CN116562124A - Method and device for predicting influence of high-speed aircraft radome ablation on electromagnetic performance - Google Patents

Method and device for predicting influence of high-speed aircraft radome ablation on electromagnetic performance Download PDF

Info

Publication number
CN116562124A
CN116562124A CN202310342223.5A CN202310342223A CN116562124A CN 116562124 A CN116562124 A CN 116562124A CN 202310342223 A CN202310342223 A CN 202310342223A CN 116562124 A CN116562124 A CN 116562124A
Authority
CN
China
Prior art keywords
model
prediction
data set
test data
proxy model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202310342223.5A
Other languages
Chinese (zh)
Other versions
CN116562124B (en
Inventor
黄寒砚
谢珊
刘泽苁
李启喆
于哲峰
徐春光
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ultra High Speed Aerodynamics Institute China Aerodynamics Research and Development Center
Sun Yat Sen University
Original Assignee
Ultra High Speed Aerodynamics Institute China Aerodynamics Research and Development Center
Sun Yat Sen University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ultra High Speed Aerodynamics Institute China Aerodynamics Research and Development Center, Sun Yat Sen University filed Critical Ultra High Speed Aerodynamics Institute China Aerodynamics Research and Development Center
Priority to CN202310342223.5A priority Critical patent/CN116562124B/en
Publication of CN116562124A publication Critical patent/CN116562124A/en
Application granted granted Critical
Publication of CN116562124B publication Critical patent/CN116562124B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/08—Thermal analysis or thermal optimisation
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Pure & Applied Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Optimization (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Operations Research (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Artificial Intelligence (AREA)
  • Medical Informatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Geometry (AREA)
  • Algebra (AREA)
  • Computer Hardware Design (AREA)
  • Databases & Information Systems (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention discloses a method and a device for predicting the influence of ablation of a radome of a high-speed aircraft on electromagnetic performance, comprising the steps of acquiring a test data set of two sub-stages of thermal ablation and thermal wave transmission in the ablation process and a plurality of test data sets of different fidelity of a full system level; constructing a minimum fidelity proxy model by using the sub-stage level data set based on an associated Gaussian proxy model method; based on a layered kriging method, gradually fusing full-system-level data sets from low to high according to fidelity, and iteratively updating the proxy model to obtain a target prediction proxy model; when corresponding input parameters of antenna housing ablation are received, response prediction is carried out through a target prediction agent model, and an ablation electromagnetic performance prediction response value and a prediction error corresponding to the antenna housing are obtained. Therefore, based on test data from various sources, a proxy model of the influence of the antenna housing ablation on the electromagnetic performance is constructed more accurately, and further, the quick prediction of the influence of the antenna housing ablation on the electromagnetic performance is realized.

Description

Method and device for predicting influence of high-speed aircraft radome ablation on electromagnetic performance
Technical Field
The invention relates to the technical field of antennas, in particular to a method and a device for predicting influence of ablation of a radome of a high-speed aircraft on electromagnetic performance.
Background
In the high-speed flight process of the aircraft, the flight speed is extremely high, kinetic energy is converted into internal energy of air molecules to enable the temperature near the front end of the air molecules to rise, and high-temperature shock waves far higher than the pyrolysis temperature of the material of the radome body are formed around the radome body, so that the thermal ablation phenomenon is generated; the antenna housing material is ablated, and then the physical structure is destroyed, so that heat wave transmission is formed when internal electromagnetic waves penetrate the housing, at this time, the change of the thickness of the housing material can influence the beam direction of the antenna, and the change of the temperature can influence the dielectric property of the housing material, so that malignant influence (including antenna impedance mismatch, antenna gain reduction, frequency offset and the like) can be caused on the working performance of the antenna, and finally, the signal transmission is problematic. Since the ablation of the radome cannot be directly avoided, it is important to analyze how the electromagnetic performance is affected by the initial factors (such as the flying speed) in the ablation process of the radome, so that the error caused by the ablation can be corrected, and the normal operation of the antenna is ensured.
However, since the flight test is complicated to implement, is expensive, and is not feasible to implement in a large amount, the present research generally relies on a large number of simulation tests for analysis. Most of researches focus on simulating and researching conditions of a radome ablation physical process, such as calculating thermal response of the radome in a thermal ablation stage through pneumatic heating software, optimizing the radome by simulating an antenna array in a thermal wave transmission stage through a high-order moment method, and analyzing the overall effect of the whole system through combined heat and power simulation.
However, the scheme only models from the simulation and simulation angle, calculates by means of simulation and simulation calculation with large calculation amount, can not rapidly realize the prediction of the antenna housing ablation on the electromagnetic performance, and does not utilize the information of an actual test.
Disclosure of Invention
The invention provides a method and a device for predicting electromagnetic performance influence caused by antenna housing ablation of a high-speed aircraft, which solve the technical problems that the electromagnetic performance cannot be rapidly predicted by antenna housing ablation and the existing scheme only models from the simulation angle and calculates by means of large-calculation-amount simulation calculation, and the experimental data information of various sources is not utilized.
The invention provides a method for predicting the influence of ablation of a radome of a high-speed aircraft on electromagnetic performance, which is characterized by comprising the following steps:
acquiring a thermal ablation stage test data set, a thermal wave transmission stage test data set and a plurality of full-system-level test data sets with different fidelity of an antenna housing of a high-speed aircraft in an ablation process;
based on an associated Gaussian proxy model method, constructing an initial predictive proxy model with the lowest fidelity by adopting the thermal ablation stage test data set and the thermal wave-transparent stage test data set;
based on a layered kriging method, gradually fusing the full-system-level test data set from low to high according to the fidelity, and iteratively updating the initial prediction proxy model to generate a target prediction proxy model;
and when the ablation input parameters of the radome are received, performing response prediction through the target prediction agent model, and generating an electromagnetic performance prediction response value and a prediction error.
Optionally, the step of constructing the initial predictive proxy model with minimum fidelity by adopting the thermal ablation stage test data set and the thermal wave-transparent stage test data set based on the associated gaussian proxy model method comprises the following steps:
When a deterministic test kernel function is received, a Gaussian process is combined with the deterministic test kernel function to respectively construct a thermal ablation stage model and a thermal wave-transparent stage model;
according to the thermal ablation stage model and the thermal wave-transparent stage model, an associated Gaussian agent model method is adopted to construct an uncertainty propagation relation;
substituting the thermal ablation stage test data set and the thermal wave-transparent stage test data set into the uncertainty propagation relation respectively, and calculating a response variance and a response mean;
and generating an initial predictive proxy model with the lowest fidelity by adopting the response variance and the response mean.
Optionally, the step of merging the full system level test data set step by step according to the fidelity from low to high based on the layered kriging method, and iteratively updating the initial predictive proxy model with the lowest fidelity to generate a target predictive proxy model includes:
constructing a deviation model between the initial predictive proxy model and a real response based on the initial predictive proxy model;
selecting a full-system-level test data set corresponding to the fidelity which is the smallest currently as a training data set;
calculating a plurality of prediction response values corresponding to the training data set through the initial prediction agent model and constructing a prediction vector;
Calculating an optimal linear unbiased predicted value and a deviation coefficient of the deviation model based on the predicted vector and a real response vector corresponding to the full-system-level test data set;
updating the initial prediction proxy model according to the optimal linear unbiased prediction value of the deviation model and the deviation coefficient to generate an intermediate prediction proxy model;
judging whether the unselected full-system-level test data set exists or not;
if yes, taking the intermediate predictive proxy model as a new initial predictive proxy model, and jumping to execute the step of selecting the full system level test data set corresponding to the current minimum fidelity as a training data set;
if not, determining the intermediate predictive proxy model at the current moment as a target predictive proxy model.
Optionally, the step of calculating the optimal linear unbiased prediction value and the deviation coefficient of the deviation model based on the prediction vector and the real response vector corresponding to the full system level test data set includes:
constructing a correlation matrix of the full system level test data set, and converting the correlation matrix into an inverse matrix;
transforming the prediction vector into a transpose matrix;
determining a corresponding deviation coefficient by adopting the inverse matrix, the transpose matrix, the real response vector and the prediction vector;
Calculating a multiplication value between the deviation coefficient and the prediction vector;
calculating a difference value between a real response vector corresponding to the full system level test data set and a multiplication value;
and determining an optimal linear unbiased prediction value of the deviation model by adopting the difference value, the correlation matrix and the inverse matrix.
Optionally, the step of updating the initial predictive proxy model according to the optimal linear unbiased predictive value of the bias model and the bias coefficient to generate an intermediate predictive proxy model includes:
combining the deviation coefficient with the initial predictive proxy model, and adding an optimal linear unbiased predictive value of the deviation model to generate an intermediate predictive proxy model;
calculating a mean square error corresponding to the intermediate prediction proxy model;
the intermediate prediction proxy model is as follows:
the mean square error is:
where x is the input of the proxy model,for the predicted response value, ρ, of the intermediate predicted proxy model n+1 As a deviation factor>Predictive response value for initial predictive proxy model, +.>For optimal linear unbiased predictions of the bias model, MSE n+1 (x) Is mean square error>Is the correlation vector between the unknown full system level test data set and the known full system level test data set, R n+1 N being a known full system level test dataset n+1 ×N n+1 Correlation matrix R (x i ,x j )，Is a regression matrix.
The invention provides a prediction device for influence of high-speed aircraft radome ablation on electromagnetic performance, which is characterized by comprising the following components:
the data set acquisition module is used for acquiring a thermal ablation stage test data set, a thermal wave-transparent stage test data set and a plurality of full-system-level test data sets with different fidelity of the radome of the high-speed aircraft in the ablation process;
the model construction module is used for constructing an initial prediction proxy model with the lowest fidelity by adopting the thermal ablation stage test data set and the thermal wave-transparent stage test data set based on an associated Gaussian proxy model method;
the model updating module is used for gradually fusing the full-system-level test data set from low to high according to the fidelity based on a layered kriging method, and iteratively updating the initial prediction proxy model to generate a target prediction proxy model;
and the prediction module is used for carrying out response prediction through the target prediction agent model when the ablation input parameters of the radome are received, and generating an electromagnetic performance prediction response value and a prediction error.
Optionally, the model building module is specifically configured to:
When a deterministic test kernel function is received, a Gaussian process is combined with the deterministic test kernel function to respectively construct a thermal ablation stage model and a thermal wave-transparent stage model;
according to the thermal ablation stage model and the thermal wave-transparent stage model, an associated Gaussian agent model method is adopted to construct an uncertainty propagation relation;
substituting the thermal ablation stage test data set and the thermal wave-transparent stage test data set into the uncertainty propagation relation respectively, and calculating a response variance and a response mean;
and generating an initial predictive proxy model with the lowest fidelity by adopting the response variance and the response mean.
Optionally, the model updating module includes:
the deviation model construction submodule is used for constructing a deviation model between the initial prediction agent model and the real response based on the initial prediction agent model;
the data set selecting sub-module is used for selecting the full-system-level test data set corresponding to the current minimum fidelity as a training data set;
the prediction vector calculation operator module is used for calculating a plurality of prediction response values corresponding to the training data set through the initial prediction agent model and constructing a prediction vector;
the parameter calculation sub-module is used for calculating an optimal linear unbiased predicted value and a deviation coefficient of the deviation model based on the predicted vector and a real response vector corresponding to the full-system-level test data set;
The model updating sub-module is used for updating the initial prediction proxy model according to the optimal linear unbiased prediction value of the deviation model and the deviation coefficient to generate an intermediate prediction proxy model;
the judging submodule is used for judging whether the unselected full-system-level test data set exists or not;
the circulation sub-module is used for taking the intermediate prediction agent model as a new initial prediction agent model if yes, and jumping to execute the step of selecting the full system level test data set corresponding to the current minimum fidelity as a training data set;
and the model determination submodule is used for determining the intermediate prediction agent model at the current moment as a target prediction agent model if not.
Optionally, the parameter calculation sub-module is specifically configured to:
constructing a correlation matrix of the full system level test data set, and converting the correlation matrix into an inverse matrix;
transforming the prediction vector into a transpose matrix;
determining a corresponding deviation coefficient by adopting the inverse matrix, the transpose matrix, the real response vector and the prediction vector;
calculating a multiplication value between the deviation coefficient and the prediction vector;
calculating a difference value between a real response vector corresponding to the full system level test data set and a multiplication value;
And determining an optimal linear unbiased prediction value of the deviation model by adopting the difference value, the correlation matrix and the inverse matrix.
Optionally, the model updating sub-module is specifically configured to:
combining the deviation coefficient with the initial predictive proxy model, and adding an optimal linear unbiased predictive value of the deviation model to generate an intermediate predictive proxy model;
calculating a mean square error corresponding to the intermediate prediction proxy model;
the intermediate prediction proxy model is as follows:
the mean square error is:
where x is the input of the proxy model,for the predicted response value, ρ, of the intermediate predicted proxy model n+1 As a deviation factor>Predictive response value for initial predictive proxy model, +.>For optimal linear unbiased predictions of the bias model, MSE n+1 (x) Is mean square error>Is the correlation vector between the unknown full system level test data set and the known full system level test data set, R n+1 N being a known full system level test dataset n+1 ×N n+1 Correlation matrix R (x i ,x j )，Is a regression matrix.
From the above technical scheme, the invention has the following advantages:
according to the invention, a test data set of two sub-stages of thermal ablation and thermal wave transmission in the ablation process and a plurality of test data sets of different fidelity of a full system level are obtained; constructing a minimum fidelity proxy model by using the sub-stage level data set based on an associated Gaussian proxy model method; based on a layered kriging method, gradually fusing full-system-level data sets from low to high according to fidelity, and iteratively updating the proxy model to obtain a target prediction proxy model; when corresponding input parameters of antenna housing ablation are received, response prediction is carried out through a target prediction agent model, and an ablation electromagnetic performance prediction response value and a prediction error corresponding to the antenna housing are obtained. Therefore, based on test data from various sources, a proxy model of the influence of the antenna housing ablation on the electromagnetic performance is constructed more accurately, and further, the quick prediction of the influence of the antenna housing ablation on the electromagnetic performance is realized.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained from these drawings without inventive faculty for a person skilled in the art.
Fig. 1 is a flowchart of steps of a method for predicting ablation electromagnetic performance of an antenna housing of an aircraft according to a first embodiment of the present invention;
fig. 2 is a flowchart of a method for predicting electromagnetic performance of antenna housing ablation of an aircraft according to a second embodiment of the present invention;
FIG. 3 is a flowchart illustrating steps of a method for constructing a coherence proxy model according to an embodiment of the present invention;
FIG. 4 is a flowchart of steps for predicting the electromagnetic performance of radome ablation for an aircraft, according to an embodiment of the present invention;
fig. 5 is a block diagram of an apparatus for predicting ablation electromagnetic performance of an antenna housing of an aircraft according to a third embodiment of the present invention.
Detailed Description
The radome is a structural protection device for a radar at the front end of an aircraft, and is generally used for reducing the influence of various external factors on a radar antenna, and has the main purpose of protecting the antenna from the external environment and good electromagnetic wave penetration characteristics. The analysis flow of the antenna housing ablation can be divided into two stages, namely a thermal ablation stage and a thermal wave transmission stage, and the complexity of the analysis flow is that the direct integral modeling research has certain difficulty, so that the thermodynamic performance of the antenna housing in the whole process is considered, and the electromagnetic performance of dynamic change is also considered. If the full-wave analysis simulation of the thermal-electric-force is carried out on the whole, the data accuracy is high, but the workload is huge, and a lot of calculation resources are required to be consumed; while staged simulation runs are relatively computationally less expensive, there are also few ways to link the two staged simulation runs. Therefore, it is highly desirable to find a method that can accurately calculate this process at a low cost, and that can integrate existing test data from a variety of sources.
The proxy model refers to an approximate mathematical model established for the complex system by using experimental data of the complex system to describe the relationship between input and output of the complex system, and plays a very important role in many scientific and engineering fields. The traditional agent model construction method is called an All-In-One (AIO) method, the AIO method regards the whole complex system as a black box, the whole system is directly tested, and an agent model is constructed according to test data. The traditional AIO modeling method mainly comprises a polynomial response surface method, a support vector machine method, a neural network, kriging and the like, wherein Kriging (also called a Gaussian process regression model) is suitable for nonlinear and small-sample modeling. However, the whole system test is expensive for the radome ablation system, so that the obtained test sample size is relatively small, but the fidelity is relatively high. In addition, the AIO method emphasizes that the whole is a black box function and ignores the association relation among all subsystems, and because the radome ablation analysis system is formed by connecting two sub-stages (a thermal ablation stage and a thermal wave-transparent stage) in series, the sub-stages are only in one-way association with each other through input and output, so that a proxy model can be constructed by considering simulation test data of all stages which are relatively easy to obtain. Such agent models that couple the agent models of the stages together by an analysis method to construct a full system are called associated agent models.
The embodiment of the invention provides a method and a device for predicting the influence of ablation of a radome of a high-speed aircraft on electromagnetic performance, which are used for solving the technical problems that the conventional scheme is only used for modeling from the simulation angle, calculation is carried out by means of large-calculation-amount simulation calculation, the prediction of the ablation of the radome on the electromagnetic performance cannot be rapidly realized, and fusion of test data information from multiple sources is lacking.
In order to make the objects, features and advantages of the present invention more comprehensible, the technical solutions in the embodiments of the present invention are described in detail below with reference to the accompanying drawings, and it is apparent that the embodiments described below are only some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, fig. 1 is a flowchart illustrating steps of a method for predicting ablation electromagnetic performance of an antenna housing of an aircraft according to an embodiment of the invention.
The invention provides a method for predicting the ablation electromagnetic performance of an antenna housing of an aircraft, which comprises the following steps:
Step 101, acquiring a thermal ablation stage test data set, a thermal wave-transparent stage test data set and a plurality of full-system-level test data sets with different fidelity of a radome of a high-speed aircraft in an ablation process;
in a specific implementation, the factors influencing the electromagnetic performance of the antenna housing during each stage of ablation can be analyzed, for example, during a thermal ablation stage, i.e. the process that the antenna housing of the aircraft is ablated during high-speed flight, the input variables are considered to be the incoming flow speed x respectively 1 Static temperature of incoming flow, static pressure x of incoming flow 3 Surface roughness x 4 Density x of material 5 Specific heat x of material 6 Coefficient of surface emissivity x 7 Coefficient of thermal conductivity x 8 Coefficient of liquid layer viscosity x 9 These variables all contribute to this thermal ablation processWhereby the output variable is the temperature profile w 1 And thickness w after ablation 2 The method comprises the steps of carrying out a first treatment on the surface of the In the thermal wave-transparent stage, i.e. the process of influencing the electromagnetic properties of the radome by various parameters after ablation, the input variables include the external variables-frequency z and the output of the thermal ablation stage, wherein the temperature profile w 1 Influence of dielectric constant by a known map gNumber g (w) 1 ) Plus the thickness w after ablation 2 These three variables act together on this thermal wave-transparent process +.>Finally, variable reflection coefficient y capable of reflecting electric performance is output 1 And transmission coefficient y 2 。
In the embodiment of the invention, in order to obtain the data base of the antenna housing ablation electromagnetic performance prediction, a sub-stage test data set and a plurality of full-system-level test data sets with different fidelity can be obtained first.
Wherein the sub-phase test data set comprises a sub-phase test data set corresponding to a thermal ablation phase and a sub-phase test data set corresponding to a thermal wave-transparent phase, namelyWherein->Test data set representing the thermal ablation phase, +.>A test dataset representing a heat wave-transparent phase. Whereas the full system level test dataset refers to the full system test dataset +.>Wherein the fidelity is ordered from low to high, and the highest fidelity is 100%, namely the real test data of the whole system level
Step 102, constructing an initial predictive proxy model with minimum fidelity by adopting a thermal ablation stage test data set and a thermal wave-transparent stage test data set based on an associated Gaussian proxy model method;
in the embodiment of the invention, after a sub-stage test data set is obtained, a Gaussian process model of a thermal ablation stage and a Gaussian process model of a thermal transmission stage can be established for the data set, and after an uncertainty propagation relationship is analyzed by using a correlation Gaussian proxy model method, a response variance and a response mean value are calculated by combining the sub-stage test data set, so that a minimum fidelity prediction proxy model of the whole system is obtained through Gaussian process modeling.
It should be noted that the thermal ablation stage model refers to a mathematical model reflecting the thermal ablation process of the radome of the high-speed aircraft during the high-speed flight. The thermal wave-transparent stage model refers to a mathematical model reflecting the process of influencing the electromagnetic performance of the radome by various parameters after ablation. The initial predictive proxy model refers to a proxy model that is built by computing the response (output) of the original model through a finite number of points (inputs) describing the relationship between the overall system inputs and outputs.
Step 103, based on a layered kriging method, gradually fusing a full system level test data set from low to high according to fidelity, and iteratively updating an initial prediction agent model to generate a target prediction agent model;
after the initial predictive proxy model is established, full-system set test data sets with various sources can be fused on the basis of the initial predictive proxy model, the constructed proxy model is updated, the predictive performance of the constructed proxy model is further improved, all full-system-level test data sets can be sequentially fused from low to high according to fidelity, and the full-system-level test data sets are adopted to update the predictive proxy model in an iterative mode to generate the target predictive proxy model.
And 104, when ablation input parameters of the radome are received, performing response prediction through a target prediction proxy model, and generating an electromagnetic performance prediction response value and a prediction error.
After the target prediction proxy model is obtained, if the corresponding input parameters of the antenna housing ablation are received at the moment, the target prediction proxy model can be adopted to conduct response prediction so as to determine a predicted response value of the antenna housing ablation under the input to the electromagnetic performance influence, and meanwhile, a prediction error is calculated so as to determine the uncertainty of the electromagnetic performance influence predicted response value.
In the embodiment of the invention, a test data set of two sub-stages of thermal ablation and thermal wave transmission in the ablation process and a plurality of test data sets of different fidelity of the whole system level are obtained; constructing a minimum fidelity proxy model by using the sub-stage level data set based on an associated Gaussian proxy model method; based on a layered kriging method, gradually fusing full-system-level data sets from low to high according to fidelity, and iteratively updating the proxy model to obtain a target prediction proxy model; when corresponding input parameters of antenna housing ablation are received, response prediction is carried out through a target prediction agent model, and an ablation electromagnetic performance prediction response value and a prediction error corresponding to the antenna housing are obtained. Therefore, based on test data from various sources, a proxy model of the influence of the antenna housing ablation on the electromagnetic performance is constructed more accurately, and further, the quick prediction of the influence of the antenna housing ablation on the electromagnetic performance is realized.
Referring to fig. 2, fig. 2 is a flowchart illustrating steps of a method for predicting ablation electromagnetic performance of an antenna housing of an aircraft according to a second embodiment of the present invention.
The invention provides a method for predicting the ablation electromagnetic performance of an antenna housing of an aircraft, which comprises the following steps:
step 201, acquiring a thermal ablation stage test data set, a thermal wave-transparent stage test data set and a plurality of full-system-level test data sets with different fidelity of a radome of a high-speed aircraft in an ablation process;
in the embodiment of the invention, the radome can undergo a thermal ablation stage and a thermal wave transmission stage during high-speed flight, so that a thermal ablation stage test data set and a thermal wave transmission stage test data set of the radome of the high-speed aircraft during ablation can be obtained, wherein the thermal ablation stage test data set comprises input variables in the thermal ablation stage, namely the incoming flow speed x 1 Static temperature x of incoming flow 2 Static pressure x of incoming flow 3 Surface roughness x 4 Density x of material 5 Specific heat x of material 6 Coefficient of surface emissivity x 7 Coefficient of thermal conductivity x 8 Coefficient of liquid layer viscosity x 9 And an output variable temperature profile w obtained after completion of the thermal ablation stage 1 And thickness w after ablation 2 The method comprises the steps of carrying out a first treatment on the surface of the The thermal ablation stage test data set comprises input variables and output variables of the thermal wave-transparent stage, namely the input variables are the temperature distribution w 1 The dielectric constant g (w 1 ) External input frequency z and post-ablation thickness w 2 The output variable is the variable reflection coefficient y reflecting the electromagnetic performance of the radome 1 And transmission coefficient y 2 。
Step 202, when a deterministic test kernel function is received, a Gaussian process is combined with the deterministic test kernel function to respectively construct a thermal ablation stage model and a thermal wave-transparent stage model;
in the embodiment of the invention, when a deterministic test kernel function, such as a normal kernel function or a Laplacian function, is received, a Gaussian process can be adopted to combine the deterministic test kernel function to construct a thermal ablation stage model and a thermal wave-transparent stage model. The thermal ablation stage model is a model constructed by reflecting the ablation process of the aerial cover of the aircraft in the high-speed flight process, and the observation error e is taken into consideration i And truncation error u i In the case of (a), the relationship between the input variable and the output variable can be modeled as:
wherein x= [ x ] 1 ,…,x 9 ]，μ i (x) Is a function of the mean value,is a model error subject to normal distribution.
The thermal wave-transparent stage model is a model reflecting the process that various parameters after antenna housing ablation influence the electromagnetic performance of the antenna housing, and the relation between input variables and output variables can be modeled as follows:
Wherein mu j (x) Is a function of the mean value,is subject to normal distribution of model errors +.>g(w 1 ) Is a dielectric constant.
The sub-phase model can be obtained by sub-phase test data setSolving to obtain the final product.
Step 203, constructing an uncertainty propagation relation by adopting a correlation Gaussian agent model method according to the thermal ablation stage model and the thermal wave-transparent stage model;
the uncertainty propagation relationship is:
y～N(E(y),D(y)),y＝{y 1 ,y 2 }
and no coupling relation exists between the input variables of the two stages, and the uncertainty propagation relation is deduced at the moment as follows:
thereby obtaining the output y= [ y ] of the second layer of the system 1 ,y 2 ]Input x= [ x ] to the first layer 1 ,...,x 9 ]And the relationship of the external input z of the second layer is thatWherein->Representing the output variable y j Is a function determined only by the input variables (x, z), the +.>Representing the output variable y j Is a function determined only by the input variables (x, z).
Step 204, substituting the test data set in the thermal ablation stage and the test data set in the thermal wave-transparent stage into an uncertainty propagation relation respectively, and calculating a response variance and a response mean;
and 205, generating an initial predictive proxy model with the lowest fidelity by adopting the response variance and the response mean.
Generating an initial predictive proxy model with minimum fidelity by adopting response variance and response mean Where x= { X, z } is the input of the whole system, E (X) is the response mean, and D (X) is the response variance. In a specific implementation, according to the expression form of the kernel function selected in the Gaussian process and whether the test is a deterministic test, the calculation method is as follows:
a) In the deterministic test, the kernel function is selected as a normal kernel, a Laplace kernel or an index kernel, and the smooth index in the kernel function is 1 or 2, so that the expressions of E (y) and D (y) can be directly calculated.
b) In other cases, the calculation may be performed using a numerical simulation method such as Monte Carlo.
Step 206, based on a layered kriging method, gradually fusing the full system level test data set from low to high according to fidelity, and iteratively updating an initial prediction agent model to generate a target prediction agent model;
optionally, step 206 may include the following sub-steps S11-S18:
s11, constructing a deviation model between the initial prediction agent model and the real response based on the initial prediction agent model;
in the embodiment of the invention, in order to further improve the prediction precision of the initial prediction proxy model, the full system level data set D can be gradually fused from low to high according to the fidelity m M=1, …, n, and iteratively updating the proxy model to the target predictive proxy model. Firstly, a full-system-level test data set D corresponding to the minimum fidelity at the current moment can be selected 1 As a training data set.
Wherein, proxy model M 1 And M is as follows 0 Obeys the following deviation model:
wherein the method comprises the steps ofRepresentation model M 1 The value at variable x, +.>Representing an initial predictive proxy model M 0 Value at variable x, ρ 1 Is the deviation factor, z 1 (x)～GP(0,k 1 (.+ -.)) is a zero-mean gaussian process, and
s12, selecting a full-system-level test data set corresponding to the current minimum fidelity as a training data set;
selecting a full system level test data set corresponding to the current minimum fidelity as a training data set, inputting the training data set into an initial prediction proxy model to calculate a plurality of prediction response values corresponding to the training data set, and constructing a prediction vectorWherein->Refers to the initial predictive proxy model at variable +.>The following predicted response values.
S13, calculating a plurality of prediction response values corresponding to the training data set through an initial prediction agent model and constructing a prediction vector;
s14, calculating an optimal linear unbiased predicted value and a deviation coefficient of the deviation model based on the predicted vector and a real response vector corresponding to the full-system-level test data set;
further, step S14 may comprise the sub-steps of:
constructing a correlation matrix of the full system level test data set, and converting the correlation matrix into an inverse matrix;
Transforming the prediction vector into a transpose matrix;
determining a corresponding deviation coefficient by adopting an inverse matrix, a transposed matrix, a real response vector and a prediction vector;
calculating a multiplication value between the deviation coefficient and the prediction vector;
calculating a difference value between a real response vector corresponding to the full system level test data set and the multiplied value;
determining an optimal linear unbiased predicted value of the deviation model by adopting a difference value, a correlation matrix and an inverse matrix;
in the embodiment of the invention, the correlation matrix R corresponding to the full system level test data set is constructed 1 Which is N 1 ×N 1 Is (x) i ,x j ) Further transformed into an inverse matrix. At the same time, transform the predictive vector into the form of transposed matrixDetermining corresponding deviation coefficient by adopting inverse matrix, transpose matrix, true response vector and predictive vector>Wherein (1)>Is the true response vector for the full system level test.
Further, calculating the multiplication value between the deviation coefficient and the prediction vector, and calculating a deviation model between the multiplication value and the true response vector corresponding to the full-system-level test data set to obtainCombining the correlation matrix and the inverse matrix to determine an optimal linear unbiased predictor of the bias model>Wherein is the correlation vector between the unknown sample and the training sample.
The intermediate predictive proxy model is:
the mean square error is:
wherein,,for the predicted response value, ρ, of the intermediate predicted proxy model n As a deviation factor>Predictive response value for initial predictive proxy model, +.>For optimal linear unbiased predictions of the bias model, MSE n (x) In the form of a mean square error,r is the correlation vector between the unknown sample and the training sample n N for training samples n ×N n Correlation matrix R (x i ,x j )， Is a regression matrix.
S15, updating an initial prediction proxy model according to the optimal linear unbiased prediction value and the deviation coefficient of the deviation model to generate an intermediate prediction proxy model;
further, S15 may include the sub-steps of:
combining the deviation coefficient with the initial predictive proxy model, and adding an optimal linear unbiased predictive value of the deviation model to generate an intermediate predictive proxy model;
calculating a mean square error corresponding to the intermediate prediction proxy model;
the intermediate predictive proxy model is:
the mean square error is:
where x is the input of the proxy model,for the predicted response value, ρ, of the intermediate predicted proxy model n+1 As a deviation factor>Predictive response value for initial predictive proxy model, +.>For optimal linear unbiased predictions of the bias model, MSE n+1 (x) Is mean square error >Is the correlation vector between the unknown full system level test data set and the known full system level test data set, R n+1 N being a known full system level test dataset n+1 ×N n+1 Correlation matrix R (x i ,x j )，Is a regression matrix.
S16, judging whether an unselected full-system-level test data set exists or not;
s17, if so, taking the intermediate prediction proxy model as a new initial prediction proxy model, and jumping to execute the step of selecting a full-system-level test data set corresponding to the current minimum fidelity as a training data set;
and S17, if not, determining the intermediate prediction proxy model at the current moment as a target prediction proxy model.
In the embodiment of the invention, after the intermediate prediction proxy model is obtained, whether an unselected full-system-level test data set exists can be further judged, if so, the selection is executed again to perform iterative updating until the true test data, namely, the full-system-level test data set with 100% of fidelity is selected, if not, the intermediate prediction proxy model at the moment is completely trained, and the intermediate prediction proxy model at the moment can be determined as the target prediction proxy model.
In a specific implementation, the construction of the target prediction proxy model is implemented by the following example:
Full-system different fidelity simulation test data D based on antenna housing ablation electrical property analysis system m M=1, …, n, gradually build the proxy model M n 。
Assuming full system-level simulation experiments with n fidelity (i.e. radome ablation electrical performance simulation analysis systems with different precision), the experimental fidelity is from low to high, D 1 ,D 2 ,...,D n Is the corresponding data set.
1) Build proxy model M 1 And M is as follows 0 Is a fusion framework of (2);
suppose a proxy model M 1 And M is as follows 0 Obeys the following deviation model:
wherein the method comprises the steps ofRepresentation model M 1 The value at variable x, +.>Representing the model M created in step 205 0 Value at variable x, ρ 1 Is the deviation factor, z 1 (x)～GP(0,k 1 (.+ -.)) is a zero-mean gaussian process, and
based on the built agent model M 0 Calculate D 1 Predictive value of input data in (a)
Vector registrationThereby obtaining a deviation model z 1 (x) The value of the optimal linear unbiased prediction (BLUP) is +.>Coefficient->Wherein the method comprises the steps ofR 1 N for training samples 1 ×N 1 Correlation matrix R (x i ,x j )。Is a regression matrix>Is the response value of the full system level test.
Finally, the agent model M is obtained 1 ：Wherein:
2) Fusion of full System level Experimental data D 2 And proxy model M 1 Build agent model M 2 ：The specific procedure is similar to 1).
Data D with different test precision are gradually fused 3 ,D 4 ,...,D n Obtaining a proxy model:
true test data D based on radome ablation electrical property analysis system n+1 Establishing a proxy model M based on multi-source test data n+1 。
3) Build model M n+1 And M is as follows n Is a fusion framework of (2);
hypothesis model M n+1 And M is as follows n Obeys the following deviation model:wherein: z n+1 (x)～GP(0,k n+1 (.+ -.)) and +.>Using a trained proxy model M n A prediction dataset is calculated.
Vector registrationThereby obtaining a deviation model z n+1 (x) The value of the optimal linear unbiased prediction (BLUP) is +.>Deviation factorWherein->R n+1 N for training samples n+1 ×N n+1 Correlation matrix R (x i ,x j )。Is a regression matrix that is a function of the regression matrix,is the response value of the true test. Finally, a consistent agent model of multisource test data based on antenna housing ablation electrical property analysis system is obtained>
Step 207, when ablation input parameters of the radome are received, response prediction is performed through the target prediction proxy model, and an electromagnetic performance prediction response value and a prediction error are generated.
In the embodiment of the invention, after corresponding input parameters of antenna housing ablation are obtained, namely a new test design point x 0 The estimated value can be predicted to be by proxy model calculationIts prediction error (i.e. uncertainty) is
The invention uses Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) to measure the error of the proxy model, and is specifically defined as follows:
Firstly, for a thermal ablation stage, input variables are incoming flow speed, incoming flow static temperature, incoming flow static pressure, surface roughness, material density, material specific heat, surface emissivity coefficient, heat conductivity coefficient and liquid layer viscosity coefficient, output responses are ablation thickness and temperature values, 1000 test points are obtained for the stage through simulation software, a Gaussian process is used for constructing a proxy model, and finally another 400 test points are used for verifying the proxy model of the stage, so that an error table between a predicted value and a true value of the proxy model is shown in table 1.
Secondly, for a thermal wave transmission stage, the input variable is the output variable ablation thickness and temperature value of the thermal ablation stage, the reflection coefficient and the transmission coefficient are influenced together by the external variable, 600 test points are obtained at the stage through simulation software, a Gaussian process is used for constructing a proxy model, and finally another 200 test points are used for verifying the proxy model at the stage, so that an error table between the predicted value and the true value of the proxy model is shown in table 1.
TABLE 1 construction of associated proxy model error Table based on sub-phase trial data
Temperature distribution w 1 Thickness w after ablation 2 Reflection coefficient y 1 Transmission coefficient y 2
MAE 0.0075 0.1704 0.0128 0.0142
RMSE 0.0174 0.3754 0.0193 0.0222
And finally, combining a group of simulation test data of the whole system, and fusing the agent model with the simulation test data. The error table between the predicted value and the true value of the obtained proxy model is shown in table 2:
Table 2 multi-fidelity proxy model error table based on full system level and sub-phase correlation proxy model
Reflection coefficient y 1 Transmission coefficient y 2
MAE 0.0010 9.3691e-4
RMSE 0.0015 0.0013
From the perspective of errors of the proxy model, the associated proxy model constructed based on the sub-stage test data has high enough precision, and reflects the accuracy of the associated proxy model method constructed by the antenna housing ablation system; in addition, the error of the multi-fidelity proxy model constructed based on the test data of different sources is relatively smaller, so that the accuracy and the reliability of the multi-fidelity proxy model method fusing the multi-source test data are demonstrated.
Referring to fig. 3, fig. 3 is a flowchart illustrating steps of a method for constructing a coherence proxy model according to an embodiment of the present invention.
Assuming a full system-level simulation experiment with n fidelity (namely a radome ablation electrical performance simulation analysis system with different precision), the test data precision is from low to high, D 1 ,D 2 ,...,D n Is the corresponding data set.
1) Build proxy model M 1 And M is as follows 0 Is a fusion framework of (2);
suppose a proxy model M 1 And M is as follows 0 Obeys the following deviation model:
wherein the method comprises the steps ofRepresentation model M 1 The value at variable x, +.>Representing the model M created in step 301 0 Value at variable x, ρ 1 Is the deviation factor, z 1 (x)～GP(0,k 1 (.+ -.)) is a zero-mean gaussian process, and
based on the proxy model M established in step 205 0 Calculate D 1 Predictive value of input data in (a)
Vector registrationThereby obtaining a deviation model z 1 (x) The value of the optimal linear unbiased prediction (BLUP) is +.>Coefficient->Wherein->R 1 N for training samples 1 ×N 1 And (5) a correlation matrix.Is a regression matrix>Is the response value of the full system level test.
Finally, the agent model M is obtained 1 ：Wherein:
，
2) Fusion of full System level Experimental data D 2 And proxy model M 1 Build agent model M 2 ：The specific procedure is similar to 1).
Data D with different test precision are gradually fused 3 ,D 4 ,...,D n Obtaining a proxy model:
1) Build model R (x) i ,x j ).M n+1 And M n Is a fusion framework of (2);
hypothesis model M n+1 And M is as follows n Obeys the following deviation model:wherein: z n+1 (x)～GP(0,k n+1 (.+ -.)) and +.>
Using a trained proxy model M n Computing a prediction dataset
Vector registrationThereby obtaining a deviation model z n+1 (x) The value of the optimal linear unbiased prediction (BLUP) is +.>Deviation factorWherein->R n+1 N for training samples n+1 ×N n+1 Correlation matrix R (x i ,x j )。Is a regression matrix>Is the response value of the true test. Finally, a consistent agent model of multisource test data based on antenna housing ablation electrical property analysis system is obtained >Wherein the method comprises the steps of
，
Referring to fig. 4, fig. 4 is a flowchart showing steps for predicting ablation electromagnetic performance of an aircraft radome according to an embodiment of the present invention.
Step 1: and (3) analyzing system influence factors: and analyzing influence factors of each stage of the antenna housing ablation electromagnetic performance analysis system.
Step 2: obtaining test data: the invention considers the following three kinds of test data from different sources, namely, test data generated by respectively carrying out simulation tests at different sub-stagesWherein->Y l 0 Representing input and output of the system as a whole, W l 0 Representing the output of the intermediate layer; secondly, simulation test data of different fidelity of the whole systemFinally, the true test data of the whole system(if practically allowed, such as a full flight test).
Step 3: construction of proxy model
Step 301: constructing a proxy model based on sub-stage test data of the antenna housing ablation electrical property analysis system;
step 302: full-system different fidelity simulation test data D based on antenna housing ablation electrical property analysis system m M=1, …, n, gradually build the proxy model M n 。
1) Build proxy model M 1 And M is as follows 0 Is a fusion framework of (2);
2) Fusion of full System level Experimental dataD 2 And proxy model M 1 Build agent model M 2 ：The specific procedure is similar to 1).
Step 303: true test data D based on radome ablation electrical property analysis system n+1 Establishing a proxy model M based on multi-source test data n+1 。
Step 4: computational prediction of electrical performance of radome ablation system using established proxy models
Using the electrical performance proxy model M of the radome ablation system based on the multi-source test data established in the step 3 n+1 The calculation prediction is performed and gives the uncertainty of the prediction. For a new design of experiment point x 0 The estimated value can be predicted to be by proxy model calculationIts prediction error (i.e. uncertainty) is
In the embodiment of the invention, a test data set of two sub-stages of thermal ablation and thermal wave transmission in the ablation process and a plurality of test data sets of different fidelity of the whole system level are obtained; constructing a minimum fidelity proxy model by using the sub-stage level data set based on an associated Gaussian proxy model method; based on a layered kriging method, gradually fusing full-system-level data sets from low to high according to fidelity, and iteratively updating the proxy model to obtain a target prediction proxy model; when corresponding input parameters of antenna housing ablation are received, response prediction is carried out through a target prediction agent model, and an ablation electromagnetic performance prediction response value and a prediction error corresponding to the antenna housing are obtained. Therefore, based on test data from various sources, a proxy model of the influence of the antenna housing ablation on the electromagnetic performance is constructed more accurately, and further, the quick prediction of the influence of the antenna housing ablation on the electromagnetic performance is realized.
Referring to fig. 5, fig. 5 is a block diagram illustrating a device for predicting ablation electromagnetic performance of an antenna housing of an aircraft according to a third embodiment of the present invention.
The invention also provides a device for predicting the influence of the ablation of the radome of the high-speed aircraft on the electromagnetic performance, which comprises the following steps:
the data set acquisition module 501 is used for acquiring a thermal ablation stage test data set, a thermal wave-transparent stage test data set and a plurality of full-system-level test data sets with different fidelity of the radome of the high-speed aircraft in the ablation process;
the model construction module 502 is configured to construct an initial predictive proxy model with minimum fidelity by adopting a thermal ablation stage test data set and a thermal wave-transparent stage test data set based on an associated gaussian proxy model method;
the model updating module 503 is configured to gradually fuse the full system level test data set from low to high according to fidelity based on a layered kriging method, and iteratively update the initial prediction proxy model to generate a target prediction proxy model;
the prediction module 504 is configured to, when ablation input parameters of the radome are received, perform response prediction through the target prediction proxy model, and generate an electromagnetic performance prediction response value and a prediction error.
Optionally, the model building module 502 is specifically configured to:
When a deterministic test kernel function is received, a Gaussian process is combined with the deterministic test kernel function to respectively construct a thermal ablation stage model and a thermal wave-transparent stage model;
according to the thermal ablation stage model and the thermal wave-transparent stage model, an associated Gaussian agent model method is adopted to construct an uncertainty propagation relation;
substituting the thermal ablation stage test data set and the thermal wave-transparent stage test data set into an uncertainty propagation relation respectively, and calculating a response variance and a response mean;
and generating an initial predictive proxy model with the lowest fidelity by adopting the response variance and the response mean.
Optionally, the model updating module 503 includes:
the deviation model construction submodule is used for constructing a deviation model between the initial prediction agent model and the real response based on the initial prediction agent model;
the data set selecting sub-module is used for selecting the full-system-level test data set corresponding to the current minimum fidelity as a training data set;
the prediction vector calculation operator module is used for calculating a plurality of prediction response values corresponding to the training data set through the initial prediction agent model and constructing a prediction vector;
the parameter calculation sub-module is used for calculating an optimal linear unbiased predicted value and a deviation coefficient of the deviation model based on the predicted vector and a real response vector corresponding to the full-system-level test data set;
The model updating sub-module is used for updating the initial prediction proxy model according to the optimal linear unbiased prediction value and the deviation coefficient of the deviation model to generate an intermediate prediction proxy model;
the judging submodule is used for judging whether an unselected full-system-level test data set exists or not;
the circulation sub-module is used for taking the intermediate prediction proxy model as a new initial prediction proxy model if yes, and performing the step of selecting a full-system-level test data set corresponding to the current minimum fidelity as a training data set in a jumping manner;
and the model determination submodule is used for determining the intermediate prediction agent model at the current moment as a target prediction agent model if not.
Optionally, the parameter calculation submodule is specifically configured to:
constructing a correlation matrix of the full system level test data set, and converting the correlation matrix into an inverse matrix;
transforming the prediction vector into a transpose matrix;
determining a corresponding deviation coefficient by adopting an inverse matrix, a transposed matrix, a real response vector and a prediction vector;
calculating a multiplication value between the deviation coefficient and the prediction vector;
calculating a difference value between a real response vector corresponding to the full system level test data set and the multiplied value;
and determining an optimal linear unbiased predicted value of the deviation model by adopting the difference value, the correlation matrix and the inverse matrix.
Optionally, the model update sub-module is specifically configured to:
combining the deviation coefficient with the initial predictive proxy model, and adding an optimal linear unbiased predictive value of the deviation model to generate an intermediate predictive proxy model; calculating a mean square error corresponding to the intermediate prediction proxy model;
the intermediate predictive proxy model is:
the mean square error is:
where x is the input of the proxy model,for the predicted response value, ρ, of the intermediate predicted proxy model n+1 As a deviation factor>Predictive response value for initial predictive proxy model, +.>For optimal linear unbiased predictions of the bias model, MSE n+1 (x) Is mean square error>Is the correlation vector between the unknown full system level test data set and the known full system level test data set, R n+1 N being a known full system level test dataset n+1 ×N n+1 Correlation matrix R (x i ,x j )，Is a regression matrix. />
It will be clearly understood by those skilled in the art that, for convenience and brevity of description, specific working procedures of the apparatus, modules and sub-modules described above may refer to corresponding procedures in the foregoing method embodiments, which are not described herein again.
In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. A method for predicting the effect of high speed aircraft radome ablation on electromagnetic performance, comprising:
acquiring a thermal ablation stage test data set, a thermal wave transmission stage test data set and a plurality of full-system-level test data sets with different fidelity of an antenna housing of a high-speed aircraft in an ablation process;
Based on an associated Gaussian proxy model method, constructing an initial predictive proxy model with the lowest fidelity by adopting the thermal ablation stage test data set and the thermal wave-transparent stage test data set;
based on a layered kriging method, gradually fusing the full-system-level test data set from low to high according to the fidelity, and iteratively updating the initial prediction proxy model to generate a target prediction proxy model;
and when the ablation input parameters of the radome are received, performing response prediction through the target prediction agent model, and generating an electromagnetic performance prediction response value and a prediction error.
2. The method of claim 1, wherein the step of constructing the initial predictive proxy model of minimum fidelity using the thermal ablation stage test dataset and the thermal wave-transparent stage test dataset based on an associated gaussian proxy model method comprises:
when a deterministic test kernel function is received, a Gaussian process is combined with the deterministic test kernel function to respectively construct a thermal ablation stage model and a thermal wave-transparent stage model;
according to the thermal ablation stage model and the thermal wave-transparent stage model, an associated Gaussian agent model method is adopted to construct an uncertainty propagation relation;
Substituting the thermal ablation stage test data set and the thermal wave-transparent stage test data set into the uncertainty propagation relation respectively, and calculating a response variance and a response mean;
and generating an initial predictive proxy model with the lowest fidelity by adopting the response variance and the response mean.
3. The method of claim 1, wherein the step of generating a target predictive proxy model based on a layered kriging method by gradually fusing the full system level trial data set from low to high in the fidelity and iteratively updating the initial predictive proxy model of the lowest fidelity comprises:
constructing a deviation model between the initial predictive proxy model and a real response based on the initial predictive proxy model;
selecting a full-system-level test data set corresponding to the fidelity which is the smallest currently as a training data set;
calculating a plurality of prediction response values corresponding to the training data set through the initial prediction agent model and constructing a prediction vector;
calculating an optimal linear unbiased predicted value and a deviation coefficient of the deviation model based on the predicted vector and a real response vector corresponding to the full-system-level test data set;
updating the initial prediction proxy model according to the optimal linear unbiased prediction value of the deviation model and the deviation coefficient to generate an intermediate prediction proxy model;
Judging whether the unselected full-system-level test data set exists or not;
if yes, taking the intermediate predictive proxy model as a new initial predictive proxy model, and jumping to execute the step of selecting the full system level test data set corresponding to the current minimum fidelity as a training data set;
if not, determining the intermediate predictive proxy model at the current moment as a target predictive proxy model.
4. A method according to claim 3, wherein the step of calculating an optimal linear unbiased prediction value and a deviation coefficient of the deviation model based on the prediction vector and a true response vector corresponding to the full system level test dataset, comprises:
constructing a correlation matrix of the full system level test data set, and converting the correlation matrix into an inverse matrix;
transforming the prediction vector into a transpose matrix;
determining a corresponding deviation coefficient by adopting the inverse matrix, the transpose matrix, the real response vector and the prediction vector;
calculating a multiplication value between the deviation coefficient and the prediction vector;
calculating a difference value between a real response vector corresponding to the full system level test data set and a multiplication value;
and determining an optimal linear unbiased prediction value of the deviation model by adopting the difference value, the correlation matrix and the inverse matrix.
5. A method according to claim 3, wherein the step of updating the initial predictive proxy model based on the optimal linear unbiased prediction value of the bias model and the bias coefficient, and generating an intermediate predictive proxy model, comprises:
combining the deviation coefficient with the initial predictive proxy model, and adding an optimal linear unbiased predictive value of the deviation model to generate an intermediate predictive proxy model;
calculating a mean square error corresponding to the intermediate prediction proxy model;
the intermediate prediction proxy model is as follows:
the mean square error is:
where x is the input of the proxy model,for the predicted response value, ρ, of the intermediate predicted proxy model n+1 As a deviation factor>Predictive response value for initial predictive proxy model, +.>For optimal linear unbiased predictions of the bias model, MSE n+1 (x) Is mean square error>Is the correlation vector between the unknown full system level test data set and the known full system level test data set, R n+1 N being a known full system level test dataset n+1 ×N n+1 Correlation matrix R (x i ,x j )，Is a regression matrix.
6. A device for predicting the effect of high speed aircraft radome ablation on electromagnetic performance, comprising:
The data set acquisition module is used for acquiring a thermal ablation stage test data set, a thermal wave-transparent stage test data set and a plurality of full-system-level test data sets with different fidelity of the radome of the high-speed aircraft in the ablation process;
the model construction module is used for constructing an initial prediction proxy model with the lowest fidelity by adopting the thermal ablation stage test data set and the thermal wave-transparent stage test data set based on an associated Gaussian proxy model method;
the model updating module is used for gradually fusing the full-system-level test data set from low to high according to the fidelity based on a layered kriging method, and iteratively updating the initial prediction proxy model to generate a target prediction proxy model;
and the prediction module is used for carrying out response prediction through the target prediction agent model when the ablation input parameters of the radome are received, and generating an electromagnetic performance prediction response value and a prediction error.
7. The apparatus of claim 6, wherein the model building module is specifically configured to:
when a deterministic test kernel function is received, a Gaussian process is combined with the deterministic test kernel function to respectively construct a thermal ablation stage model and a thermal wave-transparent stage model;
According to the thermal ablation stage model and the thermal wave-transparent stage model, an associated Gaussian agent model method is adopted to construct an uncertainty propagation relation;
substituting the thermal ablation stage test data set and the thermal wave-transparent stage test data set into the uncertainty propagation relation respectively, and calculating a response variance and a response mean;
and generating an initial predictive proxy model with the lowest fidelity by adopting the response variance and the response mean.
8. The apparatus of claim 6, wherein the model update module comprises:
the deviation model construction submodule is used for constructing a deviation model between the initial prediction agent model and the real response based on the initial prediction agent model;
the data set selecting sub-module is used for selecting the full-system-level test data set corresponding to the current minimum fidelity as a training data set;
the prediction vector calculation operator module is used for calculating a plurality of prediction response values corresponding to the training data set through the initial prediction agent model and constructing a prediction vector;
the parameter calculation sub-module is used for calculating an optimal linear unbiased predicted value and a deviation coefficient of the deviation model based on the predicted vector and a real response vector corresponding to the full-system-level test data set;
The model updating sub-module is used for updating the initial prediction proxy model according to the optimal linear unbiased prediction value of the deviation model and the deviation coefficient to generate an intermediate prediction proxy model;
the judging submodule is used for judging whether the unselected full-system-level test data set exists or not;
the circulation sub-module is used for taking the intermediate prediction agent model as a new initial prediction agent model if yes, and jumping to execute the step of selecting the full system level test data set corresponding to the current minimum fidelity as a training data set;
and the model determination submodule is used for determining the intermediate prediction agent model at the current moment as a target prediction agent model if not.
9. The apparatus of claim 8, wherein the parameter calculation sub-module is specifically configured to:
constructing a correlation matrix of the full system level test data set, and converting the correlation matrix into an inverse matrix;
transforming the prediction vector into a transpose matrix;
determining a corresponding deviation coefficient by adopting the inverse matrix, the transpose matrix, the real response vector and the prediction vector;
calculating a multiplication value between the deviation coefficient and the prediction vector;
Calculating a difference value between a real response vector corresponding to the full system level test data set and a multiplication value;
and determining an optimal linear unbiased prediction value of the deviation model by adopting the difference value, the correlation matrix and the inverse matrix.
10. The apparatus of claim 8, wherein the model update sub-module is specifically configured to:
combining the deviation coefficient with the initial predictive proxy model, and adding an optimal linear unbiased predictive value of the deviation model to generate an intermediate predictive proxy model;
calculating a mean square error corresponding to the intermediate prediction proxy model;
the intermediate prediction proxy model is as follows:
the mean square error is:
where x is the input of the proxy model,for the predicted response value, ρ, of the intermediate predicted proxy model n+1 As a deviation factor>Predictive response value for initial predictive proxy model, +.>For optimal linear unbiased predictions of the bias model, MSE n+1 (x) Is mean square error>Is the correlation vector between the unknown full system level test data set and the known full system level test data set, R n+1 N being a known full system level test dataset n+1 ×N n+1 Correlation matrix R (x i ,x j )，Is a regression matrix.
CN202310342223.5A 2023-03-31 2023-03-31 Method and device for predicting influence of high-speed aircraft radome ablation on electromagnetic performance Active CN116562124B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310342223.5A CN116562124B (en) 2023-03-31 2023-03-31 Method and device for predicting influence of high-speed aircraft radome ablation on electromagnetic performance

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310342223.5A CN116562124B (en) 2023-03-31 2023-03-31 Method and device for predicting influence of high-speed aircraft radome ablation on electromagnetic performance

Publications (2)

Publication Number Publication Date
CN116562124A true CN116562124A (en) 2023-08-08
CN116562124B CN116562124B (en) 2024-02-06

Family

ID=87485136

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310342223.5A Active CN116562124B (en) 2023-03-31 2023-03-31 Method and device for predicting influence of high-speed aircraft radome ablation on electromagnetic performance

Country Status (1)

Country Link
CN (1) CN116562124B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106372278A (en) * 2016-08-19 2017-02-01 电子科技大学 Sensitivity analysis method jointly considering input parameter uncertainty and proxy model uncertainty
US20170352948A1 (en) * 2016-06-02 2017-12-07 The Boeing Company Frequency-selective Surface Composite Structure
CN109786961A (en) * 2018-12-05 2019-05-21 航天特种材料及工艺技术研究所 A kind of high temperature resistant frequency-selective surfaces antenna house and preparation method
WO2021063136A1 (en) * 2019-09-30 2021-04-08 江苏大学 Data-driven high-precision integrated navigation data fusion method
CN114117840A (en) * 2021-10-28 2022-03-01 中国运载火箭技术研究院 Structural performance prediction method based on simulation and test data hybrid drive
CN115395240A (en) * 2022-08-30 2022-11-25 西安电子科技大学 Wave-transparent window switch type liquid metal ATFSS device

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170352948A1 (en) * 2016-06-02 2017-12-07 The Boeing Company Frequency-selective Surface Composite Structure
CN106372278A (en) * 2016-08-19 2017-02-01 电子科技大学 Sensitivity analysis method jointly considering input parameter uncertainty and proxy model uncertainty
CN109786961A (en) * 2018-12-05 2019-05-21 航天特种材料及工艺技术研究所 A kind of high temperature resistant frequency-selective surfaces antenna house and preparation method
WO2021063136A1 (en) * 2019-09-30 2021-04-08 江苏大学 Data-driven high-precision integrated navigation data fusion method
CN114117840A (en) * 2021-10-28 2022-03-01 中国运载火箭技术研究院 Structural performance prediction method based on simulation and test data hybrid drive
CN115395240A (en) * 2022-08-30 2022-11-25 西安电子科技大学 Wave-transparent window switch type liquid metal ATFSS device

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
QIZHE LI等: "Data-Driven Global Sensitivity Analysis Using theArbitrary Polynomial Chaos Expansion Mode", 2022 6TH INTERNATIONAL CONFERENCE ON SYSTEM RELIABILITY AND SAFETY, pages 274 - 278 *
刘泽苁等: "多类型体系贡献率评估的综合问题研究", 《系统工程与电子技术》, vol. 44, no. 5, pages 1572 - 1580 *
宗利平: "复杂条件下头罩电性能分析及综合优化方法研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》, no. 1, pages 032 - 323 *
杨洁颖;吕毅;张春波;郝强;郭世峰;: "飞行器用透波材料及天线罩技术研究进展", 宇航材料工艺, no. 04, pages 12 - 15 *
赵良;刘秀祥;苏汉生;: "高超声速飞行器等离子鞘套相关问题研究与展望", 遥测遥控, no. 05, pages 30 - 34 *

Also Published As

Publication number Publication date
CN116562124B (en) 2024-02-06

Similar Documents

Publication Publication Date Title
Wu et al. Machine-learning-assisted optimization and its application to antenna designs: Opportunities and challenges
Shields et al. A simple and efficient methodology to approximate a general non-Gaussian stationary stochastic process by a translation process
Tamayo et al. Multilevel adaptive cross approximation (MLACA)
Sumant et al. Reduced-order models of finite element approximations of electromagnetic devices exhibiting statistical variability
Blonigan et al. Least-squares shadowing sensitivity analysis of chaotic flow around a two-dimensional airfoil
Yoon et al. Extended particle difference method for weak and strong discontinuity problems: part II. Formulations and applications for various interfacial singularity problems
Cook et al. Robust airfoil optimization and the importance of appropriately representing uncertainty
Jin et al. Quantum simulation for partial differential equations with physical boundary or interface conditions
Szymanski et al. Inverse design of multi-input multi-output 2-D metastructured devices
Sharma et al. Machine learning methods-based modeling and optimization of 3-D-printed dielectrics around monopole antenna
CN114384518B (en) Sea surface SAR image simulation method and device based on Ku wave band actual measurement data
Rapaić et al. Stability regions of fractional systems in the space of perturbed orders
Felsen et al. Electromagnetic Engineering in the 21^{st} Century: Challenges and Perspectives
CN116562124B (en) Method and device for predicting influence of high-speed aircraft radome ablation on electromagnetic performance
Rafiee Alavi et al. RWG MoM‐via‐locally corrected Nyström method in near‐field to far‐field transformation using very‐near‐field measurement
Sun et al. Improved hybrid FEM/MOM combining MLFMA for composite electromagnetic scattering
Tang et al. Pointing error compensation of electro-optical detection systems using Gaussian process regression
Jemai et al. New adaptive multi-expansion frequencies approach for SP-MORe techniques with application to the well-conditioned asymptotic waveform evaluation
CN111901026B (en) Arrival angle estimation method in communication
Wang et al. Towards a unifying computational platform with the node-based meshless method
Warecka et al. Hybrid method analysis of unshielded guiding structures
De Ridder et al. Adaptive frequency sampling using linear Bayesian vector fitting
CN112182739A (en) Aircraft structure non-probability credible reliability topological optimization design method
Tsitsas et al. Analysis of truncated gratings and a novel technique for extrapolating their characteristics to those of infinite gratings
Plaza et al. Assessment of FEM simulations in EMC test setups for small aeronautical platforms

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant