Method and device for predicting fatigue life of welding structure of high-temperature alloy vacuum electron beam welding

Abstract

The invention discloses a method and device for predicting the fatigue life of a high-temperature alloy vacuum electron beam welding structure, and belongs to the technical field of aero-engines. The method comprises the following steps: acquiring fatigue experiment data of a target high-temperature alloy standard sample; extracting the maximum area of an initial defect, a defect shape factor and crack source position information; obtaining a stress intensity factor range considering defect morphological characteristics and a fatigue crack propagation threshold value of a material structure containing initial defects, and correcting a damage tolerance theoretical formula to obtain simulation data; fatigue test data and simulation data are fused to form an initial data set, a fatigue limit is calculated and expanded to the data set, the expanded data set serves as input, training is conducted through a support vector regression method, hyper-parameters are determined through Bayesian optimization, and a fatigue life prediction result of the high-temperature alloy vacuum electron beam welding structure is output. The method provided by the invention can well consider the influence of different morphology defects on the fatigue performance of the welded joint.

Description

Technical Field

[0001] The present invention relates to the technical field of aircraft engine structure safety, and in particular to a method and device for predicting fatigue life of a high-temperature alloy vacuum electron beam welded structure. Background Art

[0002] High-temperature alloys have excellent mechanical stability, as well as oxidation resistance and corrosion resistance in high-temperature environments. They are widely used in different parts of aircraft engines, such as turbine blades, turbine disks, and combustion chamber casings. For combustion chamber casings, advanced welding processes such as laser welding, vacuum electron beam welding, and brazing are generally used to weld the casings. In particular, for vacuum electron beam welding, the welding environment is generally under relatively vacuum conditions (vacuum degree: 10 -4 Pa-10 - 2 Pa), under these welding conditions, compared to welding processes in air, weld porosity defects are much smaller in both quantity and size. Therefore, the welding quality of the combustion chamber casing structure obtained by vacuum electron beam welding is generally superior to that of other welding processes. Although the level of porosity welding defects is better than that of other welding processes, the GH4169 welded structure obtained by vacuum electron beam welding will inevitably produce weld defects such as weld microcracks, pores, and lack of penetration. The introduction of these defects will lead to increased discreteness of fatigue properties such as fatigue life and fatigue limit of GH4169 high-temperature alloy welded joints under alternating load conditions, thus bringing new challenges to fatigue life assessment and fatigue reliability design of casing welded structures.

[0003] In fact, not only welding processes, but also additive manufacturing processes, casting processes and forging processes, will inevitably produce defects of different sizes and numbers inside the processed materials or structures. How to accurately predict the fatigue life of materials / structures containing initial defects is one of the most critical issues in engineering. Japanese scholar Murakami conducted fatigue performance tests on a large number of different materials containing different types of initial defects, and fitted the fatigue life results, fatigue limit, maximum initial equivalent defect size of the material, and Vickers hardness of the material. Based on the damage tolerance, he obtained the empirical formula for predicting the fatigue limit of materials containing initial defects and the empirical formula for predicting life based on fatigue limit. The establishment of this set of methods provides an effective method for fatigue performance evaluation of materials / structures containing initial defects in engineering. However, this method simply performs square root processing on the projected area of ​​the maximum initial defect perpendicular to the loading direction, and converts the maximum initial defect into As the core damage parameter, fatigue life of materials / structures is evaluated. However, the most obvious shortcoming of the above formula is that it does not fully consider the influence of defect geometry on fatigue performance.

[0004] On the other hand, with the development of artificial intelligence technology, machine learning algorithms are increasingly being applied to fatigue performance assessment of materials and structures. However, these algorithms are highly dependent on the quality and scale of the input data. If the data set is not of good quality, the fatigue performance assessment results obtained by machine learning methods will not be very good. Furthermore, when it comes to fatigue testing of structural components in actual engineering, large-scale fatigue testing is often impossible due to the difficulty and cost of the tests. Summary of the Invention

[0005] The purpose of the present invention is to overcome the problems in the prior art and to provide a method and device for predicting the fatigue life of a high-temperature alloy vacuum electron beam welded structure.

[0006] The fatigue life prediction method of the high-temperature alloy vacuum electron beam welding structure of the present invention comprises obtaining fatigue test data of a target high-temperature alloy standard sample; Fatigue fracture analysis of target high-temperature alloy standard specimens extracts the maximum area of ​​the initial defect, defect shape factor, and crack source location information. Based on the extracted maximum area, defect shape factor, and crack source location information, static simulation and crack growth simulation are used to obtain the stress intensity factor range that takes into account the defect morphology characteristics. A quantitative relationship between the stress intensity factor range and the defect shape factor is established, the damage tolerance theory formula is modified, and simulation data is obtained. The fatigue test data and simulation data are fused to form an initial data set. Based on the modified damage tolerance theory formula, the fatigue limit is calculated and expanded to the data set. With the expanded data set as input, the support vector regression method is used to determine the hyperparameters through Bayesian optimization, and the fatigue life prediction results of the high-temperature alloy vacuum electron beam welding structure are output.

[0007] Preferably, the defect shape factor f is the ratio of the length of the defect's minor semi-axis to the length of the defect's major semi-axis, and f ∈(0,1].

[0008] Preferably, when obtaining simulation data, a finite element analysis model of the target high-temperature alloy standard specimen is first established, and then a static simulation analysis of the target high-temperature alloy standard specimen at the peak load is performed. The finite element analysis model output file is then imported into the crack propagation model, and a crack of a specific shape and size is inserted to obtain the simulation analysis results of the stress intensity factor of the corresponding defect tip. By fitting the relationship between different shape factors and stress intensity factors, the morphology influence coefficient is determined, and the damage tolerance theoretical formula is modified.

[0009] Preferably, when determining the expanded data set, a Spearman rank correlation analysis is first performed on the factors affecting fatigue life to determine the main factors affecting fatigue life, and the determined main factors affecting fatigue life are used as the input data set.

[0010] Preferably, the input data set includes temperature, stress amplitude, equivalent defect size, Vickers hardness, fatigue life obtained from fatigue testing, and physical parameters obtained from simulation analysis of fatigue experimental data, including stress intensity factor range and fatigue limit.

[0011] Preferably, the modified damage tolerance formula is: (1) (2) (3) Where, ∆K th is the fatigue crack growth threshold of the material structure containing initial defects, C is the defect position influence coefficient, for internal defects C Take 1.56, for sub-surface defects, C Generally, 1.43 is taken, HV is Vickers hardness, is the defect equivalent size, R is the stress ratio, f is the defect shape factor, z 1 is the length of the defect's minor semi-axis; z 1 is the length of the defect's minor semi-axis; ∆ K def The stress intensity factor range considering the defect morphology characteristics is ( K def,max - K def,min ), S is the influence coefficient of stress intensity factor and defect morphology obtained by numerical simulation, C 1 and m is the material constant, which can be obtained by optimizing the fatigue test result data. a is the crack size, N is the fatigue life.

[0012] The present invention also discloses a fatigue life prediction device for an aero-engine high-temperature alloy vacuum electron beam welding structure based on the above fatigue life prediction method, comprising: A receiving module is configured to receive an input data set of a target standard sample, wherein the input data value includes fatigue test data and physical parameters obtained by simulation analysis based on the fatigue test data, wherein the physical parameters include a stress intensity factor range and a fatigue limit; The prediction module is used to predict the fatigue life of the high-temperature alloy vacuum electron beam welded structure by using a support vector regression model and an input data set.

[0013] The present invention also discloses an electronic device, comprising: at least one processor and a memory; The memory stores computer-executable instructions; The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes any one of the above-mentioned methods for predicting fatigue life of a high-temperature alloy vacuum electron beam welded structure.

[0014] The present invention also discloses a computer-readable storage medium, which stores computer-executable instructions. When a processor executes the computer-executable instructions, the fatigue life prediction method for a high-temperature alloy vacuum electron beam welding structure described in any one of the above items is implemented.

[0015] The present invention also discloses a computer program product, which, when executed by a processor, implements any of the above-mentioned methods for predicting fatigue life of a high-temperature alloy vacuum electron beam welded structure.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention firstly quantitatively determines the evolution law of the stress intensity factor at the notch tip under different shape factor ratios based on numerical simulation technology. Furthermore, according to the damage tolerance theory and the empirical relationship between defect morphology and the stress intensity factor at the notch tip, the traditional Murakami life prediction model is modified. This method can well consider the influence of defects with different morphologies on the fatigue performance of GH4169 electron beam welded joints.

[0017] This paper combines traditional fatigue life prediction with machine learning methods (SVR, RF, and XGBoost). Using physical parameters related to GH4169 welded joints derived from numerical simulations, the traditional machine learning method's input dataset is expanded and updated. This results in a new physics-guided machine learning method for GH4169 welded joints. This expanded physical information enhances the interpretability of traditional machine learning methods, while also improving their generalization and life prediction accuracy. By combining this method with nondestructive testing techniques used in engineering, it can effectively predict fatigue life for welded and additively manufactured components in aeroengines. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a drawing of a fatigue specimen of a GH4169 electron beam welded joint according to an embodiment of the present invention.

[0019] Figure 2This is the actual fatigue specimen of the GH4169 electron beam welded joint according to the embodiment of the present invention.

[0020] Figure 3 This is a scatter diagram of fatigue load-life of GH4169 according to an embodiment of the present invention.

[0021] Figure 4 This is the fatigue fracture analysis result of a partial GH4169 electron beam welded joint in an embodiment of the present invention.

[0022] Figure 5 This is the finite element simulation analysis result of the GH4169 electron beam welded joint in the embodiment of the present invention.

[0023] Figure 6 The stress intensity factor simulation analysis results of the GH4169 electron beam welded joint in the embodiment of the present invention are shown.

[0024] Figure 7 This is a fitting curve of the mechanism of the influence of defect morphology characteristics on the stress intensity factor at the notch tip in an embodiment of the present invention.

[0025] Figure 8 This is the fatigue life sensitivity analysis result of the GH4169 electron beam welded joint in the embodiment of the present invention.

[0026] Figure 9 This is a flow chart of the implementation of the data-driven and physically guided data-driven method according to an embodiment of the present invention.

[0027] Figure 10 This is the fatigue life prediction result of the GH4169 welded joint considering the welding defect morphology characteristics in the embodiment of the present invention.

[0028] Figure 11 This is the fatigue life prediction result of the GH4169 electron beam welded joint based on the traditional machine learning method in an embodiment of the present invention.

[0029] Figure 12 This is the fatigue life prediction result of the GH4169 electron beam welded joint based on the physics-guided machine learning method in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] Unless otherwise defined, the technical or scientific terms used herein shall have the ordinary meaning understood by persons of ordinary skill in the field to which the invention belongs. The words "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Words such as "include" or "comprise" and similar words mean that the elements or objects appearing before "include" or "comprise" include the elements or objects listed after "include" or "comprise" and their equivalents, and do not exclude other elements or objects. Words such as "connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0032] The present invention is mainly aimed at welded structures such as combustion chamber casings and flame tubes of aircraft engines, which serve under high temperature and alternating load conditions. Welding defects are inevitably introduced during the welding process. Defects of different sizes, positions, and morphological characteristics have completely different effects on the fatigue performance of welded structures. The current life prediction methods do not fully consider the impact of defect geometric morphological characteristics on fatigue performance.

[0033] This paper addresses the issue of weld defects, such as microcracks, incomplete penetration, and porosity, that are inevitably introduced during the welding process of electron-beam welded structures in GH4169 aeroengine combustion chamber casings. Furthermore, addressing the phenomenon of weld defects increasing the discreteness of fatigue life in GH4169 welded joints, the paper fully considers the influence of weld defect location and geometry. Using finite element simulation and Franc 3D crack growth analysis software, the paper quantitatively analyzes the influence of weld defect morphology characteristics, identifies the mechanism by which defect morphology influences the stress intensity factor at the defect tip, and establishes a new fatigue life prediction model for GH4169 welded joints. Furthermore, the paper trains the fatigue dataset (applied load, temperature, Vickers hardness, and defect equivalent size) of GH4169 welded joints using methods such as support vector regression (SVR), random forest (RF), and extreme gradient boosting (XGBoost), thereby predicting the fatigue life of GH4169 welded joints. On this basis, a new dataset was created using the physical parameters obtained during the numerical simulation. The original dataset was expanded to obtain an expanded dataset (applied load, temperature, Vickers hardness, defect equivalent size, defect area, stress intensity factor range, fatigue limit, and stress intensity factor threshold). The stress intensity factor range is the value that takes into account the defect morphology. Using support vector regression and the expanded dataset, a new fatigue life prediction method for GH4169 welded joints was established, combining physical and support vector regression algorithms.

[0034] The fatigue life prediction method for high-temperature alloy vacuum electron beam welded structures, which takes into account defect morphology, is based on the following principles: Using testing machine fractographic analysis, the weld defect is confirmed to be the source of fatigue cracks leading to structural failure. When the defect expands from its initial size to a critical size, fatigue failure is considered to have occurred. The fatigue life prediction method of the present invention involves the following steps:

[0035] The first step: Electron beam welding (EBW) of the GH4169 superalloy using a vacuum electron beam welding (EBW) process. Based on company standards and navigational aids, the welded joints were inspected using X-rays and fluorescent light. Following this inspection, a two-stage aging heat treatment was performed to obtain the GH4169 EB welded joints under the standard heat treatment regime. Standard tensile and fatigue test specimens were then processed using wire cutting. The surface quality of the test specimens, particularly the welds, was maintained consistent with that of the actual casing components.

[0036] Step 2: Based on national standards such as GB / T228.1-2010 "Metallic Materials - Tensile Tests - Part 1: Room Temperature Test Methods," GB / T228.2-2015 "Metallic Materials - Tensile Tests - Part 2: Elevated Temperature Test Methods," and GB / T3075-2008 "Metallic Materials - Fatigue Tests - Axial Force Control Method," static and fatigue properties of GH4169 welded joints were tested at room temperature and elevated temperatures (400°C, 500°C, and 550°C). The fatigue test data were plotted as SN curves. Simultaneously, SEM analysis of each fatigue failure fracture surface was performed using fractographic analysis to clarify the fatigue failure mechanism and fatigue crack source of the GH4169 electron beam welded joints. The maximum initial defect area, shape, and location of each failed specimen were also calculated.

[0037] The third step: perform finite element analysis on the GH4169 electron beam welded specimens, and use numerical simulation to clarify the influence of defect morphology on the fatigue fracture performance of GH4169 welded joints, especially the influence of stress intensity factor at the defect tip. Mainly through the combination of ABAQUS + Franc 3D, first use ABAQUS to perform static simulation analysis of the GH4169 welded joint at the peak load of each cycle load, and then import the inp file output by ABAQUS into the Franc 3D crack extension software, insert cracks of specific shape and size, and perform simulation analysis of the stress intensity factor at the defect tip. This step is mainly aimed at defects with similar area and position, and analyzes the specific influence of defect morphology. Specifically, the defect shape factor is introduced. f , f is the ratio of the defect semi-minor axis length to the semi-major axis length, f ∈(0,1], when f When it is 1, the defect is a standard circular shape. This step aims to conduct an in-depth study on the influence mechanism of slender defects, elliptical defects and circular defects on the tip stress intensity factor when the equivalent size, position and other features are similar based on numerical simulation technology.

[0038] Step 4: After clarifying the influence mechanism of defect morphology on the fatigue performance of GH4169 welded joints, a fatigue life prediction method for GH4169 welded joints can be established based on damage tolerance theory, which takes into account the initial geometric characteristics of defects. The main damage parameters involved are area 、 、 f、K max 、 K min ,∆ K th ,∆ K def Etc. The main formulas involved in the above parameters are as follows:

[0039] (1) (2) (3) Where, ∆K th The fatigue crack growth threshold value of the material structure with initial defects proposed by Murakami et al. C is the defect position influence coefficient, for internal defects C Take 1.56, for sub-surface defects, C Generally, 1.43 is taken, HV is Vickers hardness, R is the stress ratio, f is the defect shape factor, ∆K def The effective stress intensity factor range considering the defect morphology characteristics is ( K def,max - K def,min ), S is the influence coefficient of stress intensity factor and defect morphology obtained by numerical simulation, C 1 and m is the material constant, which can be obtained by optimizing the fatigue test result data. a is the crack size. Based on the damage tolerance theory, a life prediction method for GH4169 welded joints considering the defect morphology characteristics can be obtained through the above formula.

[0040] Step 5: Spearman rank correlation analysis is performed on the factors affecting the fatigue life of GH4169 welded joints. In order to clarify the factors affecting the fatigue life of GH4169 electron beam welded joints and their sensitivity, it is necessary to use Spearman rank correlation analysis to compare the fatigue life of GH4169 welded joints with temperature. T , test load σ , Vickers hardness HV, maximum initial defect area area , equivalent defect size 、 ∆ K th 、 σ 0.1 The main factors affecting the GH4169 weld joint can be preliminarily determined through Spearman rank correlation analysis, which can be used to determine the initial input data set for the next step of machine learning. Spearman rank correlation analysis is a nonlinear correlation analysis, and its formula is as follows:

[0041] (4); Where y is the fatigue life of GH4169 welded joint, x i are parameters such as temperature, applied load, Vickers hardness, defect area, and stress intensity factor.

[0042] Step 6: Use support vector regression (SVR), random forest (RF), extreme gradient boosting (XGBoost) and other methods to predict the fatigue life of GH4169 welded joints. First, establish an initial data set. The main data include test temperature T , test load σ , Vickers hardness HV, maximum initial defect area area , equivalent defect size There are 75 fatigue test data of GH4169 electron beam welded joints at three temperatures of 400℃, 500℃ and 550℃. The training set and test set are divided into a ratio of 7:3, with the training set accounting for 70% of all data. The hyperparameters of the support vector regression method are optimized by Bayesian optimization using the training data set. The main hyperparameters include the penalty parameter C , Error tolerance range parameters ε , kernel function control parameters γ After obtaining the hyperparameters of the SVR method using the training set, the hyperparameters were applied to the test data set to predict the fatigue life of GH4169 electron beam welded joints, thereby verifying the generalization capabilities of the SVR, RF, and XGBoost methods.

[0043] Step 7: The stress intensity factor, crack extension threshold, fatigue limit and fatigue life obtained by improving the physical model that take into account the defect morphology obtained by numerical simulation are used as an extended data set and expanded into the initial data set. Similarly, there are 75 fatigue test data of GH4169 electron beam welded joints at three temperatures of 400℃, 500℃ and 550℃. The training set and test set are divided in a ratio of 7:3, with the training set accounting for 70% of all data. Different from the sixth step, the expanded data set contains more data information with actual physical significance. The hyperparameters of the support vector regression method are solved by Bayesian optimization of the updated physical-training data set. The main hyperparameters include the penalty parameter C , Error tolerance range parameters ε , kernel function control parameters γ After using the physical dataset to obtain the hyperparameters of physics-driven SVR, RF, XGBoost and other methods, the hyperparameters were applied to the test dataset to predict the fatigue life of GH4169 electron beam welded joints, thereby verifying the generalization ability of physics-guided PI-SVR, PI-RF, PI-XGBoost and other methods.

[0044] First, vacuum electron beam welding technology was used to weld the solid solution state GH4169 high temperature alloy. The welding sample drawings and actual pictures are as follows: Figure 1 、 Figure 2 According to the relevant national test standards, the high temperature fatigue performance test of GH4169 welded joints was carried out at 400℃, 500℃ and 550℃, and the test results were plotted as a scatter plot. The results are shown in the figure below. Figure 3 Damage, from Figure 3 It can be clearly found that at low stress levels, some fatigue data have a relatively large dispersion. Further fracture analysis of fatigue failure fracture of GH4169 welded joints is carried out, and some results are as follows: Figure 4 As shown in the figure, based on the fracture direction, it can be judged that welding defects are the root cause of the increase in the discreteness of fatigue life of GH4169 electron beam welded joints.

[0045] In order to further study the effect of defect morphology on the fatigue behavior of GH4169 electron beam welded joints, we combined ABAQUS and Franc 3D software to simulate and analyze GH4169 welded joints with different defect morphology characteristics. The relevant results are as follows: Figure 4 、 Figure 5 、 Figure 6 and Figure 7 As shown, under the same initial area conditions, as the morphological characteristic parameters f ( z 1 / z2 ) changes, the effective driving stress intensity factor range ∆ of the defect tip under each working condition was calculated using ABAQUS finite element analysis software and Franc 3D crack growth analysis software. K def , it can be clearly found that the stress intensity factor at the defect tip has changed significantly. Therefore, it is very necessary to consider the effect of defect morphology on the fatigue performance of GH4169 electron beam welded joints. Based on damage tolerance theory and simulation analysis results, we have established a new method for predicting the fatigue life of welded structures that considers defect morphology characteristics. So far, we have basically determined the main factors affecting the fatigue life of GH4169 electron beam welded joints. In order to quantitatively clarify the mechanism of action between fatigue life and related influencing factors, the GH4169 welded joints and their influencing factors were analyzed through Spearman rank correlation analysis, and the following results were obtained: Figure 8 Fatigue life sensitivity analysis results are shown.

[0046] Finally, the fatigue life prediction problem of GH4169 electron beam welded joints based on traditional machine learning methods and physics-guided machine learning methods is presented in the following flowchart: Figure 9Before this, we first used the fatigue life prediction method considering the defect morphology characteristics to predict the fatigue life of GH4169 welded joints and compared it with the test results. The results are shown in Figure 10 As shown, the fatigue test results of GH4169 electron beam welded joints at three temperatures of 400℃, 500℃, and 550℃ as shown in Table 1 and the physical parameters obtained from the related numerical simulation analysis are established. Secondly, we use machine learning methods such as support vector regression (SVR), random forest (RF), and extreme gradient boosting (XGBoost) to train the GH4169 related data, optimize the hyperparameters, and finally predict the life of the test set. The relevant input data is the information from the second to sixth columns of Table 1. The relevant prediction results are shown in Figure 11 Finally, we expand the relevant physical information to the initial data set and establish a physical information guided machine learning method, namely the PI-SVR method. The physical data sets used for expansion mainly include ∆ K def ,∆ σ 0.1 (i.e., the information in the ninth to tenth columns of Table 1), and the fatigue life of the GH4169 welded joint predicted by considering the defect morphology characteristics. Finally, the fatigue life prediction effect of the GH4169 electron beam welded joint based on the physics-guided machine learning method is obtained as follows: Figure 12 As shown, further targeting Figure 10 、 Figure 11 、 Figure 12 A comparative analysis shows that the prediction results of the physics-guided machine learning method are better than the life prediction method that simply considers the defect morphology and the traditional machine learning method. Among them, the physics-guided random forest method (PI-RF) is superior to the physics-guided support vector regression method (PI-SVR) and the extreme gradient boosting (PI-XGBoost) method in terms of fatigue life prediction accuracy and model generalization ability.

[0047] Table 1 Input data sets related to GH4169 electron beam welded joints While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A fatigue life prediction method for high-temperature alloy vacuum electron beam welding structures, characterized in that: include: Obtain fatigue test data of target high-temperature alloy standard specimens; Fatigue fracture analysis of target high-temperature alloy standard specimens extracts the maximum area of ​​the initial defect, defect shape factor, and crack source location information. Based on the extracted maximum area, defect shape factor, and crack source location information, static simulation and crack growth simulation are used to obtain the stress intensity factor range that takes into account the defect morphology characteristics. A quantitative relationship between the stress intensity factor range and the defect shape factor is established, the damage tolerance theory formula is modified, and simulation data is obtained. The fatigue test data and simulation data are fused to form an initial data set. Based on the modified damage tolerance theory formula, the fatigue limit is calculated and expanded to the data set. With the expanded data set as input, the support vector regression method is used to determine the hyperparameters through Bayesian optimization, and the fatigue life prediction results of the high-temperature alloy vacuum electron beam welding structure are output.

2. The fatigue life prediction method for high-temperature alloy vacuum electron beam welding structure according to claim 1, characterized in that: The defect shape factor f is the ratio of the length of the defect's minor semi-axis to the length of the defect's major semi-axis, and f ∈(0,1].

3. The fatigue life prediction method for high-temperature alloy vacuum electron beam welding structure according to claim 1, characterized in that: When obtaining simulation data, a finite element analysis model of the target high-temperature alloy standard specimen is first established, and then a static simulation analysis is performed on the target high-temperature alloy standard specimen. The output file of the finite element analysis model is then imported into the crack propagation model, and cracks of specific shapes and sizes are inserted to obtain the simulation analysis results of the stress intensity factor at the corresponding defect tip. By fitting the relationship between different defect shape factors and stress intensity factors, the morphology influence coefficient is determined and the damage tolerance theory formula is corrected.

4. The fatigue life prediction method for high-temperature alloy vacuum electron beam welding structure according to claim 1, characterized in that: When determining the expanded data set, a Spearman rank correlation analysis is first performed on the factors affecting fatigue life to determine the main factors affecting fatigue life, and the determined main factors affecting fatigue life are used as the input data set.

5. The fatigue life prediction method for high-temperature alloy vacuum electron beam welding structure according to claim 4, characterized in that: The input data set includes temperature, stress amplitude, equivalent flaw size, Vickers hardness, fatigue life obtained from fatigue tests, and physical parameters obtained from simulation analysis of fatigue test data, including stress intensity factor range and fatigue limit.

6. The fatigue life prediction method for high-temperature alloy vacuum electron beam welding structure according to claim 1, characterized in that: The revised damage tolerance formula is: （1） （2） （3） Where, ∆K th is the fatigue crack growth threshold of the material structure containing initial defects, C is the defect position influence coefficient, for internal defects C Take 1.56, for sub-surface defects, C Generally, 1.43 is taken, HV is Vickers hardness, is the defect equivalent size, R is the stress ratio, f is the defect shape factor, z 1 is the length of the defect's minor semi-axis; z 1 is the length of the defect's minor axis; ∆ K def The stress intensity factor range considering the defect morphology characteristics is ( K def,max - K def,min ), S is the influence coefficient of stress intensity factor and defect morphology obtained by numerical simulation, C 1 and m is the material constant, which can be obtained by optimizing the fatigue test result data. a is the crack size, N is the fatigue life.

7. A fatigue life prediction device for vacuum electron beam welding of high-temperature alloys of aircraft engines, characterized in that: include: A receiving module is configured to receive an input data set of a target standard sample, wherein the input data value includes fatigue test data and physical parameters obtained by simulation analysis based on the fatigue test data, wherein the physical parameters include a stress intensity factor range and a fatigue limit; The prediction module is used to predict the fatigue life of the high-temperature alloy vacuum electron beam welded structure by using a support vector regression model and an input data set.

8. An electronic device, characterized in that: include: at least one processor and memory; The memory stores computer-executable instructions; The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the fatigue life prediction method for a high-temperature alloy vacuum electron beam welding structure according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions. When the processor executes the computer-executable instructions, the fatigue life prediction method for the high-temperature alloy vacuum electron beam welding structure according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the fatigue life prediction method for a high-temperature alloy vacuum electron beam welded structure according to any one of claims 1 to 6 is implemented.

Images