Surface crack fatigue propagation morphology and life prediction method based on machine learning

Abstract

The invention relates to the technical field of structural engineering, and provides a surface crack fatigue propagation morphology and life prediction method based on machine learning, which comprises the following steps: establishing a finite element simulation model containing a surface crack, calculating a stress intensity factor of the front edge of the surface crack, and generating a finite element simulation data set; performing dimensionless processing on parameters in the simulation data set to generate a machine learning data set; optimizing the back propagation neural network through a hybrid optimization algorithm, and constructing a surface crack stress intensity factor prediction model; the surface crack stress intensity factor prediction model is combined with a fatigue crack growth rate formula to cyclically predict the surface crack growth morphology and the service life of the load bearing structure; establishing a visual interaction interface of parameter input and output and image output based on the surface crack stress intensity factor prediction model and the fatigue crack growth rate formula; a user can quickly predict the propagation morphology of the surface crack and calculate the service life without mathematical analysis and code writing.

Description

Technical Field

[0001] The present invention relates to the technical field of structural engineering, and particularly to a method for predicting the fatigue crack growth morphology and life of surface cracks based on machine learning. Background Art

[0002] Components such as pressure vessels and pipelines in the energy industry usually operate under relatively high working pressures, and often under alternating loads. Surface cracks are a common defect, which will continuously expand under the action of alternating loads, eventually leading to fatigue fracture of the structure. In order to ensure the safe service of the structure, it is particularly important to accurately predict the shape change and life assessment during crack propagation.

[0003] Finite element analysis can more accurately simulate the fatigue crack growth shape in a structure. However, the computational workload of finite element analysis is large. Especially during crack propagation, a large number of iterative calculations are required. At the same time, the simulation of crack propagation by finite element analysis usually requires fine mesh division, especially in the crack tip region, which results in a long calculation time and high calculation cost, further increasing the demand for computing resources and reducing the computing efficiency.

[0004] To solve the above existing problems, for example, Chinese Patent with publication number CN111859716A discloses a method for predicting the fatigue crack growth shape of a semi-elliptical surface crack. By introducing the initial and extended crack geometric parameters and the stress intensity factor amplitude, and combining with the crack growth amount formula, an iterative method is used to determine the shape and position of the extended crack. This method does not rely on finite element simulation, reducing the calculation time. The stress intensity factor amplitude of regular cracks is calculated by existing formulas, and for irregular cracks, it is calculated by database interpolation. However, the existing formulas and linear interpolation assumptions limit the scope of application and accuracy of this method. Especially when dealing with irregular cracks, large errors may be introduced. Another example is Chinese Patent with publication number CN115470675A, which discloses a method for predicting the crack propagation path by combining machine learning and finite element simulation. By importing the current crack position parameters into the finite element simulation model, crack propagation data is output and input into the trained machine learning model to predict the next crack position, and the iteration is repeated to predict the crack propagation path. This method combines the advantages of high accuracy of finite element simulation and high efficiency of machine learning. However, grid redivision and simulation calculations are required for each iteration, and there is still room for improvement in the overall calculation efficiency.

[0005] Therefore, how to improve the computational efficiency of predicting the propagation shape and life of surface cracks under non-standard crack geometries or complex load conditions has become a technical problem to be solved urgently. Summary of the Invention

[0006] In view of this, the main object of the present invention is to provide a method for predicting the surface crack fatigue growth morphology and life based on machine learning, aiming at the technical problem of predicting the computational efficiency of the surface crack growth morphology and life.

[0007] To achieve the above object, in a first aspect, the present invention provides a method for predicting the surface crack fatigue growth morphology and life based on machine learning, including the following steps:

[0008] S1, establish a finite element simulation model containing a surface crack, calculate the stress intensity factor at the front edge of the surface crack, and generate a finite element simulation data set;

[0009] S2, perform dimensionless processing on the size of the surface crack, the size of the load-bearing structure, and the stress intensity factor at the front edge of the surface crack in the finite element simulation data set to generate a machine learning data set;

[0010] S3, optimize the backpropagation neural network through a hybrid optimization algorithm to construct a surface crack stress intensity factor prediction model;

[0011] S4, the surface crack stress intensity factor prediction model combines with the fatigue crack growth rate formula to cyclically predict the surface crack growth morphology of the load-bearing structure and calculate the surface crack growth life of the load-bearing structure;

[0012] S5, based on the surface crack stress intensity factor prediction model and the fatigue crack growth rate formula, establish a visual interactive interface for parameter input and output and image output.

[0013] Optionally, the cyclically predicting the surface crack growth morphology of the load-bearing structure and calculating the life of the surface crack growth morphology of the load-bearing structure includes the following steps:

[0014] S41, based on the initial size of the crack, calculate the stress intensity factor change range of the crack through the stress intensity factor prediction model;

[0015] S42, substitute the value of the stress intensity factor change range into the fatigue crack growth rate formula to solve the crack growth amount in this cycle;

[0016] S43, automatically update the crack size to the sum of the initial crack size and the crack growth amount in this cycle;

[0017] S44, use the updated crack size as the initial crack size, cycle steps S41 to S44 until the crack size expands from the initial depth to the final depth to obtain the value of the number of cycles, and the value of the number of cycles is the life of the surface crack growth morphology.

[0018] Optionally, the fatigue crack growth rate formula is the Paris formula.

[0019] Optionally, the visual interaction interface for establishing parameter input and output and image output is specifically:

[0020] Establish a visual interaction interface for crack parameter input, load parameter input, neural network hyperparameter input, load-bearing structure parameter input with surface cracks, image output of the load-bearing structure with surface cracks, surface crack propagation morphology parameter output, surface crack stress intensity factor output, fatigue life prediction result output, and empirical formula comparison result output.

[0021] Optionally, establishing the finite element simulation model with surface cracks includes the following steps:

[0022] S11, establish a finite element model of the load-bearing structure through ABAQUS software and export the output file of the finite element model of the load-bearing structure;

[0023] S12, insert a semi-elliptical surface crack into the model of the output file of the load-bearing structure model and output the finite element model of the load-bearing structure with surface cracks.

[0024] Optionally, inserting the semi-elliptical surface crack into the model of the output file of the load-bearing structure model includes the following steps:

[0025] S121, divide the finite element model of the load-bearing structure into an overall model and a local model;

[0026] S122, define the short semi-axis size and long semi-axis size of the semi-elliptical surface crack, introduce the semi-elliptical surface crack into the local model to form a local model with surface cracks;

[0027] S123, mesh the local model with surface cracks;

[0028] S124, merge the overall model and the local model with surface cracks to form a finite element model of the load-bearing structure with surface cracks.

[0029] Optionally, the dimensionless processing of the size of the surface crack, the size of the load-bearing structure, and the stress intensity factor at the front edge of the surface crack in the finite element simulation dataset to generate the finite element simulation dataset includes the following steps:

[0030] S21. Perform dimensionless processing on the short semi-axis size of the crack, the long semi-axis size of the crack, the thickness of the load-bearing structure, the inner radius of the load-bearing structure, and the stress intensity factor. Use the normalized distance of the crack front to replace the parametric angle of the ellipse to form dimensionless quantities. The dimensionless quantities are specifically: the ratio of the short semi-axis of the crack to the long semi-axis of the crack, the ratio of the short semi-axis of the crack to the thickness of the load-bearing structure, the ratio of the thickness of the load-bearing structure to the characteristic size of the load-bearing structure, the normalized distance of the crack front, and the surface crack boundary correction factor.

[0031] S22. Generate an axial semi-elliptical crack data set and a circumferential semi-elliptical crack data set.

[0032] Optionally, the optimizing the backpropagation neural network by the hybrid optimization algorithm includes the following steps:

[0033] S31. Optimize the weights and biases of the backpropagation neural network through the particle swarm algorithm to obtain the optimal combination of weights and biases of the backpropagation neural network.

[0034] S32. Apply the crossover and mutation operations of the genetic algorithm to the particle swarm algorithm to enhance the population diversity and global search ability.

[0035] Optionally, the step S1 is:

[0036] S11a. Based on the position of the surface crack in the load-bearing structure, generate a finite element simulation model containing surface cracks at different positions.

[0037] Among them, the position of the surface crack in the load-bearing structure is specifically the axial direction of the outer surface of the load-bearing structure, the axial direction of the inner surface of the load-bearing structure, the circumferential direction of the outer surface of the load-bearing structure, and the circumferential direction of the inner surface of the load-bearing structure.

[0038] Among them, the parameter range includes circular cracks to long elliptical cracks, shallow cracks to deep cracks, thin-walled load-bearing structures to thick-walled load-bearing structures.

[0039] S12a. The finite element simulation model containing surface cracks includes a load-bearing structure model with surface cracks under the action of internal pressure load and uniform tensile load at both ends.

[0040] Optionally, the step S3 further includes:

[0041] S33. Adjust the neural network hyperparameters, verify the fitting effect of the stress intensity factor prediction model based on machine learning according to the evaluation index, and save the neural network whose fitting accuracy meets the requirements for subsequent calling during visualization interaction; among them, the evaluation criteria include at least one of the coefficient of determination, mean absolute error, mean bias error, and mean absolute percentage error.

[0042] Second aspect, the present invention provides a device for predicting the morphology and life of surface crack fatigue propagation based on machine learning, the device comprising:

[0043] A data set acquisition module, configured to receive a finite element simulation model with surface cracks established by a user, calculate the stress intensity factor at the front edge of the surface crack, verify the stress intensity factor solved by the finite element simulation model through the Newman-Raju empirical formula, and generate a finite element simulation data set;

[0044] A data preprocessing module, configured to perform dimensionless processing on the size of the surface crack, the size of the load-bearing structure, and the stress intensity factor at the front edge of the surface crack in the finite element simulation data set to generate a machine learning data set;

[0045] A machine learning model generation module, configured to optimize a backpropagation neural network through a hybrid optimization algorithm to construct a stress intensity factor prediction model based on machine learning;

[0046] A crack propagation morphology and life prediction module, configured to combine the surface crack stress intensity factor prediction model with the fatigue crack propagation rate formula to cyclically predict the surface crack propagation morphology of the load-bearing structure and calculate the life of the surface crack propagation morphology of the load-bearing structure;

[0047] A graphical user interface output module, configured to establish a visual interaction interface for parameter input and output and image output based on the surface crack stress intensity factor prediction model and the fatigue crack propagation rate formula.

[0048] Optionally, the crack propagation morphology and life prediction module includes:

[0049] A first prediction sub-module, configured to calculate the stress intensity factor change range of the crack based on the initial size of the crack through the stress intensity factor prediction model;

[0050] A second prediction sub-module, configured to substitute the value of the stress intensity factor change range into the fatigue crack propagation rate formula to solve the crack propagation amount in this cycle;

[0051] A third prediction sub-module, configured to automatically update the size of the crack to the sum of the initial size of the crack and the crack propagation amount in this cycle;

[0052] A cyclic expansion and life prediction module, configured to use the updated crack size as the initial size of the crack, cycle through the first prediction sub-module, the second prediction sub-module, the third prediction sub-module, and the cyclic expansion and life prediction module until the crack size expands from the initial depth to the final depth to obtain the value of the number of cycles, and the value of the number of cycles is the life of the surface crack propagation morphology.

[0053] Optionally, the fatigue crack growth rate formula disclosed by the above module is the Paris formula.

[0054] Optionally, the graphical user interface output module includes:

[0055] A crack parameter input module;

[0056] A load parameter input module;

[0057] A neural network hyperparameter input module;

[0058] A load-bearing structure parameter input module;

[0059] An image output module of the load-bearing structure with a surface crack;

[0060] A surface crack propagation morphology parameter output module;

[0061] A surface crack stress intensity factor output module;

[0062] A fatigue life prediction result output module and an empirical formula comparison result output module.

[0063] Optionally, the establishment of the finite element simulation model with a surface crack includes:

[0064] A load-bearing structure modeling module, which is used to establish a finite element model of the load-bearing structure through ABAQUS software and export the finite element model output file of the load-bearing structure;

[0065] A crack insertion modeling module, which is used to insert a semi-elliptical surface crack into the model of the load-bearing structure model output file and output the finite element model of the load-bearing structure with a surface crack.

[0066] In a third aspect, the present invention provides an electronic device, including a memory and a processor, wherein:

[0067] The memory is used to store a computer program;

[0068] The processor is used to execute the computer program to implement the surface crack stress intensity factor prediction method based on machine learning disclosed in the first aspect above.

[0069] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and the computer program is used to be run by a processor to implement the surface crack fatigue propagation morphology and life prediction method based on machine learning disclosed in the first aspect above.

[0070] Fifth aspect, the present invention provides a computer program product, including a computer program / instructions, which when executed by a processor, implement the method for predicting the fatigue crack growth morphology and life of surface cracks based on machine learning disclosed in the first aspect as described above.

[0071] In the technical solution provided by the present invention, first, a finite element model of a load-bearing structure model with surface cracks covering different positions, sizes, under internal pressure loads, and under uniform tensile loads at both ends of a surface semi-elliptical crack is established in advance, and a machine learning data set is formed by a neural network model learning the finite element model data set, and a stress intensity factor prediction model based on machine learning is formed. When a user inputs the size parameters of the crack through a graphical visualization interface, the prediction model can be directly called to intuitively obtain the predicted stress intensity factor of the surface crack, avoiding the process of finite element modeling, mesh generation, and calculation when the user uses it, avoiding the long time-consuming of numerical simulation and the result being easily affected by the mesh quality, and improving the calculation efficiency of the stress intensity factor of the surface crack.

[0072] Second, the present invention combines the stress intensity factor prediction model with the Paris formula to achieve accurate prediction of the crack growth morphology, and automatically calculates the stress intensity factor after crack growth through the model to realize automatic crack growth. Through a graphical user interface (GUI) developed based on MATLAB, the user can input crack positions, crack parameters, and cylinder size parameters through simple and intuitive operations, quickly complete the prediction of the stress intensity factor, the prediction of the crack growth morphology, and the calculation of the crack life, without manual calculation and mastering programming languages, greatly reducing the technical threshold.

[0073] Third, the machine learning training data set is generated by the combined simulation calculation of the finite element software ABAQUS and the crack analysis software FRANC3D, avoiding the problems of insufficient accuracy and limited application range of existing empirical formulas, and can realize the prediction of the stress intensity factor of surface cracks with different sizes and positions under complex loads. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] One or more embodiments are exemplarily illustrated by corresponding drawings. These exemplary illustrations do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings are represented as similar elements, unless otherwise stated, and the drawings in the drawings do not constitute a proportional limitation.

[0075] Figure 1 It is a schematic flowchart of the method for predicting the fatigue crack growth morphology and life of surface cracks based on machine learning disclosed in one or more embodiments of the present application;

[0076] Figure 2Schematic diagram of modules of a surface crack fatigue propagation morphology and life prediction device based on machine learning disclosed in one or more embodiments of the present application.

[0077] Figure 3 Schematic diagram of modules of a crack propagation morphology and life prediction module disclosed in one or more embodiments of the present application;

[0078] Figure 4 Schematic diagram of the structure of a finite element model of a load-bearing structure containing surface cracks disclosed in one or more embodiments of the present application;

[0079] Figure 5 Schematic diagram of the composition of a BP neural network disclosed in one or more embodiments of the present application;

[0080] Figure 6 Schematic diagram of the regression value between the predicted value and the simulated value of a PSOGA-BP neural network model disclosed in one or more embodiments of the present application;

[0081] Figure 7 Schematic diagram of a graphical user interface disclosed in one or more embodiments of the present application;

[0082] In the figure: 100 - dataset acquisition module; 200 - data preprocessing module; 300 - machine learning model generation module; 400 - crack propagation morphology and life prediction module; 410 - first prediction sub-module; 420 - second prediction sub-module; 430 - third prediction sub-module; 440 - cyclic expansion and life prediction module; 500 - graphical user interface output module. Detailed implementation manners

[0083] For ease of understanding of the present invention, the present invention will be described in more detail below with reference to the accompanying drawings and specific embodiments. It should be noted that when an element is expressed as "fixed to" another element, it can be directly on the other element, or there can be one or more intermediate elements therebetween. When an element is expressed as "connected to" another element, it can be directly connected to the other element, or there can be one or more intermediate elements therebetween. The terms "vertical", "horizontal", "left", "right", "inside", "outside" and similar expressions used in this specification are only for the purpose of illustration. In the description of the present invention, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating relative importance or implicitly indicating the quantity of the indicated technical features. Thus, unless otherwise stated, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features; the meaning of "a plurality" is two or more. The term "comprising" and any deformation thereof mean non-exclusive inclusion, and there may be or be added one or more other features, integers, steps, operations, units, components and / or their combinations.

[0084] In addition, unless otherwise clearly stipulated and defined, the terms "install", "connect", and "join" shall be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or the communication inside two components. All the technical and scientific terms used in this specification have the same meanings as those commonly understood by those skilled in the technical field to which the present invention belongs. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not used to limit the present invention. The term "and / or" used in this specification includes any and all combinations of one or more of the related listed items.

[0085] In addition, the technical features involved in different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0086] Embodiment 1

[0087] To achieve the above object, an embodiment of the present invention provides a method for predicting the surface crack fatigue propagation morphology and life based on machine learning. Please refer to the appendix Figures 1-7 where the appendix Figure 1 is a schematic flow chart of a method for predicting the surface crack fatigue propagation morphology and life based on machine learning. Specifically, it includes the following steps:

[0088] S1. Establish a finite element simulation model containing surface cracks, calculate the stress intensity factor at the front edge of the surface crack, and generate a finite element simulation data set;

[0089] S2. Nondimensionalize the size of the surface crack, the size of the load-bearing structure, and the stress intensity factor at the front edge of the surface crack in the finite element simulation data set to generate a machine learning data set;

[0090] S3. Optimize the backpropagation neural network through a hybrid optimization algorithm to construct a surface crack stress intensity factor prediction model;

[0091] S4. The surface crack stress intensity factor prediction model combines with the fatigue crack growth rate formula to cyclically predict the surface crack propagation morphology of the load-bearing structure and calculate the life of the surface crack propagation morphology of the load-bearing structure;

[0092] S5. Based on the surface crack stress intensity factor prediction model, establish a visual interaction interface for parameter input and output and image output.

[0093] Specifically, first, a finite element simulation model with a surface crack needs to be established, and the stress intensity factor at the front edge of the surface crack is solved. Through finite element simulation analysis, a machine learning dataset is constructed, which includes surface crack size parameters, load-bearing structure size parameters, the normalized distance at the crack front edge, and the dimensionless stress intensity factor at the crack front edge. Then, a prediction model for the stress intensity factor of the surface crack is established through a machine learning algorithm. By combining the stress intensity factor prediction model with the fatigue crack growth rate formula, the crack growth morphology and growth life are predicted. Finally, a graphical user interface (GUI) based on the above model is developed to improve the operability and application of the model. This interface integrates functions such as data input, model operation, result visualization, and crack growth morphology prediction. Users can input relevant parameters (such as crack size, cylinder size, and load magnitude) through an intuitive interactive interface and quickly obtain the prediction results of the stress intensity factor and the calculation results of the crack growth morphology and life.

[0094] Taking the internally pressurized cylinder with a semi-elliptical surface crack as an example, the machine learning dataset consists of input data (the ratio of the cylinder thickness to the inner radius of the cylinder t / R i , the ratio of the short semi-axis to the long semi-axis of the semi-elliptical crack a / c, the ratio of the short semi-axis of the semi-elliptical crack to the cylinder thickness a / t, the normalized distance at the crack front edge), and output data (the dimensionless stress intensity factor).

[0095] A graphical user interface (GUI) is developed to facilitate access to and visualization of relevant problems and existing machine learning (ML) solutions; the PSOGA-BP prediction model is designed as a graphical user interface using MATLAB. This graphical user interface can achieve input of crack parameters, setting of hyperparameters of the PSOGA-BP neural network training model, output of the stress intensity factor of the surface crack, comparison of the schematic diagram of the cylinder with a surface semi-elliptical crack and the calculation results and prediction results of the stress intensity factor by the Newman-Raju empirical formula, etc. In the graphical user interface, users do not need to learn complex codes but can operate through the graphical objects therein, greatly reducing the usage threshold.

[0096] Specifically, in this embodiment, the load-bearing structure includes a cylinder structure and also includes load-bearing structures of different shapes in other actual usage scenarios, such as cube, cuboid, cone, triangular prism, etc.; the load-bearing structure in this embodiment and in the following text is a cylinder structure;

[0097] Please refer to Appendix Figure 4 , Appendix Figure 4 At 4a in Appendix Figure 4 is the finite element model of the cylinder with an external surface circumferential semi-elliptical crack, and at 4b in Appendix Figure 4At 4c is the finite element model of a cylinder with an axial semi-elliptical crack on the outer surface, attached Figure 4 At 4d is the finite element model of a cylinder with an axial semi-elliptical crack on the inner surface. The positions of the built surface semi-elliptical cracks on the cylinder include the outer surface axial, the inner surface axial, the outer surface circumferential, and the inner surface circumferential. The parameter range includes circular cracks (a / c = 1) to long elliptical cracks (a / c = 0.1), shallow cracks (a / t = 0.1) to deep cracks (a / t = 0.9), thin-walled cylinders (t / R i = 0.1) to thick-walled cylinders (t / R i = 1.0). The established finite element models include two load conditions: the cylinder model with surface cracks under internal pressure load, and the cylinder model with surface cracks under uniform tensile loads at both ends, with a total of 3,600 different samples.

[0098] In some embodiments, step S3 further includes:

[0099] S33, adjusting the neural network hyperparameters, verifying the fitting effect of the stress intensity factor prediction model based on machine learning according to the evaluation index, and saving the neural network that meets the requirements of the fitting accuracy for direct calling during subsequent visual interaction;

[0100] In some embodiments, the evaluation criteria in step S33 include at least one of the coefficient of determination (R 2 ), mean absolute error (MAE), mean bias error (MBE), and mean absolute percentage error (MAPE). In this embodiment, the evaluation criterion is the correlation coefficient (R 2 ).

[0101] Specifically, please refer to the attachment Figure 5 attachment Figure 5As shown in the schematic diagram of the BP neural network composed of an input layer, a hidden layer, and an output layer, this application established a BP neural network with four inputs, a single output, and multiple hidden layers. The weights and biases of the BP neural network are usually initialized randomly, which may affect the prediction accuracy. To solve this problem, the particle swarm optimization algorithm (PSO) is used in the present invention to optimize the weights and biases of the neural network, aiming to find the optimal solution. However, the particle swarm algorithm is prone to falling into a local optimal solution and has a low search accuracy in the later stage. To overcome this defect, the crossover and mutation operations in the genetic algorithm (GA) are introduced to enhance the diversity of particles, thereby improving the global search ability and optimization accuracy. After establishing the neural network model, adjusting the hyperparameters is a key step in improving the model performance. The adjustment of hyperparameters helps to train the neural network more scientifically, thereby obtaining a more efficient machine learning model. Among the hyperparameters, the learning rate and the number of neurons in the hidden layer have an important impact on the fitting accuracy of the model. The hyperparameters of the neural network are optimized by the exhaustive method. The results show that when the number of training times is 1000, the learning rate is 0.00001, the convergence error is 1e-06, the proportion of the training set is 0.8, the number of population iterations is 20, the particle velocity range is [-1, 1], the particle position range is [-1, 1], and the neural network structure is three hidden layers (including 15, 10, and 10 neurons respectively), the fitting accuracy of the model reaches a relatively high level. Please refer to the regression value between the predicted value and the simulated value of the neural network model shown in Figure 6 the attached

[0102] In some embodiments, establishing a finite element simulation model with a surface crack includes the following steps:

[0103] S11, establishing a finite element model of the load-bearing structure through ABAQUS software and exporting the output file of the finite element model of the load-bearing structure;

[0104] S12, inserting a semi-elliptical surface crack into the model of the output file of the load-bearing structure model through FRANC3D software and outputting a finite element model of the load-bearing structure with a surface crack.

[0105] Specifically, due to the large number of samples and the wide size range included, the finite element software ABAQUS and the crack analysis software FRANC3D are used for joint simulation. First, after completing the finite element model of the cylinder in ABAQUS, the output file (.inp) is generated and imported into FRANC3D.

[0106] In some embodiments, inserting a semi-elliptical surface crack into the model of the output file of the cylinder structure model through FRANC3D software includes the following steps:

[0107] S121, dividing the finite element model of the load-bearing structure into an overall model and a local model;

[0108] S122. Define the short semi - axis size and long semi - axis size of the semi - elliptical surface crack, and introduce the semi - elliptical surface crack into the local model to form a local model with a surface crack.

[0109] S123. Mesh the local model with a surface crack.

[0110] S124. Merge the global model and the local model with a surface crack to form a finite - element model of the load - bearing structure with a surface crack.

[0111] Specifically, please refer to the appendix Figure 4 In this application, the finite - element model established is jointly established by ABAQUS and FRANC3D. The cylindrical model is established, the material properties are set, the boundary conditions are set, and the load is applied in ABAQUS. The construction of the cylindrical finite - element model is completed in ABAQUS and exported as an.inp file, and then imported into FRANC3D. To improve the calculation efficiency and analysis accuracy, the cylindrical model is divided into a global (GLOBAL) model and a local (LOCAL) model. In Franc3D, select the semi - elliptical surface crack, define the short semi - axis size a and long semi - axis size c of the initial semi - elliptical surface crack, and introduce the surface semi - elliptical crack into the LOCAL model. Then, the LOCAL model with the added crack is meshed by FRANC3D. Finally, the GLOBAL model and the LOCAL model are merged to form a complete cylindrical finite - element model containing the initial crack.

[0112] Formula for the stress intensity factor of an axial semi - elliptical crack on the outer surface of the cylinder:

[0113]

[0114] P refers to the internal pressure load, R i refers to the inner radius of the cylinder, a refers to the short semi - axis of the crack, c refers to the long semi - axis of the crack, t refers to the thickness of the cylinder, refers to the parametric angle of the ellipse, Q refers to the second - kind elliptic integral, F e is the boundary correction factor for the crack on the outer surface of the cylinder.

[0115] Formula for the stress intensity factor of an axial semi - elliptical crack on the inner surface of the cylinder:

[0116]

[0117] F i is the boundary correction factor for the crack on the inner surface of the cylinder.

[0118] The stress intensity factor K1 is affected by the model size and the magnitude of the external load. To eliminate this part of the influence, formulas 1 and 3 are normalized:

[0119]

[0120]

[0121] Formula for the stress intensity factor of annular semi-elliptical crack on the outer surface of a cylinder:

[0122]

[0123] σ is the axial tensile load.

[0124] Formula for stress intensity factor of circumferential semi-elliptical crack on inner surface of cylinder:

[0125]

[0126] Normalize Equation 6 and Equation 7:

[0127]

[0128] The finite element simulation model is jointly established by the finite element software ABAQUS and the crack analysis software FRANC3D. ABAQUS is responsible for establishing the load-bearing structure, while FRANC3D focuses on inserting cracks. The joint simulation not only improves the calculation accuracy, but also improves the efficiency of modeling.

[0129] In some embodiments, dimensionless processing is performed on the size of the surface crack, the size of the load-bearing structure, and the stress intensity factor of the surface crack front in the finite element simulation data set to generate the finite element simulation data set, including the following steps:

[0130] S21, performing dimensionless processing on the crack semi-minor axis size, the crack semi-major axis size, the load-bearing structure thickness, the load-bearing structure inner radius, and the stress intensity factor, replacing the parameter angle of the ellipse with the normalized distance of the crack front edge to form a dimensionless quantity, the dimensionless quantity specifically being: the ratio of the crack semi-minor axis to the crack semi-major axis size, the ratio of the crack semi-minor axis to the load-bearing structure thickness, the ratio of the load-bearing structure thickness to the load-bearing structure characteristic size, the normalized distance of the crack front edge, and the surface crack boundary correction factor; in this embodiment, the ratio of the load-bearing structure thickness to the load-bearing structure characteristic size is: the ratio of the load-bearing structure thickness to the cylinder inner radius size;

[0131] S22: Generate an axial semi-elliptical crack dataset and a circumferential semi-elliptical crack dataset.

[0132] Specifically, in order to more conveniently establish a machine learning dataset, the angle Normalized distance to crack front Instead, five independent dimensionless quantities are all related, so the axial semi-elliptical crack data set adopts the following structure:

[0133] Input variables:

[0134]

[0135] Output target:

[0136]

[0137] The circumferential semi-elliptical crack dataset has the following structure:

[0138] Input variables:

[0139]

[0140] Output target:

[0141]

[0142] The machine learning dataset was generated by dimensionlessly processing crack geometry and stress intensity factors. A finite element model was established using ABAQUS and FRANC3D, and the stress intensity factors were calculated. The results were compared with the Newman-Raju empirical formula, showing an error of less than 5%, demonstrating the reliability of the finite element model and ensuring both modeling accuracy and data reliability.

[0143] A finite element model was established through joint simulation of ABAQUS and FRANC3D. ABAQUS was responsible for modeling the load-bearing structure and setting boundary conditions, while FRANC3D focused on crack insertion, meshing, and accurate solution of stress intensity factors. This approach avoids the problems of insufficient accuracy of empirical formulas, limitations of specific crack structures, and time-consuming complex crack meshing when using ABAQUS alone. Joint simulation not only improves computational efficiency and saves time costs, but also quickly generates large amounts of high-precision model data, providing reliable support for subsequent machine learning model training, while ensuring the accuracy and reliability of stress intensity factor results. The dimensionless treatment of crack geometry and stress intensity factors covers a wide range of crack sizes and locations, improving the representativeness and applicability of the data set.

[0144] In some embodiments, optimizing a back-propagation neural network using a hybrid optimization algorithm includes the following steps:

[0145] S31, optimize the weights and biases of the back-propagation neural network through the particle swarm algorithm to obtain the optimal weight and bias combination of the back-propagation neural network;

[0146] S32, by applying the crossover and mutation operations of the genetic algorithm to the particle swarm algorithm, the population diversity and global search capability are enhanced.

[0147] Specifically, the stress intensity factor prediction model is constructed by hybrid optimization of the BP neural network using the particle swarm optimization algorithm (PSO) and the genetic optimization algorithm (GA). The BP neural network consists of an input layer, a hidden layer, and an output layer. In order to improve the accuracy and generalization ability of the model, this study attempts to construct BP neural networks with different numbers of hidden layers and neuron configurations, and uses the root mean square error (RMSE) to measure the prediction accuracy of the model. In response to the problems of slow training speed, low accuracy, and easy falling into local optimality of traditional BP neural networks, this study introduces the particle swarm optimization algorithm (PSO) and genetic algorithm (GA) for improvement. The fast convergence characteristics of the PSO algorithm are used to optimize the neural network training process, and the crossover and mutation operations of GA are integrated into the PSO to increase population diversity and enhance global search capabilities, thereby effectively improving the neural network training efficiency and prediction accuracy.

[0148] In some embodiments, step S1 is:

[0149] S11a, generating a finite element simulation model containing surface cracks at different locations based on the location of the surface cracks in the load-bearing structure;

[0150] The position of the surface crack in the load-bearing structure is specifically the axial direction of the outer surface of the load-bearing structure, the axial direction of the inner surface of the load-bearing structure, the circumferential direction of the outer surface of the load-bearing structure, and the circumferential direction of the inner surface of the load-bearing structure;

[0151] Among them, the parameter range includes circular cracks to long elliptical cracks, shallow cracks to deep cracks, and thin-walled load-bearing structures to thick-walled load-bearing structures;

[0152] S12a, the finite element simulation model with surface cracks includes a load-bearing structure model with surface cracks under internal pressure load and uniform tensile load at both ends.

[0153] In some embodiments, the cyclically predicting the surface crack growth morphology of the load-bearing structure and calculating the life of the surface crack growth morphology of the load-bearing structure comprises the following steps:

[0154] S41, based on the initial size of the crack, calculate the stress intensity factor variation range of the crack using the stress intensity factor prediction model;

[0155] S42, substitute the values of the stress intensity factor variation range into the fatigue crack growth rate formula to solve for the crack growth amount in this cycle;

[0156] S43, updating the crack size to the sum of the initial crack size and the crack extension in this cycle;

[0157] S44, using the updated crack size as the initial crack size, looping steps S41 to S44 until the crack size expands from the initial depth to the final depth, and obtaining the value of the number of cycles, which is the lifespan of the surface crack extension morphology.

[0158] In some embodiments, the fatigue crack growth rate formula is the Paris formula.

[0159] In some embodiments, the establishment of a visual interactive interface for parameter input and output and image output is specifically as follows:

[0160] A visual interactive interface is established for crack parameter input, load parameter input, neural network hyperparameter input, image output of the load-bearing structure with surface cracks, surface crack extension morphology parameter output, fatigue life prediction result output, and empirical formula comparison result output.

[0161] Specifically, the Paris Law is one of the classic empirical formulas in fracture mechanics that describes fatigue crack growth behavior. This formula shows that the crack growth rate da / dN in a material is related to a certain power relationship of the stress intensity factor range ΔK. The stress intensity factor of the surface semi-elliptical crack front obtained by the prediction model is combined with the formula to obtain the next crack growth morphology. The prediction model will automatically calculate the stress intensity factor of the crack front after expansion to achieve automatic crack expansion. The formula is expressed as:

[0162]

[0163] Where: da / dN is the crack length a growth rate with the number of cycles N; ΔK is the stress intensity factor range, defined as ΔK = K max -K min , that is, the difference between the maximum stress intensity factor and the minimum stress intensity factor in the loading cycle; C and m are material-related parameters, usually determined by experiments.

[0164] The steps for calculating crack life using the Paris formula combined with the stress intensity factor prediction model are as follows: To determine the crack expansion from a0 (initial crack depth) to a N(final crack depth) at the life (number of cycles N), first, based on the initial size of the crack (including the initial crack depth a0 and the initial major semi-axis of the crack c0), the stress intensity factor variation range Δk of the crack is calculated by the stress intensity factor prediction model. The Δk value is substituted into the Paris formula to solve the crack extension Δa=da and Δc=dc in the first cycle. Subsequently, the GUI automatically updates the crack size to a=a0+Δa and c=c0+Δc, and uses the new crack size to calculate the stress intensity factor variation range Δk again. Substitute this new Δk into the Paris formula to obtain the crack extension Δa=da and Δc=dc in the second cycle. This process is iterated continuously until the crack extends to the final depth a N , the number of cycles N at this time is the total life of the crack.

[0165] The graphical user interface (GUI) was developed using the MATLAB platform. When users interact with the system through the GUI interface, they can input relevant physical parameters such as the initial position of the crack, the size of the cylinder (such as cylinder thickness, inner radius, etc.), and the geometric parameters of the crack (such as the ratio of the semi-long axis to the short axis of the crack) through simple and intuitive operations. The system will automatically process the input data and calculate the stress intensity factor through the constructed machine learning prediction model, and automatically calculate the crack expansion morphology in combination with the Paris formula. The system will display the crack expansion morphology of each step and the corresponding crack front stress intensity factor value, intuitively reflecting the crack expansion process and the changing trend of the stress intensity factor.

[0166] The stress intensity factor prediction model established in this application realizes the accurate prediction of the stress intensity factor of the semi-elliptical crack on the surface of the cylinder under complex loads. By setting the crack and cylinder size parameters, the stress intensity factor of the crack front can be quickly obtained without the need to establish and calculate the finite element simulation model, which significantly improves the calculation efficiency. At the same time, the machine learning data set is obtained by joint simulation calculation using the finite element software ABAQUS and the professional crack analysis software FRANC3D, which provides a strong guarantee for the reliability and accuracy of the model. The crack stress intensity factor prediction model is combined with the crack propagation law formula Paris formula to achieve accurate prediction of the crack propagation morphology on the cylinder surface, and the stress intensity factor after crack propagation is automatically calculated by the model to realize automatic crack propagation. The graphical user interface formed by MATLAB makes the application of the model more intuitive and convenient. The user only needs to input the relevant parameters to complete the stress intensity factor prediction and crack propagation morphology analysis.

[0167] Finally, by using the graphical user interface of MATLAB, users can easily complete key tasks such as adjusting the hyperparameters of the neural network, setting the surface crack size, calculating the stress intensity factor, inputting the parameters of the Paris formula, and predicting the crack propagation morphology through a graphical operation interface. This GUI design greatly reduces the usage threshold and learning cost, enabling non-professional users to conveniently use the system for predicting crack propagation morphology and life assessment. In addition, the GUI also provides some common model evaluation metrics, including the mean bias error (MBE), coefficient of determination (R 2 ), MAE (mean absolute error), and MAPE (mean absolute percentage error). These metrics help users intuitively judge the prediction accuracy and effect of the model, thereby effectively evaluating the reliability of the prediction results. To further enhance the user experience, the interface graphically displays the geometric shape and location of the surface cracks on the cylinder, helping users clearly understand the crack distribution and its propagation morphology.

[0168] Specifically, please refer to Appendix Figure 7 , Appendix Figure 7 is the schematic diagram of the graphical user interface in this embodiment. At 7a of Appendix Figure 7 , the predicted result of the stress intensity factor at the crack front is displayed. At 7b of Appendix Figure 7 , the performance of the prediction index evaluation model is displayed. At 7c of Appendix Figure 7 , the hyperparameters of the neural network are set. At 7d of Appendix Figure 7 , the prediction of the stress intensity factor of the axial semi-elliptical crack on the cylinder surface is made. At 7e of Appendix Figure 7 , the prediction of the stress intensity factor of the circumferential semi-elliptical crack on the cylinder surface is made. At 7f of Appendix Figure 7 , the image of the cylinder with a semi-elliptical crack is displayed. At 7g of Appendix Figure 7 , the parameters of the Paris formula are input and the fatigue life of the surface crack on the cylinder is predicted. At 7h of Appendix Figure 7 , the stress intensity factor of the internal surface axial semi-elliptical crack is solved by the empirical formula Newman-Raju. At 7i of Appendix Figure 7 , the stress intensity factor of the internal surface axial semi-elliptical crack is solved by the empirical formula Newman-Raju. At 7j of Appendix Figure 7 , the stress intensity factor of the external surface circumferential semi-elliptical crack is solved by the empirical formula Newman-Raju. At 7k of Appendix Figure 7 , the stress intensity factor of the external surface circumferential semi-elliptical crack is calculated. At 7l of Appendix Figure 7 , the propagation morphology of the surface semi-elliptical crack is displayed.

[0169] The operation process of the MATLAB graphical user interface of the present invention is as follows:

[0170] When training a new model, users need to set hyperparameters in the graphical user interface. Please refer to the attached Figure 7 The settings of the neural network hyperparameters at 7c include hidden layer neuron settings, number of training times, learning rate, convergence error, training set ratio, activation function, number of population iterations, population size, particle velocity boundary and position boundary, crossover probability and mutation probability.

[0171] After the settings are completed, please refer to the attached Figure 7 The prediction indicators at 7b are displayed to evaluate the model performance and start training. After the training is completed, the root mean square error (RMSE), correlation coefficient (R 2 ) and other evaluation indicators to observe the performance of the model. The neural network model with the required fitting accuracy is saved for direct access.

[0172] Predict the crack growth morphology on the cylinder surface.

[0173] The stress intensity factor prediction model for the axial semi-elliptical crack on the outer surface of the cylinder and the stress intensity factor prediction model for the axial semi-elliptical crack on the inner surface are saved as net1.mat and net2.mat, and the stress intensity factor prediction model for the circumferential semi-elliptical crack on the outer surface of the cylinder and the stress intensity factor prediction model for the circumferential semi-elliptical crack on the inner surface are saved as net3.mat and net4.mat. The user only needs to enter the ratio of the cylinder thickness to the inner radius t / R i , the ratio of the crack semi-minor axis to the semi-major axis a / c, the ratio of the crack semi-minor axis to the cylinder thickness a / t, the crack semi-minor axis a, the normalized distance to the crack front and the load conditions, please refer to the attached Figure 7 The stress intensity factor of the axial semi-elliptical crack on the cylinder surface at 7d can be predicted immediately, and the surface crack stress intensity factor can be predicted immediately.

[0174] Input the Paris formula parameters and combine them with the stress intensity factor to automatically predict the surface crack growth morphology and calculate the life span. Please refer to the attached Figure 7 7g and 7j.

[0175] The interface also displays the geometric morphology and location of the cylinder surface cracks in a graphical form, which allows users to intuitively understand the distribution characteristics of the cracks and their expansion trends. Please refer to the attached Figure 7 7f.

[0176] Embodiment 2

[0177] A device for predicting surface crack fatigue growth morphology and life based on machine learning, comprising:

[0178] The data set acquisition module 100 is used to receive a finite element simulation model with a surface crack established by a user, calculate the stress intensity factor of the surface crack front, verify the stress intensity factor solved by the finite element simulation model using the Newman-Raju empirical formula, and generate a finite element simulation data set;

[0179] A data preprocessing module 200 is used to perform dimensionless processing on the size of the surface crack, the size of the load-bearing structure, and the stress intensity factor of the surface crack front in the finite element simulation data set to generate a machine learning data set;

[0180] A machine learning model generation module 300 is used to optimize the back propagation neural network through a hybrid optimization algorithm to construct a stress intensity factor prediction model based on machine learning;

[0181] The crack growth morphology and life prediction module 400 is used to combine the surface crack stress intensity factor prediction model with the fatigue crack growth rate formula to cyclically predict the surface crack growth morphology of the load-bearing structure and calculate the life of the surface crack growth morphology of the load-bearing structure;

[0182] The graphical user interface output module 500 is used to establish a visual interactive interface for parameter input and output and image output based on the surface crack stress intensity factor prediction model.

[0183] In some embodiments, establishing a finite element simulation model containing surface cracks includes:

[0184] The load-bearing structure modeling module is used to establish a finite element model of the load-bearing structure through ABAQUS software and export the finite element model output file of the load-bearing structure;

[0185] The crack insertion modeling module is used to insert a semi-elliptical surface crack into the model of the load-bearing structure model output file through FRANC3D software, and output a finite element model of the load-bearing structure containing the surface crack.

[0186] In some embodiments, the crack growth morphology and life prediction module 400 includes:

[0187] A first prediction submodule 410 is configured to calculate a stress intensity factor variation range of the crack based on the initial size of the crack using a stress intensity factor prediction model;

[0188] The second prediction submodule 420 is used to substitute the value of the stress intensity factor variation range into the fatigue crack growth rate formula to solve the crack growth amount in this cycle;

[0189] The third prediction submodule 430 is used to update the crack size as the sum of the initial crack size and the crack extension in this cycle;

[0190] The cyclic expansion and life prediction module 440 is configured to use the updated crack size as the initial crack size, and cycle through the first prediction sub-module 410, the second prediction sub-module 420, the third prediction sub-module 430, and the cyclic expansion and life prediction module 440 until the crack size expands from the initial depth to the final depth, obtaining the value of the number of cycles, and the value of the number of cycles is the life of the surface crack propagation morphology.

[0191] In some embodiments, the fatigue crack growth rate formula disclosed by the above module is the Paris formula.

[0192] In some embodiments, the graphical user interface output module 500 includes:

[0193] A crack parameter input module is established;

[0194] A load parameter input module;

[0195] A neural network hyperparameter input module;

[0196] A load-bearing structure parameter input module for a structure with a surface crack;

[0197] An image output module of a load-bearing structure with a surface crack;

[0198] A surface crack propagation morphology parameter output module;

[0199] A surface crack stress intensity factor output module;

[0200] A fatigue life prediction result output module and an empirical formula comparison result output module.

[0201] The present invention utilizes finite element software ABAQUS and crack analysis software FRANC3D to build a finite element simulation model, establishes a surface crack stress intensity factor data set, establishes a surface crack stress intensity factor prediction model through a machine learning algorithm, and combines the stress intensity factor prediction model with the Paris formula to realize accurate prediction of crack extension morphology. The method is combined with finite element software ABAQUS and crack analysis software FRANC3D for joint simulation to generate a machine learning training data set, thereby effectively avoiding the shortcomings of existing formulas in accuracy and scope of application. It is not necessary to establish a finite element simulation model for calculation and analysis, and it is possible to accurately and quickly predict the surface crack stress intensity factor of different sizes and positions under complex loads. The Paris formula is widely used in fatigue crack extension prediction with its simplicity, wide applicability and reliability based on experimental data. The stress intensity factor prediction model can immediately realize the prediction and life calculation of surface crack extension morphology in combination with the Paris formula. Finally, the prediction model is combined with the Paris formula and implemented through a graphical user interface (GUI). The user does not need to perform mathematical analysis and write code to quickly predict the extension morphology and life calculation of surface cracks.

[0202] Embodiment 3

[0203] An electronic device comprising a memory and a processor, wherein:

[0204] Memory, used to store computer programs;

[0205] A processor is used to execute a computer program to implement the surface crack stress intensity factor prediction method based on machine learning disclosed in the aforementioned embodiment.

[0206] For the specific process of the above-mentioned surface crack stress intensity factor prediction method based on machine learning, reference can be made to the corresponding content disclosed in the aforementioned embodiments, which will not be repeated here.

[0207] Furthermore, the memory, as a carrier for storing resources, may be a read-only memory, a random access memory, a magnetic disk, or an optical disk, etc., and the storage method may be temporary storage or permanent storage.

[0208] Example 4

[0209] A computer-readable storage medium stores a computer program thereon, wherein the computer program is used to be executed by a processor to implement the surface crack stress intensity factor prediction method based on machine learning as disclosed in the aforementioned embodiment.

[0210] For the specific process of the above-mentioned surface crack fatigue extension morphology and life prediction method based on machine learning, please refer to the corresponding content disclosed in the aforementioned embodiments, which will not be repeated here.

[0211] Example 5

[0212] A computer program product includes a computer program / instruction, which, when executed by a processor, implements the surface crack fatigue growth morphology and life prediction method based on machine learning as disclosed in the above embodiments.

[0213] For the specific process of the above-mentioned surface crack fatigue extension morphology and life prediction method based on machine learning, please refer to the corresponding content disclosed in the aforementioned embodiments, which will not be repeated here.

[0214] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from the other embodiments. Reference can be made to the descriptions of the identical or similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple, and the relevant parts can be referred to the descriptions of the methods.

[0215] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0216] The above is a detailed introduction to the surface crack fatigue extension morphology and life prediction method, device, product, equipment and medium based on machine learning provided by this application. Specific examples are used in this article to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method of this application and its core idea; at the same time, for general technical personnel in this field, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on this application.

Claims

1. A method for predicting the morphology and life of surface crack fatigue propagation based on machine learning, characterized in that, It includes the following steps: S1. Establish a finite element simulation model with a surface crack, calculate the stress intensity factor at the front edge of the surface crack, and generate a finite element simulation data set; S2. Nondimensionalize the size of the surface crack, the size of the load-bearing structure, and the stress intensity factor at the front edge of the surface crack in the finite element simulation data set to generate a machine learning data set; S3. Optimize the backpropagation neural network through a hybrid optimization algorithm to construct a surface crack stress intensity factor prediction model; S4. The surface crack stress intensity factor prediction model combines with the fatigue crack growth rate formula to cyclically predict the surface crack growth morphology of the load-bearing structure and calculate the surface crack growth life of the load-bearing structure; S5. Based on the surface crack stress intensity factor prediction model and the fatigue crack growth rate formula, establish a visual interactive interface for parameter input and output and image output.

2. The surface crack fatigue growth morphology and life prediction method based on machine learning as claimed in claim 1, characterized in that: The cyclically predicting the surface crack growth morphology of the load-bearing structure and calculating the life of the surface crack growth morphology of the load-bearing structure includes the following steps: S41. Based on the initial size of the crack, calculate the stress intensity factor change range of the crack through the stress intensity factor prediction model; S42. Substitute the value of the stress intensity factor change range into the fatigue crack growth rate formula to solve the crack growth amount in this cycle; S43. Automatically update the crack size to the sum of the initial crack size and the crack growth amount in this cycle; S44. Use the updated crack size as the initial crack size, and cycle steps S41 to S44 until the crack size expands from the initial depth to the final depth to obtain the value of the number of cycles. The value of the number of cycles is the life of the surface crack growth morphology.

3. The method for predicting the surface crack fatigue propagation morphology and life based on machine learning according to claim 2, wherein The fatigue crack growth rate formula is the Paris formula.

4. The method for predicting the fatigue crack growth morphology life based on machine learning according to claim 1, characterized in that The establishment of the visual interactive interface for parameter input and output and image output is specifically: Establish a visual interactive interface for crack parameter input, load parameter input, neural network hyperparameter input, load-bearing structure parameter input, image output of the load-bearing structure with a surface crack, surface crack growth morphology parameter output, surface crack stress intensity factor output, fatigue life prediction result output, and empirical formula comparison result output.

5. The surface crack fatigue growth morphology and life prediction method based on machine learning as claimed in claim 1, characterized in that: The establishment of the finite element simulation model with a surface crack includes the following steps: S11. Establish a finite element model of the load-bearing structure through ABAQUS software and export the finite element model output file of the load-bearing structure; S12. Insert a semi-elliptical surface crack into the model of the load-bearing structure model output file and output the finite element model of the load-bearing structure with a surface crack.

6. The method for predicting the surface crack fatigue propagation morphology and life based on machine learning according to claim 5, characterized in that, The insertion of the semi-elliptical surface crack into the model of the load-bearing structure model output file includes the following steps: S121. Divide the finite element model of the load-bearing structure into an overall model and a local model; S122. Define the short semi-axis size and long semi-axis size of the semi-elliptical surface crack, and introduce the semi-elliptical surface crack into the local model to form a local model with a surface crack; S123. Mesh the local model with a surface crack. S124. Combine the overall model and the local model with surface cracks to form a finite element model of the load-bearing structure with surface cracks.

7. The method for predicting the surface crack fatigue propagation morphology and life based on machine learning according to claim 1, wherein The dimensionless processing of the size of the surface cracks, the size of the load-bearing structure, and the stress intensity factor at the front of the surface cracks in the finite element simulation dataset, and generating the finite element simulation dataset includes the following steps: S21. Perform dimensionless processing on the short semi-axis size of the crack, the long semi-axis size of the crack, the thickness of the load-bearing structure, the inner radius of the load-bearing structure, and the stress intensity factor, and use the normalized distance at the crack front to replace the parameter angle of the ellipse to form dimensionless quantities. The dimensionless quantities are specifically: the ratio of the short semi-axis to the long semi-axis size of the crack, the ratio of the short semi-axis of the crack to the thickness of the load-bearing structure, the ratio of the thickness of the load-bearing structure to the characteristic size of the load-bearing structure, the normalized distance at the crack front, and the surface crack boundary correction factor. S22. Generate an axial semi-elliptical crack dataset and a circumferential semi-elliptical crack dataset.

8. The method for predicting the surface crack fatigue propagation morphology and life based on machine learning according to claim 1, wherein The optimization of the backpropagation neural network by the hybrid optimization algorithm includes the following steps: S31. Optimize the weights and biases of the backpropagation neural network through the particle swarm algorithm to obtain the optimal combination of weights and biases of the backpropagation neural network. S32. Apply the crossover and mutation operations of the genetic algorithm to the particle swarm algorithm to enhance the population diversity and global search ability.

9. The method for predicting the surface crack fatigue propagation morphology and life based on machine learning according to claim 1, characterized in that, The step S1 is: S11a. Based on the position of the surface crack in the load-bearing structure, generate a finite element simulation model with surface cracks at different positions. Among them, the position of the surface crack in the load-bearing structure is specifically the axial direction of the outer surface of the load-bearing structure, the axial direction of the inner surface of the load-bearing structure, the circumferential direction of the outer surface of the load-bearing structure, and the circumferential direction of the inner surface of the load-bearing structure. Among them, the parameter range includes circular cracks to long elliptical cracks, shallow cracks to deep cracks, thin-walled load-bearing structures to thick-walled load-bearing structures. S12a. The finite element simulation model with surface cracks includes a load-bearing structure model with surface cracks under the action of internal pressure load and uniform tensile load at both ends.

10. The method for predicting the surface crack fatigue propagation morphology and life based on machine learning according to claim 1, wherein The step S3 also includes: S33. Adjust the neural network hyperparameters, verify the fitting effect of the stress intensity factor prediction model based on machine learning according to the evaluation index, and save the neural network that meets the fitting accuracy requirements for subsequent calls during visualization interaction; among them, the evaluation criteria include at least one of the coefficient of determination, mean absolute error, mean bias error, and mean absolute percentage error.

Images