Wing load uncertainty quantitative model correction method based on credible Bayesian update

Abstract

The invention discloses a wing load uncertainty quantitative model correction method based on credible Bayesian updating. The method adopts a non-probability interval mode to quantify the uncertainty of some parameters of the wing in a flight condition, establishes an aircraft wing load prediction finite element model, and combines uncertainty propagation influence research to obtain the load uncertainty of the aircraft wing. A credible Bayesian theory is utilized, new sample points obtained by a simulation experiment are introduced to update interval parameters, a Markov chain Monte Carlo method is utilized to obtain a relational expression between a non-probability quantization set and credibility, and a load uncertainty quantization model of a typical wing structure is established and corrected. The method has remarkable advantages in processing small sample and poor information problems in engineering practice, the credibility is utilized to describe the reliability degree of the reliability obtained through calculation, and a solution is provided for the balance problem between conservative property and credibility of the interval model.

Description

Technical Field

[0001] The present invention relates to the field of aircraft wing load analysis considering uncertainty, and in particular to the field of credible reliability analysis of aircraft wing loads under small sample and poor information conditions. Specifically, it relates to a method for correcting a wing load uncertainty quantification model based on credible Bayesian updating, which provides an important theoretical basis for the prediction of aircraft wing uncertainty loads and the verification of structural strength. Background Art

[0002] With the continuous advancement of computing power, simulation technology has become deeply integrated into engineering design and analysis. Finite element simulation and multibody dynamics analysis serve as core digital tools in aircraft structural design, and their predictions directly impact the reliability of design solutions. Traditional deterministic analysis methods typically treat structural parameters and external loads as fixed values, failing to effectively address the random factors prevalent in real-world engineering. For critical components like wings, which are subject to complex aerodynamic loads, sources of uncertainty are multidimensional: the aerodynamic load spectrum exhibits time-varying characteristics due to random fluctuations in the flight envelope; material constitutive parameters vary between batches due to manufacturing process fluctuations; and measurement noise and numerical discretization errors further amplify model deviations. These uncertainties, through multi-physics coupling, ultimately widen the confidence intervals of aerodynamic load predictions, threatening flight safety. These uncertainties often lead to uncertainty in the mechanical response of the aircraft structure, further threatening both structural and flight safety. Therefore, to ensure safety, it is essential to account for these uncertainties and consider structural reliability. Due to numerous uncertainties, there is often a discrepancy between simulation model predictions and actual aerodynamic loads, leading to doubts about the confidence in simulation model-based structural behavior predictions, robustness, and reliability assessments. Therefore, to measure and improve the credibility of simulation models, it is necessary to rationally quantify various uncertainties.

[0003] Traditional probabilistic reliability analysis methods often require the assumption of a probability density function for uncertain parameters. However, in the aerospace field, the cost of a single test is relatively high, the number of samples that can be used is small, and there is a lack of true statistical information on the uncertain parameters. In order to solve the difficulties faced by probabilistic methods in the case of small samples and poor information, the present invention adopts a non-probabilistic interval method to quantify the uncertainty of the input parameters of the aircraft wing structure, uses propagation analysis to obtain the load uncertainty of the wing structure, and establishes a load uncertainty prediction model for the aircraft wing structure. Then, a polynomial distribution is used as the prior distribution of the interval radius. After obtaining the credible interval of the uncertainty parameter, the load prediction model is corrected to obtain a credible corrected wing load uncertainty quantification model. Summary of the Invention

[0004] This paper provides a method for correcting a wing load uncertainty quantification model based on credible Bayesian updating. This method fully considers the uncertainties of multiple input parameters of an aircraft wing structure, uses propagation analysis to determine the wing load uncertainty, and establishes a load uncertainty prediction model for the aircraft wing structure. Based on Bayesian theory, a polynomial distribution is used as the prior distribution for the interval radius. After obtaining credible intervals for the uncertainty parameters, the load prediction model is modified to obtain a credible corrected wing load uncertainty quantification model. The resulting result includes a credibility indicator, is more consistent with real-world conditions, and has enhanced engineering applicability.

[0005] The technical solution adopted by the present invention to solve the above technical problems is:

[0006] A wing load uncertainty quantification model correction method based on credible Bayesian updating includes the following steps:

[0007] Step 1: Quantify the uncertainty of input parameters for aircraft wing structures under different calculation conditions and quantify parameter uncertainty into interval variables;

[0008] Step 2: Establish an aircraft wing load prediction model, use interval variables to perform uncertainty propagation analysis on the load prediction model, obtain the load uncertainty interval of the aircraft wing structure, and obtain an uncertainty quantification model;

[0009] Step 3: Use the aircraft wing load prediction model to conduct simulation experiments to obtain computational fluid dynamics (CFD) analysis results, which include sample point data.

[0010] Step 4: Introduce sample point data to update the load response interval, use the Markov Chain Monte Carlo algorithm to build a connection between the load uncertainty interval and the credibility, and obtain the load response interval with credibility;

[0011] Step 5: Combine the load response interval with credibility and compare the load response interval with the initial load uncertainty interval to obtain the uncertainty quantification model based on the credible Bayesian correction.

[0012] The present invention has the following beneficial effects:

[0013] This paper establishes a modified load prediction model for aircraft wing structures, using interval uncertainty quantification with confidence levels, taking into account multi-source uncertainty. To address the issue of inaccurate load estimation due to insufficient sample information for aircraft wing input variables, this model, based on Bayesian theory and interval models, is updated with limited sample data points obtained from simulation experiments. Using the Markov Chain Monte Carlo method, a relationship between interval radius and confidence level is derived. Comparative results show that the reliability of the proposed method continues to improve with the introduction of new sample points, providing a wider design space for high-performance structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The present invention is a flow chart for establishing a credible uncertainty quantification model for wing loads;

[0015] Figure 2 This is a schematic diagram of an aircraft CAD model used in the present invention;

[0016] Figure 3 It is a residual convergence diagram of the CFD calculation process of the aircraft model used in the present invention;

[0017] Figure 4 The present invention uses the aerodynamic cloud map of the aircraft at the lower limit of the flight angle of attack before the update;

[0018] Figure 5 The present invention uses the aerodynamic cloud map of the aircraft at the upper limit of the flight attack angle before the update;

[0019] Figure 6 It is the posterior distribution and posterior interval of the flight angle of attack of the aircraft wing structure under different prior distributions;

[0020] Figure 7 The present invention uses the aerodynamic cloud map of the aircraft at the lower limit of the flight attack angle after the update range;

[0021] Figure 8 The present invention uses the aerodynamic cloud map of the aircraft at the upper limit of the flight attack angle after the update range;

[0022] Figure 9 It is the change of the updated value of the flight angle of attack interval radius of the aircraft wing structure under different prior distributions with the credibility. DETAILED DESCRIPTION

[0023] In order to make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other. To achieve the above-mentioned objectives, the present invention adopts the following technical solutions.

[0024] like Figure 1 As shown, the present invention proposes a wing load uncertainty quantification model correction method based on credible Bayesian updating, comprising the following steps:

[0025] In the first step, for the aircraft wing structure under different computational conditions, the non-probabilistic interval method is used to quantify the uncertainty of the input parameters caused by multiple uncertainty sources in the finite sample case.

[0026] Due to the various uncertainties in the dispersion of wing materials, the manufacturing process, and external excitation, the wing will be affected by the uncertainty of multiple input parameters caused by multiple sources of uncertainty, including flight altitude, incoming flow pressure, incoming flow density, incoming flow temperature, incoming flow viscosity coefficient, gas specific heat ratio, local speed of sound, flight Mach number, flight speed, flight angle of attack, and flight sideslip angle. In order to avoid mutual interference between uncertain sources, independent parameters should be selected as uncertain sources. At the same time, quantities that are more basic in physical nature should be selected. Finally, the suitability of the parameters as uncertain sources should be judged based on the actual working conditions. Considering the above three aspects, the flight angle of attack was finally selected as the input parameter in the embodiment.

[0027] In the case of finite samples, the interval method is used to describe the flight angle of attack. According to the interval theory, it is assumed that the flight angle of attack to be described is expressed in the following form:

[0028] (1)

[0029] in, is the interval vector of the flight attack angle. and are the corresponding lower and upper bounds respectively. is the first A portion. and are the lower and upper bounds respectively. m is the number of uncertainty parameters, which is 1 in this case. According to interval theory, formula (1) can be expressed in another form:

[0030] (2)

[0031] (3)

[0032] in, is the interval mean, is the interval radius. 、 Respectively The interval mean and interval radius of the uncertainty parameter satisfy the following relationship with the upper and lower bounds of the interval:

[0033] (4)

[0034] Based on this, the process of quantifying uncertainty parameters into interval variables in the non-probabilistic interval method is completed.

[0035] The second step is to establish an aircraft wing load prediction model. Using the interval variables obtained in the first step, the uncertainty propagation impact is estimated based on the load prediction model to obtain the load uncertainty interval of the aircraft wing structure. First, the aircraft wing load prediction model is constructed. On this basis, the load response interval of the aircraft wing structure is obtained using the collocation analysis method based on Legendre polynomials. The collocation method is a dimensional analysis method. For multiple uncertain input parameters, response proxy models are selected one by one to obtain the response extreme value under the uncertain input parameter. Finally, the response extreme values ​​under all uncertain input parameters are combined to obtain the response uncertainty interval. Specifically:

[0036] First, assume that the given reference point is the center value of the interval , through the reference point in the interval parameter vector to form the subspace Aerodynamic load response of wing structure intercepted by a 3D hyperplane The resulting curve is

[0037] (5)

[0038] in The response curve represents the response, Consists of 6 response components , Is the mapping relationship between input parameters and response parameters. Now assume The form of is Legendre polynomial, by transforming the input parameter interval into the following form, for any input parameter value :

[0039] (6)

[0040] This transforms the input parameter into an interval Parameters The Legendre polynomials are:

[0041] (7)

[0042] in is the differential operator, is the order of Legendre polynomial, using Legendre polynomial, Written in the following form:

[0043] (8)

[0044] is the Legendre polynomial, where , using the Gauss-Legendre quadrature formula, the final Legendre polynomial approximation can be obtained as:

[0045] (9)

[0046] Using formula (9), the load uncertainty interval of the aircraft wing structure is obtained, where is the zero point of Legendre polynomial, and the truncation error of the above formula is:

[0047] (10)

[0048] in , is the order of the Legendre polynomial used. In order to make the Legendre approximation work best, it is necessary to minimize the truncation error, which can be achieved by taking an appropriate order. to achieve the desired truncation error.

[0049] The third step is to conduct simulation experiments using the aircraft wing load prediction model to quantify the sample point data of the input parameters through simulation experiments;

[0050] Using an aircraft wing load prediction model and ANSYS Fluent simulations, high-precision computational fluid dynamics (CFD) analysis results for the aircraft wing were obtained. These CFD results include sample point data. Calculations were performed for a selection of input parameters, including flight altitude, incoming flow pressure, incoming flow density, incoming flow temperature, incoming flow viscosity, gas specific heat ratio, local speed of sound, flight Mach number, flight speed, flight angle of attack, and flight sideslip angle. To simulate the harsh mechanical and thermal environment, several parameters were selected for simulation according to the principles in the first step. Among these parameters, flight altitude is the independent variable, typically determining the local speed of sound, incoming flow pressure, incoming flow density, incoming flow viscosity, and incoming flow temperature. However, considering that the range of uncertainty sources should not be too large, small changes in flight altitude will not significantly change the incoming flow parameters and, therefore, will not affect the hypersonic aerodynamic and aerothermal characteristics. Therefore, flight altitude is not suitable as an uncertainty source.

[0051] For a calorimetrically perfect gas, if the incoming flow temperature can be determined, the incoming flow viscosity coefficient, the gas specific heat ratio, and the local sound speed can also be determined. Conversely, the incoming flow temperature cannot be determined from the other three parameters. Therefore, the incoming flow temperature is a more suitable uncertainty source. For the local sound speed, flight Mach number, and flight speed, knowing any two of them allows the other to be determined. Considering that in real-world situations, the aircraft directly controls flight speed, and the main factors causing uncertainty include control accuracy, incoming flow speed changes, gusts, and so on, flight speed is a suitable uncertainty source. For the flight angle of attack and flight sideslip angle, both are independent variables, so the flight angle of attack can be selected as an uncertainty source.

[0052] Finally, the flight angle of attack is selected as the uncertainty source. The interval quantification results of the flight angle of attack are obtained through simulation experiments.

[0053] In the fourth step, the polynomial distribution is used to introduce the CFD analysis results and sample point data to update the load response interval. The Markov chain Monte Carlo algorithm is used to build the connection between the load uncertainty interval and the credibility, and the load response interval with credibility is obtained.

[0054] First, assume that the flight angle of attack follows a certain prior distribution. Then, introduce new sample points based on the simulation results in the third step to update the posterior distribution of the flight angle of attack. According to Bayesian theory, the posterior probability density function of the flight angle of attack can be expressed as:

[0055] (11)

[0056] in, is the prior probability density, is the sampling density of the sample, is a normalization constant that represents the marginal density of the sample.

[0057] The first thing to determine is the prior distribution of the flight angle of attack This parameter is typically obtained empirically, based on known information or understanding of the model. Several options exist for the prior distribution: a standard non-informative prior distribution; an informative conjugate distribution; or a more general informative prior distribution. Since this problem addresses engineering scenarios with small sample sizes and limited information, in the absence of prior information, in order to reliably model the structure, we assume that the uncertainties in the upper and lower bounds of the flight angle of attack follow a polynomial distribution. This polynomial distribution is used as the prior distribution for the radius of the flight angle of attack interval.

[0058] The probability density function of the multinomial distribution is selected as:

[0059] (12)

[0060] In the formula are all constants. The probability density function of the multinomial distribution in the interval is:

[0061] (13)

[0062] In the formula For the upper and lower bounds, , The distribution functions of the multinomial distribution are all constants. The interval method assumes that all samples are equally likely to be distributed in the interval, and samples are drawn from the uniform distribution with a sampling density of:

[0063] (14)

[0064] The marginal density is:

[0065] (15)

[0066] Based on Bayesian theory, the posterior distribution probability density function is:

[0067] (16)

[0068] Given credibility , integrating the posterior distribution:

[0069] (17)

[0070] in, is the solution of the exponential integral, is the updated value of the interval parameter under given sample and confidence level. There is no analytical solution, so the Markov Chain Monte Carlo method is used to find a numerical solution to this integral.

[0071] The Markov Chain Monte Carlo method assumes that the probability of a state transition at a given moment depends solely on the previous state. Because a state transition at a given moment depends solely on the previous state, if the transition probabilities between any two states in the system can be determined, a state transition probability matrix can be obtained. These simulated values ​​are then used to infer possible parameter values ​​or functions of parameter values. Using the Markov Chain Monte Carlo method, a plausible range of aircraft wing load response is obtained. The specific implementation process is as follows:

[0072] Assumptions is a dimensional real parameter vector, first generate a candidate point , defining the proposal density Used from generate , a common method is to generate a single component Add a random uniformly distributed bias:

[0073] (18)

[0074] in, is the standard uniform distribution, is an arbitrary constant. Then calculate the probability that the candidate value is accepted as the next simulated value in the sequence and define the acceptance probability:

[0075] (19)

[0076] in, For the new sample data obtained in the experiment, the acceptance probability represents the product of the ratio of the posterior density evaluated at the candidate parameter value to the current parameter value and the ratio of the proposal density of the current point and the candidate point. Finally, draw a random variable that follows a uniform distribution , and and For comparison, if Then accept the candidate value and set ,if Then reject the candidate value and set According to the above mentioned method, a series of Value, take For a series of values ​​in the flight angle of attack interval, the updated aircraft wing load response interval can be obtained [ , ].

[0077] Step 5: Combine the load response interval with credibility, compare the load response interval with the initial load uncertainty interval, and obtain the uncertainty quantification model based on credible Bayesian correction.

[0078] Based on the results of the fourth step, the credibility is The load response range [ , ] and compare this interval with the initial load uncertainty interval. Combining all the above results and processes, a modified credible Bayesian uncertainty quantification model is established.

[0079] Example:

[0080] In order to more fully understand the characteristics of the invention and its applicability to actual engineering, the present invention first establishes a CAD model of the aircraft wing structure, uses the flight angle of attack as the uncertainty input parameter, adopts the Legendre polynomial expansion method to perform uncertainty propagation analysis, uses CFD simulation to obtain new samples, and calculates credible interval parameters based on Bayesian theory, which are then compared with the initial interval parameters to establish a credible Bayesian uncertainty quantification model.

[0081] Aircraft wing model Figure 2 As shown, the aircraft flow field and angle of attack parameters are set as follows: the flow field fluid is air under normal conditions, the aircraft skin material is aluminum alloy, and the density is set to 2719kg / m 3 , the specific heat capacity is 871 J / kg·K, and the thermal conductivity is 202.4 W / m·K. Set the aircraft flight speed to 1.5 Mach, the far-field flow temperature to 300K, the turbulence intensity to 5%, the turbulence viscosity ratio to 10, the pressure outlet return turbulence intensity to 5%, and the return turbulence viscosity ratio to 10. The uncertainty range of the aircraft flight angle of attack is , perform CFD calculation on the aircraft. In this example, the calculation converges at about 550 steps, as shown in Figure 3 The aerodynamic force cloud diagrams of the aircraft at 0° and 2° angles of attack are shown in Figure 4 and Figure 5 As shown. Using Legendre polynomial approximation, the aerodynamic force range of the wing can be obtained as , the Legendre polynomials can also be used to approximate the load uncertainty interval at any position of the aircraft wing structure.

[0082] The center of the interval is considered fixed, and the interval radius is considered to be an uncertain parameter to be updated. Based on the non-probabilistic credible Bayesian model update method, new sample points are introduced to update the interval radius. The method proposed in this invention uses the multinomial distribution as the prior distribution of the parameter to be updated. Given a series of credibility, under different prior distributions, the posterior distribution and posterior interval of the interval radius of the aircraft flight angle of attack parameter after the update are respectively as follows: Figure 6 As shown in the figure, it can be seen that, except for the Pareto prior distribution, the interval radius after the new sample is introduced into other distributions is larger than the original interval radius and can envelop all sample points. Although the use of Pareto distribution in the prior distribution makes the update process easier to handle mathematically, it will cause the updated interval to no longer contain all sample points, reducing the accuracy. However, the updated interval range after the use of polynomial distribution in the prior distribution is smaller than the uniform distribution, which shows the accuracy of this method. The updated result of the uncertainty interval of the flight angle of attack in this example is , Figure 7 and Figure 8 The aerodynamic cloud diagrams of the aircraft at attack angles of 0.136064° and 2.136064° are given. Figure 9The radius of the interval of flight angle of attack after the three prior distributions are updated changes with the credibility level. As the credibility level increases, the updated interval radius of the three prior distributions will increase, which is consistent with actual engineering experience, that is, ensuring a high credibility level will lead to a broad estimate of the uncertainty and increased conservatism. In addition, the interval radius after the prior distribution polynomial distribution is updated is smaller than that of the uniform distribution and can include all sample points, indicating that the method proposed in the present invention can reduce conservatism while ensuring the quantitative accuracy of uncertainty. Re-propagation analysis of the flight angle of attack range that includes credibility can obtain the aerodynamic results of the aircraft in flight with credibility, and finally the aerodynamic resultant force range of the wing is obtained. By comparing the aerodynamic range of direct propagation analysis, it can be found that the aerodynamic range becomes larger due to the introduction of new sample points, which is consistent with actual engineering experience. This shows that the credible Bayesian uncertainty quantification model proposed in this paper can more accurately characterize the uncertainty of interval parameters.

[0083] The above are only specific steps of the present invention and do not constitute any limitation to the scope of protection of the present invention; it can be extended to the field of reliability analysis of other composite materials structures, and any technical solutions formed by equivalent transformation or equivalent replacement fall within the scope of protection of the present invention.

[0084] Some parts of the present invention are well known to those skilled in the art and are not described in detail.

Claims

1. A wing load uncertainty quantification model correction method based on credible Bayesian updating, characterized by: The steps include: Step 1: Quantify the uncertainty of input parameters for aircraft wing structures under different calculation conditions and quantify parameter uncertainty into interval variables; Step 2: Establish an aircraft wing load prediction model, use interval variables to perform uncertainty propagation analysis on the load prediction model, obtain the load uncertainty interval of the aircraft wing structure, and obtain an uncertainty quantification model; Step 3: Use the aircraft wing load prediction model to conduct simulation experiments to obtain computational fluid dynamics (CFD) analysis results, which include sample point data. Step 4: Introduce sample point data to update the load response interval, use the Markov Chain Monte Carlo algorithm to build a connection between the load uncertainty interval and the credibility, and obtain the load response interval with credibility; Step 5: Combine the load response interval with credibility and compare the load response interval with the initial load uncertainty interval to obtain the uncertainty quantification model based on the credible Bayesian correction.

2. The method for correcting the wing load uncertainty quantification model based on credible Bayesian updating according to claim 1, characterized in that: In the first step, the flight angle of attack is selected as the input parameter.

3. The method for correcting the wing load uncertainty quantification model based on credible Bayesian updating according to claim 1, characterized in that: The interval method is used to describe the input parameters, and the input parameters are expressed in the following form: (1) in, is the interval vector of the flight attack angle, and are the corresponding lower and upper bounds respectively; is the first A quantity, and are the lower bound and upper bound respectively; m is the number of uncertainty parameters, which completes the process of quantifying uncertainty parameters into interval variables.

4. The method for correcting the uncertainty quantification model of wing loads based on credible Bayesian updating according to claim 1, wherein the second step include: First, assume that the given reference point is the center value of the interval , through the reference point in the interval variable subspace 3D aerodynamic load response The resulting curve is: (5) in The response curve represents the response, Consists of 6 response components , Is the mapping relationship between input parameters and response parameters, assuming The form of is Legendre polynomial, by transforming the input parameter interval into the following form, for any input parameter value : (6) Transform the input parameter into an interval Parameters , the Legendre polynomials are: (7) in is the differential operator, is the order of Legendre polynomial, using Legendre polynomial, Written in the following form: (8) in , using the Gauss-Legendre quadrature formula, the final Legendre polynomial approximation is: (9) Using formula (9), the load uncertainty interval of the aircraft wing structure is obtained.

5. The method for correcting the wing load uncertainty quantification model based on credible Bayesian updating according to claim 1, characterized in that: In the third step, the aircraft wing load prediction model is used to conduct simulation experiments using ANSYS Fluent to obtain the computational fluid dynamics (CFD) analysis results of the aircraft wing, where the computational fluid dynamics (CFD) analysis results include sample point data.

6. The method for correcting the wing load uncertainty quantification model based on credible Bayesian updating according to claim 1, characterized in that: In the fourth step, we first assume that the flight angle of attack follows a certain prior distribution. Then, we introduce the sample point data and update the posterior distribution of the flight angle of attack. According to Bayesian theory, the posterior probability density function of the flight angle of attack is expressed as: (11) in, is the prior probability density, is the sampling density of the sample, is a normalization constant, which represents the marginal density of the sample; The multinomial distribution is used as the prior distribution of the radius of the flight angle of attack interval; The probability density function of the multinomial distribution is selected as: (12) In the formula Are all constants, and the probability density function of the multinomial distribution in the interval is: (13) In the formula For the upper and lower bounds, , Both are distribution functions of multinomial distribution. The interval method assumes that all samples are equally likely to be distributed in the interval. Samples are drawn from the uniform distribution, and the sampling density is: (14) The marginal density is: (15) Based on Bayesian theory, the posterior distribution probability density function is: (16) Given credibility , integrating the posterior distribution: (17) in, is the solution of the exponential integral, The Markov Chain Monte Carlo method is used to obtain the numerical solution of the integral for the updated value of the interval parameter under given sample and credibility conditions.

7. The method for correcting the wing load uncertainty quantification model based on credible Bayesian updating according to claim 6, characterized in that: The specific implementation process of the Markov chain Monte Carlo method is as follows: Assumptions is a dimensional real parameter vector, first generate a candidate point , defining the proposal density Used from generate , for a single component Add a random uniformly distributed bias: (18) in, is the standard uniform distribution, is an arbitrary constant, and then the probability of the candidate value being accepted as the next simulated value in the sequence is calculated, defining the acceptance probability: (19) in, For the new sample data obtained in the experiment, the acceptance probability represents the product of the ratio of the posterior density evaluated at the candidate parameter value to the current parameter value and the ratio of the proposal density of the current point and the candidate point. Finally, a random variable obeying the uniform distribution is drawn. , and and For comparison, if Then accept the candidate value and set ,if Then reject the candidate value and set , and obtain a series of Value, take is a series of values ​​in the flight angle of attack interval, and the updated aircraft wing load response interval is obtained [ , ].

8. The method for correcting the wing load uncertainty quantification model based on credible Bayesian updating according to claim 1, characterized in that: The fifth step includes: Based on the credibility The load response range [ , ], and compared this interval with the initial load uncertainty interval to obtain a revised uncertainty quantification model based on credible Bayesian.

9. The method for correcting the wing load uncertainty quantification model based on credible Bayesian updating according to claim 1, characterized in that: Formula (1) can be expressed in another form: (2) (3) in, is the interval mean, is the interval radius, 、 Respectively The interval mean and interval radius of the uncertainty parameter satisfy the following relationship with the upper and lower bounds of the interval: (4)。 10. A device for correcting wing load uncertainty quantification model based on credible Bayesian updating, characterized in that: include: The interval variable acquisition module quantifies the uncertainty of input parameters for aircraft wing structures under different calculation conditions and quantifies the parameter uncertainty into interval variables; The uncertainty quantification model establishment module establishes an aircraft wing load prediction model, uses interval variables to perform uncertainty propagation analysis on the load prediction model, obtains the load uncertainty interval of the aircraft wing structure, and obtains the uncertainty quantification model; The sample point data acquisition module uses the aircraft wing load prediction model to conduct simulation experiments to obtain the computational fluid dynamics analysis CFD results, which include sample point data; Credibility load response interval acquisition module introduces sample point data to update the load response interval, uses Markov chain Monte Carlo algorithm to build the relationship between load uncertainty interval and credibility, and obtains the load response interval with credibility; The uncertainty quantification model correction module combines the load response interval with credibility, compares the load response interval with the initial load uncertainty interval, and obtains the uncertainty quantification model after credibility Bayesian correction.

Images