Physical information neural network steel fatigue life prediction method based on genetic algorithm

By introducing a genetic algorithm-optimized physical information neural network in multi-axis fatigue life prediction, combined with the physical information item weight of the critical surface model, the problems of low interpretability and insufficient accuracy in the existing technology are solved, and higher predictions are achieved Accuracy and interpretability.

CN119069051APending Publication Date: 2024-12-03TIANJIN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411284906.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-13
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

The prior art has problems of low interpretability and insufficient accuracy in multi-axis fatigue life prediction, especially due to the ‘black box’ characteristics of machine learning models and the scarcity of experimental data, it is difficult to achieve accurate predictions under small samples.

Method used

The physical information neural network method based on genetic algorithm is adopted, and by introducing physical information term weights and loss functions, combined with the critical surface model, the parameters of the neural network are optimized to improve prediction accuracy and interpretability.

Benefits of technology

It realizes improving prediction accuracy and interpretability in multi-axis fatigue life prediction, ensuring guidance on physical meaning, and improving the generalization ability of the model and small sample prediction ability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119069051A_ABST
    Figure CN119069051A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of machine learning, and discloses a physical information neural network steel fatigue life prediction method based on a genetic algorithm, which constructs a physical information loss function with a physical information item weight by selecting a critical plane model suitable for to-be-predicted steel, thereby constructing a physical information neural network. Through genetic algorithm optimization, physical information item weights are taken as target parameters, a fitness function related to prediction performance of a physical information neural network is defined for genetic algorithm optimization, and for a specific steel material, an optimal individual is obtained through optimization by balancing characteristics and prediction performance of different critical plane models on the steel material. The physical information item weights are correspondingly represented as a group of optimal physical information item weights, so that the fitness function is minimum. Under the weight, the physical information neural network has the best performance in the aspect of fatigue life prediction on the specific steel, and the physical information neural network is used for predicting the fatigue life of the steel.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of machine learning, and in particular relates to a method for predicting fatigue life of steel. Background Art

[0002] Fatigue-induced failure is one of the most common types of failure for metal parts under cyclic loading. Compared with the simpler uniaxial fatigue life prediction, multiaxial fatigue prediction is more complicated due to the influence of non-proportional additional cyclic hardening effect on metal structural parts. In order to ensure the reliability and safe service of metal structural parts, multiaxial fatigue life prediction of metal materials has become a major research hotspot. Fatigue testing is a common method for studying the fatigue characteristics of materials, but it often takes a lot of time and cost, and due to the diversity of multiaxial paths and loading conditions, relying on fatigue experiments to obtain the fatigue life of metal materials in complex situations has limitations.

[0003] Over the years, a large number of empirical methods have been used to construct various models based on strain, stress, energy, and fracture mechanism using the damage parameters of materials. Among them, the critical surface method, as a representative model of the energy method, has outstanding prediction performance in empirical models due to its comprehensive consideration of metal materials in many aspects such as stress, strain, energy, and fracture mechanism. However, for a specific critical surface model, its applicable metal materials often have limitations; and although the prediction accuracy is higher than that of other empirical methods, the predicted life distribution can only be concentrated within a five-fold error band.

[0004] In recent years, with the advancement of computer computing power and the maturity of multiaxial fatigue experimental technology, a certain amount of multiaxial fatigue data sets for specific metal materials have gradually accumulated. Through data-driven algorithms such as machine learning, artificial neural networks and other models are trained using data sets, and methods for multiaxial fatigue life prediction have received a lot of attention and development. However, due to the black box characteristics of pure data-driven algorithms such as machine learning, the life prediction methods based on multiaxial fatigue lack sufficient interpretability and physical background, which has become a major obstacle to their widespread application. Summary of the invention

[0005] The purpose of the present invention is to provide a physical information neural network steel fatigue life prediction method based on genetic algorithm. Based on the critical surface method and its damage parameters, a physical information neural network and a physical information loss function are constructed to introduce physical meaning and background into the data-driven method, increase the interpretability, and ensure that the prediction ability of the prediction method is further improved.

[0006] In order to achieve the above-mentioned object of the invention, the present invention is implemented by the following technical solutions:

[0007] According to one aspect of the present invention, a method for predicting steel fatigue life based on a physical information neural network using a genetic algorithm is provided, wherein the method comprises:

[0008] S1. Collect the fatigue test data set of the steel to be predicted, and divide the data of the fatigue test data set into a training set and a validation set;

[0009] S2. After selecting the critical surface model according to the steel to be predicted, the parameters required for the critical surface model are calculated according to the fatigue test data set, and the damage parameters are calculated;

[0010] S3, determining the input features and output features of the physical information neural network, and performing data preprocessing on the input features and the output features; wherein the input features use the damage parameters obtained in S2, and the output features are fatigue life prediction results;

[0011] S4, build a physical information neural network based on the artificial neural network as the basic model, define the physical information loss function according to the critical surface model selected in S2, and introduce the weight of the physical information item;

[0012] S5. Set the basic parameters and fitness function of the genetic algorithm, use the weight of the physical information item as the target parameter of the genetic algorithm, and initialize the population information of the genetic algorithm and the individuals in the first generation;

[0013] S6. Setting the corresponding number of cycles of each training of the physical information neural network to ensure the convergence of the results, performing a physical information neural network training for each set of physical information item weights, and calculating the fitness function value after training; each set of physical information item weights is derived from an individual in each generation of the genetic algorithm;

[0014] S7, according to the fitness function value obtained in S6, select the individual with the smallest fitness function value in the current generation;

[0015] S8, cross the individual selected in S7 with other individuals to obtain a new individual;

[0016] S9, mutate the new individual obtained in S8 to obtain the next generation individual, and increase the generation number by 1;

[0017] S10, repeat S6-S9 until the current generation number exceeds the set generation number;

[0018] S11. Select the individual with the smallest fitness function value in all generations as the best individual, use the physical information item weight obtained from the best individual as the best physical information item weight, and obtain the fatigue life prediction result through the physical information neural network.

[0019] Further, in S2, for steel whose fatigue fracture type is shear type, the critical surface model adopts at least two of the FS model, MGS Yu model, MGSE Zhu model, WB model, Liu model, CXH model, Zhu model or GSE model; for steel whose fatigue fracture type is tensile type, the critical surface model adopts at least two of the SWT model, Liu model, CXH model, Zhu model and GSE model.

[0020] Furthermore, in S2, for carbon steel, the critical surface model adopts FS model, SWT model, MGSEZhu model, WB model and CXH(T) model:

[0021] FS Model:

[0022] SWT Model:

[0023] MGSEZhu Model:

[0024] WB Model:

[0025] CXH(T) Model:

[0026] Among them, σ nmax , τ nmax , Δε nmax , Δγ nmax They are the maximum normal stress, maximum shear stress, maximum normal strain range, and maximum normal stress range corresponding to the critical surface when the plane where the maximum shear strain range is located is taken as the critical surface; σ smax , τ smax , Δε smax , Δγ smax They are the maximum normal stress, maximum shear stress, maximum normal strain range, and maximum normal stress range corresponding to the critical surface when the plane where the maximum normal strain range is located is taken as the critical surface; Δε n is the positive strain stroke; σ' f is the fatigue strength coefficient, τ' f is the shear fatigue strength coefficient, ε' f is the fatigue ductility coefficient, γ' f is the shear fatigue ductility coefficient, b is the fatigue strength index, c is the fatigue ductility index, b0 is the shear fatigue strength index, c0 is the shear fatigue ductility index, k is the FS model constant, k MGSE is the MGSEZhu model constant, S is the WB model constant; E is Young's modulus, G is the shear modulus, σ y is the yield strength, N f is fatigue life; ve is the Poisson's ratio in the elastic stage, which is taken as 0.3; v p is the Poisson's ratio in the plastic stage, which is taken as 0.5;

[0027] Among them, the critical surface model requires parameters σ' f , τ' f , ε' f , γ' f ,b,c,b0,c0,k,k MGSE and S;

[0028] Wherein, the damage parameter is: nmax , τ nmax , Δε nmax , Δγ nmax , σ smax , τ smax , Δε smax , Δγ smax and Δε n .

[0029] Furthermore, in S3, the input feature of the physical information neural network is σ nmax , τ nmax , Δε nmax , Δγ nmax , σ smax , τ smax , Δε smax , Δγ smax .

[0030] Furthermore, in S3, the data preprocessing of the input features and the output features includes: normalizing the input features and the output features after logarithmic processing:

[0031]

[0032] Among them, x refers to each input feature or logarithmic output feature, x * represents the normalized value, μ represents the mean of each input feature or logarithmic output feature, and σ represents the standard deviation of each input feature or logarithmic output feature.

[0033] Further, in S4, the physical information loss function is:

[0034] L PINN =MSE+ω1L FS +ω2L SWT +ω3L MGSEZhu +ω4L WB +ω5L CXH(T)

[0035] in:

[0036]

[0037] ω1, ω2, ω3, ω4 and ω5 are the weights of physical information items, i is the sample number, n is the number of samples, N exp,i is the experimental life of the i-th sample, N pre,i is the predicted life span of the sample with sample number i;

[0038] Among them, σ nmax , τ nmax , Δε nmax , Δγ nmax They are the maximum normal stress, maximum shear stress, maximum normal strain range, and maximum normal stress range corresponding to the critical surface when the plane where the maximum shear strain range is located is taken as the critical surface; σ smax , τ smax , Δε smax , Δγ smax They are the maximum normal stress, maximum shear stress, maximum normal strain range, and maximum normal stress range corresponding to the critical surface when the plane where the maximum normal strain range is located is taken as the critical surface; Δε n is the positive strain stroke; σ' f is the fatigue strength coefficient, τ' f is the shear fatigue strength coefficient, ε' f is the fatigue ductility coefficient, γ' f is the shear fatigue ductility coefficient, b is the fatigue strength index, c is the fatigue ductility index, b0 is the shear fatigue strength index, c0 is the shear fatigue ductility index, k is the FS model constant, k MGSE is the MGSEZhu model constant, S is the WB model constant; E is Young's modulus, G is the shear modulus, σ y is the yield strength, N f is fatigue life; v e is the Poisson's ratio in the elastic stage, which is taken as 0.3; v p is the Poisson's ratio in the plastic stage and is taken as 0.5.

[0039] Furthermore, in S5, the basic parameters of the genetic algorithm are the number of generations and the number of individuals in each generation; the fitness function of the genetic algorithm is:

[0040]

[0041] Among them, N exp,i is the experimental life of the i-th sample, N pre,i is the predicted life span of the i-th sample, and n is the number of samples. Further, in S5, an individual in each generation of the genetic algorithm is represented as:

[0042] I pg =(ω1pg ,ω 2pg ,ω 3pg ,ω 4pg ,ω 5pg )

[0043] Among them, p represents the individual serial number, and g represents the generation number;

[0044] Taking the weight contained in the individual as the weight of the physical information item, then:

[0045] ω apg =ω a (a=1,2..5)

[0046] Where a is the serial number of the physical information item weight.

[0047] According to another aspect of the present invention, there is provided a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned physical information neural network steel fatigue life prediction method based on genetic algorithm.

[0048] According to another aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-mentioned physical information neural network steel fatigue life prediction method based on genetic algorithm.

[0049] The beneficial effects of the present invention are:

[0050] (I) The physical information neural network steel fatigue life prediction method based on genetic algorithm of the present invention selects a critical surface model suitable for steel, constructs a physical information loss function with physical information item weights, and thus constructs a physical information neural network. Through genetic algorithm optimization, the physical information item weights are used as target parameters, and the fitness function related to the prediction performance of the physical information neural network is defined for genetic algorithm optimization. For a specific steel, by balancing the characteristics and prediction performance of different critical surface models on the steel, an optimal individual is optimized, which is represented by a set of optimal physical information item weights, so that the fitness function is minimized. Under this weight, the physical information neural network has the best performance in fatigue life prediction on this specific steel, and the physical information neural network is used to predict the fatigue life of the steel.

[0051] (ii) The physical information neural network steel fatigue life prediction method based on genetic algorithm of the present invention, wherein the genetic algorithm, as a traditional numerical optimization method, can simulate the evolution of biological populations, and use individuals carrying optimization parameters as a medium. Under the guidance of the fitness function, through the algorithm of population evolution, individuals with poor performance on the fitness function are gradually eliminated, and individuals with good performance are retained, so as to efficiently search in the parameter search space, and can effectively adjust the hyperparameters related to the physical information neural network to improve the performance of the physical information neural network.

[0052] (III) The physical information neural network steel fatigue life prediction method based on genetic algorithm of the present invention, wherein the physical information neural network loss function has good generalization for multi-axis fatigue life prediction of different types of steels due to the introduction of weights of physical information items.

[0053] (IV) The method for predicting steel fatigue life based on a physical information neural network using a genetic algorithm of the present invention applies the physical information neural network optimized based on a genetic algorithm to the field of multi-axial fatigue life prediction of steel, and by introducing the physical information term in the loss function, uses the empirical model in the field as a physical guide for model training, thereby effectively improving the accuracy of the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0055] Figure 1 A schematic diagram of the framework of the physical information neural network steel fatigue life prediction method based on genetic algorithm provided by the present invention;

[0056] Figure 2 A schematic diagram of the genetic algorithm optimization process in the method provided by the present invention;

[0057] Figure 3 The genetic algorithm optimization-physical neural network ablation experimental performance in the method provided by the present invention;

[0058] Figure 4 This is a comparison between the predicted value and experimental value of multi-axial fatigue life of the physical information neural network steel fatigue life prediction method in three metal material data sets: 316LN, 316L and 304. DETAILED DESCRIPTION

[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] In order to solve the problems in the background technology, the present invention provides the following solution ideas in predicting the multiaxial fatigue life of steel with both accuracy and interpretability:

[0061] Traditional empirical models have good physical meaning, but due to experimental deviations and the theoretical nature of empirical models, the accuracy of multi-axial fatigue life prediction of steel is average. Data-driven methods have good accuracy in multi-axial fatigue life prediction of steel, but due to the "black box" characteristics of data-driven models such as machine learning, the physical meaning and interpretability of such prediction models are low. At the same time, since fatigue data of complex paths are relatively scarce, improving the accuracy of fatigue life prediction under small samples has also become a direction for the development of data-driven methods.

[0062] Among them, fatigue life refers to: the material performance expressed by the number of stress or strain cycles loaded when the material is subjected to fatigue test under uniaxial load or multiaxial load.

[0063] In order to propose a method that has both interpretability and accuracy under small samples, the present invention constructs a loss function with the critical surface model in the traditional empirical method as the source of physical information, and constructs a physical information neural network on this basis. The weights of each sub-item in the physical information item of the loss function of the physical information neural network can be adjusted according to the different materials to adapt to the multi-axial fatigue characteristics of different materials; wherein, the weight adjustment is performed through genetic algorithm optimization. In the genetic algorithm optimization, through the iterative process of initializing the population size, constructing the first generation, calculating the fitness function, selection, crossover, mutation and the next generation, the best individual with the smallest fitness function is optimized as the genetic algorithm optimization result, which is used for the weight of the physical information neural network on this material, and the physical information network is used to complete the multi-axial fatigue life prediction task of steel.

[0064] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0065] Example 1

[0066] like Figure 1 As shown, this embodiment provides a method for predicting multi-axial fatigue life of steel based on a physical neural network optimized by a genetic algorithm, which specifically includes:

[0067] S1. Collect the fatigue test data set of the steel to be predicted, and divide the data of the fatigue test data set into a training set and a validation set.

[0068] The fatigue experimental data set of the steel to be predicted can be obtained from published literature or fatigue experiments.

[0069] The fatigue test data set specifically includes multi-axial loading hysteresis loops, real-time normal stress, normal strain, shear stress, shear strain signal values, and corresponding life cycles.

[0070] Steel refers to carbon steel and alloy steel, etc.; fatigue test data is obtained based on experiments on steel specimens of certain shape, size and performance made by pressure processing in the form of ingots, billets, etc.

[0071] In this embodiment, 80% of the data is used for the training set and 20% is used for the validation set.

[0072] S2. Select the critical surface model according to the type of steel to be predicted, and then calculate the parameters required for the critical surface model according to the fatigue test data set, and calculate the damage parameters.

[0073] Generally, for steels with shear fatigue fracture type, the selectable critical surface models include FS model, MGS Yu model, MGSE Zhu model, WB model, Liu model, CXH model, Zhu model, and GSE model; for steels with tensile fatigue fracture type, the selectable critical surface models include SWT model, Liu model, CXH model, Zhu model, and GSE model; for each type, the critical surface model can be a combination of any two or more of the above models.

[0074] In this embodiment, for 316LN, 316L, and 304 steels, the selected critical surface model is:

[0075] FS Model:

[0076] SWT Model:

[0077] MGSEZhu Model:

[0078] WB Model:

[0079] CXH(T) Model:

[0080] Among them, σ nmax , τ nmax , Δε nmax , Δγ nmaxThe plane with the maximum shear strain range is taken as the critical surface, the maximum normal stress, maximum shear stress, maximum normal strain range, maximum normal stress range corresponding to the critical surface, σ smax , τ smax , Δε smax , Δγ smax The plane with the maximum normal strain range is taken as the critical surface, the maximum normal stress, maximum shear stress, maximum normal strain range, maximum normal stress range corresponding to the critical surface, Δε n is the positive strain stroke; σ' f is the fatigue strength coefficient, τ' f is the shear fatigue strength coefficient, ε' f is the fatigue ductility coefficient, γ' f is the shear fatigue ductility coefficient, b is the fatigue strength index, c is the fatigue ductility index, b0 is the shear fatigue strength index, c0 is the shear fatigue ductility index, k is the FS model constant, k MGSE is the MGSEZhu model constant, S is the WB model constant; E is Young's modulus, G is the shear modulus, σ y is the yield strength, N f is the fatigue life.

[0081] In addition, the above five critical surface models can also be applied to other carbon steels except 316LN, 316L, and 304 steels.

[0082] The fatigue parameter τ' on the right side of the above critical surface model equation is f , γ' f , b0 and c0 are obtained by fitting the torsion data; σ' f , ε' f , b and c are obtained by fitting the uniaxial data. Due to the lack of torsion data for 316L, the Mises equivalent stress criterion is used to obtain:

[0083]

[0084] b o =b

[0085] c o =c

[0086] E, G, σ y Please check the data sheet of the corresponding steel material to obtain it.

[0087] The table lists the parameters required for the above critical surface model for 316LN, 316L, and 304 stainless steel:

[0088] Table 1 316LN, 316L, 304 stainless steel material parameters

[0089]

[0090] And the corresponding damage parameters are calculated according to the following formula:

[0091]

[0092] Among them, θ is the angle between the calculation plane and the axis of the steel sample, ε θ is the stress value on the calculation plane, ε x is the axial normal strain value, γ θ is the calculated plane shear strain value, τ xy is the axial shear stress value, σ θ is the calculated plane normal stress value, σ x is the axial normal stress value.

[0093] Thus, the normal strain and shear strain range are obtained:

[0094] Δε θ =ε θmax -ε θmin

[0095] Δγ θ =γ θmax -γ θmin

[0096] And normal stress, shear stress range:

[0097] Δσ θ =σ θmax -σ θmin

[0098] Δτ θ =τ θmax -τ θmin

[0099] According to the critical surfaces defined by different critical surface methods, the damage function σ is obtained. nmax , τ nmax , Δε nmax , Δγ nmax , σ smax , τ smax , Δε smax , Δγ smax and Δε n .

[0100] S3. Determine the input features of the physical information neural network, and perform data preprocessing on the input features and output features of the physical information neural network.

[0101] Among them, the input feature uses the damage parameter obtained by S2, and the output feature is the fatigue life prediction result.

[0102] The input features selected in this embodiment are: nmax, τ nmax , Δε nmax , Δγ nmax , σ smax , τ smax , Δε smax , Δγ smax .

[0103] The data preprocessing of input features and output features is specifically to normalize the input features and the output features after logarithmic processing:

[0104]

[0105] Among them, x refers to each input feature and the logarithmic output feature, x * represents the normalized value, μ represents the mean of each input feature or logarithmic output feature, and σ represents the standard deviation of each input feature or logarithmic output feature.

[0106] S4. Construct a physical information neural network based on the artificial neural network as the basic model, define the physical information loss function according to the critical surface model selected in S2, and introduce the weight of the physical information item.

[0107] σ nmax , τ nmax , Δε nmax , Δγ nmax , σ smax , τ smax , Δε smax and Δγ smax As 8 input features, an artificial neural network is constructed. Among them, an artificial neural network refers to a mathematical model in the field of machine learning and cognitive science that imitates the structure and function of a biological neural network (the central nervous system of animals, especially the brain).

[0108] The structure of the artificial neural network in this embodiment is 5 hidden layers, and the number of neurons in each layer is 16, 16, 8, 8, and 4. The output is the predicted life span after logarithmic normalization.

[0109] On this basis, the loss function of physical information is defined, thereby constructing a physical information neural network, which is expressed as:

[0110] L PINN =MSE+ω1L FS +ω2L SWT +ω3L MGSEZhu +ω4L WB +ω5L CXH(T)

[0111] in:

[0112]

[0113] ω1, ω2, ω3, ω4 and ω5 are the physical information weights introduced by the above physical information loss function, i is the sample number, n is the number of samples, N exp,i is the experimental life of the i-th sample, N pre,i is the predicted life span of the ith sample.

[0114] S5. Set the basic optimization parameters and fitness function of the genetic algorithm, use the corresponding weights of the physical information items in the physical information neural network as the target parameters of the genetic optimization algorithm, and initialize the population information and the individuals of the first generation of the genetic optimization algorithm.

[0115] In this embodiment, the number of generations is set to 50, the number of individuals in each generation is set to 30, and the population information and the individuals of the first generation are randomly initialized.

[0116] The fitness function is defined as:

[0117]

[0118] Among them, N exp,i is the experimental life of the ith sample, N pre,i is the predicted life span of the ith sample, and n is the number of samples.

[0119] The corresponding physical information item weights ω1, ω2, ω3, ω4 and ω5 in the physical information neural network are taken as variables and optimized according to the fitness function in the genetic algorithm.

[0120] Initialize the individuals in the first generation randomly.

[0121] S6. Set the corresponding number of cycles for each training of the physical information neural network to ensure the convergence of the results. A set of physical information item weights comes from an individual in each generation of the genetic algorithm. Each set of physical information item weights is trained once for the physical information neural network, and the fitness function value after training is calculated.

[0122] During training, the training set is used as the training sample, and the prediction set is used to evaluate the model performance.

[0123] According to trial calculations, the final number of cycles was determined to be 1500, at which point the training of the physical information neural networks for the three materials can converge.

[0124] Each individual in the genetic algorithm consists of a list of five weights:

[0125] I pg =(ω 1pg ,ω 2pg ,ω 3pg ,ω 4pg ,ω5pg )

[0126] Among them, p represents the number of individuals, which is 1, 2...30; g represents the number of generations, which is 1, 2...50.

[0127] Taking the weight contained in the individual as the weight of the physical information item, then:

[0128] ω apg =ω a (a=1,2..5)

[0129] Among them, a is the serial number of the weight of the physical information item, p represents the number of individuals, taking 1, 2...30, and g represents the number of generations, taking 1, 2...50.

[0130] The set of weights corresponding to the above individuals are used as the weights of the physical information items introduced in S4 in turn. The physical information neural network corresponding to each set of weights is trained once, and the fitness function after training is calculated:

[0131]

[0132] Among them, N exp,i is the experimental life of the i-th sample, N pre,i is the predicted life span of the ith sample, and n is the number of samples.

[0133] S7. According to the fitness function value obtained in S6, select the individual with the smallest fitness function value in the current generation.

[0134] The selection method may be tournament selection method, roulette selection, ranking selection, robust selection, elite selection, random selection, elimination selection method, and the like.

[0135] In this embodiment, three individuals are randomly selected using the tournament selection method, and the individual with the smallest fitness function value is selected from them. The selection formula is:

[0136]

[0137] Where T is a randomly selected set of individuals of size 3.

[0138] S8. Cross the individuals selected in S7 with other individuals to obtain new individuals.

[0139] This embodiment adopts a hybrid crossover method.

[0140] The interval of the new individual is:

[0141] X p =[(1+α)·P1 gene -α·P2 gene -α·P1 gene+(1+α)·P2 gene ]

[0142] Where X p is the new gene of the new individual, P1 gene is the gene of parent individual 1, P2 gene is the gene of parent individual 2, α is the hybrid crossover interval parameter (0-1), set to 0.5.

[0143] S9: Make the new individual obtained in S8 mutate, finally get the individual in the next generation, and increase the number of generations by 1.

[0144] Individual mutation uses Gaussian mutation:

[0145] M p =k p +N(μ,σ)

[0146] Among them, M p is the new individual after mutation, k p is the individual before mutation, N(μ,σ) represents a normally distributed random number with mean μ and standard deviation σ. Assume that the parameters are μ=0.5, σ=0.15, and the mutation probability indpb=0.2 (20% probability of this mutation).

[0147] S10, repeat S6-S9 until the current generation number exceeds the set generation number;

[0148] S11. Select the individual with the minimum fitness function in all generations as the best individual, use the physical information item weight obtained from the best individual as the best physical information item weight, and obtain the fatigue life prediction result through the current physical information neural network.

[0149] For 316LN, the best individual appears at g=5, p=8, with corresponding weights of 0.135, 0.168, 0.355, 0.097, and 0.433; for 316L, the best individual appears at g=36, p=17, with corresponding weights of 0.488, 0.296, 0.676, 0.549, and 0.613; for 304, the best individual appears at g=1, p=6, with corresponding weights of 0.821, 0.056, 0.526, 0.118, and 0.498. Finally, the corresponding weights are used as the weights of physical information items and used in physical information neural networks to predict the fatigue life of the corresponding steels.

[0150] Figure 2 The specific process of genetic algorithm optimization in the physical information neural network steel fatigue life prediction method is demonstrated.

[0151] Figure 3The ablation experiment performance of the method proposed in this invention on three metal materials: 316LN, 316L and 304. Figure 3 It can be seen that the genetic optimization algorithm-physical information neural network performs better than the artificial neural network and the physical information neural network with unoptimized weights (weight is 1) in repeated experiments on multiaxial fatigue data sets of three metal materials.

[0152] Figure 4 The comparison between the predicted value and experimental value of multiaxial fatigue life of three metal material data sets of 316LN, 316L and 304 by the method proposed in this invention. Figure 4 It can be seen that the prediction values ​​of the genetic algorithm optimization-physical information neural network on both the test set and the training set are within the 2-fold error band, and most of them are concentrated within the 1.5-fold error band.

[0153] Example 2

[0154] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the physical information neural network steel fatigue life prediction method based on genetic algorithm in Example 1.

[0155] Example 3

[0156] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the physical information neural network steel fatigue life prediction method based on genetic algorithm in Example 1.

[0157] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0158] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above description of the embodiments is only used to help understand the method and core ideas of the present invention; at the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A physical information neural network steel fatigue life prediction method based on genetic algorithm, It is characterized in that The method includes: S1. Collect the fatigue test data set of the steel to be predicted, and divide the data of the fatigue test data set into a training set and a validation set; S2. After selecting the critical surface model according to the steel to be predicted, the parameters required for the critical surface model are calculated according to the fatigue test data set, and the damage parameters are calculated; S3, determining the input features and output features of the physical information neural network, and performing data preprocessing on the input features and the output features; wherein the input features use the damage parameters obtained in S2, and the output features are fatigue life prediction results; S4, build a physical information neural network based on the artificial neural network as the basic model, define the physical information loss function according to the critical surface model selected in S2, and introduce the weight of the physical information item; S5. Set the basic parameters and fitness function of the genetic algorithm, use the weight of the physical information item as the target parameter of the genetic algorithm, and initialize the population information of the genetic algorithm and the individuals in the first generation; S6. Setting the corresponding number of cycles of each training of the physical information neural network to ensure the convergence of the results, performing a physical information neural network training for each set of physical information item weights, and calculating the fitness function value after training; each set of physical information item weights is derived from an individual in each generation of the genetic algorithm; S7, according to the fitness function value obtained in S6, select the individual with the smallest fitness function value in the current generation; S8, cross the individual selected in S7 with other individuals to obtain a new individual; S9, mutate the new individual obtained in S8 to obtain the next generation individual, and increase the generation number by 1; S10, repeat S6-S9 until the current generation number exceeds the set generation number; S11. Select the individual with the smallest fitness function value in all generations as the best individual, use the physical information item weight obtained from the best individual as the best physical information item weight, and obtain the fatigue life prediction result through the physical information neural network.

2. According to claim 1, a physical information neural network steel fatigue life prediction method based on genetic algorithm, It is characterized in that In S2, for steels whose fatigue fracture type is shear type, the critical surface model adopts at least two of the FS model, MGS Yu model, MGSE Zhu model, WB model, Liu model, CXH model, Zhu model or GSE model; for steels whose fatigue fracture type is tensile type, the critical surface model adopts at least two of the SWT model, Liu model, CXH model, Zhu model and GSE model.

3. According to claim 2, a physical information neural network steel fatigue life prediction method based on genetic algorithm, It is characterized in that In S2, for carbon steel, the critical surface model adopts FS model, SWT model, MGSEZhu model, WB model and CXH(T) model: FS Model: SWT Model: MGSEZhu Model: WB Model: CXH(T) Model: Among them, σ nmax , τ nmax , Δε nmax , Δγ nmax They are the maximum normal stress, maximum shear stress, maximum normal strain range, and maximum normal stress range corresponding to the critical surface when the plane where the maximum shear strain range is located is taken as the critical surface; σ smax , τ smax , Δε smax , Δγ smax They are the maximum normal stress, maximum shear stress, maximum normal strain range, and maximum normal stress range corresponding to the critical surface when the plane where the maximum normal strain range is located is taken as the critical surface; Δε n is the positive strain stroke; σ' f is the fatigue strength coefficient, τ' f is the shear fatigue strength coefficient, ε' f is the fatigue ductility coefficient, γ' f is the shear fatigue ductility coefficient, b is the fatigue strength index, c is the fatigue ductility index, b 0 is the shear fatigue strength index, c 0 is the shear fatigue ductility index, k is the FS model constant, k MGSE is the MGSEZhu model constant, S is the WB model constant; E is Young's modulus, G is the shear modulus, σ y is the yield strength, N f is fatigue life; v e is the Poisson's ratio in the elastic stage, which is taken as 0.3; v p is the Poisson's ratio in the plastic stage, which is taken as 0.5; Among them, the critical surface model requires parameters σ' f , τ' f , ε' f , γ' f , b, c, b 0 , c 0 , k, k MGSE and S; Among them, the damage parameters are: p nmax ,t nmax ,No nmax ,Dg nmax ,s smax ,t smax ,No smax ,Dg smax No n 。 4. According to claim 1, a physical information neural network steel fatigue life prediction method based on genetic algorithm, It is characterized in that In S3, the input feature of the physical information neural network is σ nmax , τ nmax , Δε nmax , Δγ nmax , σ smax , τ smax , Δε smax , Δγ smax .

5. According to claim 1, a physical information neural network steel fatigue life prediction method based on genetic algorithm, It is characterized in that In S3, data preprocessing of input features and output features includes: normalizing the input features and the output features after logarithmic processing: Among them, x refers to each input feature or logarithmic output feature, x * represents the normalized value, μ represents the mean of each input feature or logarithmic output feature, and σ represents the standard deviation of each input feature or logarithmic output feature.

6. The method for predicting steel fatigue life based on a physical information neural network using a genetic algorithm according to claim 1, It is characterized in that In S4, the physical information loss function is: L PINN =MSE+ω 1 L FS +oh 2 L SWT +oh 3 L MGSEZhu +oh 4 L WB +oh 5 L CXH(T) in: ω 1 ,ω 2 ,ω 3 ,ω 4 and ω 5 is the weight of the physical information item, i is the sample number, n is the number of samples, N exp,i is the experimental life of the i-th sample, N pre,i is the predicted life span of the sample with sample number i; Among them, σ nmax , τ nmax , Δε nmax , Δγ nmax They are the maximum normal stress, maximum shear stress, maximum normal strain range, and maximum normal stress range corresponding to the critical surface when the plane where the maximum shear strain range is located is taken as the critical surface; σ smax , τ smax , Δε smax , Δγ smax They are the maximum normal stress, maximum shear stress, maximum normal strain range, and maximum normal stress range corresponding to the critical surface when the plane where the maximum normal strain range is located is taken as the critical surface; Δε n is the positive strain stroke; σ' f is the fatigue strength coefficient, τ' f is the shear fatigue strength coefficient, ε' f is the fatigue ductility coefficient, γ' f is the shear fatigue ductility coefficient, b is the fatigue strength index, c is the fatigue ductility index, b 0 is the shear fatigue strength index, c 0 is the shear fatigue ductility index, k is the FS model constant, k MGSE is the MGSEZhu model constant, S is the WB model constant; E is Young's modulus, G is the shear modulus, σ y is the yield strength, N f is fatigue life; v e is the Poisson's ratio in the elastic stage, which is taken as 0.3; v p is the Poisson's ratio in the plastic stage and is taken as 0.

5.

7. The method for predicting steel fatigue life based on a physical information neural network using a genetic algorithm according to claim 1, It is characterized in that In S5, the basic parameters of the genetic algorithm are the number of generations and the number of individuals in each generation; the fitness function of the genetic algorithm is: Among them, N exp,i is the experimental life of the i-th sample, N pre,i is the predicted life span of the ith sample, and n is the number of samples.

8. The method for predicting steel fatigue life based on a physical information neural network using a genetic algorithm according to claim 1, It is characterized in that In S5, an individual in each generation of the genetic algorithm is represented as: I pg =(ω 1pg Oh, oh 2pg Oh, oh 3pg Oh, oh 4pg Oh, oh 5pg ) Among them, p represents the individual serial number, and g represents the generation number; Taking the weight contained in the individual as the weight of the physical information item, then: oh apg =ω a (a=1,2..5) Where a is the serial number of the physical information item weight.

9. A computer device, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the physical information neural network steel fatigue life prediction method based on genetic algorithm as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the physical information neural network steel fatigue life prediction method based on genetic algorithm described in any one of claims 1 to 7 is implemented.