Fatigue crack growth prediction method for structural parts based on digital twin

Through digital twin technology and dynamic Bayesian network model, combined with particle filtering algorithm, the problem of low prediction accuracy of fatigue crack propagation in complex equipment is solved, and efficient and accurate prediction of crack propagation processes under complex loads is achieved.

CN114282709BActive Publication Date: 2024-11-15OCEAN UNIV OF CHINA +1
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202111417327.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-26
Publication Date
2024-11-15
Estimated Expiration
2041-11-26

AI Technical Summary

Technical Problem

The prior art is difficult to accurately predict the fatigue crack propagation trend of key structures under complex alternating loads in complex equipment. Traditional fracture mechanics methods are low in accuracy and are greatly affected by uncertain factors, making it difficult to apply to crack propagation prediction of complex equipment.

Method used

Using the fatigue crack propagation prediction method of structural parts based on digital twins, a dynamic Bayesian network model under local sample density is established with a dynamic Bayesian network model under composite amplitude load for crack propagation, combined with a particle filtering algorithm , gradually update uncertain parameters to achieve dynamic follow-up and accurate prediction of crack propagation.

Benefits of technology

The stress strength factor calculation efficiency of complex equipment during crack propagation is improved, the impact of uncertainty factors on crack propagation process is reduced, and the accurate prediction of crack propagation life and path under complex loads is achieved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114282709B_ABST
    Figure CN114282709B_ABST
Patent Text Reader

Abstract

The present invention proposes a method for predicting fatigue crack propagation of structural parts based on digital twins, establishes a stress intensity factor-Gaussian process proxy model based on local sample densification, adopts a sample generation method of local sample densification, establishes a dynamic Bayesian network model under a composite variable amplitude load for crack propagation, and uses the uncertainty factors existing in the crack propagation process as the state nodes of the model. At the same time, the stress intensity factor-Gaussian process proxy model and the load order effect model can also be integrated to realize the prediction of the crack propagation life and path under variable angle and amplitude of load; establishes an uncertainty parameter inference algorithm based on particle filtering, and uses the key data collected by the structural part data acquisition system based on digital twins to gradually reduce the influence of multiple uncertainty factors on the crack propagation process when they work together.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of fatigue crack propagation of structural parts, and in particular, to a method for predicting fatigue crack propagation of structural parts based on digital twins. Background Art

[0002] When complex equipment is subjected to complex alternating loads, fatigue cracks are easily generated in some key structures, which weakens the overall strength. Therefore, how to accurately know the current crack state at the key structure position and analyze and predict its expansion trend has great research significance and engineering value for ensuring the safe operation of complex equipment.

[0003] The current crack propagation prediction methods are mainly divided into online monitoring methods and methods based on theoretical analysis. The former directly senses the current state of the structure through non-destructive testing methods (such as acoustic emission and Lamb waves). Its disadvantage is that the equipment is prone to failure in complex environments, and it is difficult to deploy sensors throughout the structure. The latter is based on the theory of fracture mechanics to calculate the crack propagation of key structural parts. Although it can avoid the problem of insufficient structural damage information caused by sensor reliability, it is often affected by uncertain factors during the crack propagation process due to the complexity of the environment, resulting in low crack propagation prediction accuracy. Therefore, traditional fracture mechanics methods cannot be directly applied to the crack propagation prediction process of complex equipment. Summary of the invention

[0004] In order to reduce the influence of uncertainty factors on complex equipment structures during crack propagation, the present invention provides a method for predicting fatigue crack propagation of structural parts based on digital twins, which is implemented by the following technical scheme: A method for predicting fatigue crack propagation of structural parts based on digital twins, characterized by comprising:

[0005] Step A, establishing a stress intensity factor-Gaussian process proxy model based on local sample densification;

[0006] Step A1: Establish a digital twin model based on the physical structure and environmental information of complex equipment;

[0007] Step A2: for possible crack shapes, a proxy model method is used to calculate the stress intensity factor at the crack tip, the inflection point of the crack is used as a feature point to establish a sample space format, and a local sample densification method is used to generate a sample data space;

[0008] Step A3: Based on Gaussian process theory, a stress intensity factor-Gaussian process proxy model is established according to a stress intensity factor sample database;

[0009] Step B: establishing a dynamic Bayesian network model under a composite variable amplitude load for crack propagation;

[0010] Step B1, based on the unified fatigue life prediction model, using the control variable method, taking the crack propagation prediction length as the judgment quantity, a sensitivity analysis is performed on the uncertainty factors in the fatigue crack propagation process;

[0011] Step B2: Based on the sensitivity analysis results, high-sensitivity parameters are set as state nodes in the dynamic Bayesian network, and low-sensitivity parameters are set as fixed nodes, so as to establish a complex equipment performance degradation model for crack propagation.

[0012] Step B3: Based on the dynamic Bayesian theory and the sensitivity analysis results of the uncertainty factors in the unified fatigue life prediction model, a dynamic Bayesian network model for crack propagation is established;

[0013] Step C performs uncertainty parameter reasoning based on particle filtering

[0014] Step C1: Based on expert experience, the probability distribution of the state nodes in the dynamic Bayesian network is set, and the distribution particle groups are generated by a random combination method using a variety of uncertainty parameters;

[0015] Step C2: Based on the crack information of the current structure collected by the digital twin structural component data acquisition system, a likelihood estimation is performed on each particle;

[0016] Step C3: Sampling The importance sampling algorithm updates the weights of the particles and outputs the updated state node parameter values ​​in the dynamic Bayesian network.

[0017] Furthermore, in step A2, the sample space format is:

[0018] Where F is the load size, β is the load angle, A is the crack extension length under the current loading angle, α is the crack deflection angle, n is the number of crack angle changes, j is the number of crack samples, K is the stress intensity factor, and i is the stress intensity factor type.

[0019] Furthermore, in step A3, in order to improve the prediction accuracy of the proxy model, a sample database is generated by the following method: generating corresponding coarse sample points with a large distribution density according to the crack path in the previous time step, and then encrypting the sparse sample points around the crack propagation path.

[0020] Furthermore, in step B2, the parameters A, m, ΔK are selected. effth, γ and a0 are regarded as highly sensitive uncertain parameter state nodes in the dynamic Bayesian network model for crack extension, and other variables are regarded as a deterministic fixed node. Furthermore, the Bayesian network model in step B3 is constructed using four types of nodes: (1) the observation node is the input part of the system, that is, the load information and crack length information collected by the structural component data acquisition system based on digital twins; (2) the state node indicates that the variable is random with respect to the parent node and has uncertainty, that is, the highly sensitive uncertainty factor of the model; (3) the fixed node is the uncertainty factor parameter with low sensitivity to the crack extension prediction accuracy; (4) the function node represents the deterministic function.

[0021] Furthermore, in step C1, when generating the distribution particle group, first, each of the S uncertainty factors independently generates N particles, and the particles generated by each uncertainty factor are randomly combined to form N model parameter combinations, that is, X 1:N ={x1,x2,...,x s}, where x t (t=1:S) is the random value of the particle generated by the tth uncertain factor.

[0022] Furthermore, in step C2, when likelihood estimation is performed, when the prediction time interval ΔN is very small, the state equation of the fatigue crack growth process is:

[0023] The measurement equation of the system is: k =h k (x k ,n k ), where k represents the time; y is the actual measured crack information; h is the measurement equation of the system, that is, the crack information of the sensor; and n is the measurement noise of the system.

[0024] Furthermore, in step C3, the deviation degree of each particle calculation is compared by inputting the observation information into the model, and the particle weight is iteratively updated.

[0025] Compared with the prior art, the advantages and positive effects of the present invention are:

[0026] The present invention can gradually update the uncertainty parameters in the crack propagation process for the real data collected by the sensor, such as environmental load data, crack state information, etc., to achieve dynamic tracking and accurate prediction of crack propagation. A stress intensity factor-Gaussian process proxy model based on local sample densification is established, and a sample generation method based on local sample densification is adopted to greatly improve the efficiency of stress intensity factor calculation in the crack propagation process of complex equipment; in order to realize the performance degradation of complex equipment under complex loads under time series, a dynamic Bayesian network model under composite variable amplitude load for crack propagation is established, and the uncertainty factors existing in the crack propagation process are used as the state nodes of the model. At the same time, the stress intensity factor-Gaussian process proxy model and the load order effect model can also be integrated to realize the prediction of crack propagation life and path under load variable angle and amplitude; in order to reduce the influence of uncertainty factors on crack propagation, an uncertainty parameter inference algorithm based on particle filtering is established, and the key data collected by the structural component data acquisition system based on digital twins is used to gradually reduce the influence of multiple uncertainty factors on the crack propagation process when they work together. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 This is a schematic diagram of the structural component data acquisition system based on digital twins;

[0028] Figure 2 It is a flow chart of fatigue crack growth consistency maintenance strategy based on digital twin;

[0029] Figure 3 It is a method for generating a stress intensity factor sample database based on local sample densification;

[0030] Figure 4 It is a flow chart of crack growth calculation based on stress intensity factor-Gaussian process proxy model;

[0031] Figure 5 It is a dynamic Bayesian network model for crack propagation under composite variable amplitude load;

[0032] Figure 6 It is the flow chart of particle filter inference algorithm. DETAILED DESCRIPTION

[0033] In order to facilitate the understanding of the present invention, the present invention is further described below in conjunction with embodiments, but the protection scope of the present invention is not limited to the following specific embodiments.

[0034] As an important marine resource development equipment, offshore platforms often suffer from fatigue crack damage at the pipe node positions. The data acquisition system based on the digital twin of the offshore platform key structure of this embodiment, such as Figure 1As shown, it includes a data acquisition module, a data transmission module, a data storage module and a data processing module at the key structure of the offshore platform. In the offshore platform structure, the data acquisition module mainly includes a crack information acquisition device set at the pipe node and other stress concentration positions; a marine environment information acquisition device is set at a position close to the sea surface and on the top of the platform; and an offshore platform overall response acquisition device is set at the intersection of the pipe nodes of the offshore platform. In this embodiment, the crack information acquisition device includes a piezoelectric ceramic sensor, a distributed optical fiber sensor, a stress strain sensor, etc. The marine environment information acquisition device includes a wave pressure sensor, a wave direction sensor, a temperature sensor, a wind speed sensor, a wind direction sensor, etc. The offshore platform overall response acquisition device includes a displacement sensor, an inclination sensor, etc.

[0035] After collecting key data information on the offshore platform, the data transmission module is used to output the relevant data to the data storage module for processing and use. The data transmission module mainly adopts a combination of wired and wireless transmission. Wired transmission mainly adopts optical fiber transmission and other methods, while wireless transmission can adopt technologies such as 5G and WIFI. Due to the large amount of collected data, the data storage module mainly adopts a combination of media storage and cloud storage. Media storage includes hard disks, disk arrays, etc., which mainly store crack status information for subsequent timely calls; cloud storage mainly stores marine environmental information and marine platform status data. Subsequently, these data are processed, mainly including data cleaning, denoising, and classification. Among them, a digital twin model of the overall structure is established based on the offshore platform structure data and environmental data, and crack monitoring data is used to support the crack extension life and path prediction under complex equipment.

[0036] In order to reduce the impact of various uncertain factors on the crack propagation process during the operation of offshore platforms and improve the computational efficiency of crack propagation prediction, a fatigue crack propagation consistency maintenance strategy based on digital twins is proposed. Figure 2 As shown in the figure, it mainly includes three steps: the establishment of a stress intensity factor-Gaussian process proxy model based on local sample densification, the establishment of a dynamic Bayesian network model under composite variable amplitude load for crack propagation, and an uncertainty parameter inference algorithm based on particle filtering.

[0037] In order to overcome the problem of computational efficiency and computational cost caused by the conventional method that the crack prediction results need to be calculated on one side and the results are continuously brought back to the finite element simulation software to calculate the key parameter of the stress intensity factor under the current crack information, the present invention establishes a proxy model based on the stress intensity factor of local sample densification-Gaussian process, realizes the calculation of the stress intensity factor at the crack tip, establishes the sample space format with the inflection point of the crack as the characteristic point, adopts the method of local sample densification to generate the sample data space, and improves the crack extension calculation efficiency of complex structures under complex loads. The specific steps are as follows:

[0038] (1) A digital twin model for the offshore platform is established based on the physical geometric structure of the offshore platform, the marine environment information, and the damage status of the structure.

[0039] (2) According to the characteristics of crack propagation, when the load angle of the structure remains unchanged, the crack propagation direction remains unchanged. Therefore, the sample space format of the stress intensity factor-Gaussian process proxy model is defined as:

[0040]

[0041] Among them, F is the load size, β is the load angle, A is the crack extension length under the current loading angle, α is the crack deflection angle, n is the number of crack angle changes, j is the number of crack samples, K is the stress intensity factor, and i is the stress intensity factor type.

[0042] The established samples are brought into the digital twin model of the offshore platform to obtain the stress intensity factor at the crack tip and establish a complete stress intensity factor sample database. In this process, in order to reduce the computational cost and time of the sample generation process, the following Figure 3 The method shown in the figure is used to generate a sample database. First, the corresponding coarse sample points with large distribution density are generated according to the crack path in the previous time step, and then the sparse sample set is encrypted around the crack propagation path to improve the prediction accuracy of the proxy model.

[0043] (3) Based on the Gaussian process theory, a corresponding proxy model is established according to the stress intensity factor sample database, and the covariance function determined by automatic correlation is selected as the covariance kernel function k(x i ,x j ):

[0044]

[0045] in, is the kernel function signal variance that controls the local correlation of the input variables; l d is the variance scale, D is the dimension of the input data, is the i-th input data.

[0046] like Figure 6 As shown in Figure 1, the stress intensity factor calculation part in the crack growth rate model is replaced by the stress intensity factor-Gaussian process proxy model to calculate the crack growth life.

[0047] The crack propagation process of offshore platforms is often affected by a variety of uncertain factors, such as the error of the crack propagation rate model, the error of material properties, and the error of environmental detection. As time goes by, the predicted results of crack propagation will gradually deviate greatly from the actual crack. Therefore, a dynamic Bayesian network model for crack propagation under composite variable amplitude load is proposed to realize the performance degradation calculation of complex equipment under complex loads, such as Figure 2 The specific steps of establishing the model are as follows:

[0048] (1) There are three types of uncertainty factors in the crack propagation process, namely, the uncertainty of the detection data, such as the load monitoring accuracy and crack monitoring accuracy; the uncertainty of the material, such as the uncertainty of the material properties; the uncertainty of the model, such as the model coefficient, etc. After determining the uncertainty variable parameters, the control variable method is used to analyze the sensitivity of the uncertainty factors with the crack propagation prediction length as the judgment quantity.

[0049] In order to consider the uncertainty parameters in the crack growth rate model, a sensitivity analysis of the uncertainty parameters in the rate model was carried out based on the unified fatigue life prediction model.

[0050] Unified fatigue life prediction model:

[0051]

[0052] Among them, A is the material constant, m is the slope coefficient of the crack growth rate curve, n is the unstable fracture coefficient, K max is the maximum stress intensity factor, ΔK effth is the effective stress intensity factor threshold, f op is the crack opening state function, K C is the actual fracture toughness, and φ is the load order correction factor.

[0053] When the cyclic load changes, the crack growth shows the characteristics of "overload hysteresis and low load acceleration", so the load order correction factor φ can be expressed as:

[0054] Overload:

[0055] Low load:

[0056] Among them, γ is the variable amplitude load effect coefficient; ai is the crack length at the current moment; a OL is the crack length under overload; r OL is the size of the plastic zone at the crack tip when overloaded; r UL is the size of the plastic zone at the crack tip under low load; r yi is the size of the plastic zone at the crack tip at the current moment.

[0057] (2) Based on the unified fatigue life prediction model, select parameters A, m, ΔK effth , variable amplitude load effect coefficient γ and initial crack length a0 are used as highly sensitive uncertain parameter state nodes in the dynamic Bayesian network model for crack propagation, and other variables are regarded as deterministic fixed nodes. Then an initial crack value is given, and the detected real load data is input into the model to realize the crack propagation calculation of the structure under time series.

[0058] (3) Based on the dynamic Bayesian network theory, we establish Figure 5 A dynamic Bayesian network model for crack propagation under composite variable amplitude loads is proposed. The model is constructed using four types of nodes: (1) The observation node is the input part of the model, namely the marine environmental load information and crack length information collected by the digital twin-based structural component data acquisition system; (2) The state node represents the variable is random with respect to the parent node and has uncertainty, that is, the highly sensitive uncertainty factor of the model; (3) The fixed node is the uncertainty factor parameter with low sensitivity to the crack propagation prediction accuracy; (4) The function node represents the deterministic function, such as the load order correction factor and the crack propagation rate formula.

[0059] When the crack is detected at time t-1, the observed value The crack length at the next moment is If no crack observation is detected at time t-1, the crack length at the next moment is

[0060] There are usually multiple uncertain factors that affect crack propagation, and these uncertain factors also have a correlation with each other. Therefore, in order to reduce the influence of these uncertain parameters on crack propagation, such as Figure 6 , an uncertainty parameter inference algorithm for particle filtering is proposed, and the specific steps are as follows:

[0061] (1) Based on expert experience, the probability distribution range of highly sensitive parameters in the dynamic Bayesian network model under composite variable amplitude load for crack propagation is determined by defining these particle weights, and then comparing the degree of deviation of each particle calculation based on the observation information input into the model. The particle weights are iteratively updated, and particles with very small weights are removed, gradually approaching the true posterior distribution.

[0062] (2) After setting the probability distribution of the uncertainty parameters, it is necessary to estimate the likelihood of the particles. According to the unified life prediction model, if the prediction time interval ΔN is very small, the state equation of the fatigue crack growth process is:

[0063]

[0064] The measurement equation is:

[0065] y k =h k (x k ,n k )

[0066] Where k represents the time; y is the actual measured crack information; h is the measurement equation of the system, that is, the crack information of the sensor; and n is the measurement noise of the system.

[0067] There are multiple uncertain factors in the dynamic Bayesian network model under composite variable amplitude load for crack propagation, and these factors may affect each other. Therefore, when generating particles, first, each of the S uncertain factors independently generates N particles, and the particles generated by each uncertain factor are randomly combined to form N model parameter combinations, that is, X 1:N ={x1,x2,...,x s}, where x t (t=1:S) is the random value of the particle generated by the tth uncertainty factor. By bringing the model parameter combination Xk into the DBN model, the state a of the structural crack extension is obtained. k .

[0068] (3) The particle filter inference algorithm mainly estimates the posterior distribution p(a) of the state through the existing observations. 0:k |y 1:k-1 ), and if enough particles with different weights are obtained from this posterior distribution, the structural crack state value at time k can be expressed as

[0069]

[0070] In the formula, The superscript of represents the i-th model parameter combination, and the subscript represents the k-th moment; w represents the weight of the particle; δ is The increment function of .

[0071] Importance Weight Use sequence important sampling to obtain values:

[0072]

[0073] As the update process continues to iterate, particle degradation will occur, that is, the weights of a few particles will be very large, while the weights of other particles will have a negligible effect on the prior distribution of the state, making it difficult for the algorithm to effectively estimate the state of the system. Therefore, the resampling method is used to avoid this situation, that is, particles with large weights will be copied, and particles with small weights will be removed. This method can ensure that the weights of different particles are relatively even, and the weights of all particles after resampling become 1 / N.

[0074] In the dynamic Bayesian network model process under the composite variable amplitude load for crack extension, the crack information data collected by the digital twin-based structural component data acquisition system is input, and the uncertainty factor parameter node generates N sample particles, where the initial weight of each particle is the same. According to the value of each particle, the crack extension prediction value at the current time is calculated in the Bayesian model of the current time slice, and the prediction value is compared with the observed real data and the weight is updated. The resampling particle filter algorithm is used to update the reasoning between the state nodes. After the update, the particle sample mean is obtained, and the calculation of the next time step is continued until the observation data is input into the model again. The advantage of this method is that as the number of observations increases, the number of particle updates of the uncertain node also increases, so this makes the uncertainty of the state node gradually decrease, realizing the dynamic tracking and accurate prediction of crack extension.

[0075] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for predicting fatigue crack growth of structural parts based on digital twins, characterized in that include: Step A: Establishing a Gaussian process proxy model for stress intensity factor based on local sample densification Step A1: Establish a digital twin model based on the physical structure and environmental information of complex equipment; Step A2: for possible crack shapes, a proxy model method is used to calculate the stress intensity factor at the crack tip, the inflection point of the crack is used as a feature point to establish a sample space format, and a local sample densification method is used to generate a sample data space; Step A3: Based on Gaussian process theory, a stress intensity factor-Gaussian process proxy model is established according to a stress intensity factor sample database; Step B: establishing a dynamic Bayesian network model under a composite variable amplitude load for crack propagation; Step B1, based on the unified fatigue life prediction model, adopting the control variable method, taking the crack propagation prediction length as the judgment quantity, performing sensitivity analysis on the uncertainty factors in the fatigue crack propagation process; Step B2: Based on the sensitivity analysis results, high sensitivity parameters are set as state nodes in the dynamic Bayesian network, and low sensitivity parameters are set as fixed nodes, so as to establish a complex equipment performance degradation model for crack propagation; Step B3: Based on the dynamic Bayesian theory and the sensitivity analysis results of the uncertainty factors in the unified fatigue life prediction model, a dynamic Bayesian network model for crack propagation is established; Step C performs uncertainty parameter reasoning based on particle filtering Step C1: Based on expert experience, the probability distribution of the state nodes in the dynamic Bayesian network is set, and the distribution particle groups are generated by a random combination method using a variety of uncertainty parameters; Step C2: Based on the crack information of the current structure collected by the digital twin structural component data acquisition system, a likelihood estimation is performed on each particle; Step C3: update the weights of the particles and output the updated state node parameter values ​​in the dynamic Bayesian network.

2. The method for predicting fatigue crack growth of structural parts based on digital twin according to claim 1, characterized in that: In step A2, the sample space format is: Where F is the load size, β is the load angle, A is the crack extension length under the current loading angle, α is the crack deflection angle, n is the number of crack angle changes, j is the number of crack samples, K is the stress intensity factor, and i is the stress intensity factor type.

3. The method for predicting fatigue crack growth of structural parts based on digital twin according to claim 1, characterized in that: In step A3, in order to improve the prediction accuracy of the proxy model, a sample database is generated by the following method: generating corresponding coarse sample points with a large distribution density according to the crack path in the previous time step, and then encrypting the sparse sample points around the crack propagation path.

4. The method for predicting fatigue crack growth of structural parts based on digital twin according to claim 1, characterized in that: In step B2, select parameters A, m, ΔK effth , γ and a0 are regarded as highly sensitive uncertain parameter state nodes in the dynamic Bayesian network model for crack propagation, and other variables are regarded as a deterministic fixed node. A is the material constant, m is the slope coefficient of the crack propagation rate curve, ΔK effth is the effective stress intensity factor threshold, γ is the variable amplitude load effect coefficient, and a0 is the initial crack length.

5. The method for predicting fatigue crack growth of structural parts based on digital twin according to claim 1, characterized in that: The Bayesian network model in step B3 is constructed using four types of nodes: (1) the observation node is the input part of the system, namely, the load information and crack length information collected by the structural component data acquisition system based on the digital twin; (2) the state node indicates that the variable is random with respect to the parent node; (3) the fixed node is an uncertain factor parameter with low sensitivity to the crack extension prediction accuracy; (4) the function node represents a deterministic function.

6. The method for predicting fatigue crack growth of structural parts based on digital twin according to claim 1, characterized in that: In step C1, when generating a distributed particle group, first, each of the S uncertainty factors independently generates N particles, and the particles generated by each uncertainty factor are randomly combined to form N model parameter combinations, that is, X 1:N ={x1,x2,...,x s }, where x t (t=1:S) is the random value of the particle generated by the tth uncertain factor.

7. The method for predicting fatigue crack growth of structural parts based on digital twin according to claim 4, characterized in that: In step C2, when likelihood estimation is performed, when the prediction time interval ΔN is very small, the state equation of the fatigue crack growth process is: The measurement equation of the system is: k =h k (x k ,n k ), where k represents the moment; y is the actual measured crack information; h is the measurement equation of the system, that is, the crack information measured by the sensor; n is the measurement noise of the system, Kmax is the maximum stress intensity factor, fop is the crack opening state function, KC is the actual fracture toughness, and φ is the load order correction factor.

8. The method for predicting fatigue crack growth of structural parts based on digital twin according to claim 1, characterized in that: In step C3, the deviation degree of each particle calculation is compared by inputting the observation information into the model, and the particle weight is iteratively updated.

Citation Information

Cited By

  • Variable-working-condition multi-stage deep-sea pressure-resistant cabin fatigue crack propagation prediction method

    CN120372976A

  • A multi-stage fatigue crack growth prediction method for deep-sea pressure vessels with variable operating conditions

    CN120372976B