Fatigue life prediction method for structures under broadband non-Gaussian random loads

Abstract

The present invention relates to the technical field of structural health monitoring and fault diagnosis, and specifically to a method for predicting the fatigue life of a structure under a broadband non-Gaussian random load, comprising: S1, solving the fatigue damage under a non-Gaussian random vibration load and a Gaussian random vibration load by a time domain method; S2, calculating a damage correction coefficient according to the fatigue damage under the non-Gaussian random vibration load and the Gaussian random vibration load; S3, fitting the calculated damage correction coefficient with the load kurtosis, skewness, structural damping ratio and material parameters to determine a functional expression of the damage correction coefficient; S4, applying a power spectrum load to a finite element model of the structure, and calculating the fatigue damage under the current load by a frequency domain method; S5, determining the damage correction coefficient under the current load according to the functional expression of the damage correction coefficient; S6, multiplying the damage correction coefficient calculated by S5 with the fatigue damage calculated by S4, taking the product result as the fatigue damage under the non-Gaussian random vibration load, and taking the inverse of the calculated damage value as the fatigue predicted life of the structure. The present invention combines the characteristics of the accuracy of the time domain method and the simplicity of the frequency domain method, so that the calculation of the fatigue damage of the structure under the non-Gaussian random vibration load and the life prediction results have high accuracy and low dispersion.

Description

Technical Field

[0001] The present invention relates to the technical field of structural health monitoring and fault diagnosis, and more particularly to a method for predicting structural fatigue life under broadband non-Gaussian random loads. Background Art

[0002] With the rapid development and progress of modern aerospace and special vehicle equipment, core components are often subjected to random vibration loads caused by various complex movements. The fatigue problems caused by these random vibration loads put forward higher requirements on the service life and reliability of the structure. Complex random vibration loads are an important cause of equipment failure. Gaussian random vibration loads are often used for calculations in engineering. However, during the actual operation of equipment components, due to the complexity of environmental conditions, the excitations to the equipment have strong non-Gaussian characteristics. In particular, for fatigue damage caused by non-Gaussian random vibration loads, in many engineering cases it is simplified to Gaussian conditions for processing. However, a large number of research results show that such simplifications will lead to greater errors in the calculation of fatigue damage life. This will also lead to a certain probability of fatigue failure of key core components on the equipment during field use after passing the traditional steady Gaussian load assessment.

[0003] So far, there has been a deep accumulation of research on fatigue of structures under Gaussian random vibration loads. Researchers hope to find a method that can reasonably and reliably predict the fatigue damage calculation and life of structures under non-Gaussian random vibration loads. The current methods are mainly based on two processing methods: time domain and frequency domain.

[0004] For the time domain method, it is necessary to know the stress response at the dangerous point. However, a long stress history can include random vibration loads of various working conditions. Although this type of method has accurate calculation results and small errors, it is not suitable for actual engineering due to problems such as large calculation amount and long operation time.

[0005] For the frequency domain method, since the current frequency domain method is proposed for Gaussian random vibration loads, direct use of the frequency domain method to calculate damage under non-Gaussian random vibration loads will result in large errors. At present, the main method is to decompose the non-Gaussian random vibration load into a series of Gaussian random vibration loads, and then solve the fatigue damage life under each Gaussian random vibration load. Or first solve the stress amplitude probability density function according to the frequency domain method, and then convert the Gaussian stress amplitude probability density function into the non-Gaussian stress amplitude probability density function through a nonlinear transformation function to calculate the corresponding damage life.

[0006] For the response of linear structures under non-Gaussian random vibration loads with a given power spectral density function, the kurtosis and skewness of the input load and the damping ratio of the structure are closely related to the output characteristics of the response, and these characteristics directly determine the damage life of the structure. Therefore, when predicting the damage and life of structures under broadband non-Gaussian random vibration loads, the influence of these key parameters needs to be considered. Summary of the invention

[0007] In view of this, the present invention provides a method for predicting structural fatigue life under broadband non-Gaussian random loads, which combines the accuracy of the time domain method and the simplicity of the frequency domain method, so that the structural fatigue damage calculation and life prediction results under non-Gaussian random vibration loads have higher accuracy and less dispersion.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] A method for predicting fatigue life of a structure under broadband non-Gaussian random loads comprises the following steps:

[0010] S1. Solve the fatigue damage under non-Gaussian random vibration load and Gaussian random vibration load by time domain method;

[0011] S2, calculate the damage correction factor based on fatigue damage under non-Gaussian random vibration load and Gaussian random vibration load;

[0012] S3, fitting the calculated damage correction coefficient with the load kurtosis, skewness, structural damping ratio and material parameters to determine the functional expression of the damage correction coefficient;

[0013] S4, applying power spectrum load to the finite element model of the structure, and calculating the fatigue damage under the current load by frequency domain method;

[0014] S5. determining the damage correction coefficient under the current load according to the functional expression of the damage correction coefficient;

[0015] S6. Multiply the damage correction factor calculated by S5 by the fatigue damage calculated by S4, use the product as the fatigue damage under non-Gaussian random vibration load, and use the inverse of the calculated damage value as the fatigue prediction life of the structure.

[0016] Furthermore, S1 includes:

[0017] S11. Based on the selected broadband power spectrum, a Gaussian random vibration load is generated by using independent random phase angle and discrete inverse Fourier transform method, and a non-Gaussian random vibration load with specific kurtosis, skewness and root mean square value is modulated by a nonlinear transformation function;

[0018] S12, using the non-Gaussian random vibration load and Gaussian random vibration load generated in S11 to perform transient analysis in the finite element model, determine the dangerous point with the largest equivalent stress value in the finite element model, and extract the equivalent stress-time history for the dangerous point;

[0019] S13. The damage of the equivalent stress-time history extracted in S12 is calculated using the rain flow counting method and the Miner damage accumulation criterion, and the fatigue damage under non-Gaussian random vibration load and Gaussian random vibration load are obtained respectively.

[0020] Furthermore, the nonlinear transformation function used in S11 is a monotonically increasing function.

[0021] Furthermore, in S2, the damage correction coefficient is calculated as follows:

[0022]

[0023] in, represents fatigue damage under Gaussian random vibration load, Represents fatigue damage under non-Gaussian random vibration loading.

[0024] Furthermore, S3 includes:

[0025] S31, based on the damage correction coefficient calculated in S2, establish the damage correction coefficient and skewness γ 3 , kurtosisγ 4 , a quadratic polynomial response surface model between the structural damping ratio ξ and the material parameter m;

[0026] S32. Use the damage correction coefficient calculated in S2 and the finite element model parameters to solve the coefficients of the quadratic polynomial response surface model and determine the functional expression of the damage correction coefficient.

[0027] Furthermore, the functional expression of the damage correction coefficient is:

[0028] λ=a 0 +a 1 γ 3 +a 2 γ 4 +a 3 m+a 4 ξ+a 5 γ 3 2 +a 6 γ 4 2 +a 7 m 2 +a 8 ξ 2 +a 9 γ 3 γ4 +a 10 γ 3 m+a 11 γ 3 ξ+a 12 γ 4 m+a 13 γ 4 ξ+a 14 mξ

[0029] Among them, γ 3 represents the load skewness, γ 4 represents the load kurtosis, ξ represents the structural damping ratio, m represents the material parameter, a i (i=0,1,2…14) represents the coefficients of the quadratic polynomial response surface model.

[0030] Furthermore, S4 includes:

[0031] S41. Apply acceleration power spectrum load to the finite element model, extract the equivalent stress power spectrum density function at the dangerous point through random vibration analysis, and calculate the equivalent stress amplitude probability density function of the power spectrum according to the Dirlik method;

[0032] S42. Calculate the fatigue damage under random vibration load based on the probability density function of equivalent stress amplitude.

[0033] Furthermore, in S41, the probability density function formula of the equivalent stress amplitude calculated by the Dirlik method is:

[0034]

[0035]

[0036]

[0037] in, D 3 =1-D 1 -D 2 ,

[0038] m 0 ,m 1 ,m 2 ,m 4 The four moments of inertia of the power spectral density function are expressed as:

[0039]

[0040] Where G(f) is the power spectrum density function and f is the power spectrum frequency.

[0041] Further, in S42, the expression for calculating fatigue damage under random vibration load according to the probability density function of equivalent stress amplitude is:

[0042]

[0043] Among them, E[0 + ] represents the frequency of the stress amplitude crossing the value of 0, A represents the material parameter, T represents the total time, p(S) is the probability density function of the equivalent stress amplitude obtained by the Dirlik method, and S is the stress amplitude.

[0044] Furthermore, in S6, the calculation formula for structural fatigue damage under non-Gaussian random vibration load is:

[0045]

[0046] The fatigue prediction life of the structure is expressed as:

[0047]

[0048] in, represents fatigue damage under Gaussian random vibration load, represents fatigue damage under non-Gaussian random vibration load, and λ is the damage correction factor.

[0049] It can be seen from the above technical solutions that, compared with the prior art, the present invention has the following beneficial effects:

[0050] (1) The present invention establishes the relationship between non-Gaussian random vibration load and fatigue damage under Gaussian random vibration load process through non-Gaussian damage correction coefficient, estimates fatigue damage under Gaussian random process based on Gaussian approximation method, and obtains structural fatigue damage under non-Gaussian random process, and establishes an analysis process for estimating structural fatigue damage life under broadband non-Gaussian random vibration load.

[0051] (2) The present invention establishes a quadratic polynomial response surface model between the correction coefficient and the skewness, kurtosis, structural damping ratio, and material parameters (the slope of the SN double logarithmic curve), and uses the correction coefficient to fit the response surface model coefficient, thereby determining the functional expression of the correction coefficient. The calculation process is relatively simple and the accuracy is high;

[0052] (3) The present invention combines the accurate results of the time domain method and the simple calculation of the frequency domain method, so that the fatigue damage calculation and life prediction results of the structure under non-Gaussian random vibration load have small dispersion and high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0054] Figure 1 A flow chart of a method for predicting structural fatigue life under broadband non-Gaussian random loads provided by the present invention;

[0055] Figure 2 A schematic diagram of the fatigue test piece structure provided by the present invention;

[0056] FIG3( a ) is a stress-time history curve diagram under Gaussian random vibration load obtained by the present invention;

[0057] FIG3( b ) is a stress-time history curve diagram under non-Gaussian random vibration load obtained by the present invention;

[0058] Figure 4 The power spectrum density function curve diagram of the dangerous point equivalent stress extracted by the present invention;

[0059] Figure 5 It is a schematic diagram comparing the prediction results and simulation results of the structural specimen by the method of the present invention. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0061] like Figure 1 As shown, the embodiment of the present invention discloses a method for predicting fatigue life of a structure under a broadband non-Gaussian random load, comprising the following steps:

[0062] S1. Solve the fatigue damage under non-Gaussian random vibration load and Gaussian random vibration load by time domain method;

[0063] S2, calculate the damage correction factor based on fatigue damage under non-Gaussian random vibration load and Gaussian random vibration load;

[0064] S3, fitting the calculated damage correction coefficient with the load kurtosis, skewness, structural damping ratio and material parameters to determine the functional expression of the damage correction coefficient;

[0065] S4, applying power spectrum load to the finite element model of the structure, and calculating the fatigue damage under the current load by frequency domain method;

[0066] S5. determining the damage correction coefficient under the current load according to the functional expression of the damage correction coefficient;

[0067] S6. Multiply the damage correction factor calculated by S5 by the fatigue damage calculated by S4, use the product as the fatigue damage under non-Gaussian random vibration load, and use the inverse of the calculated damage value as the fatigue prediction life of the structure.

[0068] Below, each step of the present invention is further described.

[0069] S1. Through transient analysis of the specimen model, damage calculation is performed by extracting the equivalent stress-time history of the dangerous point. The specimen structure and size are as follows: Figure 1 The material parameters used in the specimen model are shown in Table 1.

[0070] Table 1 Material parameters of the specimen model

[0071] parameter value Young's modulus 210GPa density <![CDATA[7890Kg / m 3 ]]> Poisson's ratio 0.3 Yield Strength 400MPa Ultimate tensile strength 700MPa

[0072] S11. Based on the selected broadband power spectrum, a Gaussian random vibration load X(T) is generated using independent random phase angles and discrete inverse Fourier transform methods. A nonlinear transfer function G(·) is introduced so that the random process Z(t) of Z(t)=G(X(T)) is a non-Gaussian random vibration load with specific kurtosis, skewness and root mean square value, and multiple groups of non-Gaussian random vibration loads with different kurtosis and skewness are generated.

[0073] The present invention introduces a nonlinear transformation function (Hermite function) for modulation, which is a monotonically increasing function, and its expression is:

[0074]

[0075] Among them, κ is the coefficient that ensures that the non-Gaussian random process and the Gaussian random process have the same root mean square value, γ 3 is the skewness of the non-Gaussian process, γ 4 is the kurtosis of the non-Gaussian process, X 0 is a standard normal random process, X 0 =(X-μ x ) / σ x . The expressions of kurtosis and skewness are:

[0076]

[0077] Among them, μ z and σ Zare the mean and variance of the non-Gaussian random vibration load, respectively.

[0078] S12. Use the non-Gaussian random vibration load and Gaussian random vibration load generated in S11 to perform transient analysis in the finite element model, determine the dangerous point with the maximum equivalent stress value of the finite element model, and extract the equivalent stress-time history of the dangerous point.

[0079] S13. The damage of the equivalent stress-time history extracted in S12 is calculated using the rain flow counting method and the Miner damage accumulation criterion, and the fatigue damage under non-Gaussian random vibration load and Gaussian random vibration load are obtained respectively.

[0080]

[0081] S2. Calculate the damage correction coefficient based on the damage value of the structure under non-Gaussian random vibration load and Gaussian random vibration load, and establish the functional relationship between the correction coefficient and the skewness, kurtosis, damping ratio, and the slope of the material SN double logarithmic curve:

[0082]

[0083] in, represents fatigue damage under Gaussian random vibration load, represents fatigue damage under non-Gaussian random vibration load. λ represents damage correction factor. γ 3 represents the load skewness, γ 4 represents the load kurtosis, ξ represents the structural damping ratio, and m represents the material parameter (the slope of the SN double logarithmic curve). A total of multiple groups of samples were simulated, 12 of which are given here. The non-Gaussian load related parameters and damage correction coefficients are shown in Table 2.

[0084] Table 2 Non-Gaussian load related parameters and damage correction coefficients

[0085]

[0086]

[0087] S3. Fit the calculated damage correction coefficient with the load kurtosis, skewness, structural damping ratio and material parameters to determine the functional expression of the damage correction coefficient; specifically including:

[0088] S31, based on the damage correction coefficient calculated in S2, establish the damage correction coefficient and skewness γ 3 , kurtosisγ 4 , structural damping ratio ξ, and material parameter m (slope of the SN double logarithmic curve):

[0089] λ=a 0 +a1 γ 3 +a 2 γ 4 +a 3 m+a 4 ξ+a 5 γ 3 2 +a 6 γ 4 2 +a 7 m 2 +a 8 ξ 2 +a 9 γ 3 γ 4 +a 10 γ 3 m+a 11 γ 3 ξ+a 12 γ 4 m+a 13 γ 4 ξ+a 14 mξ

[0090] Among them, a i (i=0,1,2…14) represents the coefficients of the quadratic polynomial response surface model.

[0091] S32. Use the damage correction coefficient calculated in S2 and the finite element model parameters to solve the coefficients of the quadratic polynomial response surface model and determine the functional expression of the damage correction coefficient.

[0092] Since the damage correction coefficient and the parameters of the finite element model are known, these parameters are substituted into the quadratic polynomial response surface model equation. By solving or fitting at least 15 equations, a can be obtained. 0 —a 14 These 15 unknown coefficients.

[0093] The coefficients of the polynomial response surface model are equivalent to the constant coefficients obtained by solving or fitting. The relationship between the model parameters and the damage correction coefficient is established through this coefficient. The damage correction coefficient is the dependent variable that needs to be obtained, and the finite element model coefficient is equivalent to the input independent variable.

[0094] The embodiment of the present invention uses the least square method to fit the coefficients of the polynomial response surface model according to the damage correction coefficient. Table 3 shows the coefficients of the fitted quadratic polynomial response surface model.

[0095] Table 3 Coefficients of quadratic polynomial response surface model

[0096] <![CDATA[a 0 ]]> <![CDATA[a 1 ]]> <![CDATA[a 2 ]]> <![CDATA[a 3 ]]> <![CDATA[a 4 ]]> -20.04 145.96 50.35 -37.63 -349.12 <![CDATA[a 5 ]]> <![CDATA[a 6 ]]> <![CDATA[a 7 ]]> <![CDATA[a 8 ]]> <![CDATA[a 9 <!-- 6 -->]]> 198.21 -7.52 -0.22 -69.01 -101.71 <![CDATA[a 10 ]]> <![CDATA[a 11 ]]> <![CDATA[a 12 ]]> <![CDATA[a 13 ]]> <![CDATA[a 14 ]]> 13.19 4392 6.93 -100.86 121.27

[0097] S4. Apply power spectrum load to the finite element model of the structure and calculate the fatigue damage under the load by frequency domain method, including:

[0098] S41. Apply acceleration power spectrum load to the finite element model, and extract the equivalent stress power spectrum density function at the dangerous point through random vibration analysis. Figure 4 The power spectrum density function curve of the equivalent stress extracted at the dangerous point position is used, and then the probability density function of the equivalent stress amplitude of the power spectrum is calculated according to the Dirlik method. The expression is:

[0099]

[0100] in,

[0101] m 0 ,m 1 ,m 2 ,m 4 The four moments of inertia of the power spectral density function are expressed as:

[0102]

[0103] Where G(f) is the power spectrum density function and f is the power spectrum frequency.

[0104] S42. Calculate the fatigue damage under random vibration load based on the probability density function of equivalent stress amplitude. The calculation formula is:

[0105]

[0106] Among them, E[0 + ] represents the frequency of the stress amplitude crossing the value of 0, A represents the material parameter, T represents the total time, p(S) is the probability density function of the equivalent stress amplitude obtained by the Dirlik method, and S is the stress amplitude.

[0107] S5. Determine the damage correction coefficient according to the functional expression of the damage correction coefficient in S3.

[0108] S6. The damage correction coefficient is multiplied by the damage value calculated by the frequency domain method as the structural fatigue damage under broadband non-Gaussian random vibration load. The calculation formula is:

[0109]

[0110] The inverse of the damage value is taken as the fatigue prediction life of the structure, and the expression is:

[0111]

[0112] in, represents fatigue damage under Gaussian random vibration load, represents fatigue damage under non-Gaussian random vibration load, and λ is the damage correction factor.

[0113] Figure 5 The figure is a comparison chart of the prediction results and simulation results according to the method of the present invention. The simulation results are obtained by applying non-Gaussian random time domain loads to perform transient analysis in finite element simulation software, extracting the equivalent stress-time history at the dangerous point, and then calculating the life in the time domain according to the rain flow counting method.

[0114] pass Figure 5 The comparison between the simulation results and the prediction results in FIG. 1 shows that the prediction results obtained by combining the frequency domain analysis results and the correction coefficients are close to the simulation results. Since the simulation results are calculated based on the time domain method, the results are relatively accurate. Therefore, the relatively accurate results obtained by combining the frequency domain analysis and the correction coefficients in the present invention can prove the rationality and effectiveness of the method of the present invention.

[0115] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0116] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting structural fatigue life under broadband non-Gaussian random loads, characterized in that: The following steps are involved: S1. Solve the fatigue damage under non-Gaussian random vibration load and Gaussian random vibration load by time domain method; S2, calculate the damage correction factor based on fatigue damage under non-Gaussian random vibration load and Gaussian random vibration load; S3, fitting the calculated damage correction coefficient with the load kurtosis, skewness, structural damping ratio and material parameters to determine the functional expression of the damage correction coefficient; S3 includes: S31, based on the damage correction coefficient calculated in S2, a quadratic polynomial response surface model between the damage correction coefficient and the skewness γ3, kurtosis γ4, structural damping ratio ξ, and material parameter m is established; S32, using the damage correction coefficient calculated in S2 and the finite element model parameters to solve the coefficients of the quadratic polynomial response surface model, and determine the function expression of the damage correction coefficient; S4, applying power spectrum load to the finite element model of the structure, and calculating the fatigue damage under the current load by frequency domain method; S5. determining the damage correction coefficient under the current load according to the functional expression of the damage correction coefficient; S6. Multiply the damage correction coefficient calculated by S5 by the fatigue damage calculated by S4, and use the product as the fatigue damage under non-Gaussian random vibration load. Use the inverse of the calculated damage value as the fatigue prediction life of the structure. The calculation formula for structural fatigue damage under non-Gaussian random vibration load is: The fatigue prediction life of the structure is expressed as: in, represents fatigue damage under Gaussian random vibration load, represents fatigue damage under non-Gaussian random vibration load, and λ is the damage correction factor.

2. The method for predicting structural fatigue life under broadband non-Gaussian random loads according to claim 1 is characterized in that: S1 includes: S11. Based on the selected broadband power spectrum, a Gaussian random vibration load is generated by using independent random phase angle and discrete inverse Fourier transform method, and a non-Gaussian random vibration load with specific kurtosis, skewness and root mean square value is modulated by a nonlinear transformation function; S12, using the non-Gaussian random vibration load and Gaussian random vibration load generated in S11 to perform transient analysis in the finite element model, determine the dangerous point with the largest equivalent stress value in the finite element model, and extract the equivalent stress-time history for the dangerous point; S13. The damage of the equivalent stress-time history extracted in S12 is calculated using the rain flow counting method and the Miner damage accumulation criterion, and the fatigue damage under non-Gaussian random vibration load and Gaussian random vibration load are obtained respectively.

3. The method for predicting structural fatigue life under broadband non-Gaussian random loads according to claim 2 is characterized in that: The nonlinear transformation function used in S11 is a monotonically increasing function.

4. The method for predicting structural fatigue life under broadband non-Gaussian random loads according to claim 1 is characterized in that: In S2, the damage correction factor is calculated as: in, represents fatigue damage under Gaussian random vibration load, Represents fatigue damage under non-Gaussian random vibration loading.

5. The method for predicting structural fatigue life under broadband non-Gaussian random loads according to claim 1 is characterized in that: The functional expression of the damage correction coefficient is: λ=a0+a1γ3+a2γ4+a3m+a4ξ+a5γ3 2 +a6γ4 2 +a7m 2 +a8ξ 2 +a9γ3γ4+a 10 γ3m+a 11 γ3ξ+a 12 c4m+a 13 γ4ξ+a 14 max Among them, γ3 represents the load skewness, γ4 represents the load kurtosis, ξ represents the structural damping ratio, m represents the material parameter, and a i represents the coefficients of the quadratic polynomial response surface model, where i = 0, 1, 2, ..., 14.

6. The method for predicting structural fatigue life under broadband non-Gaussian random loads according to claim 1, characterized in that S4 include: S41. Apply acceleration power spectrum load to the finite element model, extract the equivalent stress power spectrum density function at the dangerous point through random vibration analysis, and calculate the equivalent stress amplitude probability density function of the power spectrum according to the Dirlik method; S42. Calculate the fatigue damage under random vibration load based on the probability density function of equivalent stress amplitude.

7. The method for predicting structural fatigue life under broadband non-Gaussian random loads according to claim 6 is characterized in that: In S41, the probability density function formula of the equivalent stress amplitude calculated by the Dirlik method is: in, The four moments of inertia of the power spectral density function of m0, m1, m2, m4 are expressed as: Where G(f) is the power spectrum density function and f is the power spectrum frequency.

8. The method for predicting structural fatigue life under broadband non-Gaussian random loads according to claim 6 is characterized in that: In S42, the expression for calculating fatigue damage under random vibration load based on the probability density function of equivalent stress amplitude is: Among them, E[0 + ] represents the frequency of the stress amplitude crossing the zero value, A represents the material parameter, T represents the total time, p(S) is the probability density function of the equivalent stress amplitude obtained by the Dirlik method; S is the stress amplitude.

Images