An atmospheric density inverse identification method and system based on aerodynamic load of an aircraft
By conducting sensitivity analysis and constructing neural network models for the external loads of the aircraft, the problem of accurately acquiring atmospheric density data under hypersonic conditions was solved, achieving efficient and accurate density identification and improving related design and safety.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SOUTHEAST UNIV
- Filing Date
- 2023-02-24
- Publication Date
- 2026-04-17
AI Technical Summary
Existing technologies struggle to accurately acquire atmospheric density data at hypersonic speeds, impacting the orbital design of satellites and missiles, the aerodynamic design of high-speed aircraft, and significantly affecting engine design, flight simulation, and the safety of manned spaceflight return.
By conducting sensitivity analysis on the external loads of the aircraft, a surrogate model based on neural networks is constructed. The atmospheric density is then identified by using the responses of external loads such as aerodynamic forces, aerodynamic heat, and aerodynamic noise. This establishes the correlation between atmospheric density and the external loads of the aircraft, enabling efficient and accurate density identification.
Atmospheric density can be efficiently and accurately identified with only small sample simulation calculations, improving the accuracy of orbital and aerodynamic design, enhancing the accuracy of engine design and flight simulation, and ensuring the safety of manned spaceflight return.
Smart Images
Figure CN116227383B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of atmospheric density identification technology, and in particular to a method and system for reverse atmospheric density identification based on aircraft aerodynamic loads. Background Technology
[0002] The density of Earth's atmosphere varies greatly, and standard atmosphere and reference atmosphere are often used in engineering as reference standards for atmospheric density. Although standard atmosphere roughly represents the annual average atmospheric density of mid-latitude regions, and reference atmosphere takes into account the changes in atmospheric density with geographical latitude and season, as well as with time and solar activity, and to some extent represents the dynamic characteristics of atmospheric density, the variation of Earth's atmospheric density is extremely complex, and it is extremely difficult to obtain the true distribution of physical quantities entirely based on theory or limited measured data.
[0003] Traditional air pressure data systems based on pitot tubes and embedded air pressure data systems can provide information such as free-flowing atmospheric pressure, but atmospheric density is derived from altitude data using empirical formulas, which cannot provide accurate atmospheric density data at hypersonic speeds. The accuracy of atmospheric density data directly impacts the orbital design of satellites and missiles, affecting the accuracy of precision-guided weapons. Furthermore, it is crucial for the aerodynamic design of high-speed aircraft, engine design, flight simulation, and the safety of manned spaceflight reentry. Therefore, there is an urgent need to develop efficient and accurate atmospheric density identification methods suitable for high-speed aircraft. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides a method and system for reverse atmospheric density identification based on aircraft aerodynamic loads, with the technical objective of achieving efficient and accurate atmospheric density identification suitable for high-speed aircraft.
[0005] The technical solution adopted in this invention is as follows:
[0006] This application provides a method for reverse identification of atmospheric density based on aircraft aerodynamic loads, including:
[0007] Sensitivity analysis of atmospheric density and external loads on the aircraft was conducted.
[0008] Using atmospheric density and other relevant parameters as analysis parameters, a sample space of analysis parameters is constructed within the set numerical range of the analysis parameters. Based on the sample space, different analysis conditions are sampled and aircraft fluid simulation calculations are carried out under each analysis condition to obtain the external load response of multiple types of aircraft. A dataset of analysis parameters and corresponding external load responses of multiple types of aircraft is established. The dataset is analyzed to obtain the external load type that is most sensitive to changes in atmospheric density.
[0009] Constructing an atmospheric density reverse identification surrogate model based on external aircraft loads:
[0010] Using the external load response most sensitive to changes in atmospheric density under each working condition and other relevant analytical parameters under the corresponding working condition as inputs, and the atmospheric density under the corresponding working condition as outputs, a proxy model between inputs and outputs is established by training, learning and testing based on a neural network method.
[0011] Based on the aforementioned proxy model, atmospheric density is inversely identified through external loads on the aircraft.
[0012] The further technical solution is as follows:
[0013] Establish a dataset containing the analysis parameters and the corresponding aircraft external load response, and analyze the dataset, including:
[0014] A relational model Y = f(X) is established using the analysis parameters as input parameters and the corresponding external load response of the aircraft as output parameters. The expansion equation of the relational model is as follows:
[0015]
[0016] In the above formula, Y represents the external load response of the aircraft, f0 is the constant term in the expansion formula, and X i X j Let f represent the i-th and j-th analysis parameters, respectively. i f represents the contribution of the i-th analysis parameter to the aircraft's external load response when acting alone. ij This represents the contribution of the i-th and j-th analysis parameters to the external load response of the aircraft when they interact; the meanings of the remaining terms follow the same logic.
[0017] Substituting the dataset into the expansion equation and performing variance analysis and normalization on both sides of the equation, we obtain:
[0018]
[0019] In the above formula, S i To analyze parameter X i The main effect index, i.e., the first-order sensitivity index, S ij To analyze parameter X i and X j The coupling effect index;
[0020] The overall sensitivity index of each analytical parameter to atmospheric density is calculated based on the main effect index and the coupling effect index:
[0021]
[0022] In the above formula, S Ti To analyze parameter X i The comprehensive sensitivity index, S ～i To exclude the analysis parameter X i Sensitivity indices for other analytical parameters;
[0023] By comparing the comprehensive sensitivity index of analytical parameters to atmospheric density for different types of external loads, the aircraft external load response that is most sensitive to changes in atmospheric density is obtained.
[0024] The Latin hypersolution method is used to sample the sample space to obtain different analytical conditions.
[0025] Other relevant analytical parameters include atmospheric temperature and flight status parameters.
[0026] External load types include aerodynamic forces, aerodynamic heat, and aerodynamic noise.
[0027] The various types of external load responses of the aircraft obtained through aircraft fluid simulation calculations include external load responses at different locations on the aircraft.
[0028] Another aspect of this application provides an atmospheric density reverse identification system based on aircraft aerodynamic loads, comprising:
[0029] The sensitivity analysis module is used to perform sensitivity analysis of atmospheric density and external loads on aircraft. It uses atmospheric density and other relevant parameters as analysis parameters, constructs a sample space of analysis parameters within the set numerical range of the analysis parameters, samples different analysis conditions based on the sample space, performs fluid simulation calculations under each analysis condition, obtains the response of multiple types of external loads on aircraft, establishes a dataset of analysis parameters and corresponding responses of multiple types of external loads on aircraft, analyzes the dataset, and obtains the type of external load that is most sensitive to changes in atmospheric density.
[0030] The proxy model construction module is used to construct an atmospheric density reverse identification proxy model based on the external load of the aircraft: taking the external load response that is most sensitive to changes in atmospheric density under each working condition and other relevant analysis parameters under the corresponding working condition as inputs, and the atmospheric density under the corresponding working condition as outputs, the module is trained, learned and tested based on the neural network method to establish a proxy model between input and output.
[0031] The identification module is used to perform reverse identification of atmospheric density based on the proxy model through the external loads of the aircraft.
[0032] The beneficial effects of this invention are as follows:
[0033] This invention analyzes and establishes the correlation between atmospheric density and external loads on an aircraft. Only a small sample of simulation calculations on the aircraft are needed to efficiently and accurately identify the atmospheric density through the external loads on the aircraft, which has important engineering application value.
[0034] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. Attached Figure Description
[0035] Figure 1 This is a flowchart illustrating the method described in this application.
[0036] Figure 2 This is a schematic diagram illustrating the sensitivity of different external load types to atmospheric density in embodiments of this application.
[0037] Figure 3 This is a schematic diagram of the atmospheric density identification result obtained from the proxy model in an embodiment of this application. Detailed Implementation
[0038] The specific embodiments of the present invention are described below with reference to the accompanying drawings.
[0039] See Figure 1 This application provides a method for reverse identification of atmospheric density based on aerodynamic loads of an aircraft, including:
[0040] S1. Conduct a sensitivity analysis of atmospheric density and external loads on the aircraft:
[0041] Using atmospheric density and other relevant parameters as analysis parameters, a sample space of analysis parameters is constructed within the set numerical range of the analysis parameters. Based on the sample space, different analysis conditions are sampled and fluid simulation calculations are carried out under each analysis condition to obtain the external load response of multiple types of aircraft. A dataset of analysis parameters and corresponding external load responses of multiple types of aircraft is established. The dataset is analyzed to obtain the external load type that is most sensitive to changes in atmospheric density.
[0042] S2. Construct an atmospheric density reverse identification proxy model based on external loads of the aircraft:
[0043] Using the external load response most sensitive to changes in atmospheric density under each working condition and other relevant analytical parameters under the corresponding working condition as inputs, and the atmospheric density under the corresponding working condition as outputs, a proxy model between inputs and outputs is established by training, learning and testing based on a neural network method.
[0044] S3. The atmospheric density is reverse-identified through the external load of the aircraft using the proxy model.
[0045] The method described in this application analyzes and establishes the correlation between atmospheric density and external loads on an aircraft. It can efficiently and accurately identify atmospheric density through external loads on an aircraft by performing simulation calculations on a small sample of the aircraft, which has significant engineering application value.
[0046] Specifically, in step S1, the parameter vector x = [x1,...,x] is analyzed. i ,...,x n ], where x1 represents atmospheric density, x2-x n Other relevant parameters are indicated, including atmospheric temperature and flight status parameters.
[0047] Based on the range of atmospheric density variation corresponding to the aircraft's flight altitude and the upper and lower limits of other relevant analytical parameters, a sample space for the analytical parameters is defined. N design samples are generated from the sample space of each analytical parameter using Latin hypercube sampling. The parameter vector corresponding to the j-th design sample is then x. j =[x j1 ,...,x ji ,...,x jn ], where x i l ≤x ji ≤x i u , 1≤j≤N, i∈[1,n], n is the number of sample points, and The analysis parameters are x respectively i The upper and lower bounds.
[0048] N design samples represent N analysis conditions. Fluid simulation calculations are performed under each analysis condition to obtain the external load response of the aircraft. The external load response at different locations corresponding to the j-th design sample (analysis condition) is y. j =[y j1 ,...,y ji' ,...,y jm ], where 1≤i'≤m, and then construct the external load response dataset y=[y1,...,y j ,...,y m ], where m is the data sequence number.
[0049] The external loads include aerodynamic forces, aerodynamic heat, and aerodynamic noise.
[0050] Specifically, in step S2, a dataset is established regarding the analysis parameters and the corresponding external load response of the aircraft. This dataset is then analyzed using the Sobol method based on analysis of variance, including:
[0051] A relationship model Y = f(X) is established between the analysis parameters as input parameters and the corresponding external load response of the aircraft as output parameters. The expansion equation of the relationship model is as follows:
[0052]
[0053] In the above formula, Y represents the external load response of the aircraft, f0 is the constant term in the expansion formula, and X i X j Let f represent the i-th and j-th analysis parameters, respectively. i f represents the contribution of the i-th analysis parameter to the aircraft's external load response when acting alone. ij This represents the contribution of the i-th and j-th analysis parameters to the external load response of the aircraft when they interact; the meanings of the remaining terms follow the same logic.
[0054] Substituting the dataset obtained in step S1 into the expansion equation, and performing variance analysis and normalization on both sides of the equation, we can obtain:
[0055]
[0056] In the above formula, S i To analyze parameter X i The main effect index, i.e., the first-order sensitivity index, S ij To analyze parameter X i and X j The coupling effect index;
[0057] The overall sensitivity index of each analytical parameter to atmospheric density is calculated based on the main effect index and the coupling effect index:
[0058]
[0059] In the above formula, S Ti To analyze parameter X i The comprehensive sensitivity index to atmospheric density, S ～i To exclude the analysis parameter X i The sensitivity index of other analytical parameters to atmospheric density shows that S Ti It is the analysis parameter X i The sum of the main effect index and the coupling effect index with the remaining analytical parameters;
[0060] By comparing the comprehensive sensitivity index of analytical parameters to atmospheric density for different types of external loads, the aircraft external load response that is most sensitive to changes in atmospheric density is obtained.
[0061] The following specific embodiments further illustrate the atmospheric density reverse identification method based on aircraft aerodynamic loads according to the present application.
[0062] We selected an atmospheric density variation range of 20km to 50km and a typical aircraft wing and rudder model as the subjects. The specific steps are as follows:
[0063] (1) Using atmospheric density x1, and atmospheric temperature x2, flight speed x3 and flight angle of attack x4 as other relevant analysis parameters, construct the analysis parameter vector x = [x1, x2, x3, x4].
[0064] The design space for each analysis parameter is set separately.
[0065] Within the design space, 100 design samples are generated using Latin hypercube sampling. Let x be the vector of analytical parameters under the working condition corresponding to the j-th design sample. j =[x j1 ,x j2 ,x j3 ,x j4 ], where x i l ≤x ij ≤x i u , 1≤j≤100.
[0066] (2) Establish a fluid simulation analysis model for the aircraft wing and rudder model, conduct fluid simulation analysis based on simulation software, and obtain the external load response at different locations as y. j =[y j1 ,...,y ji' ,...,y jm ], where j is the j-th design sample, 1≤i'≤m, and m is the number of sample points, construct the response dataset y=[y1,...,y j ,...,y n ], where n is the total number of design samples, and then an atmospheric density and external load response dataset is constructed;
[0067] (3) Based on the dataset, the sensitivity of different types of external loads to atmospheric density changes was analyzed using the Sobol method based on analysis of variance. The analysis included structures such as... Figure 2 As shown in the figure, the sensitivity of different types of external loads to atmospheric density is as follows: aerodynamic force, aerodynamic heat, and aerodynamic noise are 64%, 9%, and 27% respectively. Aerodynamic force is the most sensitive to changes in atmospheric density; therefore, there is a strong correlation between aerodynamic force and atmospheric density, which can be used as a reverse identification of atmospheric density.
[0068] (4) Combine the aerodynamic data and other relevant analysis parameters [y1, x2, x3, x4] under 100 operating conditions. TThe atmospheric density [x1] under the corresponding operating condition will be used as input to the neural network. T As the output of the neural network, the surrogate model is trained, learned, and tested using the BP (Back Propagation) neural network method, with 70% of the samples used for training, 15% for validation, and 15% for testing. The number of hidden neuron layers is set to 10, and the model is trained using the Levenberg-Marquardt method to build the surrogate model.
[0069] This surrogate model can then be used for reverse identification of atmospheric density: by inputting external load data into the surrogate model, the corresponding atmospheric density can be obtained. For example... Figure 3 The diagram shown is a schematic representation of the atmospheric density identification result based on the surrogate model in this embodiment. Figure 3 The straight line represents the actual value, and the point represents the recognition result obtained by the proxy model in this embodiment. It can be seen that the recognition result of this embodiment is accurate and has high precision.
[0070] This application also provides an atmospheric density reverse identification system based on aircraft aerodynamic loads, including:
[0071] The sensitivity analysis module is used to perform sensitivity analysis of atmospheric density and external loads on aircraft. It uses atmospheric density and other relevant parameters as analysis parameters, constructs a sample space of analysis parameters within the set numerical range of the analysis parameters, samples different analysis conditions based on the sample space, performs fluid simulation calculations under each analysis condition, obtains the response of multiple types of external loads on aircraft, establishes a dataset of analysis parameters and corresponding responses of multiple types of external loads on aircraft, analyzes the dataset, and obtains the type of external load that is most sensitive to changes in atmospheric density.
[0072] The proxy model construction module is used to construct an atmospheric density reverse identification proxy model based on the external load of the aircraft: taking the external load response that is most sensitive to changes in atmospheric density under each working condition and other relevant analysis parameters under the corresponding working condition as inputs, and the atmospheric density under the corresponding working condition as outputs, the module is trained, learned and tested based on the neural network method to establish a proxy model between input and output.
[0073] The identification module is used to perform reverse identification of atmospheric density based on the proxy model through the external loads of the aircraft.
[0074] It will be understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. An atmospheric density inverse identification method based on aircraft aerodynamic load, characterized in that, include: Sensitivity analysis of atmospheric density and external loads on the aircraft was conducted. Using atmospheric density and other relevant analytical parameters as analytical parameters, a sample space of analytical parameters is constructed within the set numerical range of analytical parameters. Based on the sample space, different analytical conditions are sampled and obtained. Aircraft fluid simulation calculations are carried out under each analytical condition to obtain the external load response of multiple types of aircraft. A dataset of analytical parameters and corresponding external load responses of multiple types of aircraft is established. The dataset is analyzed to obtain the external load type that is most sensitive to changes in atmospheric density. Constructing an atmospheric density reverse identification surrogate model based on external aircraft loads: Using the external load response most sensitive to changes in atmospheric density under each working condition and other relevant analytical parameters under the corresponding working condition as inputs, and the atmospheric density under the corresponding working condition as outputs, a proxy model between input and output is established by training, learning and testing based on a neural network method. Based on the aforementioned proxy model, atmospheric density is inversely identified through external loads on the aircraft; The process of establishing a dataset of analysis parameters and corresponding external load responses of various types of aircraft, and analyzing the dataset, includes: A relational model is established using the analysis parameters as input parameters and the corresponding external load response of the aircraft as output parameters. Y = f ( X The expansion equation of the relational model is: ， In the above formula, Represents the aircraft's external load response. f 0 For the constant term in the expansion, X i , X j They represent the first i The first analysis parameter and the first j One analysis parameter, f i Indicates the first i The contribution of each analytical parameter to the aircraft's external load response when acting alone. f ij Indicates the first i The first analysis parameter and the first j The contribution of the interaction of the analysis parameters to the external load response of the aircraft, and the meaning of the remaining terms is deduced accordingly; Substituting the dataset into the expansion equation and performing variance analysis and normalization on both sides of the equation, we obtain: ， In the above formula, S i For analysis parameters X i The main effect index, i.e., the first-order sensitivity index, S ij For analysis parameters X i and X j The coupling effect index; The overall sensitivity index of each analytical parameter to atmospheric density is calculated based on the main effect index and the coupling effect index: ， In the above formula, S Ti is the overall sensitivity index of the analysis parameters X i S ~i is the sensitivity index of the remaining analysis parameters except for the analysis parameter X i By comparing the comprehensive sensitivity index of analytical parameters to atmospheric density for different types of external loads, the aircraft external load response that is most sensitive to changes in atmospheric density is obtained.
2. The atmospheric density back-identification method based on aircraft aerodynamic load according to claim 1, characterized in that, The Latin hypersolution method is used to sample the sample space to obtain different analytical conditions.
3. The atmospheric density inverse identification method based on aircraft aerodynamic load according to claim 1, characterized in that, Other relevant analytical parameters include atmospheric temperature and flight status parameters.
4. The atmospheric density reverse identification method based on aircraft aerodynamic loads according to claim 1, characterized in that, External load types include aerodynamic forces, aerodynamic heat, and aerodynamic noise.
5. The atmospheric density inverse identification method based on aircraft aerodynamic load according to claim 1, wherein, The various types of external load responses of the aircraft obtained through aircraft fluid simulation calculations include external load responses at different locations on the aircraft.
6. A system for atmospheric density reverse identification based on aircraft aerodynamic loads according to the method of any one of claims 1 to 5, characterized in that, include: The sensitivity analysis module is used to perform sensitivity analysis of atmospheric density and external loads on aircraft. It uses atmospheric density and other relevant analytical parameters as analytical parameters, constructs a sample space of analytical parameters within the set numerical range of the analytical parameters, samples different analytical conditions based on the sample space, performs fluid simulation calculations under each analytical condition, obtains the responses of multiple types of external loads on aircraft, establishes a dataset of analytical parameters and corresponding responses of multiple types of external loads on aircraft, analyzes the dataset, and obtains the type of external load that is most sensitive to changes in atmospheric density. The proxy model construction module is used to construct an atmospheric density reverse identification proxy model based on the external load of the aircraft: taking the external load response that is most sensitive to changes in atmospheric density under each working condition and other relevant analysis parameters under the corresponding working condition as inputs, and the atmospheric density under the corresponding working condition as outputs, the module is trained, learned and tested based on the neural network method to establish a proxy model between input and output. The identification module is used to perform reverse identification of atmospheric density based on the proxy model through the external loads of the aircraft.
Citation Information
Patent Citations
PLS local fitting-based SCR reactor information rapid acquisition method
CN109766577A
Aircraft force load online prediction method based on multi-source data fusion
CN113821869A