Method and system for evaluating fatigue crack life of welding structure under random load

Abstract

The invention provides a method and system for evaluating fatigue crack life of a welding structure under a random load, and relates to the technical field of rail transit, and the method comprises the steps: recognizing a load peak value and a load valley value through a point-by-point extreme value extraction method, and generating an equivalent variable-amplitude load data set; initial crack propagation rate parameters C and m and stress intensity factor distribution are calculated, and then a deterministic crack propagation increment result is output; a hierarchical Bayesian framework is combined with an MCMC sampling method, and global parameter distribution considering material heterogeneity and observation noise is generated through a posterior distribution formula; predicting crack depth evolution curves under different confidence coefficients, and outputting fatigue life probability distribution and confidence intervals; and finally generating a fatigue life reliability evaluation report of the welding structure under the random load. The method has the beneficial effects that the problem of fatigue crack propagation of the railway vehicle welding structure under the action of the complex random load can be better solved, and more accurate and reasonable technical support is provided for reliability evaluation of the welding structure.

Description

Technical Field

[0001] The present invention relates to the field of rail transportation technology, and in particular to a method and system for evaluating fatigue crack life of a welded structure under random loads. Background Art

[0002] In the field of rail transit, especially in the design and operation of rail vehicles such as subways, light rail and high-speed railways, welded structures are key parts that bear important loads, and their fatigue life assessment is of vital importance. Accurately predicting the fatigue crack growth life of rail vehicle welded structures under random loads in service is of great significance to ensuring the safe operation of rail transit, rationally planning maintenance cycles and improving the service life of structures. However, existing fatigue life assessment methods for welded structures have certain limitations when facing complex random loads during the service life of rail vehicles. Traditional fatigue life assessment methods are mostly based on simple constant amplitude or idealized variable amplitude load models. These methods fail to fully consider the random fluctuations and complexity of actual loads when considering the crack growth process. Although Mine's cumulative damage theory is widely used in the analysis of fatigue crack growth, it ignores the interaction between load sequences and the influence of small-amplitude load cycles, which may lead to deviations in the prediction of fatigue life.

[0003] Although the current mainstream fatigue crack growth prediction tools, such as NASGRO and FASTRAN, can handle the crack growth problem under variable amplitude load, they still face the following major challenges under random load conditions in service: (1) Insufficient load history simplification: Existing methods often simplify complex random load histories into equivalent variable amplitude load cycles through extreme point extraction. Although this can reduce the amount of calculation to a certain extent, this simplification method usually ignores the time series effect of the load, that is, the influence of the order and alternation of the load sequence on crack growth. For example, the crack growth rate will vary significantly under different load sequences, and existing models often assume that each load cycle is independent, which reduces the accuracy of the model. (2) Insufficient consideration of crack closure effect: Many existing fatigue crack growth models have a relatively simple consideration of crack closure effect, which plays an important role under complex load conditions (especially random load). Due to the frequent and irregular load fluctuations, cracks may close during the growth process, thereby slowing down the crack growth rate. However, current analysis tools, such as FASTRAN, although taking crack closure into account, often rely on simple load peak-valley pair decomposition, which cannot fully reflect the details of load fluctuations and the complex nonlinear effects in the crack propagation process. (3) Ignoring the uncertainty of model parameters: Most existing fatigue crack propagation models rely on deterministic assumptions. The crack propagation rate parameters in the model (such as C and m) are usually considered to be constants, while ignoring the changes in material properties, manufacturing defects and the influence of service environment on the crack propagation rate. In fact, there is a large uncertainty in these parameters, especially under complex service conditions, the impact of their volatility on crack propagation cannot be ignored. (4) Lack of accurate simulation of random loads: Although some studies have attempted to introduce randomness, most methods have failed to fully consider the complexity of crack propagation under random loads. Therefore, existing technologies are usually unable to effectively simulate the effects of load amplitude fluctuations, period changes and load sequence on fatigue crack propagation. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for assessing fatigue crack life of welded structures under random loads to improve the above-mentioned problems. To achieve the above-mentioned purpose, the technical solutions adopted by the present invention are as follows:

[0005] In a first aspect, the present application provides a method for evaluating fatigue crack life of a welded structure under random loads, comprising:

[0006] Based on the historical data of random loads in service, the load peaks and valleys are identified through point-by-point extreme value extraction. A variable amplitude load cycle is constructed, and a threshold compression algorithm is used to eliminate loads below the crack growth threshold to generate an equivalent variable amplitude load data set.

[0007] Using the equivalent variable amplitude load data set, the stress intensity factor range is calculated point by point, and combined with the two-stage crack growth formula, the initial crack growth rate parameters C, m and the stress intensity factor distribution are calculated, and then the deterministic crack growth increment result is output;

[0008] Based on the crack growth parameters C and m and the observed crack depth data, a hierarchical Bayesian framework combined with the MCMC sampling method is used to dynamically update the parameters C, m and the random fluctuation factor χ through the posterior distribution formula. A global parameter distribution that takes into account material heterogeneity and observation noise is generated through the global posterior distribution aggregation formula, where the parameters of each plate thickness group are independently defined.

[0009] Based on the global parameter distribution, the Monte Carlo simulation method is used to generate multiple random crack extension trajectories. Combined with the cumulative growth formula, the crack depth evolution curves under different confidence levels are predicted, and the fatigue life probability distribution and confidence interval are output;

[0010] The model accuracy of the fatigue life probability distribution and confidence interval is evaluated through error analysis and confidence verification algorithms. Combined with the initial crack size, plate thickness and load condition data, a fatigue life reliability assessment report of the welded structure under random loads is finally generated.

[0011] Preferably, based on the historical data of random loads in service, load peaks and valleys are identified by a point-by-point extreme value extraction method, a variable amplitude load cycle is constructed, and a threshold compression algorithm is used to eliminate loads below the crack growth threshold to generate an equivalent variable amplitude load data set, which includes:

[0012] Based on the historical data of random loads in service, the peaks and valleys in the load time history are identified by point-by-point extreme value extraction. The key load cycle information is extracted to form an equivalent variable amplitude load cycle, and then preliminary variable amplitude load cycle data is obtained. Each load cycle consists of a peak value and a valley value following it, representing the tensile and compressive process of the load.

[0013] For the loads below the crack growth threshold in the preliminary variable amplitude load cycle data, a threshold compression algorithm is used to process them. By comparing the stress intensity factor range and the crack growth threshold, the part below the threshold is compressed to remove its influence on crack growth, thus obtaining the optimized equivalent variable amplitude load cycle data.

[0014] The peak-valley matching is realized by Matlab program, and the optimized equivalent variable amplitude load cycle data are integrated to form an equivalent variable amplitude load data set.

[0015] Preferably, the equivalent variable amplitude load data set is used to calculate point by point through the stress intensity factor range, and combined with the two-stage crack growth formula, the initial crack growth rate parameters C, m and the stress intensity factor distribution are calculated, and then the deterministic crack growth increment result is output, including:

[0016] Using the equivalent variable amplitude load data set, the stress intensity factor range is calculated point by point, and combined with the two-stage crack growth formula, the initial crack growth rate parameters C, m and the stress intensity factor distribution are calculated to obtain the preliminary calculation results of the crack growth rate parameters.

[0017] Based on the preliminary calculation results of the crack growth rate parameters, the stress intensity factor range at the deepest point of the surface crack was gradually calculated under given initial surface cracks and geometric parameters of the welded structure. The crack growth increment was solved by an iterative method, and the preliminary results of the crack growth increment were obtained.

[0018] Based on the preliminary results of the crack growth increment, the relative size is determined by comparing the stress intensity factor range with the crack growth threshold of the full stress ratio range point by point. For loads less than the threshold, they are compressed, and then a new variable-amplitude driven stress intensity factor range is constructed to realize the iterative calculation of the crack growth increment and obtain the deterministic crack growth increment result.

[0019] Preferably, based on the crack growth parameters C, m and observed crack depth data, a hierarchical Bayesian framework combined with an MCMC sampling method is used to dynamically update the parameters C, m and the random fluctuation factor χ through the posterior distribution formula, and generate a global parameter distribution that takes into account material heterogeneity and observation noise through the global posterior distribution aggregation formula, wherein the parameters of each plate thickness group are independently defined, including:

[0020] Based on the crack growth parameters C, m and observed crack depth data, a hierarchical Bayesian framework combined with MCMC sampling method is used to dynamically update the parameters C, m and random fluctuation factor χ through the posterior distribution formula.

[0021] Define the parameters C, m, σ for each plate thickness group ∈ are independent random variables, for thickness group t k The parameters of the posterior distribution are calculated based on the observed data, where the likelihood function p(D k |θ k ) represents the given parameter θ k When the observation data D is generated k The probability of the prior distribution p(θ k ) describes θ k Prior information of

[0022] Combined with observation noise σ ∈, optimize the likelihood function, and generate a global parameter distribution that takes into account material heterogeneity and observation noise through the global posterior distribution aggregation formula, where the parameters of each plate thickness group are defined independently.

[0023] In a second aspect, the present application also provides a fatigue crack life assessment system for welded structures under random loads, comprising:

[0024] Construction module: Based on the historical data of random loads in service, the module uses a point-by-point extreme value extraction method to identify load peaks and valleys, constructs a variable amplitude load cycle, and uses a threshold compression algorithm to eliminate loads below the crack growth threshold to generate an equivalent variable amplitude load dataset.

[0025] Calculation module: Used to calculate the initial crack growth rate parameters C, m and stress intensity factor distribution using the equivalent variable amplitude load data set, point by point through the stress intensity factor range, and combined with the two-stage crack growth formula, and then output the deterministic crack growth increment result;

[0026] Generation module: Based on the crack growth parameters C and m and the observed crack depth data, a hierarchical Bayesian framework is used in conjunction with the MCMC sampling method to dynamically update the parameters C, m and the random fluctuation factor χ through the posterior distribution formula. The global posterior distribution aggregation formula is then used to generate a global parameter distribution that takes into account material heterogeneity and observation noise. The parameters of each plate thickness group are defined independently.

[0027] Prediction module: Based on the global parameter distribution, it uses the Monte Carlo simulation method to generate multiple random crack extension trajectories, and combines the cumulative growth formula to predict the crack depth evolution curves under different confidence levels, and output the fatigue life probability distribution and confidence interval;

[0028] Evaluation module: This module is used to evaluate the model accuracy of the fatigue life probability distribution and confidence interval through error analysis and confidence verification algorithms. It also combines the initial crack size, plate thickness, and load condition data to ultimately generate a fatigue life reliability assessment report for the welded structure under random loads.

[0029] In a third aspect, the present application also provides a device for evaluating fatigue crack life of welded structures under random loads, comprising:

[0030] Memory for storing computer programs;

[0031] A processor is configured to implement the steps of the method for evaluating fatigue crack life of a welded structure under random loads when executing the computer program.

[0032] In a fourth aspect, the present application also provides a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned method for evaluating fatigue crack life of welded structures under random loads.

[0033] The beneficial effects of the present invention are:

[0034] The present invention adopts a point-by-point extreme value extraction method to convert the complex random load history in the service process into an equivalent variable amplitude load cycle. By extracting the peak and valley values ​​in the load history, the complex load sequence can be simplified while retaining important load information, providing a more accurate input for crack growth analysis. Combining Bayesian inference and Markov chain Monte Carlo (MCMC) method, the parameters in the crack growth model (model parameters C, m, σ ∈ ) is dynamically updated, taking into account the randomness and uncertainty of the parameters. By introducing the hierarchical Bayesian method, it is possible to model according to the data of different plate thicknesses and initial crack states, and to quantify the uncertainty of the model parameters, thereby improving the prediction accuracy of the crack propagation life.

[0035] When dealing with crack propagation, the present invention fully considers the influence of crack closure effect and load amplitude fluctuation on crack propagation rate. Through the improved variable amplitude load analysis method, it can accurately simulate the nonlinear behavior during crack propagation, especially in the case of large load fluctuations, and provide more reliable crack propagation prediction results.

[0036] By combining random load characteristics, crack propagation models and Bayesian updating technology, the present invention can comprehensively model the fatigue crack propagation process of welded structures, and sample and adjust the random fluctuation factors in the crack propagation process through the MCMC method. It can effectively deal with the randomness caused by factors such as environmental changes and load fluctuations, and provide more comprehensive and accurate prediction results for the fatigue life assessment of welded structures.

[0037] The present invention can accurately predict the fatigue crack growth life of welded structures under complex random load conditions, providing reliable technical support for the safety assessment of engineering structures such as rail vehicles. At the same time, it can better deal with the fatigue crack growth problem of rail vehicle welded structures under complex random loads, and provide more accurate and reasonable technical support for the reliability assessment of welded structures.

[0038] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the embodiments of the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0040] Figure 1 Schematic diagram of the process of the fatigue crack life assessment method of a welded structure under random loads according to an embodiment of the present invention;

[0041] Figure 2 Schematic diagram of the fatigue crack life assessment system for welded structures under random loads according to an embodiment of the present invention;

[0042] Figure 3 Schematic diagram of the structure of the equipment for evaluating fatigue crack life of welded structures under random loads according to an embodiment of the present invention;

[0043] Figure 4(a) is a schematic diagram of random load feature extraction during the process of extracting variable amplitude loads point by point from the extreme values ​​of random loads, and Figure 4(b) is a schematic diagram of equivalent variable amplitude load during the process of extracting variable amplitude loads point by point from the extreme values ​​of random loads;

[0044] Figure 5(a) is a schematic diagram of a two-dimensional curve of the threshold compression in the average stress threshold of dynamic crack growth, and Figure 5(b) is a schematic diagram of a three-dimensional surface of the threshold compression in the average stress threshold of dynamic crack growth;

[0045] Figure 6 Schematic diagram of crack growth test under variable amplitude load and life prediction results of random crack growth in an embodiment of the present invention;

[0046] Figure 7 Schematic diagram of the experimental data of surface fatigue crack growth of a butt joint subjected to random loads according to an embodiment of the present invention;

[0047] Figure 8 The posterior distribution and probability density curve of the variable amplitude load model parameters described in the embodiment of the present invention are (a) parameter C, (b) parameter m, and (c) parameter σ ∈ Schematic diagram;

[0048] Figure 9 Schematic diagram of the posterior sampling time series of the variable amplitude load model parameters described in the embodiment of the present invention (a) parameter C, (b) parameter m, and (c) parameter σ∈;

[0049] Figure 10Schematic diagram of the comparison between the crack growth experiment and the predicted life under variable amplitude load described in the embodiment of the present invention, wherein the experimental comparison object is a 10mm thick aluminum alloy butt joint. (a) The figure shows specimen #1; (b) specimen #2; (c) specimen #3; (d) specimen #4; (e) specimen #5; (f) specimen #6;

[0050] Figure 11 Schematic diagram of crack propagation experiment and predicted life verification under variable amplitude load described in an embodiment of the present invention.

[0051] In the figure: 701, construction module; 702, calculation module; 703, generation module; 704, prediction module; 705, evaluation module; 800, fatigue crack life evaluation equipment for welded structures under random loads; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0053] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are used only to distinguish the description and should not be understood as indicating or implying relative importance.

[0054] Example 1:

[0055] This embodiment provides a method for evaluating fatigue crack life of a welded structure under random loads.

[0056] See also Figure 1 , the figure shows that the method includes step S100, step S200, step S300, step S400 and step S500.

[0057] S100, based on the service random load history data, the peak and valley of the load are identified by the point-by-point extreme value extraction method, the variable amplitude load cycle is constructed, and the threshold compression algorithm is used to eliminate the load below the crack propagation threshold to generate the equivalent variable amplitude load data set.

[0058] It can be understood that the step S100 includes S101, S102 and S103.

[0059] S101, based on the service random load history data, the peak and valley of the load are identified by the point-by-point extreme value extraction method, the variable amplitude load cycle is constructed, and the threshold compression algorithm is used to eliminate the load below the crack propagation threshold to generate the equivalent variable amplitude load data set.

[0060] S102, for the load below the crack propagation threshold in the preliminary variable amplitude load cycle data, the threshold compression algorithm is used for processing, by comparing the stress intensity factor range and the crack propagation threshold, the part below the threshold is compressed, the influence on the crack propagation is removed, and the optimized equivalent variable amplitude load cycle data is obtained.

[0061] S103, the peak-valley matching is realized by the Matlab program, the optimized equivalent variable amplitude load cycle data is integrated to form the equivalent variable amplitude load data set.

[0062] It should be noted that in order to accurately evaluate the fatigue crack propagation life of the welded structure in the service process, the complex random load history needs to be converted into the variable amplitude load suitable for fatigue analysis. This process is realized by the point-by-point extreme value extraction method.

[0063] It can be understood that under the condition of random load, the peak and valley of the load history are identified by point-by-point, and the key load cycle information is extracted. These peak and valley represent the maximum stress and minimum stress of the load change, thereby forming the equivalent variable amplitude load cycle. Each load cycle is composed of a peak and a valley immediately following it, representing the tensile and compression process of the load. By this method, the complex random load history is simplified into a series of variable amplitude load cycles, which provides effective load input for crack propagation analysis. Then, when dealing with crack propagation, especially for the load less than the crack propagation threshold, its influence on crack propagation is small, and even may cause the delay of crack propagation. Therefore, the threshold compression method is adopted to compress the load below the crack propagation threshold, and the influence on crack propagation is removed. This process compares the stress intensity factor range and the crack propagation threshold, compresses the part below the threshold, forms a new equivalent variable amplitude load, and this method can more accurately simulate the actual influence of the load history on crack propagation.

[0064] It should be noted that, in this step, point-by-point stress extreme value extraction from the random load time history is one of the key steps in converting random load into variable amplitude load. The peak value (maximum stress) and valley value (minimum stress) in the time period are identified by Matlab program, and the complex load time history is simplified into a series of variable amplitude load cycles. Each marked point represents a peak value or valley value of a load, and these extreme values ​​are used to determine the range of stress amplitude and load cycle, forming different tension and compression cycles. By extracting extreme values ​​point by point, the entire load time history can be decomposed into multiple variable amplitude load cycles, each cycle consisting of a peak value and a valley value immediately following it. By extracting extreme values ​​point by point, the key stress points in the random load history can be effectively identified, thereby constructing a series of variable amplitude load cycles for further analyzing the fatigue crack growth behavior of the weld joint surface. Given the initial surface crack and geometric parameters of the weld structure, the stress intensity factor range of the deepest point of the surface crack is calculated step by step and the crack growth increment is solved by an iterative method. By comparing the stress intensity factor range with the crack growth threshold of the full stress ratio range point by point, the relative size is determined. For loads less than the threshold, they are compressed, and then a new variable amplitude driven stress intensity factor range is constructed to realize the iterative calculation of the crack growth increment.

[0065] It should be noted that the structural stress range Δσ s is a key parameter in these cycles, representing the difference between the maximum stress and the minimum stress. The stress range has a direct impact on the fatigue behavior of the material, especially in the prediction of fatigue crack growth under variable amplitude loading. The green and pink segments in Figure 4(a) represent load cycles of different ranges, respectively. These segments connect the extreme points to form multiple variable amplitude loads, and each cycle represents a complete process of tension or compression. Figure 4(b) shows the statistics of the variable amplitude load under random load history, including the minimum and maximum stresses and the corresponding number of cycles. Where σ min is the minimum structural stress, σ max is the maximum structural stress, and Δσ is the structural stress range of the welded structure.

[0066] S200, using the equivalent variable amplitude load data set, calculate the initial crack growth rate parameters C, m and stress intensity factor distribution point by point through the stress intensity factor range, and combine with the two-stage crack growth formula, and then output the deterministic crack growth increment result.

[0067] It can be understood that step S200 includes S201, S202 and S203, wherein:

[0068] S201, using the equivalent variable amplitude load data set, calculating the initial crack propagation rate parameters C, m and stress intensity factor distribution by stress intensity factor range point-by-point calculation, and combining two-stage crack propagation formula, obtaining the preliminary calculation results of crack propagation rate parameters;

[0069] S202, based on the preliminary calculation results of crack propagation rate parameters, under the given initial surface crack and geometric parameters of the welded structure, the stress intensity factor range of the deepest point of the surface crack is calculated step by step, and the crack propagation increment is solved by iteration method, obtaining the preliminary results of crack propagation increment;

[0070] S203, according to the preliminary results of crack propagation increment, by comparing the stress intensity factor range and the full stress ratio range crack propagation threshold value point by point, the relative size is determined, for the load less than the threshold value, it is compressed, and then a new variable amplitude driving stress intensity factor range is constructed, the iteration calculation of crack propagation increment is realized, and the deterministic crack propagation increment result is obtained.

[0071] It should be noted that first, the uncertainty of the deterministic model parameters is quantified, the non-uniformity of the material, the state updating process of the crack propagation parameters and the fluctuation of the crack propagation increment are considered comprehensively, and the random process model of the surface fatigue crack growth of the welded structure is constructed. Then, based on the prior distribution of the model parameters and the fatigue crack monitoring data, the hierarchical Bayesian method and Markov chain Monte Carlo simulation are used to dynamically update the fatigue crack propagation model parameters, so as to realize the prediction of the random process of the fatigue crack propagation of the welded structure.

[0072] Among them, based on the deterministic two-stage crack propagation model, the model parameters are dynamically updated by combining Bayesian method and MCMC simulation. Through the combination of prior distribution, likelihood function and observation data, the uncertainty of crack propagation rate model (such as C, m) is quantified, and the deterministic two-stage surface fatigue crack crack propagation model of welded joint is shown in formula (2-1):

[0073] da / dN=C(M kn ) 2 (ΔK n ) m (2-1)

[0074] Among them, ΔK nThe stress intensity factor range, parameters C and m are used as random parameters of the crack growth rate model. By solving the stress intensity factor range under each cyclic load using the equivalent variable amplitude load input in the above steps, the crack growth increment under the current load can be calculated. The Bayesian method is used to infer the posterior distribution of the model parameters through the prior distribution and observation data to analyze the uncertainty of the parameters. The random fluctuation factor χ in the crack growth process is used to characterize the random disturbance that is not explicitly modeled in the physical process. The model can be expanded to a random form as shown in formula (2-2):

[0075] da / dN=C(M kn ) 2 (ΔK n ) m +χ i (2-2)

[0076] Combining the initial crack and cumulative growth formula, the increment of crack depth can be discretized into formula (2-3):

[0077]

[0078] in, It is used to describe the fluctuation amplitude during crack propagation, reflecting the physical randomness in crack growth, and is a process-level uncertainty.

[0079] S300, based on the crack propagation parameters C, m and observed crack depth data, adopts a hierarchical Bayesian framework combined with MCMC sampling method, and dynamically updates the parameters C, m and random fluctuation factor χ through the posterior distribution formula. And through the global posterior distribution aggregation formula, a global parameter distribution considering material heterogeneity and observation noise is generated, in which the parameters of each plate thickness group are independently defined.

[0080] It can be understood that step S300 includes S301, S302 and S303, wherein:

[0081] S301. Based on the crack growth parameters C and m and the observed crack depth data, a hierarchical Bayesian framework combined with an MCMC sampling method is used to dynamically update the parameters C, m and the random fluctuation factor χ through the posterior distribution formula. The posterior distribution formula is as follows:

[0082] P(θ k |D)∝P(D|θ k )P(θ k )

[0083] Where θ k is the model parameter, D is the experimental observation data, P(D|θ k ) is the likelihood function, which means that given the parameter θ kThe probability of data appearing when k ) is a likelihood function based on historical experience or initial assumptions;

[0084] S302, define the parameters C, m, σ of each plate thickness group ∈ are independent random variables, for thickness group t k The parameters of the posterior distribution are calculated based on the observed data, where the likelihood function p(D k |θ k ) represents the given parameter θ k When the observation data D is generated k The probability of the prior distribution p(θ k ) describes θ k Prior information of

[0085] S303, combined with observation noise σ ∈ , optimize the likelihood function, and generate a global parameter distribution that takes into account material heterogeneity and observation noise through the global posterior distribution aggregation formula, where the parameters of each plate thickness group are defined independently.

[0086] It should be noted that MCMC sampling is used to sample from the posterior distribution of each plate thickness group, which enables dynamic parameter updates during the crack propagation process. The expression for updating the crack growth model parameters in formula (2-3) in the above steps based on the Bayesian formula is shown in formula (2-4):

[0087] P(θ k |D)∝P(D|θ k )P(θ k ) (2-4)

[0088] Among them, θ k is the model parameter, and D is the experimental observation data (the observed value of the crack depth a under multiple cycles N). k ) is the likelihood function, which means that given the parameter θ k The probability of the data appearing when P(θ k ) is the likelihood function based on historical experience or initial assumptions. k ) plays the role of connecting the prior distribution and the posterior distribution in Bayesian analysis, describing the ability of the current model parameters to explain the observed data. k ) is updated to the posterior distribution P(θ k |D) Infer the distribution of parameters to obtain uncertainty information about these parameters. Defined in N1, N2, ..., N k The crack length a was observed at the cycle number obs =a1,a2,...,a kThe expression of the likelihood function is formula (2-5):

[0089]

[0090] where a model,i is the crack depth calculated according to the model, a obs,i is the crack depth observed in the experiment. The parameters C and m in formula (2-2) in step S1 are defined as In the expression of the likelihood function (2-5), σ ∈ is the standard deviation used to describe the error between the model prediction value and the observed data.

[0091] Further define the parameters C, m, σ ∈ of each plate thickness group. k The parameters θ k ={C k ,m k ,σ ∈,k} of each plate thickness group are defined independently, and the observed data is D k ={a obs ,N obs}, and the expression of the posterior distribution is (2-8):

[0092] p(θ k |D k )∝p(D k |θ k )·p(θ k ) (2-8)

[0093] Where the likelihood function p(D k |θ k ) represents the probability of generating the observed data D k given the parameters θ k , and the prior distribution p(θ k ) describes the prior information of θ k . Combined with the observation noise σ ∈ , the likelihood function can be optimized as formula (2-9):

[0094]

[0095] Where N si,k ; C k , m k , t k are random variables of each plate thickness group, N si,k is the number of cycles corresponding to each crack depth obtained by monitoring a single sample of the same plate thickness sample, C k is the crack propagation model propagation rate constant of each plate thickness sample, and m kis the crack growth model growth rate index for each plate thickness sample, t k For grouping by plate thickness. For each group parameter, σ ∈,k ~JeffreysPrior(p(σ ∈,k )∝1 / σ ∈,k ), that is: p(θ k )=p(C k )·p(m k )·p(σ ∈,k ), the sampling process of these parameters is completely independent. The overall posterior distribution of the hierarchical Bayesian model can be expressed as formula (2-10):

[0096]

[0097] Among them, θ k is the model parameter, D k is the experimental observation data (the observed value of the crack depth a under multiple cycles N), and the parameter θ for each plate thickness group is k ={C k ,m k ,σ ∈,k}, the prior distribution parameter is fixed to p(θ k )=p(C k )p(m k )p(σ c,k ). Finally, all sampling points are aggregated to generate global distribution results, thereby constructing a more comprehensive and reliable model parameter of the random fatigue crack growth process.

[0098] For the crack growth samples contained in the observation data of each plate thickness group, the model parameters are θ k ={C k ,m k ,σ ∈,k}, the posterior distribution is calculated based on the Bayesian formula (2-10), and the acceptance probability α of the candidate sample is calculated based on the likelihood function and the prior distribution as shown in formula (2-11):

[0099]

[0100] in, are candidate values ​​of model parameters, is the current value of the model parameter, the numerator is the probability density of the target distribution (candidate value), and the denominator is the probability density of the target distribution (current value).

[0101] S400, based on the global parameter distribution, uses the Monte Carlo simulation method to generate multiple random crack extension trajectories, and combines the cumulative growth formula to predict the crack depth evolution curve under different confidence levels, and outputs the fatigue life probability distribution and confidence interval.

[0102] It can be understood that in this step, based on the deterministic weld joint surface fatigue crack growth model in step S100, the random process life assessment model is constructed by combining Bayesian inference and MCMC sampling. First, the deterministic crack growth model parameters (C, m, σ ∈ ) is set as a random variable, and combined with the Bayesian inference in step S200, a reasonable prior distribution is defined for it to reflect the uncertainty of the parameters. A hierarchical Bayesian model is used to model each plate thickness group separately, and the acceptance probability of the candidate samples is calculated using the likelihood function and the prior distribution to determine the parameter update. The random crack propagation trajectory is generated by Monte Carlo simulation using the global posterior distribution, and the observation error and the randomness of the physical process are comprehensively considered to achieve probabilistic prediction of the structural fatigue life at different confidence levels.

[0103] It should be noted that the threshold compression results for a typical 10 mm thick aluminum alloy butt joint structure during dynamic crack growth are shown in Figures 5(a) and 5(b). The green shaded area represents the threshold value, and the random stress intensity factor range within the shaded area is compressed, while the range outside the green shaded area is retained.

[0104] S500: Evaluate the model accuracy of the fatigue life probability distribution and confidence interval through error analysis and confidence verification algorithms, and combine the initial crack size, plate thickness and load condition data to finally generate a fatigue life reliability assessment report for the welded structure under random loads.

[0105] It is understandable that in this step, based on the first part step S100, key load characteristics are extracted from the actual service random load and converted into equivalent variable amplitude loads suitable for fatigue analysis. By using the point-by-point extreme value extraction method, the maximum stress and minimum stress points are identified from the random load history to form a load cycle. For load cycles below the fatigue crack growth threshold, the influence of the load cycle on crack growth is removed by compression treatment. Based on the second part step S200, the deterministic crack growth model of the stress intensity factor is used to describe the crack growth process. Then, the model parameters (such as C and m of the crack growth rate) are regarded as random variables, and the second part step S200 is used to introduce Bayesian inference and MCMC methods to dynamically update the model parameters to deal with the uncertainty of the parameters. After completing the establishment of the model and the dynamic update of the parameters in the above steps, the next step is to perform fatigue crack growth life assessment. The Monte Carlo simulation method is used to generate multiple crack growth trajectories, simulate the crack growth process under different random loads, and provide life predictions under different confidence levels based on the posterior distribution. By calculating the error and confidence interval, the accuracy and reliability of the model are verified, and the reliability prediction of the fatigue crack growth life of the welded structure under random loads under different crack initial sizes, plate thicknesses and load conditions is achieved.

[0106] It should be noted that if Figure 6 As shown, Figure 6 The crack growth test data of 6 10mm thick samples under random fatigue loads are used, and the crack growth data of these six samples are used as model samples to output the random process model to predict the crack growth fatigue cycle life curves with different confidence levels under the current random load. The 95% confidence interval generally accepted in engineering is used as the reliability boundary value. The results of the random process prediction model can better cover the dispersion of crack growth life of different samples under random loads.

[0107] Example 2:

[0108] This embodiment provides a method for evaluating fatigue crack life of welded structures under random loads, including the following steps:

[0109] Step S1: First, based on the aluminum alloy butt joint welding structure, an experiment is carried out under random cyclic fatigue tensile load to obtain the evolution data of the surface crack depth direction. Figure 7 As shown. Figure 7 From the crack growth curves of different samples of the same plate thickness, it can be seen that the fluctuation and dispersion of the crack growth curve under variable amplitude load are more obvious. The deterministic two-stage surface fatigue crack growth model of the welded joint under variable amplitude load is shown in formula (4-1):

[0110]

[0111] Where ΔK eff is the effective stress intensity factor range (unit: MPa·mm 0.5 ), parameter m eff =4.82 and C eff =8.2038×10 -16 is the random parameter of the crack growth rate model.

[0112] Step S2, by solving the effective stress intensity factor range under each cyclic load and substituting it into the model to calculate the crack extension increment under the current load. Bayesian is used to infer the posterior distribution of model parameters through prior distribution and observation data. The MCMC method is used to sample from the complex posterior distribution and estimate the characteristics of the posterior distribution. In order to analyze the distribution law of model parameters, the fluctuation during the sampling process and the stable convergence trend, the posterior distribution distribution form of model parameters under variable amplitude load characteristics and the Markov chain trajectory are obtained. Figure 8 and Figure 9 Therefore, the histogram of parameter C shows that its posterior distribution is 2.5×10 -16 ~2×10 -15The distribution is wider than that of the constant amplitude load, which proves that the crack growth data of variable amplitude load has a strong dispersion. The histogram of parameter m shows that the posterior distribution of this parameter fluctuates between 4.75 and 4.95, and the PDF curve shows a relatively smooth distribution, indicating that its value is mainly concentrated around 4.82 and the fluctuation is relatively small. ∈ The posterior distribution of is concentrated between 0.12 and 0.32 and the sampling frequency in this interval is relatively high. MCMC sampling is performed on each sample, and 10 4 Sampling times, a total of 6×10 4 The parameter trajectory plot shows some fluctuations, and the uncertainty of different samples under the variable amplitude load condition is stronger. There are no large jumps in the trajectory plot, indicating that the MCMC chain has converged and stabilized in a certain region.

[0113] Step S3, the global distribution results of the posterior samples are used to describe the uncertainty of the overall model parameters. During the crack propagation simulation, the initial cracks, plate thickness, geometric characteristics, crack shape, etc. of different experimental samples have an impact on the crack propagation path. The confidence boundary curve data obtained by the random process model is compared with all the experimental sample data. Figure 10 As shown. Figure 10 It can be seen that each specimen generates a different crack propagation life curve according to its initial crack parameters and number of fatigue cycles. The confidence boundary curve of each sample reflects the randomness and uncertainty of the crack propagation process of the sample. Due to the different initial crack depths, number of fatigue cycles and geometric characteristics of different specimens, the randomness of the crack propagation process is more dispersed under variable amplitude loads. Although the crack propagation process is highly uncertain, the model can provide reliable estimation results and can make effective predictions in different actual situations. Even if there are certain fluctuations or uncertainties in the experimental data, the model can still provide a reasonable prediction range. The experimental data of all samples are compared with the different confidence prediction life curves of the variable amplitude crack propagation model. Figure 11 shown.

[0114] Therefore, in crack growth prediction under variable-amplitude loading, changes in the order of load application can cause fluctuations in the crack growth rate, thereby affecting the fatigue life of the structure. Under the given experimental loading conditions, the 95% confidence level curve effectively captures the randomness of the experimental data, demonstrating that the variable-amplitude loading stochastic process model can effectively capture the actual range of crack growth. Due to load fluctuations, differences in geometric characteristics, and the randomness of crack evolution, the fatigue crack growth rate of welded joints under variable-amplitude loading has significant uncertainty. In actual crack growth experiments, the crack growth life of different specimens can vary significantly, primarily due to differences in initial crack state, number of fatigue cycles, and geometric characteristics. Comparing the experimental data and predicted results for each specimen in the figure shows significant variability in the crack growth life between specimens, but all results fall within the confidence interval predicted by the model, further validating the effectiveness of the variable-amplitude loading stochastic process model in capturing these variations. The crack growth life predicted by the stochastic model is relatively broad, and the experimental data all fall within the predicted range, demonstrating that the model effectively simulates actual crack growth behavior. Unlike traditional deterministic models, the random process model with random loads can capture the randomness of the crack growth process, which is particularly important for crack growth life assessment in engineering practice. The posterior distribution reflects the uncertainty of crack growth after being updated based on monitoring data and can be used to predict the crack growth life with different confidence intervals under new input parameters.

[0115] Example 3:

[0116] like Figure 2 As shown, this embodiment provides a fatigue crack life assessment system for welded structures under random loads, see Figure 2 The system comprises:

[0117] Construction module 701: used to identify load peaks and valleys based on the historical service random load data by using a point-by-point extreme value extraction method, construct a variable amplitude load cycle, and use a threshold compression algorithm to eliminate loads below the crack growth threshold to generate an equivalent variable amplitude load data set;

[0118] Calculation module 702: used to calculate the initial crack growth rate parameters C, m and the stress intensity factor distribution using the equivalent variable amplitude load data set, point by point, and in combination with the two-stage crack growth formula, and then output the deterministic crack growth increment result;

[0119] Generation module 703: Based on the crack growth parameters C and m and the observed crack depth data, a hierarchical Bayesian framework is used in conjunction with the MCMC sampling method to dynamically update the parameters C, m and the random fluctuation factor χ through the posterior distribution formula. A global parameter distribution that takes into account material heterogeneity and observation noise is generated through the global posterior distribution aggregation formula, where the parameters of each plate thickness group are independently defined;

[0120] Prediction module 704: used to generate multiple random crack extension trajectories based on the global parameter distribution using the Monte Carlo simulation method, and combine it with the cumulative growth formula to predict the crack depth evolution curves under different confidence levels, and output the fatigue life probability distribution and confidence interval;

[0121] Evaluation module 705: is used to evaluate the model accuracy of the fatigue life probability distribution and confidence interval through error analysis and confidence verification algorithm, and finally generate a fatigue life reliability evaluation report of the welded structure under random loads by combining the initial crack size, plate thickness and load condition data.

[0122] Specifically, the building block 701 includes:

[0123] Extraction unit: Based on the service random load history data, it uses the point-by-point extreme value extraction method to identify the peaks and valleys in the load time history, extract key load cycle information, form equivalent variable amplitude load cycles, and then obtain preliminary variable amplitude load cycle data, where each load cycle consists of a peak and a valley value following it, representing the tension and compression process of the load;

[0124] Processing unit: used to process the load below the crack growth threshold in the preliminary variable amplitude load cycle data using a threshold compression algorithm. By comparing the stress intensity factor range and the crack growth threshold, the part below the threshold is compressed to remove its influence on crack growth, and the optimized equivalent variable amplitude load cycle data is obtained;

[0125] Integration unit: used to achieve peak-valley matching through Matlab program, integrate the optimized equivalent variable amplitude load cycle data to form an equivalent variable amplitude load data set.

[0126] Specifically, the calculation module 702 includes:

[0127] The first calculation unit is used to calculate the initial crack growth rate parameters C, m and stress intensity factor distribution using the equivalent variable amplitude load data set, point by point, and in combination with the two-stage crack growth formula, to obtain the preliminary calculation results of the crack growth rate parameters;

[0128] The second calculation unit is used to gradually calculate the stress intensity factor range of the deepest point of the surface crack based on the preliminary calculation results of the crack growth rate parameter under the given initial surface crack and geometric parameters of the welded structure, and solve the crack growth increment through an iterative method to obtain the preliminary result of the crack growth increment;

[0129] The third calculation unit is used to determine the relative size based on the preliminary results of the crack growth increment by comparing the stress intensity factor range with the crack growth threshold of the full stress ratio range point by point. For loads less than the threshold, they are compressed and a new variable amplitude driven stress intensity factor range is constructed to realize the iterative calculation of the crack growth increment and obtain the deterministic crack growth increment result.

[0130] Specifically, the generating module 703 includes:

[0131] Sampling update unit: Based on the crack growth parameters C, m and observed crack depth data, a hierarchical Bayesian framework combined with MCMC sampling method is used to dynamically update the parameters C, m and random fluctuation factor χ through the posterior distribution formula. The posterior distribution formula is as follows:

[0132] P(θ k |D)∝P(D|θ k )P(θ k )

[0133] Where θ k is the model parameter, D is the experimental observation data, P(D|θ k ) is the likelihood function, which means that given the parameter θ k The probability of data appearing when k ) is a likelihood function based on historical experience or initial assumptions;

[0134] The fourth calculation unit: used to define the parameters C, m, σ of each plate thickness group ∈ are independent random variables, for thickness group t k The parameters of the posterior distribution are calculated based on the observed data, where the likelihood function p(D k |θ k ) represents the given parameter θ k When the observation data D is generated k The probability of the prior distribution p(θ k ) describes θ k Prior information of

[0135] Generator: used to combine observation noise σ ∈ , optimize the likelihood function, and generate a global parameter distribution that takes into account material heterogeneity and observation noise through the global posterior distribution aggregation formula, where the parameters of each plate thickness group are defined independently.

[0136] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated on here.

[0137] Example 4:

[0138] Corresponding to the above method embodiment, this embodiment also provides a fatigue crack life assessment device for a welded structure under random load. The fatigue crack life assessment device for a welded structure under random load described below and the fatigue crack life assessment method for a welded structure under random load described above can be referred to each other.

[0139] Figure 3 FIG. 8 is a block diagram of a device 800 for evaluating fatigue crack life of a welded structure under random load according to an exemplary embodiment. Figure 3 As shown, the apparatus 800 for assessing fatigue crack life of welded structures under random loads includes a processor 801 and a memory 802. The apparatus 800 also includes one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.

[0140] The processor 801 is configured to control overall operations of the welding structure fatigue crack life assessment device 800 under random load to complete all or part of the steps of the welding structure fatigue crack life assessment method under random load described above. The memory 802 is configured to store various types of data to support the operation of the welding structure fatigue crack life assessment device 800 under random load, which can include, for example, instructions for any application or method operating on the welding structure fatigue crack life assessment device 800 under random load, and application-related data, such as contact data, sent and received messages, pictures, audio, video, and the like. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 803 can include a screen and an audio component. The screen can be a touch screen, for example, and the audio component is configured to output and / or input audio signals. For example, the audio component can include a microphone configured to receive external audio signals. The received audio signals can be further stored in the memory 802 or transmitted through the communication component 805. The audio component also includes at least one speaker configured to output audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, which can be a keyboard, a mouse, or a button, etc. The buttons can be virtual buttons or physical buttons. The communication component 805 is configured to perform wired or wireless communication between the welding structure fatigue crack life assessment device 800 under random load and other devices. Wireless communication, such as Wi-Fi, Bluetooth, near field communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 can include a Wi-Fi module, a Bluetooth module or an NFC module.

[0141] In an exemplary embodiment, the fatigue crack life assessment device 800 for a welded structure under random loads can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-mentioned method for fatigue crack life assessment for a welded structure under random loads.

[0142] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided. When executed by a processor, the program instructions implement the steps of the aforementioned method for assessing fatigue crack life of a welded structure under random loads. For example, the computer-readable storage medium may be the aforementioned memory 802 including the program instructions. The program instructions may be executed by the processor 801 of the apparatus 800 for assessing fatigue crack life of a welded structure under random loads to implement the aforementioned method for assessing fatigue crack life of a welded structure under random loads.

[0143] Example 4:

[0144] Corresponding to the above method embodiment, this embodiment further provides a readable storage medium. The readable storage medium described below and the fatigue crack life assessment method of a welded structure under random load described above can refer to each other.

[0145] The readable storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the method for evaluating fatigue crack life of a welded structure under random loads of the above method embodiment are implemented.

[0146] The readable storage medium may specifically be any readable storage medium that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0147] In summary, the present invention provides a life assessment method for fatigue crack propagation on the surface of welded structures under random loads in service, which improves the accuracy of crack propagation life prediction: by extracting extreme value information in random loads point by point, the complex load history is converted into a simplified variable amplitude load cycle, which can effectively reduce the computational complexity and improve the accuracy of crack propagation prediction. Considering the randomness in the crack propagation process: combining Bayesian inference and MCMC methods, the crack propagation model parameters can be dynamically updated, and a more comprehensive life prediction is provided taking into account factors such as the non-uniformity of material properties and the uncertainty of crack propagation rate. Applicable to complex service load conditions: This method can handle random loads with complex fluctuation characteristics, and is applicable to common random load environments in the fields of subways, aviation, etc., and provides a reliable assessment tool for fatigue crack propagation of welded structures. Strong reliability in engineering applications: through the establishment and verification of random process models, the fatigue life of welded structures can be predicted with multiple confidence intervals under different actual conditions, thereby providing a reliable basis for fatigue life assessment for engineering practice. In summary, the method of the present invention has important application value in the evaluation of fatigue crack growth life of welded structure surfaces, can effectively solve the shortcomings of existing methods, and provide advanced technical support for safety assessment and life prediction in related fields.

[0148] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

[0149] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for evaluating fatigue crack life of welded structures under random loads, characterized in that: include: Based on the historical data of random loads in service, the load peaks and valleys are identified through point-by-point extreme value extraction. A variable amplitude load cycle is constructed, and a threshold compression algorithm is used to eliminate loads below the crack growth threshold to generate an equivalent variable amplitude load data set. Using the equivalent variable amplitude load data set, the stress intensity factor range is calculated point by point, and combined with the two-stage crack growth formula, the initial crack growth rate parameters C, m and the stress intensity factor distribution are calculated, and then the deterministic crack growth increment result is output; Based on the crack growth parameters C and m and the observed crack depth data, a hierarchical Bayesian framework combined with the MCMC sampling method is used to dynamically update the parameters C, m and the random fluctuation factor χ through the posterior distribution formula. A global parameter distribution that takes into account material heterogeneity and observation noise is generated through the global posterior distribution aggregation formula, where the parameters of each plate thickness group are independently defined. Based on the global parameter distribution, the Monte Carlo simulation method is used to generate multiple random crack extension trajectories. Combined with the cumulative growth formula, the crack depth evolution curves under different confidence levels are predicted, and the fatigue life probability distribution and confidence interval are output; The model accuracy of the fatigue life probability distribution and confidence interval is evaluated through error analysis and confidence verification algorithms. Combined with the initial crack size, plate thickness and load condition data, a fatigue life reliability assessment report of the welded structure under random loads is finally generated.

2. The method for evaluating fatigue crack life of welded structures under random loads according to claim 1, characterized in that: Based on the historical data of random loads in service, the load peaks and valleys are identified by point-by-point extreme value extraction, a variable amplitude load cycle is constructed, and a threshold compression algorithm is used to eliminate loads below the crack growth threshold to generate an equivalent variable amplitude load data set, which includes: Based on the historical data of random loads in service, the peaks and valleys in the load time history are identified by point-by-point extreme value extraction. The key load cycle information is extracted to form an equivalent variable amplitude load cycle, and then preliminary variable amplitude load cycle data is obtained. Each load cycle consists of a peak value and a valley value following it, representing the tensile and compressive process of the load. For the loads below the crack growth threshold in the preliminary variable amplitude load cycle data, a threshold compression algorithm is used to process them. By comparing the stress intensity factor range and the crack growth threshold, the part below the threshold is compressed to remove its influence on crack growth, thus obtaining the optimized equivalent variable amplitude load cycle data. The peak-valley matching is realized by Matlab program, and the optimized equivalent variable amplitude load cycle data are integrated to form an equivalent variable amplitude load data set.

3. The method for evaluating fatigue crack life of welded structures under random loads according to claim 1, wherein: The method utilizes the equivalent variable amplitude load data set, calculates the initial crack growth rate parameters C, m and the stress intensity factor distribution point by point through the stress intensity factor range, and combines the two-stage crack growth formula to calculate the initial crack growth rate parameters C, m and the stress intensity factor distribution, and then outputs the deterministic crack growth increment results, including: Using the equivalent variable amplitude load data set, the stress intensity factor range is calculated point by point, and combined with the two-stage crack growth formula, the initial crack growth rate parameters C, m and the stress intensity factor distribution are calculated to obtain the preliminary calculation results of the crack growth rate parameters. Based on the preliminary calculation results of the crack growth rate parameters, the stress intensity factor range at the deepest point of the surface crack was gradually calculated under given initial surface cracks and geometric parameters of the welded structure. The crack growth increment was solved by an iterative method, and the preliminary results of the crack growth increment were obtained. Based on the preliminary results of the crack growth increment, the relative size is determined by comparing the stress intensity factor range with the crack growth threshold of the full stress ratio range point by point. For loads less than the threshold, they are compressed, and then a new variable-amplitude driven stress intensity factor range is constructed to realize the iterative calculation of the crack growth increment and obtain the deterministic crack growth increment result.

4. The method for evaluating fatigue crack life of welded structures under random loads according to claim 1, wherein: Based on the crack growth parameters C and m and the observed crack depth data, a hierarchical Bayesian framework combined with an MCMC sampling method is used to dynamically update the parameters C, m and the random fluctuation factor χ through the posterior distribution formula. A global parameter distribution that takes into account material heterogeneity and observation noise is generated through the global posterior distribution aggregation formula, wherein the parameters of each plate thickness group are independently defined, including: Based on the crack growth parameters C, m and observed crack depth data, a hierarchical Bayesian framework combined with the MCMC sampling method is used to dynamically update the parameters C, m and the random fluctuation factor χ through the posterior distribution formula. The posterior distribution formula is as follows: P(θ k |D)∝P(D|θ k )P(θ k ) Where θ k is the model parameter, D is the experimental observation data, P(D|θ k ) is the likelihood function, which means that given the parameter θ k The probability of data occurrence when P(θ k ) is a likelihood function based on historical experience or initial assumptions; Define the parameters C, m, σ for each plate thickness group ∈ are independent random variables, for thickness group t k The parameters of the posterior distribution are calculated based on the observed data, where the likelihood function p(D k |θ k ) represents the given parameter θ k When the observation data D is generated k The probability of the prior distribution p(θ k ) describes θ k Prior information of Combined with observation noise σ ∈ , optimize the likelihood function, and generate a global parameter distribution that takes into account material heterogeneity and observation noise through the global posterior distribution aggregation formula, where the parameters of each plate thickness group are defined independently.

5. A system for evaluating fatigue crack life of welded structures under random loads, based on the method for evaluating fatigue crack life of welded structures under random loads according to claim 1, characterized in that: include: Construction module: Based on the historical data of random loads in service, the module uses a point-by-point extreme value extraction method to identify load peaks and valleys, constructs a variable amplitude load cycle, and uses a threshold compression algorithm to eliminate loads below the crack growth threshold to generate an equivalent variable amplitude load dataset. Calculation module: Used to calculate the initial crack growth rate parameters C, m and stress intensity factor distribution using the equivalent variable amplitude load data set, point by point through the stress intensity factor range, and combined with the two-stage crack growth formula, and then output the deterministic crack growth increment result; Generation module: Based on the crack growth parameters C and m and the observed crack depth data, a hierarchical Bayesian framework is used in conjunction with the MCMC sampling method to dynamically update the parameters C, m and the random fluctuation factor χ through the posterior distribution formula. The global posterior distribution aggregation formula is then used to generate a global parameter distribution that takes into account material heterogeneity and observation noise. The parameters of each plate thickness group are defined independently. Prediction module: Based on the global parameter distribution, it uses the Monte Carlo simulation method to generate multiple random crack extension trajectories, and combines the cumulative growth formula to predict the crack depth evolution curves under different confidence levels, and output the fatigue life probability distribution and confidence interval; Evaluation module: This module is used to evaluate the model accuracy of the fatigue life probability distribution and confidence interval through error analysis and confidence verification algorithms. It also combines the initial crack size, plate thickness, and load condition data to ultimately generate a fatigue life reliability assessment report for the welded structure under random loads.

6. The fatigue crack life assessment system for welded structures under random loads according to claim 5, characterized in that: The building blocks include: Extraction unit: Based on the service random load history data, it uses the point-by-point extreme value extraction method to identify the peaks and valleys in the load time history, extract key load cycle information, form equivalent variable amplitude load cycles, and then obtain preliminary variable amplitude load cycle data, where each load cycle consists of a peak and a valley value following it, representing the tension and compression process of the load; Processing unit: used to process the load below the crack growth threshold in the preliminary variable amplitude load cycle data using a threshold compression algorithm. By comparing the stress intensity factor range and the crack growth threshold, the part below the threshold is compressed to remove its influence on crack growth, and the optimized equivalent variable amplitude load cycle data is obtained; Integration unit: used to achieve peak-valley matching through Matlab program, integrate the optimized equivalent variable amplitude load cycle data to form an equivalent variable amplitude load data set.

7. The fatigue crack life assessment system for welded structures under random loads according to claim 5, characterized in that: The computing module includes: The first calculation unit is used to calculate the initial crack growth rate parameters C, m and stress intensity factor distribution using the equivalent variable amplitude load data set, point by point, and in combination with the two-stage crack growth formula, to obtain the preliminary calculation results of the crack growth rate parameters; The second calculation unit is used to gradually calculate the stress intensity factor range of the deepest point of the surface crack based on the preliminary calculation results of the crack growth rate parameter under the given initial surface crack and geometric parameters of the welded structure, and solve the crack growth increment through an iterative method to obtain the preliminary result of the crack growth increment; The third calculation unit is used to determine the relative size based on the preliminary results of the crack growth increment by comparing the stress intensity factor range with the crack growth threshold of the full stress ratio range point by point. For loads less than the threshold, they are compressed and a new variable amplitude driven stress intensity factor range is constructed to realize the iterative calculation of the crack growth increment and obtain the deterministic crack growth increment result.

8. The fatigue crack life assessment system for welded structures under random loads according to claim 5, characterized in that: The generation module includes: Sampling update unit: Based on the crack growth parameters C, m and observed crack depth data, a hierarchical Bayesian framework combined with MCMC sampling method is used to dynamically update the parameters C, m and random fluctuation factor χ through the posterior distribution formula. The posterior distribution formula is as follows: P(θ k |D)∝P(D|θ k )P(θ k ) Where θ k is the model parameter, D is the experimental observation data, P(D|θ k ) is the likelihood function, which means that given the parameter θ k The probability of data occurrence when P(θ k ) is a likelihood function based on historical experience or initial assumptions; The fourth calculation unit: used to define the parameters C, m, σ of each plate thickness group ∈ are independent random variables, for thickness group t k The parameters of the posterior distribution are calculated based on the observed data, where the likelihood function p(D k |θ k ) represents the given parameter θ k When the observation data D is generated k The probability of the prior distribution p(θ k ) describes θ k Prior information of Generator: used to combine observation noise σ ∈ , optimize the likelihood function, and generate a global parameter distribution that takes into account material heterogeneity and observation noise through the global posterior distribution aggregation formula, where the parameters of each plate thickness group are defined independently.

9. A device for evaluating fatigue crack life of welded structures under random loads, characterized in that: include: memory for storing computer programs; A processor is configured to implement the method for evaluating fatigue crack life of a welded structure under random loads as claimed in any one of claims 1 to 4 when executing the computer program.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for evaluating fatigue crack life of a welded structure under random load according to any one of claims 1 to 4 is implemented.

Images