Aerodynamic load simulation calculation system and method in aircraft ground vibration test
The non-stable aerodynamic model is fitted through the Kriging agent model, and the calculation accuracy problem of the time-varying structure of the dynamic characteristics in the aircraft ground vibration test is solved, efficient simulation of the time-varying structure is achieved, and the scope of test application is expanded.
Patent Information
- Application Number
- CN202210770406.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-30
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2042-06-30
AI Technical Summary
The existing aircraft ground vibration test simulation technology cannot effectively deal with the time-varying structure of dynamic characteristics, resulting in the non-stable aerodynamic calculation method no longer applicable, affecting the calculation accuracy and application range.
Using a pneumatic load simulation calculation system based on the Kriging agent model, a non-stable aerodynamic model at multiple discrete time points is established through the time freezing assumption, and fitted with the Kriging agent model to obtain a non-stable aerodynamic rapidity suitable for time-varying structures. Computational model, combining vibration exciters and sensors for real-time loading and verification.
The calculation accuracy of the time-varying structure of dynamic characteristics is improved, the application scope of ground flutter simulation test is expanded, the requirements of real-time and high efficiency are met, and the processing capability of the time-varying flutter system is enhanced.
Smart Images

Figure CN114970402B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of aircraft vibration testing, and in particular to a system and method for simulating and calculating aerodynamic loads in an aircraft ground vibration test. Background Art
[0002] Aircraft ground vibration test simulation technology is a semi-physical simulation flutter verification technology developed in recent years that uses real structures as verification objects. It has the advantages of low risk, short cycle and high accuracy. This technology uses sensors to measure the vibration response signal of the structure, substitutes it into the pre-calculated unsteady aerodynamic model to obtain the equivalent aerodynamic force of the structure, and then uses an exciter to load the equivalent aerodynamic force on the test structure in real time, forming a closed-loop system to simulate the structural flutter phenomenon on the ground. By applying initial disturbances to the test structure and observing its response, flutter boundary test verification for the real structure can be carried out.
[0003] In the above-mentioned aircraft ground vibration test simulation steps, considering the frequency of flutter of conventional structures, the test system should complete a cycle calculation period of no more than 0.002 seconds. In order to meet the demand for computational efficiency of the aircraft ground vibration test simulation system, the frequency domain aerodynamic influence coefficient matrix is reduced and condensed according to the structural dynamic characteristics during the aircraft ground vibration test simulation design, and then the rational function approximation and Laplace transform are used to obtain the time domain unsteady aerodynamic model composed of several low-order matrices.
[0004] A major premise for the above modeling method to be valid is that the dynamic characteristics of the structure do not change. In the actual aircraft structure, whether it is the influence of the control surface deflection or the time-varying characteristics of the structural thermal mode caused by the aerodynamic thermal effect, the existing unsteady aerodynamic calculation method will no longer be fully applicable.
[0005] The present invention proposes a system and method for simulating and calculating aerodynamic loads in an aircraft ground vibration test based on a Kriging proxy model. After establishing an unsteady aerodynamic model at multiple discrete time points based on the time freezing assumption, the present invention uses a Kriging proxy model to fit and model the unsteady aerodynamic model at discrete time points, so that the newly obtained aerodynamic model has higher calculation accuracy for structures with time-varying dynamic characteristics while ensuring the system operation efficiency, effectively expanding the application scope of ground flutter simulation tests. Summary of the invention
[0006] In view of the defect that the current ground flutter simulation test can only be applied to steady systems, the present invention provides a fast calculation method for unsteady aerodynamic forces that can be applied to structures with time-varying dynamic characteristics, effectively improving the current ground flutter simulation test technology and promoting the progress of its engineering application.
[0007] To solve the above problems, the technical solution of the present invention is as follows:
[0008] The present invention provides a pneumatic load simulation calculation system in an aircraft ground vibration test, comprising:
[0009] A computing subsystem for performing aerodynamic load simulation of an aircraft ground vibration test, the computing subsystem comprising: a loading controller for controlling the operation of the ground vibration test through a control signal and equipped with an unsteady aerodynamic fast calculation model, the loading controller also being equipped with a flutter boundary tracking module for automatically adjusting the wind speed in the unsteady aerodynamic fast calculation model according to a measured test piece response signal change trend, so as to achieve real-time identification and recording of a flutter critical speed,
[0010] A ground vibration test subsystem for verifying the unsteady aerodynamic fast calculation model, the ground vibration test subsystem comprising: an exciter located at the bottom of the test piece and providing a flutter force to the test piece, a laser displacement sensor and an acceleration sensor arranged at a control point of the test piece on the test piece, a heater arranged above the test piece, a power amplifier for amplifying a control signal,
[0011] A semi-physical simulation subsystem is used to convert the equivalent aerodynamic force calculated by the unsteady aerodynamic fast calculation model into an exciter control signal. The semi-physical simulation subsystem is connected to Matlab.simulink through NI VeriStand simulation configuration software.
[0012] The present invention also provides a method for simulating and calculating aerodynamic loads in an aircraft ground vibration test. The method is based on the above-mentioned system for simulating and calculating aerodynamic loads in an aircraft ground vibration test, and includes the following steps:
[0013] S1. Based on the time freezing assumption, the unsteady aerodynamic fast calculation model for multiple discrete time points is fitted with a Kriging proxy model to obtain an unsteady aerodynamic fast calculation model suitable for the entire time-varying process of the aircraft ground vibration test;
[0014] S2. The unsteady aerodynamic fast calculation model established in step S1 is tested and verified by an actual aircraft ground vibration test, which specifically includes the following steps:
[0015] S2-1. The interpolation points between the aerodynamic plane and the structure in the flutter calculation of the aircraft ground vibration test are condensed, and the optimal interpolation points are selected by optimization method. The optimal interpolation points are used as the control points of the test piece.
[0016] S2-2, select several sampling time points, import the data of several sampling time points into step S1, and obtain a fast calculation model of unsteady aerodynamic force suitable for the control point of the test piece,
[0017] S2-3. Build a ground vibration test subsystem in the laboratory, arrange a heater above the test piece, fix an exciter and a power amplifier electrically connected to the exciter at the bottom of the test piece, arrange a laser displacement sensor and an acceleration sensor at the control point of the test piece on the test piece, and electrically connect the power amplifier, laser displacement sensor, and acceleration sensor to the loading controller.
[0018] S2-4, the acquired control point response signal of the test piece is transmitted to the loading controller through the laser displacement sensor and the acceleration sensor, the loading controller substitutes the control point response signal of the test piece into the unsteady aerodynamic fast calculation model established in step S2-1, the unsteady aerodynamic fast calculation model calculates the equivalent aerodynamic force of the test piece in the time-varying process of the aircraft ground vibration test, and transmits the equivalent aerodynamic force to the semi-physical simulation subsystem, the semi-physical simulation subsystem is connected to Matlab.simulink through NI VeriStand simulation configuration software, and the equivalent aerodynamic force of the test piece is converted into an exciter control signal, the loading controller controls the exciter through the exciter control signal to load the equivalent aerodynamic force on the test piece in real time, and returns to step S2-2 after loading is completed until the equivalent simulation process of the aircraft vibration test is completed, during which the loading controller records the flutter critical speed in the equivalent simulation process of the aircraft vibration test,
[0019] S2-5. Conduct an actual aircraft ground vibration test through the ground vibration test subsystem, and use the flutter boundary tracking module to record the flutter critical speed of the test piece in the time-varying history of the aircraft ground vibration test, and compare it with the flutter critical speed of the aircraft ground vibration test time-varying history calculated by the unsteady aerodynamic force rapid calculation model in step S2-4 to determine the accuracy of the unsteady aerodynamic force rapid calculation model. If the accuracy of the unsteady aerodynamic force rapid calculation model meets the requirements, obtain the unsteady aerodynamic force rapid calculation model after test verification. If it does not meet the requirements, encrypt the sampling time points and return to step S2-2 until the accuracy of the unsteady aerodynamic force rapid calculation model meets the requirements.
[0020] S3. Use the unsteady aerodynamic force rapid calculation model verified by the experiment in step S2 to simulate and calculate the aerodynamic loads in the aircraft ground vibration test.
[0021] Furthermore, the optimal interpolation points in step S2-1 are: the excitation point and the vibration pickup point.
[0022] Furthermore, step S1 specifically includes the following steps:
[0023] S1-1. Establish the calculation formula for the unsteady aerodynamic force borne by the control point of the test piece.
[0024] S1-2. Establish the calculation model of the unsteady aerodynamic force that the control points of the test piece should be subjected to at discrete time points.
[0025] S1-3. Establish the corresponding random function by combining the Kriging proxy model
[0026] Kriging surrogate model setting A ij (t) is the corresponding random function The minimum unbiased estimation result of the mean square error is: It consists of two parts: a global approximate model based on the sampling time point data and a random process. The expression formula is as follows:
[0027]
[0028] In the above formula, is the corresponding random function, g(t) is the known basis function, β is the basis function coefficient, βg(t) is the global approximate model obtained by solving the sampling points in the design space, and Z(t) is the model with mean 0 and variance The static random process function of
[0029] S1-4. Establish a covariance calculation model for a static random process function between any two sampling time points
[0030] Within the time range defined by the aircraft ground vibration test, the covariance calculation model of the static random process function between any two sampling time points is as follows:
[0031]
[0032] In the above formula, is the variance, Z(t1) is the static random process function at the first sampling time point, Z(t2) is the static random process function at the second sampling time point, t1 is the first sampling time point, t2 is the second sampling time point, R(t1, t2) is the correlation function, and the correlation function is in the form of a Gaussian function or a cubic spline function.
[0033] S1-5. Establish a fast calculation model for unsteady aerodynamic forces, that is, a calculation model for the predicted value of the Kriging proxy model at any time t
[0034] The formula for defining the sample point data column vector of the form (3) is constructed by formula (4) as follows:
[0035] [A ij (t)] s =[A ij (t1)A ij (t2)…A ij (t n )] T =βG+z (5)
[0036] In the above formula, [Aij (t)] s is the column vector of sample point data, β is the basis function coefficient, [A ij (t1)A ij (t2)…A ij (t n )] is the sample point data of n sampling time points, G is the basis function column vector, G = [g(t1)g(t2)…g(t n )] T , z is a random function column vector, z=[Z(t1), Z(t2)…Z(t n )] T ,
[0037] According to the correlation function R(t1, t2) in formula (4), the correlation matrix of the correlation between n time sampling points is defined as and the correlation vector r representing the correlation between any time point and all sampling time points t , the correlation matrix The calculation formula is as follows:
[0038]
[0039] In the above formula, is the correlation matrix of the correlations between n time sampling points, t i is the i-th sampling time point among n time sampling points, t j is the jth sampling time point among n time sampling points, R(t i , t j ) is the correlation function between the i-th sampling time point and the j-th sampling time point,
[0040] Matrix function at time nodes other than sampling time points Obtained by linear weighting of the function values at known sampling points:
[0041]
[0042] In the above formula, is the weighted row vector, is a time node function outside the sampling time point, [A ij (t)] s is the column vector of sample point data,
[0043] In order to meet the requirement of unbiased estimation E(Z(t))=0 of the Kriging surrogate model, the calculation formula for the relationship between the basis function and the basis function column vector can be obtained according to formula (5) and formula (8):
[0044]
[0045] In the above formula, G is the basis function column vector, g T (t) is the row vector of basis functions, is the weighted row vector, β is the basis function coefficient, E is the mathematical expectation, is a time node function outside the sampling time point, [A ij (t)] s is the column vector of sample point data,
[0046] According to formula (3), formula (8) and formula (9), the prediction value error calculation formula of the Kriging proxy model is as follows:
[0047]
[0048] In the above formula, is the function value of the time node outside the sampling time point, A ij (t) The corresponding random function obtained by the Kriging surrogate model, A ij (t) is an element in the matrix of any time-varying aerodynamic influence coefficient, Z is a random function column vector, and Z(t) is a function with a mean of 0 and a variance of The static random process function of is the weighted row vector,
[0049] According to the Kriging proxy model, the predicted value meets the requirements of unbiased estimation and minimum mean square error, that is, T c t =g(t), find the weighted column vector c that satisfies the minimum mean square error t To this end, the Lagrangian multiplier λ is introduced to construct the Lagrangian function H(c t ,λ) are as follows:
[0050]
[0051] In the above formula, H(c t ,λ) is the Lagrangian function, is the variance, c t is the weighted column vector, is the weighted row vector, r t is the correlation vector, is the correlation matrix of the correlation between n time sampling points, g(t) is the known basis function, λ is the Lagrange multiplier, G T is the row vector of basis functions,
[0052] The weighted column vector c in formula (12) is t And Lagrange multiplier λ to find the partial derivative, and get the weighted column vector c that meets the requirements t The calculation formula is:
[0053]
[0054] In the above formula, is the correlation matrix of the correlations between n time sampling points, r t is the correlation vector, g(t) is the known basis function, G is the basis function column vector, G T is the row vector of basis functions,
[0055] Substituting formula (13) into formula (8) yields the unsteady aerodynamic fast calculation model, that is, the unsteady aerodynamic prediction value calculation model of the Kriging proxy model at any time t is as follows:
[0056]
[0057] In the above formula, is a time node function outside the sampling time point, [A ij (t)] s is the column vector of sample point data, is the correlation matrix of the correlations between n time sampling points, r t is the correlation vector, g(t) is the known basis function, G is the basis function column vector, G T is the row vector of basis functions.
[0058] From formula (5) and formula (13), we can see that in formula (14), except for c t r in the expression t Other parameters are only related to the sampling time point, so they can be calculated in advance, while r t It is necessary to substitute the current time t into the relevant function formula (7) to solve it. The input parameter of the model is time. Therefore, based on formula (14), the aerodynamic force matching the structural characteristics at any time can be obtained.
[0059] Furthermore, step S1-1 specifically includes the following contents:
[0060] The unsteady aerodynamic force borne by the control point of the test piece is calculated based on the structural vibration response of the test piece in the aircraft ground vibration test. The calculation formula for the unsteady aerodynamic force borne by the control point of the test piece is as follows:
[0061] f s =q ∞ A(k)z s (1)
[0062] In the above formula, f s is the unsteady aerodynamic force that the control point of the test piece should bear, q ∞is the incoming flow pressure, A(k) is the reduced-order aerodynamic influence coefficient matrix, A(k) is obtained by calculating the unsteady aerodynamic theory matching the Mach number and reducing the order according to the structural dynamic characteristics, Z s It represents the structural vibration displacement of the test piece, and k is the reduction frequency.
[0063] Furthermore, the calculation formula of the reduction frequency k is as follows:
[0064]
[0065] In the above formula, k is the reduced frequency, ω is the circular frequency of simple harmonic motion, b is the reference length. For aircraft wing structures, the reference length is half of the root chord length, and V is the incoming flow velocity.
[0066] Preferably, step S1-2 specifically includes the following contents:
[0067] Since the actual aircraft ground vibration test requires the time domain relationship of the unsteady aerodynamic force that the control point of the test piece should bear, formula (1) is extended to the Laplace domain through the rational function approximation of the minimum state method, and then the calculation model of the unsteady aerodynamic force that the control point of the test piece should bear at discrete time points is obtained through the Laplace inverse transform as follows:
[0068]
[0069] In the above formula, f s (t) is the unsteady aerodynamic force that the control point of the test piece should bear related to time, A0 is the displacement aerodynamic force influence coefficient matrix, A1 is the velocity aerodynamic force influence coefficient matrix, A2 is the acceleration aerodynamic force influence coefficient matrix, B1 is the first auxiliary aerodynamic force influence coefficient matrix, B2 is the second auxiliary aerodynamic force influence coefficient matrix, B3 is the third auxiliary aerodynamic force influence coefficient matrix, Z S is the control point displacement, is the control point speed, is the control point acceleration, q ∞ is the incoming flow pressure, V is the incoming flow velocity, b is the reference length. For aircraft wing structures, the reference length is half of the root chord length.
[0070] Since the modeling method of formula (2) can only be applied to time-invariant systems, and the values of the aerodynamic influence coefficient matrix at each time node in some actual aircraft ground vibration tests will change, it is necessary to introduce the Kriging proxy model on the basis of formula (2) and replace the solution process of formula (1) and formula (2) with simple matrix operations to realize the rapid calculation of unsteady aerodynamic forces for structures with time-varying dynamic characteristics.
[0071] Preferably, in step S1-5:
[0072] Correlation vector r t The calculation formula is as follows:
[0073] r t =[R(t1, t)R(t2, t)…R(t n , t)] T (7)
[0074] In the above formula, t is any sample point among the n time sampling points, t∈[t1 t2 … t n ],r t is the correlation vector, R(t1, t) is the correlation function between any sample point t among the n time sampling points and the first sampling time point, R(t2, t) is the correlation function between any sample point t among the n time sampling points and the second sampling time point, R(t n , t) is the correlation function between any sample point t among the n time sampling points and the nth sampling time point.
[0075] Further preferably, step S1-5 further includes:
[0076] According to the Kriging proxy model prediction value meets the requirements of unbiased estimation and minimum mean square error, the mean square error calculation formula of the Kriging proxy model is obtained by formula (10) as follows:
[0077]
[0078] In the above formula, is the pre-mean square error of the Kriging surrogate model, is a time node function outside the sampling time point, A ij (t) The corresponding random function obtained by the Kriging surrogate model, A ij (t) is an element in any time-varying aerodynamic influence coefficient matrix, is the weighted row vector, E is the mathematical expectation, is the variance, c t is the weighted column vector, r t is the correlation vector, is the correlation matrix of the correlations between n time sampling points.
[0079] The beneficial effects of the present invention are:
[0080] (1) The present invention introduces the Kriging proxy model into the ground flutter simulation test, obtains the sampling time point data based on the time freezing assumption, and proposes a complete modeling method based on the sampling time point data, thereby obtaining a fast calculation method for unsteady aerodynamic forces that can be applied to structures with time-varying dynamic characteristics, thereby enhancing the processing capability of such test systems for time-varying flutter systems;
[0081] (2) The working module required by the present invention is only to add several low-order matrix operations on the basis of the unsteady aerodynamic module adopted by the existing structural time-invariant system, which has little impact on the computational efficiency of the ground flutter simulation test calculation subsystem and can meet the requirements of the ground flutter simulation test for the real-time performance of aerodynamic calculation. In addition, since the Kriging model has obvious advantages in dealing with nonlinear systems, it has a high fitting accuracy when calculating unsteady aerodynamics, which can ensure that a high unsteady aerodynamic calculation accuracy is maintained during the time-varying process of the test piece. BRIEF DESCRIPTION OF THE DRAWINGS
[0082] Figure 1 is a system structure diagram of the aerodynamic load simulation calculation system in the aircraft ground vibration test in Example 1;
[0083] Figure 2 This is a flow chart of a method for simulating calculation of aerodynamic loads in an aircraft ground vibration test in Example 2;
[0084] Figure 3 is a flow chart of step S1 in Example 2;
[0085] Figure 4 is a structural diagram of a calculation model of unsteady aerodynamic forces to be borne by the control points of the test piece at discrete time points in Example 2;
[0086] Figure 5 This is a schematic diagram of the optimal interpolation point position in Example 2. DETAILED DESCRIPTION
[0087] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0088] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings, and "multiple" generally includes at least two.
[0089] Example 1
[0090] This embodiment is a simulation calculation system for aerodynamic loads in an aircraft ground vibration test. Figure 1 As shown, including:
[0091] A computing subsystem for performing aerodynamic load simulation of an aircraft ground vibration test, the computing subsystem comprising: a loading controller for controlling the operation of the ground vibration test through a control signal and equipped with an unsteady aerodynamic fast calculation model, the loading controller also being equipped with a flutter boundary tracking module for automatically adjusting the wind speed in the unsteady aerodynamic fast calculation model according to a measured test piece response signal change trend, so as to achieve real-time identification and recording of a flutter critical speed,
[0092] The model calculation formula of the unsteady aerodynamic fast calculation model, that is, the calculation model of the unsteady aerodynamic prediction value of the Kriging proxy model at any time t is as follows:
[0093]
[0094] In the above formula, is a time node function outside the sampling time point, [A ij (t)] s is the column vector of sample point data, is the correlation matrix of the correlations between n time sampling points, r t is the correlation vector, g(t) is the known basis function, G is the basis function column vector, G T is the row vector of basis functions,
[0095] A ground vibration test subsystem for verifying a fast calculation model of unsteady aerodynamic forces, the ground vibration test subsystem comprising: an exciter electrically connected to a loading controller, located at the bottom of a test piece and providing a flutter force to the test piece, a laser displacement sensor and an acceleration sensor arranged at a control point of the test piece on the test piece, a heater arranged above the test piece, a power amplifier for amplifying a control signal,
[0096] A semi-physical simulation subsystem is used to convert the equivalent aerodynamic force calculated by the unsteady aerodynamic fast calculation model into an exciter control signal. The semi-physical simulation subsystem is connected to Matlab.simulink through NI VeriStand simulation configuration software.
[0097] Example 2
[0098] This embodiment is a method for simulating and calculating aerodynamic loads in an aircraft ground vibration test, based on the aerodynamic load simulation and calculation system in an aircraft ground vibration test in Embodiment 1. Figure 2 As shown, the following steps are included:
[0099] S1. Based on the time freezing assumption, the unsteady aerodynamic fast calculation model for multiple discrete time points is fitted with a Kriging proxy model to obtain an unsteady aerodynamic fast calculation model suitable for the entire time-varying process of the aircraft ground vibration test, such as Figure 3 As shown, the specific steps include:
[0100] S1-1. Establish the calculation formula for the unsteady aerodynamic force borne by the control point of the test piece
[0101] The unsteady aerodynamic force borne by the control point of the test piece is calculated based on the structural vibration response of the test piece in the aircraft ground vibration test. The calculation formula for the unsteady aerodynamic force borne by the control point of the test piece is as follows:
[0102] f s =q ∞ A(k)z s (1)
[0103] In the above formula, f s is the unsteady aerodynamic force that the control point of the test piece should bear, q ∞ is the incoming flow pressure, A(k) is the reduced-order aerodynamic influence coefficient matrix, A(k) is obtained by calculating the unsteady aerodynamic theory matching the Mach number and reducing the order according to the structural dynamic characteristics, Z S represents the structural vibration displacement of the test piece, k is the reduction frequency, and the calculation formula of the reduction frequency k is as follows:
[0104]
[0105] In the above formula, k is the reduced frequency, ω is the circular frequency of simple harmonic motion, b is the reference length. For aircraft wing structures, the reference length is half of the root chord length, V is the incoming flow velocity,
[0106] S1-2. Establishment of the calculation model of the unsteady aerodynamic force that the control points of the test piece should be subjected to at discrete time points
[0107] Since the actual aircraft ground vibration test requires the time domain relationship of the unsteady aerodynamic force that the control point of the test piece should bear, formula (1) is extended to the Laplace domain through the rational function approximation of the minimum state method, and then the calculation model of the unsteady aerodynamic force that the control point of the test piece should bear at discrete time points is obtained through the Laplace inverse transform as follows:
[0108]
[0109] In the above formula, f s(t) is the unsteady aerodynamic force that the control point of the test piece should bear related to time, A0 is the displacement aerodynamic force influence coefficient matrix, A1 is the velocity aerodynamic force influence coefficient matrix, A2 is the acceleration aerodynamic force influence coefficient matrix, B1 is the first auxiliary aerodynamic force influence coefficient matrix, B2 is the second auxiliary aerodynamic force influence coefficient matrix, B3 is the third auxiliary aerodynamic force influence coefficient matrix, Z s is the control point displacement, is the control point speed, is the control point acceleration, q ∞ is the incoming flow pressure, V is the incoming flow velocity, b is the reference length. For aircraft wing structures, the reference length is half of the root chord length. The control point of the test piece at the discrete time point should be subjected to the unsteady aerodynamic calculation model structure as follows Figure 4 As shown,
[0110] S1-3. Establish the corresponding random function by combining the Kriging proxy model
[0111] Kriging surrogate model setting A ij (t) is the corresponding random function The minimum unbiased estimation result of the mean square error is: It consists of two parts: a global approximate model based on the sampling time point data and a random process. The expression formula is as follows:
[0112]
[0113] In the above formula, is the corresponding random function, g(t) is the known basis function, β is the basis function coefficient, βg(t) is the global approximate model obtained by solving the sampling points in the design space, and Z(t) is the model with mean 0 and variance The static random process function of
[0114] S1-4. Establish a covariance calculation model for a static random process function between any two sampling time points
[0115] Within the time range defined by the aircraft ground vibration test, the covariance calculation model of the static random process function between any two sampling time points is as follows:
[0116]
[0117] In the above formula, is the variance, Z(t1) is the static random process function at the first sampling time point, Z(t2) is the static random process function at the second sampling time point, t1 is the first sampling time point, t2 is the second sampling time point, R(t1, t2) is the correlation function, and the correlation function is in the form of a Gaussian function or a cubic spline function.
[0118] S1-5. Establish a fast calculation model for unsteady aerodynamic forces, that is, a calculation model for the predicted value of the Kriging proxy model at any time t
[0119] The formula for defining the sample point data column vector of the form (3) is constructed by formula (4) as follows:
[0120] [A ij (t)] s =[A ij (t1)A ij (t2)…A ij (t n )] T =βG+z (5)
[0121] In the above formula, [A ij (t)] s is the column vector of sample point data, β is the basis function coefficient, [A ij (t1)A ij (t2)…A ij (t n )] is the sample point data of n sampling time points, G is the basis function column vector, G = [g(t1)g(t2)…g(t n )] T , z is the random function column vector, z=[Z(t1)Z(t2)…Z(t n )] T ,
[0122] According to the correlation function R(t1, t2) in formula (4), the correlation matrix of the correlation between n time sampling points is defined as and the correlation vector r representing the correlation between any time point and all sampling time points t , the correlation matrix The calculation formula is as follows:
[0123]
[0124] In the above formula, is the correlation matrix of the correlations between n time sampling points, t i is the i-th sampling time point among n time sampling points, t j is the jth sampling time point among n time sampling points, R(t i , t j ) is the correlation function between the i-th sampling time point and the j-th sampling time point,
[0125] Correlation vector r t The calculation formula is as follows:
[0126] r t=[R(t1, t)R(t2, t)…R(t n , t)] T (7) In the above formula, t is any sample point among the n time sampling points, t∈[t1 t2…t n ],r t is the correlation vector, R(t1, t) is the correlation function between any sample point t among the n time sampling points and the first sampling time point, R(t2, t) is the correlation function between any sample point t among the n time sampling points and the second sampling time point, R(t n , t) is the correlation function between any sample point t among the n time sampling points and the nth sampling time point,
[0127] Matrix function at time nodes other than sampling time points Obtained by linear weighting of the function values at known sampling points:
[0128]
[0129] In the above formula, is the weighted row vector, is a time node function outside the sampling time point, [A ij (t)] s is the column vector of sample point data,
[0130] In order to meet the requirement of unbiased estimation E(Z(t))=0 of the Kriging surrogate model, the calculation formula for the relationship between the basis function and the basis function column vector can be obtained according to formula (5) and formula (8):
[0131]
[0132] In the above formula, G is the basis function column vector, g T (t) is the row vector of basis functions, is the weighted row vector, β is the basis function coefficient, E is the mathematical expectation, is a time node function outside the sampling time point, [A ij (t)] s is the column vector of sample point data,
[0133] According to formula (3), formula (8) and formula (9), the prediction value error calculation formula of the Kriging proxy model is as follows:
[0134]
[0135] In the above formula, is the function value of the time node outside the sampling time point, A ij(t) The corresponding random function obtained by the Kriging surrogate model, A ij (t) is an element in the matrix of any time-varying aerodynamic influence coefficient, z is a random function column vector, and Z(t) is a function with a mean of 0 and a variance of The static random process function of is the weighted row vector,
[0136] According to the Kriging proxy model prediction value meets the requirements of unbiased estimation and minimum mean square error, the mean square error calculation formula of the Kriging proxy model is obtained by formula (10) as follows:
[0137]
[0138] In the above formula, is the pre-mean square error of the Kriging surrogate model, is a time node function outside the sampling time point, A ij (t) The corresponding random function obtained by the Kriging surrogate model, A ij (t) is an element in any time-varying aerodynamic influence coefficient matrix, is the weighted row vector, E is the mathematical expectation, is the variance, c t is the weighted column vector, r t is the correlation vector, is the correlation matrix of the correlations between n time sampling points,
[0139] According to the Kriging proxy model, the predicted value meets the requirements of unbiased estimation and minimum mean square error, that is, T c t =g(t), find the weighted column vector c that satisfies the minimum mean square error t To this end, the Lagrangian multiplier λ is introduced to construct the Lagrangian function H(c t ,λ) are as follows:
[0140]
[0141] In the above formula, H(c t ,λ) is the Lagrangian function, is the variance, c t is the weighted column vector, is the weighted row vector, r t is the correlation vector, is the correlation matrix of the correlation between n time sampling points, g(t) is the known basis function, λ is the Lagrange multiplier, G T is the row vector of basis functions,
[0142] The weighted column vector c in formula (12) is t And Lagrange multiplier λ to find the partial derivative, and get the weighted column vector c that meets the requirements t The calculation formula is:
[0143]
[0144] In the above formula, is the correlation matrix of the correlations between n time sampling points, r t is the correlation vector, g(t) is the known basis function, G is the basis function column vector, G T is the row vector of basis functions,
[0145] Substituting formula (13) into formula (8) yields the unsteady aerodynamic fast calculation model, that is, the prediction value calculation model of the Kriging proxy model at any time t is as follows:
[0146]
[0147] In the above formula, is a time node function outside the sampling time point, [A ij (t)] s is the column vector of sample point data, is the correlation matrix of the correlations between n time sampling points, r t is the correlation vector, g(t) is the known basis function, G is the basis function column vector, G T is the row vector of basis functions;
[0148] S2. The unsteady aerodynamic fast calculation model established in step S1 is tested and verified by an actual aircraft ground vibration test, which specifically includes the following steps:
[0149] S2-1, such as Figure 5 As shown in the figure, the interpolation points between the aerodynamic plane and the structure in the flutter calculation of the aircraft ground vibration test are condensed, and the optimal interpolation points are selected by optimization method. The optimal interpolation points are the excitation points and the vibration pickup points. The optimal interpolation points are used as the control points of the test piece.
[0150] S2-2, select 15 sampling time points, import the data of the 15 sampling time points into step S1, and obtain a fast calculation model of unsteady aerodynamic force suitable for the control point of the test piece,
[0151] S2-3. Build a ground vibration test subsystem in the laboratory, arrange a heater above the test piece, fix an exciter and a power amplifier electrically connected to the exciter at the bottom of the test piece, arrange a laser displacement sensor and an acceleration sensor at the control point of the test piece on the test piece, and electrically connect the power amplifier, laser displacement sensor, and acceleration sensor to the loading controller.
[0152] S2-4, the acquired control point response signal of the test piece is transmitted to the loading controller through the laser displacement sensor and the acceleration sensor, the loading controller substitutes the control point response signal of the test piece into the unsteady aerodynamic fast calculation model established in step S2-1, the unsteady aerodynamic fast calculation model calculates the equivalent aerodynamic force of the test piece in the time-varying process of the aircraft ground vibration test, and transmits the equivalent aerodynamic force to the semi-physical simulation subsystem, the semi-physical simulation subsystem is connected to Matlab.simulink through NI VeriStand simulation configuration software, and the equivalent aerodynamic force of the test piece is converted into an exciter control signal, the loading controller controls the exciter through the exciter control signal to load the equivalent aerodynamic force on the test piece in real time, and returns to step S2-2 after loading is completed until the equivalent simulation process of the aircraft vibration test is completed, during which the loading controller records the flutter critical speed in the equivalent simulation process of the aircraft vibration test,
[0153] S2-5. Conduct an actual aircraft ground vibration test through the ground vibration test subsystem, and use the flutter boundary tracking module to record the flutter critical speed of the test piece in the time-varying history of the aircraft ground vibration test, and compare it with the flutter critical speed of the aircraft ground vibration test time-varying history calculated by the unsteady aerodynamic force rapid calculation model in step S2-4 to determine the accuracy of the unsteady aerodynamic force rapid calculation model. If the accuracy of the unsteady aerodynamic force rapid calculation model meets the requirements, obtain the unsteady aerodynamic force rapid calculation model after test verification. If it does not meet the requirements, encrypt the sampling time points and return to step S2-2 until the accuracy of the unsteady aerodynamic force rapid calculation model meets the requirements.
[0154] S3. Use the unsteady aerodynamic force rapid calculation model verified by the experiment in step S2 to simulate and calculate the aerodynamic loads in the aircraft ground vibration test.
Claims
1. The aerodynamic load simulation calculation system in the aircraft ground vibration test is characterized by: include: A computing subsystem for performing aerodynamic load simulation of an aircraft ground vibration test, the computing subsystem comprising: a loading controller for controlling the operation of the ground vibration test through a control signal and equipped with an unsteady aerodynamic fast calculation model, the loading controller also equipped with a flutter boundary tracking module for automatically adjusting the wind speed in the unsteady aerodynamic fast calculation model according to a measured test piece response signal change trend to achieve real-time identification and recording of a flutter critical speed, A ground vibration test subsystem for verifying the unsteady aerodynamic fast calculation model, the ground vibration test subsystem comprising: an exciter located at the bottom of the test piece and providing a flutter force to the test piece, a laser displacement sensor and an acceleration sensor arranged at a control point of the test piece on the test piece, a heater arranged above the test piece, a power amplifier for amplifying a control signal, A semi-physical simulation subsystem for converting the equivalent aerodynamic force calculated by the unsteady aerodynamic fast calculation model into an exciter control signal, wherein the semi-physical simulation subsystem is connected to Matlab.simulink through NI VeriStand simulation configuration software; Based on the aerodynamic load simulation calculation system in the aircraft ground vibration test, the aerodynamic load simulation calculation method in the aircraft ground vibration test includes the following steps: S1. Based on the time freezing assumption, the unsteady aerodynamic fast calculation model for multiple discrete time points is fitted with a Kriging proxy model to obtain an unsteady aerodynamic fast calculation model suitable for the entire time-varying process of the aircraft ground vibration test; The step S1 specifically includes the following steps: S1-1, establishing a calculation formula for the unsteady aerodynamic force borne by the control point of the test piece, wherein the step S1-1 specifically includes the following contents: The unsteady aerodynamic force borne by the control point of the test piece is calculated based on the structural vibration response of the test piece in the aircraft ground vibration test. The calculation formula for the unsteady aerodynamic force borne by the control point of the test piece is as follows: f s =q ∞ A(k)z s (1) In the above formula, f s is the unsteady aerodynamic force that the control point of the test piece should bear, q ∞ is the incoming flow pressure, A(k) is the reduced-order aerodynamic influence coefficient matrix, A(k) is obtained by calculating the unsteady aerodynamic theory matching the Mach number and reducing the order according to the structural dynamic characteristics, Z S represents the structural vibration displacement of the test piece, k is the reduction frequency; S1-2. Establish the calculation model of the unsteady aerodynamic force that the control points of the test piece should be subjected to at discrete time points. S1-3. Establish the corresponding random function by combining the Kriging proxy model Kriging surrogate model setting A ij (k) is the corresponding random function The minimum unbiased estimate of the mean square error is: It consists of two parts: a global approximate model based on the sampling time point data and a random process. The expression formula is as follows: In the above formula, is the corresponding random function, g(t) is the known basis function, β is the basis function coefficient, βg(t) is the global approximate model obtained by solving the sampling points in the design space, and Z(t) is the model with mean 0 and variance The static random process function of S1-4. Establish a covariance calculation model for a static random process function between any two sampling time points Within the time range defined by the aircraft ground vibration test, the covariance calculation model of the static random process function between any two sampling time points is as follows: In the above formula, is the variance, Z(t1) is the static random process function at the first sampling time point, Z(t2) is the static random process function at the second sampling time point, t1 is the first sampling time point, t2 is the second sampling time point, R(t1, t2) is the correlation function, and the correlation function is in the form of a Gaussian function or a cubic spline function. S1-5. Establish a fast calculation model for unsteady aerodynamic forces, that is, a calculation model for the predicted value of the Kriging proxy model at any time t The formula for defining the sample point data column vector of the form (3) is constructed by formula (4) as follows: [A ij (t)] s =[A ij (t1) A ij (t2) … A ij (t n )] T =βG+z(5) In the above formula, [A ij (t)] s is the column vector of sample point data, β is the basis function coefficient, [A ij (t1) A ij (t2) … A ij (t n )] is the sample point data of n sampling time points, G is the basis function column vector, G=[g(t1) g(t2) … g(t n )] T , z is the random function column vector, z=[Z(t1) Z(t2) ... Z(t n )] T , According to the correlation function R(t1, t2) in formula (4), the correlation matrix of the correlation between n time sampling points is defined as and the correlation vector r representing the correlation between any time point and all sampling time points t , the correlation matrix The calculation formula is as follows: In the above formula, is the correlation matrix of the correlations between n time sampling points, t i is the i-th sampling time point among n time sampling points, t j is the jth sampling time point among n time sampling points, R(t i , t j ) is the correlation function between the i-th sampling time point and the j-th sampling time point, Matrix function at time nodes other than sampling time points Obtained by linear weighting of the function values at known sampling points: In the above formula, is the weighted row vector, is a time node function outside the sampling time point, [A ij (t)] s is the column vector of sample point data, In order to meet the requirement of unbiased estimation E(Z(t))=0 of the Kriging surrogate model, the calculation formula for the relationship between the basis function and the basis function column vector can be obtained according to formula (5) and formula (8): In the above formula, G is the basis function column vector, g T (t) is the row vector of basis functions, is the weighted row vector, β is the basis function coefficient, E is the mathematical expectation, is a time node function outside the sampling time point, [A ij (t)] s is the column vector of sample point data, According to formula (3), formula (8) and formula (9), the prediction value error calculation formula of the Kriging proxy model is as follows: In the above formula, is the function value of the time node outside the sampling time point, A ij (t) The corresponding random function obtained by the Kriging surrogate model, A ij (t) is an element in the matrix of any time-varying aerodynamic influence coefficient, Z is a random function column vector, and Z(t) is a function with a mean of 0 and a variance of The static random process function of is the weighted row vector, According to the Kriging proxy model, the predicted value meets the requirements of unbiased estimation and minimum mean square error, that is, T c t =g(t), find the weighted column vector c that satisfies the minimum mean square error t To this end, the Lagrangian multiplier λ is introduced to construct the Lagrangian function H(c t ,λ) are as follows: In the above formula, H(c t ,λ) is the Lagrangian function, is the variance, c t is the weighted column vector, is the weighted row vector, r t is the correlation vector, is the correlation matrix of the correlation between n time sampling points, g(t) is the known basis function, λ is the Lagrange multiplier, G T is the row vector of basis functions, The weighted column vector c in formula (12) is t And Lagrange multiplier λ to find the partial derivative, and get the weighted column vector c that meets the requirements t The calculation formula is: In the above formula, is the correlation matrix of the correlations between n time sampling points, r t is the correlation vector, g(t) is the known basis function, G is the basis function column vector, G T is the row vector of basis functions, Substituting formula (13) into formula (8) yields the unsteady aerodynamic fast calculation model, that is, the unsteady aerodynamic prediction value calculation model of the Kriging proxy model at any time t is as follows: In the above formula, is a time node function outside the sampling time point, [A ij (t)] s is the column vector of sample point data, is the correlation matrix of the correlations between n time sampling points, r t is the correlation vector, g(t) is the known basis function, G is the basis function column vector, G T is the row vector of basis functions; According to the Kriging proxy model prediction value meets the requirements of unbiased estimation and minimum mean square error, the mean square error calculation formula of the Kriging proxy model is obtained by formula (10) as follows: In the above formula, is the pre-mean square error of the Kriging surrogate model, is a time node function outside the sampling time point, A ij (t) The corresponding random function obtained by the Kriging surrogate model, A ij (t) is an element in any time-varying aerodynamic influence coefficient matrix, is the weighted row vector, E is the mathematical expectation, is the variance, c t is the weighted column vector, r t is the correlation vector, is the correlation matrix of the correlations between n time sampling points; S2. The unsteady aerodynamic fast calculation model established in step S1 is tested and verified by an actual aircraft ground vibration test, which specifically includes the following steps: S2-1. The interpolation points between the aerodynamic plane and the structure in the flutter calculation of the aircraft ground vibration test are condensed, and the optimal interpolation points are selected by optimization method. The optimal interpolation points are used as the control points of the test piece. S2-2, select several sampling time points, import the data of several sampling time points into step S1, and obtain a fast calculation model of unsteady aerodynamic force suitable for the control point of the test piece, S2-3. Build a ground vibration test subsystem in the laboratory, arrange a heater above the test piece, fix an exciter and a power amplifier electrically connected to the exciter at the bottom of the test piece, arrange a laser displacement sensor and an acceleration sensor at the control point of the test piece on the test piece, and electrically connect the power amplifier, laser displacement sensor, and acceleration sensor to the loading controller. S2-4, the acquired control point response signal of the test piece is transmitted to the loading controller through the laser displacement sensor and the acceleration sensor, the loading controller substitutes the control point response signal of the test piece into the unsteady aerodynamic fast calculation model established in step S2-1, the unsteady aerodynamic fast calculation model calculates the equivalent aerodynamic force of the test piece in the time-varying process of the aircraft ground vibration test, and transmits the equivalent aerodynamic force to the semi-physical simulation subsystem, the semi-physical simulation subsystem is connected to Matlab.simulink through the NIVeriStand simulation configuration software, and the equivalent aerodynamic force of the test piece is converted into an exciter control signal, the loading controller controls the exciter through the exciter control signal to load the equivalent aerodynamic force on the test piece in real time, and returns to step S2-2 after the loading is completed until the equivalent simulation process of the aircraft vibration test is completed, during which the loading controller records the critical flutter speed in the equivalent simulation process of the aircraft vibration test, S2-5. Conduct an actual aircraft ground vibration test through the ground vibration test subsystem, and use the flutter boundary tracking module to record the flutter critical speed of the test piece in the time-varying history of the aircraft ground vibration test, and compare it with the flutter critical speed of the aircraft ground vibration test time-varying history calculated by the unsteady aerodynamic force rapid calculation model in step S2-4 to determine the accuracy of the unsteady aerodynamic force rapid calculation model. If the accuracy of the unsteady aerodynamic force rapid calculation model meets the requirements, obtain the unsteady aerodynamic force rapid calculation model after test verification. If it does not meet the requirements, encrypt the sampling time points and return to step S2-2 until the accuracy of the unsteady aerodynamic force rapid calculation model meets the requirements. S3. Use the unsteady aerodynamic force rapid calculation model verified by the experiment in step S2 to simulate and calculate the aerodynamic loads in the aircraft ground vibration test.
2. The aerodynamic load simulation calculation system in the aircraft ground vibration test according to claim 1, characterized in that: The optimal interpolation points in step S2-1 are: the excitation point and the vibration pickup point.
3. The aerodynamic load simulation calculation system in the aircraft ground vibration test according to claim 1, characterized in that: The calculation formula of the reduction frequency k is as follows: In the above formula, k is the reduced frequency, ω is the circular frequency of simple harmonic motion, b is the reference length. For aircraft wing structures, the reference length is half of the root chord length, and V is the incoming flow velocity.
4. The aerodynamic load simulation calculation system in the aircraft ground vibration test according to claim 1, characterized in that: The step S1-2 specifically includes the following contents: Since the actual aircraft ground vibration test requires the time domain relationship of the unsteady aerodynamic force that the control point of the test piece should bear, formula (1) is extended to the Laplace domain through the rational function approximation of the minimum state method, and then the calculation model of the unsteady aerodynamic force that the control point of the test piece should bear at discrete time points is obtained through the Laplace inverse transform as follows: In the above formula, f s (t) is the unsteady aerodynamic force that the control point of the test piece should bear related to time, A0 is the displacement aerodynamic force influence coefficient matrix, A1 is the velocity aerodynamic force influence coefficient matrix, A2 is the acceleration aerodynamic force influence coefficient matrix, B1 is the first auxiliary aerodynamic force influence coefficient matrix, B2 is the second auxiliary aerodynamic force influence coefficient matrix, B3 is the third auxiliary aerodynamic force influence coefficient matrix, Z s is the control point displacement, is the control point speed, is the control point acceleration, q ∞ is the incoming flow pressure, V is the incoming flow velocity, b is the reference length. For aircraft wing structures, the reference length is half of the root chord length.
Citation Information
Patent Citations
Aircraft landing gear and tail hook dynamic load coupled analytic method
CN104217060A
Gust response ground-based simulation test method
CN109738144A
