integrating sparse fluid dynamics and engineering data to verify cross-velocity domain reliability; Based on the global design space, a hybrid dataset is first constructed: high-precision CFD sparse data and engineering experience data are divided into training and validation sets in a 7:3 ratio. A deep neural network architecture is adopted: reliable predictions are performed across the velocity domain through validation under three sets of operating conditions: subsonic, supersonic, and hypersonic.

[0025] Among them, the proxy model is a simplified model built based on a small amount of high-fidelity sample data in the global design space. It can quickly output results that are similar to high-fidelity calculations and reduce the computational burden.

[0026] Based on the global design space, a surrogate model is trained by integrating sparse fluid dynamics and engineering data, and the reliability across velocity domains is verified, including the following sub-steps: S121: Integrate sparse fluid dynamics data with engineering data to construct a hybrid training set for training proxy models; Based on the global design space, a hybrid training set is constructed to train the surrogate model. Data preprocessing is performed: multiphysics simulation data are acquired and divided into CFD sparse datasets and engineering experience datasets according to their sources. 12-dimensional feature vectors such as Mach number, angle of attack, and wall heat flux are extracted from the CFD data, and the engineering data are subjected to dimension unification and normalization to form a standard feature space.

[0027] A hybrid training strategy was constructed: a two-layer attention mechanism was designed to fuse the network. The bottom layer network contains two parallel branches: the CFD branch uses a 3-layer CNN to extract flow field features, and the engineering branch uses LSTM to capture the temporal correlation of parameters. The upper layer network dynamically allocates the contribution of the two types of data through an attention weight matrix.

[0028] The training process employs a multi-stage optimization: the first stage freezes the CFD branch, training only the engineering branch and attention layer, using the mean squared error loss function; the second stage unfreezes all parameters, introduces physical constraint loss (such as constraints based on the second law of thermodynamics), and uses cosine annealing for learning rate scheduling. The final output is a hybrid surrogate model, with aerodynamic / structural / thermal control parameter vectors as input and six-dimensional performance metrics including lift-to-drag ratio and peak temperature.

[0029] S122: Conduct cross-speed domain operating condition verification and verify the reliability of the surrogate model through dynamic error correction.

[0030] Perform cross-velocity reliability verification and dynamic error correction. Construct a verification test case set: Select four typical test cases: subsonic, transonic, supersonic, and hypersonic. For each test case, set 10 different combinations of angle of attack and structural parameters to form a verification sample. For each sample, run high-precision CFD simulation and surrogate model prediction to collect key indicators such as drag ratio and heat flux density.

[0031] The first stage constructs an error prediction network, taking the surrogate model's predicted values ​​and operating parameters as inputs, and predicts the error correction amount through a two-layer fully connected network. The second stage employs an ensemble learning method, fusing the prediction results from five base models, and determines the final correction value through a soft voting mechanism.

[0032] The reliability assessment employs a three-indicator system: prediction error index, which calculates the root mean square error and mean absolute percentage error before and after correction; confidence interval index, which estimates prediction uncertainty using the Monte Carlo dropout method, requiring a 95% confidence interval coverage greater than or equal to 90%; and physical consistency index, which checks whether the corrected results satisfy physical constraints such as the second law of thermodynamics. The final output includes a cross-velocity reliability assessment report and a dynamic error correction surrogate model.

[0033] S130: Decouple the conflict between centroid / lift-to-drag ratio / thermal load using a proxy model to generate a velocity-domain adaptive configuration library; Based on the validated surrogate model, a multi-objective optimization framework is constructed: inputting parameters such as Mach number and angle of attack, the optimal solution set is obtained using the Pareto front algorithm. The NSGA-III algorithm is employed, incorporating physical constraints. The solution set is segmented according to the velocity domain to generate an adaptive configuration parameter library.

[0034] The adaptive configuration parameter library, based on the analysis results of surrogate models and decoupling algorithms, selects and stores the best-performing aircraft configuration parameters in each velocity domain, providing targeted configuration inputs for strong coupling analysis of fluid-structure-thermal fields. Using a surrogate model to decouple the conflicts between the center of mass / lift-to-drag ratio / thermal load, a velocity-domain adaptive configuration library is generated, including the following sub-steps: S131: Based on the surrogate model, analyze the coupling mechanism of centroid / lift drag ratio / thermal load and establish a multi-objective decoupling algorithm; A multi-objective decoupling framework is constructed to analyze the coupling mechanism of centroid / lift-drag ratio / thermal load. A parameter sensitivity analysis module is established to perform variance decomposition on samples from the entire design space and identify key coupling parameters.

[0035] A multi-objective optimization algorithm was constructed, with the objectives of maximizing the lift-to-drag ratio, minimizing thermal load, and ensuring centroid stability. Design variables included geometric parameters and material distribution parameters. The NSGA-Ⅲ algorithm was used for solving the problem, and constraints were set to ensure that the structural strength and centroid offset remained within safe limits.

[0036] The design variables are grouped and evolved independently before information exchange occurs. Based on the multi-objective decoupling objective compensation function, when the increase in drag ratio causes the thermal load to exceed the threshold, the corresponding parameter adjustments are automatically triggered.

[0037] The final output is the Pareto optimal solution set. The solution set is divided into three velocity domains: subsonic, supersonic, and hypersonic by clustering. Representative configurations are extracted from each domain to form the basis of the adaptive configuration library.

[0038] S132: Generate velocity domain segmented configurations based on the decoupling results, and construct an adaptively adjustable configuration library.

[0039] Based on the Pareto optimal solution set, velocity domain segmented configuration generation and adaptive configuration library construction are performed. A velocity domain partitioning module is constructed, which divides the solution set into three regions—subsonic, supersonic, and hypersonic—according to the Mach number threshold. Feature extraction is performed on the non-dominated solutions in each region: the correlation matrix between geometric parameters and performance indicators is calculated, and the dominant parameters of each velocity domain are identified.

[0040] Bézier curves were used to describe the leading and trailing edges of the wing, and the configuration was parameterized using control point coordinate vectors. The fuselage was represented by NURBS surfaces, and the surface curvature was controlled by weight factor vectors. For each velocity domain, cluster analysis was applied to group similar configurations, and a representative configuration from each group was selected as the baseline.

[0041] Constructing a parameter mapping model: Inputting current Mach number, angle of attack, and other flight state parameters, the optimal configuration parameter vector is predicted using Gaussian process regression. Designing a configuration deformation strategy: Based on the base configuration, local deformation is performed through morphological functions. For example, when the Mach number transitions from subsonic to supersonic, the morphological function automatically increases the sweep angle and decreases the aspect ratio. Finally, an adaptive configuration library containing multiple sets of base configurations and their deformation rules is constructed, with each base configuration having an adjustable parameter range and performance boundaries. The configuration parameters are updated through a dynamic adjustment algorithm.

[0042] S140: For the velocity domain adaptive configuration library, use shock wave-boundary layer adaptive mesh technology to perform strong coupling simulation of fluid-structure-thermal field; An adaptive mesh for shock waves and boundary layer is constructed based on an adaptive configuration library. A feature recognition algorithm is used to capture the shock wave location and boundary layer thickness in the flow field, dynamically adjusting the mesh density. An arbitrary Lagrange-Euler description is used to establish the fluid-structure-thermal coupling equations, mapping the adaptive mesh to the structural and thermal control models, and performing multi-field strongly coupled simulations.

[0043] For the velocity-domain adaptive configuration library, a strongly coupled fluid-structure-thermal field simulation is performed using shock-boundary layer adaptive meshing technology, including the following sub-steps: S141: Construct an adaptive mesh for dynamic capture of shock waves and boundary layers to improve the resolution of complex flow fields; An adaptive mesh system for dynamic capture of shock waves and boundary layers is constructed. A flow field feature recognition module is established, which identifies the shock wave location using a pressure gradient detection algorithm and determines the boundary layer thickness using a wall distance function combined with a turbulence model. The local mesh density is dynamically adjusted based on the flow field parameters.

[0044] A triangular prism mesh is deployed near the shock wave surface, and the mesh moves with the shock wave using a spring smoothing method. An orthogonal-preserving stretched mesh is used in the boundary layer region, and a sparse tetrahedral mesh is used in the far field to reduce computational cost. According to the mesh adaptive algorithm, the flow field characteristics are re-evaluated after each physical time step, and the mesh is dynamically adjusted through node insertion / deletion operations.

[0045] The system calculates indicators such as the condition number of the Jacobian matrix and the element orthogonality factor in real time, and triggers local mesh reconstruction when the threshold is exceeded.

[0046] S142: Performs strong coupling simulation of fluid-structure-thermal field based on adaptive mesh, and quantifies the dynamic response of strong coupling of fluid-structure-thermal field.

[0047] Based on an adaptive mesh, a multi-field strongly coupled solver is constructed to perform cross-domain dynamic response analysis. By solving the modified fluid-structure-thermal field strongly coupled governing equations, a fully coupled analysis of the flow field, structure, and thermal control is conducted. The flow field equations are specifically expressed as follows:

[0048] in, It represents fluid density and describes the characteristics of fluid mass distribution. The mass derivative represents the rate of change of fluid velocity over time following the motion of a fluid particle; u represents the fluid velocity vector. It represents the inertial force per unit volume of fluid, reflecting the acceleration / deceleration characteristics of the flow field; This represents the force generated by the fluid pressure gradient, and is one of the core driving forces of flow field motion; It represents the dynamic viscosity of a fluid, describing the magnitude of the fluid's viscosity; Represents the velocity gradient tensor, describing the spatial variation of velocity; This represents the transpose of the velocity gradient tensor, where T denotes the transpose, reflecting the tensor's symmetry. The Kronecker symbol is used to correct for the volume expansion effect; It represents velocity divergence, describing the expansion / compression characteristics of a fluid; This represents the divergence operator, which, when applied to a tensor, yields a vector. It represents the shear force generated by fluid viscosity, reflecting the resistance to motion caused by internal fluid friction; This represents the influence coefficient of thermally induced structural deformation, quantifying the nonlinear effect of structural deformation on the flow field; It represents the structural displacement vector and describes the deformation state of the structure; It represents the temperature gradient and describes the rate of spatial change of temperature. This represents the gravity acting on a unit volume of fluid and is retained only when gravity has a significant effect on the flow field. This represents the shock wave-vibration frequency coupling term; The shock wave oscillation frequency is represented, describing its dynamic characteristics; t represents the current time. By introducing the influence coefficient of thermally induced structural deformation ,in Let d be a material constant, and d be the structural displacement field. This quantifies the nonlinear correction of structural deformation on the aerodynamic performance of the flow field; it also incorporates a shock wave-vibration frequency coupling term. ,in, The coupling strength coefficient is... Shock wave oscillation frequency, The frequency correlation between shock wave oscillation and structural vibration is captured by using the phase difference.

[0049] Through temperature-deformation coupling stiffness matrix ,in, For initial stiffness, To quantify the thermally induced stiffness attenuation effect of the coupling coefficient, the thermal gradient-induced additional force coefficient is combined to directly correlate the thermal load with the structural dynamic response. The fluid-structure-thermal field strongly coupled control equations also include structural dynamics equations and thermal control transfer equations, collectively forming a fully coupled system. The structural dynamics equations are:

[0050] in, Indicates the density of the structural material; The second-order time derivative of the structural displacement; The inertial force represents the force that resists the accelerated motion of the structure; C represents the damping matrix of the structure, which describes the energy dissipation characteristics of the structure. The first time derivative of the structural displacement; It represents the damping force during structural vibration, which consumes vibration energy and causes the amplitude to decrease; This represents the temperature-deformation coupled stiffness matrix, which dynamically changes with structural deformation d and temperature T; T represents temperature. Indicates the initial temperature; denoted as the fluid-structure interaction interface; P represents the flow field pressure; n represents the interface normal vector, pointing outwards from the structure; dS represents the interface area element. This represents the aerodynamic load exerted on the structure by the flow field through pressure. Represents a structural field; This represents the coefficient of thermal gradient-induced additional force, which is related to the temperature gradient; Indicates the structural displacement gradient; Represents a volume element; This indicates other additional force vectors.

[0051] Introducing dynamic thermal conductivity Where V is the temperature coefficient. Assuming the deformation rate coefficient, a flow-structure thermal interaction is established, using a thermal protection efficiency feedback term. ,in, As a benchmark efficiency, The deformation attenuation coefficient is used to quantify the impact of structural deformation on thermal protection capabilities. The thermal control transfer equation is as follows:

[0052] in, This indicates the specific heat capacity at constant pressure. The first time derivative of temperature; It represents the heat storage rate per unit volume, reflecting the non-steady-state change of temperature over time; Represents the velocity vector u and the temperature gradient The dot product describes the migration rate of heat carried by the fluid; This indicates the heat transfer rate caused by convection; This represents the dynamic thermal conductivity as a function of temperature T and structural deformation rate. change; The modulus representing the rate of structural deformation; This represents the rate of change of heat caused by heat conduction; Indicates the aerodynamic heating rate, which describes the heat input generated by aerodynamic compression / friction in the flow field; This represents the efficiency feedback term of the thermal protection system, quantifying the impact of structural deformation on thermal protection capability. This indicates the heat dissipation rate of the thermal protection system; This indicates the actual heat dissipation of the thermal protection system; This indicates the internal heat generation rate of the structure.

[0053] A loosely coupled iterative strategy is adopted, and the fluid-structure interface coupling condition is used in each time step. The flow field and structural displacement are continuous; The closed-loop solution (with continuous heat flux) is used to predict the configuration evolution process during flight across the velocity domain. For example, when the Mach number increases from 3.0 to 5.0, based on the cross-domain correction criterion (the time-varying characteristics of the shock wave-vibration frequency coupling term and the thermal protection efficiency feedback term), the rate of change of heat flux density caused by leading-edge passivation and the flutter critical velocity offset caused by the decrease in structural stiffness are automatically calculated.

[0054] By comparing the solution results of the above equations with wind tunnel experimental data, the prediction accuracy of the strongly coupled fluid-structure-thermal field model is verified. For example, under hypersonic conditions, the model utilizes the flow field equations for accurate calculation of aerodynamic heating, the structural equations for dynamic capture of deformation, and the thermal control equations for feedback correction of thermal loads.

[0055] S150: Based on the strong coupling mechanism of fluid-structure-thermal field, it uses dynamic sensing technology to predict the aerodynamic performance of aircraft; Based on the simulation results of strong coupling between fluid, structure, and thermal field, a dynamic sensing and prediction system is constructed. Real-time parameters are collected, fused using a Kalman filter, and then input into a physically constrained neural network. Using the strong coupling mechanism of fluid, structure, and thermal field as constraints, real-time sensing data and flight state parameters are input, and aerodynamic performance prediction results are output. Key parameters are automatically identified through an attention mechanism, and the spatiotemporal resolution of the prediction model is dynamically adjusted to generate a full-velocity performance trend map.

[0056] Based on the strong coupling mechanism of fluid-structure-thermal field, the aerodynamic performance of aircraft is predicted using dynamic sensing technology, including the following sub-steps: S151: Collect real-time flow field sensor data, correct prediction parameters, and adjust model boundary conditions.

[0057] Based on the simulation results of strong coupling between fluid, structure, and thermal field, a dynamic parameter correction system integrating sensor network and physical constraint neural network is constructed. Real-time data collected by sensors is transmitted to the data processing unit through time-division multiplexing technology. High-frequency noise is removed by wavelet transform, and time-varying feature parameters are extracted using a sliding window algorithm.

[0058] A Physically Constrained Neural Network (PNN) is constructed to perform parameter correction. The network input layer receives real-time acquired pressure, temperature, and strain field data, while the hidden layer contains differential operator layers based on a strongly coupled fluid-structure-thermal field equation set. Physical residuals are calculated using automatic differentiation techniques. A hybrid loss function is employed during training to simultaneously minimize the deviation between predicted and measured values, as well as the residuals of the physical equations. Bayesian regularization is introduced, and the posterior distribution of network parameters is estimated through variational inference.

[0059] An online learning strategy is employed to dynamically adjust model parameters: each time new sensor data is acquired, a mini-batch gradient descent algorithm is triggered to update the network weights, with the update frequency adaptively correlated to the rate of change of flight state. For critical operating conditions such as high Mach numbers, a meta-learning mechanism is enabled to quickly adapt to drastic changes in the parameter space. The corrected set of boundary condition parameters is output, including the wall heat flux density function and aerodynamic load distribution function.

[0060] S152: Combining the strong coupling mechanism of fluid-structure-thermal field, a trend map of aerodynamic performance in the full velocity domain is generated.

[0061] Based on boundary condition parameters, the structural dynamics equations and thermal control transfer equations are transformed into a differential operator layer. By automatically differentiating and calculating the coupling relationship between the flow field, structure, and thermal control, a spatiotemporal prediction model embedding a strong coupling mechanism between the fluid, structure, and thermal field is constructed. In the supersonic region, the shock wave-vibration frequency coupling term is activated, and under high Mach number conditions, a thermal protection efficiency feedback term is introduced to form an adaptive physical constraint across the velocity domain.

[0062] A multi-scale attention module is designed to dynamically adjust the weights of each equation based on the Mach number. Inputting corrected boundary parameters and flight state vectors, a residual network is used to capture coupling effects such as thermally induced stiffness decay and the nonlinear influence of deformation on the flow field. A Gaussian process regression with spatiotemporal consistency constraints is employed to generate the spectrum. In the time dimension, a Kalman smoother preserves historical coupling effects, while in the spatial dimension, a covariance function is constructed based on the strongly coupled fluid-structure-thermal field equations. A Bayesian model averaging technique is introduced, incorporating attenuation coefficients for different thermal control efficiencies to quantify uncertainties. A three-dimensional coupled response spectrum is generated in tensor form, including key indicators such as shock wave location, rate of change of heat flux density, and flutter critical velocity offset.

[0063] S160: Integrates dynamic prediction data with digital simulation, calibrates model parameters, and outputs aerodynamic performance reports and optimization schemes.

[0064] Based on the full-velocity aerodynamic performance trend map, a multi-source data calibration framework is established. The prediction results of the Physically Constrained Neural Network (PNN) are spatiotemporally aligned with high-fidelity CFD simulation data and wind tunnel experimental data. Weighted fusion coefficients are calculated using a Bayesian model averaging algorithm to form a multi-confidence benchmark dataset. The Fourier Amplitude Sensitivity Test (FAST) method is employed to quantify the influence weight of the benchmark dataset on key performance indicators, analyzing the impact of parameters such as leading-edge radius, sweep angle, and thermal shield thickness on heat flux density, lift-to-drag ratio, and structural stability. Based on the analysis results, sampling points are iteratively optimized using an active learning strategy, generating an analysis report and optimization schemes. The optimal parameter ranges for each optimization scheme are marked with confidence ellipses, and a risk assessment matrix is ​​used to quantify the robustness of each scheme.

[0065] Corresponding to the above embodiments, the present invention provides a computer storage medium, including: at least one memory and at least one processor; The memory is used to store one or more program instructions; A processor is used to run one or more program instructions to execute a method for analyzing the aerodynamic performance of an aircraft.

[0066] Corresponding to the above embodiments, this embodiment of the invention provides a computer-readable storage medium containing one or more program instructions, which are executed by a processor to provide a method for analyzing the aerodynamic performance of an aircraft.

[0067] The embodiments disclosed in this invention provide a computer-readable storage medium storing computer program instructions that, when executed on a computer, cause the computer to perform the aforementioned method for analyzing the aerodynamic performance of an aircraft.

[0068] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0069] The various methods, steps, and logic diagrams disclosed in the embodiments of this invention can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods.

[0070] The storage medium can be memory, such as volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0071] Among them, non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.

[0072] Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (Synchlink DRAM, SLDRAM), and direct memory bus RAM (DRRAM).

[0073] The storage media described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memory.

[0074] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this invention can be implemented using a combination of hardware and software. When applied as software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0075] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for analyzing the aerodynamic performance of an aircraft, characterized in that, include: Construct a multi-dimensional parametric model integrating aerodynamics, structure, and thermal control, and output a global design space; Based on the global design space, a proxy model is trained by integrating sparse fluid dynamics and engineering data to verify the reliability across velocity domains. By using a proxy model to decouple the conflicts between the center of mass, lift-to-drag ratio, and thermal load, a velocity-domain adaptive configuration library is generated. For the velocity domain adaptive configuration library, shock wave-boundary layer adaptive mesh technology is used to perform strong coupling simulation of fluid-structure-thermal field; Based on the strong coupling mechanism of fluid-structure-thermal field, dynamic sensing technology is used to predict the aerodynamic performance of aircraft. It integrates dynamic prediction data with digital simulation, calibrates model parameters, and outputs aerodynamic performance reports and optimization schemes.

2. The method for analyzing the aerodynamic performance of an aircraft according to claim 1, characterized in that, Construct a multi-dimensional parametric model integrating aerodynamics, structure, and thermal control, and output a global design space, including: Establish a correlation model of aerodynamic / structural / thermal control parameters, and perform multi-dimensional parametric modeling with multi-physics coupling; A global design space is generated based on an integrated model, and the breadth and accuracy of coverage are verified through multiple constraints.

3. The method for analyzing the aerodynamic performance of an aircraft according to claim 1, characterized in that, Based on the global design space, a proxy model is trained by integrating sparse fluid dynamics and engineering data, and the reliability across velocity domains is verified. This includes: integrating sparse fluid dynamics data and engineering data to construct a hybrid training set to train the proxy model. Cross-speed domain operating condition verification was carried out, and the reliability of the surrogate model was verified through dynamic error correction.

4. The method for analyzing the aerodynamic performance of an aircraft according to claim 1, characterized in that, By using a surrogate model to decouple the conflicts between the center of mass, lift-to-drag ratio, and thermal load, a velocity-domain adaptive configuration library is generated, including: Based on the surrogate model, the coupling mechanism of centroid / lift-drag ratio / thermal load is analyzed, and a multi-objective decoupling algorithm is established. Based on the decoupling results, a velocity-domain segmented configuration is generated, and an adaptive configuration library is constructed.

5. The method for analyzing the aerodynamic performance of an aircraft according to claim 1, characterized in that, For the velocity-domain adaptive configuration library, shock-boundary layer adaptive meshing technology is used to perform strongly coupled fluid-structure-thermal field simulations, including: Construct an adaptive mesh for dynamic capture of shock waves and boundary layers to improve the resolution of complex flow fields; The simulation of strong coupling between fluid, structure, and thermal field is performed based on adaptive mesh, and the dynamic response of strong coupling between fluid, structure, and thermal field is quantified.

6. The method for analyzing the aerodynamic performance of an aircraft according to claim 1, characterized in that, Based on the strong coupling mechanism of fluid-structure-thermal field, dynamic sensing technology is used to predict the aerodynamic performance of aircraft, including: Collect real-time flow field sensor data, correct prediction parameters, and adjust model boundary conditions; By combining the strong coupling mechanism of fluid-structure-thermal field, a trend map of aerodynamic performance across the entire velocity domain is generated.

7. A computer-readable storage medium, characterized in that, It includes one or more program instructions, which are executed by a processor as described in any one of claims 1-6, to provide an aircraft aerodynamic performance analysis method.

Images