CN-122019496-A - Method and system for constructing pneumatic database of aircraft

CN122019496ACN 122019496 ACN122019496 ACN 122019496ACN-122019496-A

Abstract

The invention provides an aircraft pneumatic database construction method and a system thereof, wherein the method comprises the steps of determining an initial sampling point set in a design space by adopting a test design method, calculating aerodynamic data of the initial sampling point set by adopting a CFD method to obtain an initial aerodynamic data set, constructing an initial aerodynamic model based on the initial aerodynamic data set by adopting a proxy model method, solving an optimization problem by using a genetic algorithm to obtain a design point P1 with the maximum aerodynamic coefficient prediction uncertainty, sampling by adopting a Latin hypercube method in the design space, selecting a sampling point farthest from the initial point set to form a new supplementary sampling point set P2, combining the P1 and the P2 to form the supplementary sampling point set, solving the CFD problem to obtain new aerodynamic data, constructing a new aerodynamic data prediction model until a cross verification index of the aerodynamic model is superior to a specified value. The method can obviously reduce the CFD calculation times, ensure the model precision and improve the pneumatic database construction efficiency.

Inventors

ZHAO YUAN
SHI XIAOTIAN
GAO JUN
LV MENG

Assignees

中国航天空气动力技术研究院

Dates

Publication Date: 20260512
Application Date: 20251218

Claims (10)

1. The method for constructing the aerodynamic database of the aircraft is characterized by comprising the following steps of: S1, determining an initial sampling point set in a design space by adopting a test design method; S2, calculating aerodynamic data of the initial sampling point set by adopting a CFD method to obtain an initial aerodynamic data set; s3, constructing an initial aerodynamic model based on the initial aerodynamic data set by adopting a proxy model method; S4, solving an optimization problem by using a genetic algorithm to obtain a design point with the maximum uncertainty of aerodynamic coefficient prediction, and marking the design point as a point P1; S5, sampling is carried out in the design space by using a Latin hypercube method, sampling points farthest from the initial point set are selected to form a new supplementary sampling point set P2, and P1 and P2 are combined to form the supplementary sampling point set; S6, solving a CFD problem on the supplementary sampling point set to obtain new aerodynamic data; s7, constructing a new aerodynamic data prediction model based on all aerodynamic data; and S8, repeating the steps S4 to S7 until the cross verification index of the aerodynamic model is better than the specified value.
2. The method of claim 1, wherein in step S1, the test design method is one of latin hypercube design, orthogonal array design, or full factor design.
3. The method according to claim 1, wherein in step S1, the design space includes a flight altitude, a flight mach number, a flight attack angle, a flight sideslip angle, and a rudder deflection angle.
4. The method for constructing an aerodynamic database of an aircraft according to claim 1, wherein in step S2, the process of calculating aerodynamic data of the initial sampling point set by using a CFD method includes the steps of: S21, determining initial parameters for calculating a fluid domain, grid division, a numerical solver, a turbulence model and boundary conditions; S22, performing iterative CFD calculation on the flight condition and configuration of the aerodynamic profile of the selected aircraft by adopting the initial parameters to generate a CFD calculation result; and S23, carrying out post-processing on the CFD calculation result, and extracting aerodynamic coefficient data.
5. The method for constructing an aerodynamic database of an aircraft according to claim 1, wherein in step S3, the proxy model method comprises one of a standard gaussian process, a random forest regression, a sparse variation gaussian process, a sparse variation deep kernel learning, and a gradient lifting tree.
6. The aircraft pneumatic database construction method according to claim 1, wherein in step S4, the process of solving the optimization problem using the genetic algorithm comprises the steps of: S41, maximizing uncertainty of aerodynamic coefficient prediction as an objective function; s42, taking the boundary of the design space as a constraint condition; s43, iteratively searching an optimal solution through the selection, crossing and mutation operations of a genetic algorithm.
7. The method for constructing an aircraft pneumatic database according to claim 1, wherein in step S5, the method for selecting the sampling point farthest from the initial point set comprises: S51, calculating the minimum Euclidean distance from each Latin hypercube sampling point to all points in the initial point set; S52, selecting the first N with the maximum Euclidean distance, wherein N is the design space dimension sampling point to form a supplementary sampling point set P2.
8. The method for constructing an aircraft pneumatic database according to claim 1, wherein in step S8, the cross-validation index is one of a root mean square error, an average absolute error, or a decision coefficient, and the specified value is determined according to engineering accuracy requirements.
9. An aircraft pneumatic database construction system for implementing the aircraft pneumatic database construction method according to any one of claims 1 to 8, characterized by comprising: the test design module is used for determining an initial sampling point set in a design space; The CFD calculation module is used for calculating aerodynamic force data of the initial sampling point set; The agent model building module is used for building a aerodynamic model based on aerodynamic data; The optimization sampling module is used for searching a design point with maximum prediction uncertainty by using a genetic algorithm; And the database updating module is used for iteratively updating the pneumatic database until the precision requirement is met.
10. A computer-readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the steps of the aircraft pneumatic database construction method according to any one of claims 1 to 8.

Description

Method and system for constructing pneumatic database of aircraft Technical Field The invention relates to the technical field of aerodynamic characteristic analysis of aircrafts, in particular to an aircraft aerodynamic database construction method and an aircraft aerodynamic database construction system. Background The pneumatic database is the core basic data for aircraft design, simulation, and performance assessment. The traditional pneumatic database construction method mainly depends on large-scale Computational Fluid Dynamics (CFD) calculation or wind tunnel test, and has the problems of high calculation cost, long period, large resource consumption and the like. The conventional pneumatic database construction method mainly comprises (1) a method based on a full-factor test design, which needs to uniformly distribute points in a design space and perform a large number of CFD calculations, and has low calculation efficiency although the coverage is comprehensive, (2) a method based on a traditional test design (such as an orthogonal design and a Latin hypercube design), which can not adaptively adjust a sampling strategy according to model characteristics although the number of sampling points is reduced, and (3) a method based on an empirical formula, which has high calculation speed but limited precision and is difficult to meet the requirement of high-precision pneumatic analysis. The method has the main defects that CFD computing resources are consumed greatly, a high-precision model cannot be obtained under limited computing resources, a sampling strategy is fixed and cannot be dynamically adjusted according to uncertainty of the model, and balance between model precision and computing cost is difficult to optimize. Therefore, a novel method for remarkably reducing the number of times of CFD calculation and improving the construction efficiency of the pneumatic database on the premise of ensuring the precision is urgently needed in the field. Disclosure of Invention The invention aims to provide an aircraft pneumatic database construction method and system, which can obviously reduce CFD calculation times through intelligent sampling and iterative optimization, and improve the pneumatic database construction efficiency while ensuring model accuracy. In a first aspect of the invention, a method for constructing an aerodynamic database of an aircraft is provided, comprising the steps of: S1, determining an initial sampling point set in a design space by adopting a test design method; s2, calculating aerodynamic data of the initial sampling point set by adopting a Computational Fluid Dynamics (CFD) method to obtain an initial aerodynamic data set; s3, constructing an initial aerodynamic model based on the initial aerodynamic data set by adopting a proxy model method; S4, solving an optimization problem by using a genetic algorithm to obtain a design point with the maximum uncertainty of aerodynamic coefficient prediction, and marking the design point as a point P1; S5, sampling is carried out in the design space by using a Latin hypercube method, sampling points farthest from the initial point set are selected to form a new supplementary sampling point set P2, and P1 and P2 are combined to form the supplementary sampling point set; S6, solving a CFD problem on the supplementary sampling point set to obtain new aerodynamic data; s7, constructing a new aerodynamic data prediction model based on all aerodynamic data (including initial data and supplementary data); and S8, repeating the steps S4 to S7 until the cross verification index of the aerodynamic model is better than the specified value. Preferably, in step S1, the test design method uses one of latin hypercube design, orthogonal array design or full factor design to ensure uniformity and representativeness of the initial sampling points in the design space. Preferably, in step S1, the design space includes critical flight parameters such as flight altitude, flight mach number, flight attack angle, flight sideslip angle, rudder deflection angle, and the like. Preferably, in step S1, the initial sampling scale is 10n, where n is the design space dimension. Preferably, in step S2, the process of calculating aerodynamic data of the initial sampling point set by using the CFD method includes the following steps: S21, determining initial parameters for calculating a fluid domain, grid division, a numerical solver, a turbulence model and boundary conditions; S22, performing iterative CFD calculation on the flight condition and configuration of the aerodynamic profile of the selected aircraft by adopting the initial parameters to generate a CFD calculation result; and S23, carrying out post-processing on the CFD calculation result, and extracting aerodynamic coefficient data. Preferably, in step S3, the proxy model method includes one of a standard gaussian process, a random forest regression, a sparse variation gaussian process, a sparse variation d