Transportation aircraft gust load spectrum compilation method based on maximum mutual information-Bayesian fusion

Abstract

The invention discloses a transportation aircraft gust load spectrum compilation method based on maximum mutual information-Bayesian fusion. The method comprises the steps of obtaining actually measured load data of an aircraft with a to-be-compiled load spectrum; acquiring historical load data of various transportation aircrafts; establishing statistical distribution parameters of the historical load data based on a Bayesian framework; respectively introducing weight factors into the statistical distribution parameters and the actually measured load data, and determining the weight factors by adopting maximized mutual information; and based on the weight factor, fusing the statistical distribution parameter and the actually measured load data to obtain an average overload cumulative transcendental curve, and further compiling a 5 * 5 spectrum. Through combination of the Bayesian framework and the mutual information theory, the reliability problem of load spectrum compilation of transportation aircrafts under the condition of a small amount of actual measurement data is effectively solved. According to the method, weight factor dynamic balance prior information and actually measured data are innovatively introduced, parameter estimation is optimized by maximizing mutual information, and the estimation precision of load distribution parameters is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of aircraft load spectrum compilation, and in particular to a method for compiling gust load spectra of transport aircraft based on maximum mutual information-Bayesian fusion. Background Art

[0002] The compilation of measured load spectra for existing transport aircraft is typically based on field flight data from prototype aircraft. However, when the fleet is small, the sample size of measured data is limited, making it difficult to fully cover all load conditions encountered in actual aircraft operation.

[0003] The method for compiling measured spectra for military aircraft proposed by Yan Chuliang has clear requirements for the minimum number of measured takeoffs and landings, but this requirement is often difficult to meet for transport aircraft. The method for compiling fused spectra for civil aircraft proposed by He Xiaofan lacks a dynamic balance between the weights of prior information and measured data information. At present, transport aircraft mainly use the TWIST spectrum method to compile load spectra, which relies on historical gust data to construct a gust load cumulative exceedance curve. However, the actual operating environment of the aircraft may deviate from the design assumptions, resulting in limited applicability of the existing load spectra. Therefore, it is still necessary to obtain real load data through actual measurement and compile a measured spectrum that conforms to actual working conditions based on this.

[0004] Due to the urgency of transport aircraft entering service, measured data is usually scarce. At the same time, their loads (especially gust loads) are significantly dispersed, and accurate characterization of the load statistical characteristics requires a large amount of data support. If the data is insufficient, it is difficult to ensure the coverage and reliability of the spectrum compilation. Therefore, how to dynamically balance prior and measured information based on a small amount of measured data of the aircraft, combined with historical reference data, to compile a load spectrum that not only meets high reliability requirements but also reflects the load characteristics of the aircraft, remains a technical problem that needs to be solved urgently. Summary of the Invention

[0005] In response to the problems in the above-mentioned background technology, the present invention is based on the Bayesian method and mutual information theory. By weighting the prior data and the measured data, the prior and measured information are effectively integrated to estimate the load data distribution parameters, and then a more reliable gust load spectrum is compiled, providing more reliable input for aircraft fatigue and durability analysis.

[0006] To achieve the above objectives, the present invention provides a method for compiling a gust load spectrum for a transport aircraft based on maximum mutual information-Bayesian fusion, the steps comprising:

[0007] Obtain the measured load data of the aircraft for which the load spectrum is to be compiled;

[0008] Obtain historical load data for several types of transport aircraft;

[0009] Establishing statistical distribution parameters of the historical load data based on a Bayesian framework;

[0010] Introducing weight factors into the statistical distribution parameters and the measured load data respectively, and determining the weight factors by maximizing mutual information;

[0011] Based on the weight factors, the statistical distribution parameters and the measured load data are integrated to obtain an average overload cumulative exceedance curve, and then a 5×5 spectrum is compiled.

[0012] Preferably, the measured load data of the aircraft to be loaded is collected and real-time quality G i Correct the normal overload data of the aircraft center of gravity:

[0013] Δn z =(n zi -1)*G i / G0

[0014] Where Δn z is the incremental overload value corrected according to the standard mission section; n zi is the measured real-time overload value; G0 is 18000kg.

[0015] Preferably, the cumulative exceedance number data of discrete gust speeds of multiple transport aircraft are collected and the historical load data are obtained through screening and analysis; the exceedance numbers of upward and downward gusts in the historical load data are merged to form the gust speed per nautical mile U de Cumulative transcendental number curve family:

[0016]

[0017] Where N represents the cumulative number of exceedances.

[0018] Preferably, the gust speed in the historical load data is converted into gust incremental overload according to the structural response characteristics of the aircraft in the load profile to be compiled:

[0019]

[0020] Where: Δn represents the gust load coefficient increment; R represents the vertical gust response parameter of the wing and fuselage; U de represents the converted gust speed; ρ0 represents the air density at sea level; C Nα Indicates the slope of the lift line; V e represents equivalent airspeed; G / S represents wing loading; K W represents the discrete gust mitigation factor; μ g represents the aircraft mass parameter; g represents the acceleration due to gravity; represents the average geometric chord length of the wing; ρ h Indicates the air density at the altitude.

[0021] Preferably, establishing the statistical distribution parameters of the historical load data includes:

[0022] Where, σ 2 represents variance; μ1 and σ1 2 represents the mean and variance of historical load data; σ1 represents the standard deviation of historical load data; n1 represents the number of prior data; P represents μ and σ 2 The joint prior distribution of ; Γ represents the chi-square distribution; μ represents the general term for the mean.

[0023] Preferably, the weight factor α of the statistical distribution parameter and the weight factor β of the measured load data satisfy:

[0024]

[0025] Where c is a constant.

[0026] Preferably, the method of determining the weight factor by maximizing mutual information includes:

[0027] In the formula, I(μ,σ 2 , x) represents mutual information; x i represents the i-th measured data; P(μ, σ 2 , x i ) represents the joint distribution; P(μ,σ 2 ) represents the prior distribution; P(x i ) is x i The marginal distribution of .

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] This invention effectively solves the reliability problem of load spectrum compilation for transport aircraft with a small amount of measured data by combining a Bayesian framework with mutual information theory. The invention innovatively introduces weighting factors to dynamically balance prior information and measured data, optimizing parameter estimation by maximizing mutual information, significantly improving the estimation accuracy of load distribution parameters. Furthermore, based on the deep integration of historical data and measured data, the compilation results not only reflect the load characteristics of the aircraft, but also cover complex operating conditions, ensuring high coverage and reliability of the load spectrum. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0031] Figure 1 Schematic diagram of a method flow in an embodiment of the present invention;

[0032] Figure 2 Schematic diagram of discrete gust historical data at different altitudes according to an embodiment of the present invention; wherein (a) represents an altitude of 0-1500 feet; (b) represents an altitude of 1500-4500 feet; (c) represents an altitude of 4500-9500 feet; (d) represents an altitude of 9500 feet-19500 feet; (e) represents an altitude of 19500 feet-29500 feet; and (f) represents an altitude of 29500 feet-39500 feet.

[0033] Figure 3 U of the embodiment of the present invention de Schematic diagram of a family of transcendental number curves; where (a) represents an altitude of 0-1500 feet; (b) represents an altitude of 1500-4500 feet; (c) represents an altitude of 4500-9500 feet; (d) represents an altitude of 9500 feet-19500 feet; (e) represents an altitude of 19500 feet-29500 feet; and (f) represents an altitude of 29500 feet-39500 feet. DETAILED DESCRIPTION

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0035] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0036] First, some core technical points of the present invention are described:

[0037] The core of the present invention is to introduce weight factors α and β into the prior information and measured information in the posterior distribution. After determining the weight factors α and β, the maximum a posteriori estimation method is referred to, and the posterior distribution is respectively adjusted for μ and σ. 2The partial derivatives are calculated to be 0 to obtain the fusion parameters of the two (data fusion based on the Bayesian method); the weight factors α and β of the prior and measured information are determined by maximizing the mutual information (parameter estimation based on maximum mutual information optimization). Among them, the prior information in the posterior distribution refers to the statistical distribution parameters (such as mean μ1, standard deviation σ1 and variance σ1) established based on historical load data. 2 ), which is introduced through the Bayesian framework and is used to represent the existing knowledge of load distribution before the current measured data is obtained; the measured information mentioned later (including measured take-off and landing data, etc.) is the distribution parameters (such as mean μ2, standard deviation σ2 and variance σ2) obtained based on the statistical analysis of the measured load data. 2 ).

[0038] Example 1

[0039] like Figure 1 FIG. 1 is a flow chart of the method of this embodiment, and the steps include:

[0040] S1. Obtain the measured load data of the aircraft for which the load spectrum is to be compiled.

[0041] Collect the measured overload time history data (measured load data) of the aircraft to be loaded. i "Correct the normal overload data of the aircraft's center of gravity. That is:

[0042] Δn z =(n zi -1)*G i / G0 (1)

[0043] Where Δn z is the incremental overload value corrected according to the standard mission section; n zi is the measured real-time overload value; G0 is taken as 18000 kg; the real-time mass of the aircraft is calculated by subtracting the fuel consumption from the aircraft weight, and the fuel consumption is calculated as the total average (the time for calculating the average fuel consumption is from the start of takeoff taxiing to the end of landing taxiing).

[0044] Extract the peak and valley values of the overload data, retain the sampling sequence numbers of the peak and valley points and remove invalid data points. If:

[0045]

[0046] Then take the peak value (or valley value) once.

[0047] After extracting the valid peak-valley points, the limited span-average peak counting method is used for counting, with the "1g" load state of each task segment as the benchmark. When the valley value between two peaks (or the peak value between two valleys) does not exceed the upper deviation of the "1g" baseline of the task segment (the lower deviation when the valley value is overloaded), only the maximum peak value (the valley value when negative overloaded) is recorded, and the upper and lower deviations are 5% of the average overload of 1g, that is, 0.05g.

[0048] The peak-valley value sequence of gust overload (Δn zp ,Δn zt ) i , for different task segments, select several levels of incremental overload, and record the cumulative exceedance number for the positive peak incremental overload and the negative valley incremental overload, which are recorded as N(+Δn y ) and N(-Δn z Then, the geometric mean is used to obtain the average of the cumulative exceedance corresponding to the symmetrical incremental overload, and (Δn z ,N) i (i=1,…,n) data pairs, see formula (3).

[0049]

[0050] Where N represents the cumulative exceedance number.

[0051] Assume that the flight time corresponding to the measured data is t m , the standard time of the task segment is t s The cumulative exceedance number N of the i-th level load obtained during the measured time ib , then the cumulative exceedance number N of the i-th level load of the standard time is i Should be:

[0052]

[0053] The cumulative exceedance number of each level of load is replaced according to formula (4), and the cumulative exceedance number data of the task segment overload of standard time can be obtained.

[0054] The equation shown in formula (5) is used to fit the incremental overload cumulative exceedance data of each task segment (Δn z ,N) i .

[0055] lg[N(n z )]=a0+a1Δn z +a3 lg(Δn z ) (5)

[0056] Among them, a0, a1, and a3 represent fitting parameters; Δn z Indicates incremental overload.

[0057] S2. Obtain historical load data for various types of transport aircraft.

[0058] This embodiment collects the discrete gust speed U of multiple transport aircraft since the 1970s. de The cumulative exceedance data, obtained through screening and analysis, is shown in Table 1, which includes the measured data of seven aircraft models published by the FAA and the measured data of the B747 operating in Europe for nearly ten years published by KSSU.

[0059] The upward and downward gust exceedance numbers in the original data are combined using formula (6) to form the gust speed U in units of nautical miles (nm): de The family of cumulative transcendental curves, such as Figure 2 As shown in the table, the altitude ranges are 0-1500ft, 1500-4500ft, 4500-9500ft, 9500ft-19500ft, 19500ft-29500ft, and 29500ft-39500ft. Since gust intensity decreases with increasing altitude, for conservative reasons, the gust load data for altitudes of 39500-70000ft uses the data for altitudes of 29500-39500ft.

[0060] Table 1

[0061]

[0062] The upward and downward gust exceedance numbers in the original data are combined using formula (6) to form the gust speed U in units of nautical miles (nm): de Cumulative transcendental number curve family:

[0063]

[0064] In order to integrate the gust incremental overload cumulative exceedance curve generated by the aircraft sample, the gust speed in the historical load data must be converted into gust incremental overload according to the aircraft structural response characteristics to be compiled. The conversion method is derived from the simplified aircraft motion model. The model simplifies the fuselage into a two-dimensional rigid body. Under the action of gust disturbance, it only has pitch and float motion, and the period of the gust disturbance is simplified to a fixed value (25 times the average chord length of the wing). The conversion relationship between the gust incremental overload amplitude and the gust speed obtained through dynamic analysis is shown in the following formula:

[0065]

[0066] Where: Δn represents the gust load coefficient increment; R represents the vertical gust response parameter of the wing and fuselage (s / m); U de Indicates the converted gust speed (m / s) (abbreviated as U de or gust speed); ρ0 represents the air density at sea level (kg / s); CNα represents the slope of the lift line (1 / rad); V e represents the equivalent airspeed (m / s); G / S represents the wing load (N / m 2 );K W represents the discrete gust mitigation factor; μ g Indicates the aircraft mass parameter; g indicates the acceleration due to gravity (m / s 2 ); represents the average geometric chord length of the wing (m); ρ h Indicates the air density at the height (kg / m 3 ).

[0067] The gust speed cumulative exceedance number curve family of each height interval of the reference data is converted into the gust overload cumulative exceedance number curve family of the reference data.

[0068] The geometric mean method is used to combine the exceedance numbers corresponding to the upward and downward gust speeds, and then merge them into a family of gust speed exceedance number curves per unit nautical mile according to different height intervals, such as Figure 3 The curve represents a symmetrical gust speed cycle ±U de Transcendental numbers and U de Functional relationship.

[0069] S3. Based on the Bayesian framework, establish the statistical distribution parameters of historical load data.

[0070] Discretize the cumulative transcendental number curve and subtract the cumulative transcendental numbers of two adjacent levels to obtain the transcendental number ΔN:

[0071] ΔN i =N i -N i-1 (8)

[0072] It is usually considered that the specified overload Δn z The corresponding transcendental number ΔN obeys the lognormal distribution lgN(μ,σ 2 ), the distribution parameters μ and σ are estimated using the maximum likelihood estimation method 2 , the estimation formulas are shown in formula (9) and formula (10) respectively.

[0073]

[0074]

[0075] The distribution parameters of historical load data and measured take-off and landing data are μ1, σ1 and μ2, σ2 respectively. The Bayesian method is used to calculate the μ and variance σ of the two sets of data. 2 Perform fusion estimation, and the posterior distribution expression is as follows:

[0076]

[0077] Where, x = lgΔN; P(μ,σ 2 ) is the prior distribution.

[0078] According to the central limit theorem, the prior distribution of μ is taken as normal distribution Variance σ 2 The prior distribution is taken as the inverse gamma distribution Inverse-Gamma((n1-1) / 2,(n1-3)σ1 2 / 2), whose parameters are chosen to ensure that the expected value of the prior distribution is close to the estimated value of the sample variance σ1 2 The shape parameter a = (n1-1) / 2 corresponds to the degrees of freedom of the sample variance in the normal model.

[0079] μ and σ 2 The joint prior distribution of is:

[0080]

[0081] Where, n1 represents the number of prior data; P represents μ and σ 2 The joint prior distribution of Γ represents the chi-square distribution; μ represents the general term for the mean of the distribution parameters.

[0082] S4. Introduce weight factors into the statistical distribution parameters and measured load data respectively, and determine the weight factors by maximizing mutual information.

[0083] The likelihood function P(x|μ,σ) of the measured data 2 ) is shown in formula (13):

[0084]

[0085] Among them, x1 represents the first measured data; x n2 represents the n2th measured data; Π represents cumulative multiplication; x i represents the i-th measured data; n2 represents the number of measured data.

[0086] Since the denominator of formula (11) is a marginal distribution that is independent of the parameters μ and σ, and the posterior distribution depends only on the numerator, formula (11) can be rewritten as the following equivalent form:

[0087] P(μ,σ 2 |x)∝P(μ,σ 2 )P(x|μ,σ 2 )(14)

[0088] In the formula, ∝ means proportional, and the right side of the formula is the posterior distribution P(μ,σ 2 |x)'s core.

[0089] Assign weight factors α and β to the two terms on the right side of equation (14), and define the weighted kernel function as F(x):

[0090] F(x)=(P(μ,σ 2 )) α (P(x|μ,σ 2 )) β (15)

[0091] The weight factors α and β satisfy:

[0092]

[0093] Wherein, c is a constant (2 in this embodiment), corresponding to the original weight setting (i.e., α=1 and β=1). In this embodiment, α min and β min Take 0.5, α max and β max The purpose of taking 1.5 is to ensure that the minimum influence of prior and measured information can be guaranteed, so as to achieve a reasonable balance between prior and measured information in the model.

[0094] S4. Based on the weight factors, the statistical distribution parameters and the measured load data are integrated to obtain the average overload cumulative exceedance curve, and then compile a 5×5 spectrum.

[0095] After determining the weighting factors α and β, refer to the maximum a posteriori estimation method and calculate μ and σ respectively by 2 The partial derivatives are calculated to be 0 to obtain the fusion parameters of the two.

[0096] The weight factors α and β of the prior and measured information are determined by maximizing the mutual information. The mutual information is shown in formula (17):

[0097]

[0098] In the formula, I(μ,σ 2 ; x) represents mutual information; P(μ, σ 2 , x i ) represents the joint distribution function.

[0099] P(μ,σ 2 ,x i )=P(μ,σ 2 )P(x i |μ,σ 2 ) (18)

[0100] Substituting (18) into (17), we get

[0101]

[0102] In the process of maximizing mutual information, weight factors α and β are introduced to weight the prior information and measured data information respectively. The weighted mutual information is shown in formula (20).

[0103]

[0104] Where, P(x i ) is x i The marginal distribution of is shown in formula (21).

[0105]

[0106] Calculating the above integral is actually calculating the conditional probability P(x i |μ,σ 2 ) in the parameter distribution P(μ,σ 2 ) takes the expectation down, that is

[0107]

[0108] Monte Carlo sampling method is used to calculate P(x i ) to perform numerical integration. The main steps are as follows:

[0109] 1) From the prior distribution of μ and variance σ 2 Prior distribution of Inverse-Gamma((n1-1) / 2,(n1-3)σ1 2 / 2) to sample μ j and σ j Combination, this embodiment samples 100,000 times respectively;

[0110] 2) Calculate P(x) for each combination i |μ j ,σ j );

[0111] 3) By summing and averaging, we can get P(x i ) is approximate:

[0112]

[0113] At the same time, it is necessary to normalize the probability density function of the weighted prior distribution and the likelihood function of the measured data to ensure that the integral over the entire domain is 1 and maintain the basic characteristics of the probability density function.

[0114] First, the weighted prior distribution probability density function P(μ,σ 2 ) α Perform normalization processing, where the normalization parameter C1 needs to satisfy:

[0115]

[0116] Right now:

[0117]

[0118] The double integral is calculated using the Monte Carlo numerical integration method described above:

[0119]

[0120] Similarly, the likelihood function P(x i |μ,σ 2 ) β Perform normalization processing, where the normalization parameter C2 needs to satisfy:

[0121]

[0122] Right now:

[0123]

[0124] Substituting formula (14) into formula (31), we get:

[0125]

[0126] Applying Gauss's integral formula make We can get:

[0127]

[0128] The weighted mutual information after normalization is:

[0129]

[0130] Substituting equations (12), (13), (23), (26) and (30) into equation (31) yields the final mutual information expression.

[0131] To calculate the numerical integral of the mutual information and find the weight factors α and β that maximize the mutual information, the Monte Carlo sampling numerical integral combined with grid search optimization strategy is also used. The specific steps are as follows:

[0132] 1) Initialize the weight factor range: set the candidate value range of α and β, as shown in formula (15), and the step size, which is 0.01 in this paper;

[0133] 2) Numerical integration calculation: For each set of candidate weight factors α and β, the mutual information value is calculated by numerical integration method; the specific method is to use the above-mentioned Monte Carlo numerical integration method to calculate each x iThe corresponding double integral G(x i ):

[0134]

[0135] By G(x i ) and then we get the mutual information value I(μ,σ 2 ; x):

[0136]

[0137] 3) Grid search optimization: traverse all candidate weight combinations and record the mutual information value corresponding to each group (α, β);

[0138] 4) Take the maximum value: Determine the weight combination (α * ,β * ), as the final optimized solution.

[0139] Set α=α * , β=β * Substituting equations (11) and (12) into equation (17), the weighted kernel function F(x) can be expressed as:

[0140]

[0141] Refer to the maximum a posteriori estimation method, according to the function F(x) for μ and σ 2 The target parameters μ and σ are solved by taking the derivative to zero, as shown in Equations (35) and (36):

[0142]

[0143]

[0144] For the i-th level gust overload of each mission segment, calculate the exceedance number ΔN with a coverage rate of 50% 50,i :

[0145]

[0146] Where μ i and σ i are the mean and standard deviation of the logarithmic exceedance corresponding to the i-th level gust overload, z 50 is the quantile corresponding to the cumulative probability of 50% in the standard normal distribution N(0,1). Due to the symmetry of the standard normal distribution, z 50 =0, so formula (37) can be simplified to:

[0147]

[0148] Furthermore, the transcendental number ΔN50,i The cumulative transcendental number N is obtained by accumulation i :

[0149]

[0150] Where m represents the total number of gust overload levels.

[0151] Based on the above cumulative transcendence number N i , formula (5) is used to fit the average gust overload cumulative exceedance curve of each mission segment.

[0152] The 5×5 spectral load matrix obtained by satisfying the extreme load lognormal distribution criterion and load similarity criterion is shown in Table 2:

[0153] Table 2

[0154]

[0155] Here are the steps:

[0156] 1) Determine the number of occurrences y of each flight type i ;

[0157] 2) Introducing load level ratio R i , as shown in formula (40):

[0158] R i =(LB i -LB i+1 ) / (LB1-LB n ). (40)

[0159] The average overload cumulative exceedance curve based on the task segment is discretized into 4 segments, and a straight line is used to replace the curve of the discrete segment. The linear equation of the i-th segment is:

[0160] Δn z =a i lg N+b i (41)

[0161] Where a i 、b i is the fitting constant of the load spectrum curve of the i-th segment.

[0162] The equivalent overload corresponding to the load spectrum curve of the i-th segment is the abscissa corresponding to the midpoint of the curve:

[0163]

[0164] The number of equivalent load cycles is the difference between the cumulative exceedances corresponding to the two end points of the curve:

[0165]

[0166] Corresponding load level ratio R i and the number of occurrences y of each flight type in 1) i Repeated iterations are performed to make each level equivalent overload Δn yd The extreme load lognormal distribution criterion is met.

[0167] 3) According to 2) (Δn yd ,N eq ) j And the gust similarity criterion determines the number of occurrences of each level of load under each flight type B i,j .

[0168] According to the data in the compiled 5×5 spectrum based on the mission segment, the peak and valley values are randomly sorted to compile the mission segment spectrum, and the mission profile spectrum is compiled according to the order of the mission segments. The typical mission profile spectra are randomly sorted according to the proportion of the typical mission profile in the basic life unit to obtain the final flight-continuation flight spectrum.

[0169] Example 2

[0170] This embodiment also provides a gust load spectrum compilation system for transport aircraft based on maximum mutual information-Bayesian fusion, including: a measured data acquisition module, a historical data acquisition module, a construction module, an introduction module and a compilation module; the measured data acquisition module is used to obtain the measured load data of the aircraft for which the load spectrum is to be compiled; the historical data acquisition module is used to obtain historical load data of various types of transport aircraft; the construction module is used to establish statistical distribution parameters of historical load data based on a Bayesian framework; the introduction module is used to introduce weight factors into the statistical distribution parameters and the measured load data respectively, and determine the weight factors by maximizing mutual information; the compilation module is used to fuse the statistical distribution parameters and the measured load data based on the weight factors, obtain an average overload cumulative exceedance curve, and then compile a 5×5 spectrum.

[0171] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A method for compiling gust load spectra for transport aircraft based on maximum mutual information-Bayesian fusion, characterized in that the steps include: Obtain the measured load data of the aircraft for which the load spectrum is to be compiled; Obtain historical load data for several types of transport aircraft; Establishing statistical distribution parameters of the historical load data based on a Bayesian framework; Introducing weight factors into the statistical distribution parameters and the measured load data respectively, and determining the weight factors by maximizing mutual information; Based on the weight factors, the statistical distribution parameters and the measured load data are integrated to obtain an average overload cumulative exceedance curve, and then a 5×5 spectrum is compiled.

2. The method for compiling gust load spectrum for transport aircraft based on maximum mutual information-Bayesian fusion according to claim 1 is characterized in that: Collect the measured load data of the aircraft to be loaded, and use real-time mass G i Correct the normal overload data of the aircraft center of gravity: Δn z =(n zi -1)*G i / G0 Where Δn z is the incremental overload value corrected according to the standard mission section; n zi is the measured real-time overload value; G0 is 18000kg.

3. The method for compiling gust load spectrum for transport aircraft based on maximum mutual information-Bayesian fusion according to claim 1 is characterized in that: Collect the cumulative exceedance data of discrete gust speeds of multiple transport aircraft, and obtain the historical load data through screening and analysis; merge the exceedance numbers of upward and downward gusts in the historical load data to form the gust speed per nautical mile U de Cumulative transcendental number curve family: Where N represents the cumulative number of exceedances.

4. The method for compiling gust load spectrum for transport aircraft based on maximum mutual information-Bayesian fusion according to claim 3 is characterized in that: The gust speed in the historical load data is converted into gust incremental overload according to the structural response characteristics of the aircraft in the load profile to be compiled: Where: Δn represents the gust load coefficient increment; R represents the vertical gust response parameter of the wing and fuselage; U de represents the converted gust speed; ρ0 represents the air density at sea level; C Na Indicates the slope of the lift line; V e represents equivalent airspeed; G / S represents wing loading; K W represents the discrete gust mitigation factor; μ g represents the aircraft mass parameter; g represents the acceleration due to gravity; represents the average geometric chord length of the wing; ρ h Indicates the air density at the altitude.

5. The method for compiling gust load spectrum for transport aircraft based on maximum mutual information-Bayesian fusion according to claim 1 is characterized in that: Establishing the statistical distribution parameters of the historical load data includes: Where, σ 2 A general term for variance; μ1, σ1 2 represents the mean and variance of historical load data; σ1 represents the standard deviation of historical load data; n1 represents the number of historical prior data; P represents μ and σ 2 The joint prior distribution of ; Γ represents the chi-square distribution; μ represents the general term for the mean.

6. The method for compiling gust load spectrum for transport aircraft based on maximum mutual information-Bayesian fusion according to claim 5 is characterized in that: The weight factor α of the statistical distribution parameter and the weight factor β of the measured load data satisfy: Where c is a constant.

7. The method for compiling gust load spectrum for transport aircraft based on maximum mutual information-Bayesian fusion according to claim 6 is characterized in that: The method of determining the weight factor by maximizing mutual information includes: In the formula, I(μ,σ 2 ; x) represents mutual information; x i represents the i-th measured data; P(μ, σ 2 , x i ) represents the joint distribution; P(μ,σ 2 ) represents the prior distribution; P(x i ) is x i The marginal distribution of .

Images