A method, apparatus, device, and medium for determining material fatigue strength

Abstract

This disclosure provides a method, apparatus, device, and medium for determining the fatigue strength of a material. The method includes: determining fatigue strength test data of a target material under at least one index; inputting the fatigue strength test data into a pre-trained material fatigue strength prediction model; and outputting the fatigue strength corresponding to the fatigue strength test data. The fatigue strength characterizes the maximum alternating stress value that the target material can withstand under a preset load condition and / or the number of cycles that the target material can withstand without damage when the maximum alternating stress value is applied. The material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer. The technical solution of this disclosure enables the determination of the fatigue strength of a target material based on fatigue strength test data and a material fatigue strength prediction model under an index, thereby improving the efficiency and accuracy of fatigue strength determination.

Description

Technical Field

[0001] This disclosure relates to the interdisciplinary field of artificial intelligence and materials science, and more particularly to a method, apparatus, device, and medium for determining the fatigue strength of materials. Background Technology

[0002] With the deepening of industrialization and the continuous expansion of the application range of metallic materials, in-depth research on the fatigue strength of nickel alloy materials can ensure the safe and reliable operation of mechanical components throughout their service life.

[0003] The current methods for determining the fatigue strength of nickel alloys mainly involve using stress-life curves to determine the fatigue life of nickel alloys under different stress levels, thereby helping to determine the fatigue strength of the nickel alloys. However, stress-life curves can only reflect the change of fatigue strength with stress as a single dimension, providing only limited analysis and lacking the flexibility and precision required for determining fatigue strength. Summary of the Invention

[0004] This disclosure provides a method, apparatus, device, and medium for determining the fatigue strength of materials, thereby improving the efficiency and accuracy of determining the fatigue strength of nickel alloy materials.

[0005] In a first aspect, embodiments of this disclosure provide a method for determining the fatigue strength of a material, the method comprising:

[0006] Determine the fatigue strength test data of the target material under at least one index; wherein, the at least one index includes at least one heat treatment index, constituent element content index, physical property index and microstructure index;

[0007] The fatigue strength test data is input into a pre-trained material fatigue strength prediction model, which outputs the fatigue strength corresponding to the fatigue strength test data. The fatigue strength is used to characterize the maximum alternating stress value that the target material can withstand under a preset load condition and / or the number of cycles that the target material can withstand without damage when the maximum alternating stress value is applied to the target material. The material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer.

[0008] Secondly, embodiments of the present invention also provide an apparatus for determining the fatigue strength of a material, the apparatus comprising:

[0009] The fatigue strength test data determination module is used to determine the fatigue strength test data of the target material under at least one index; wherein, the at least one index includes at least one heat treatment index, constituent element content index, physical property index and microstructure index;

[0010] The fatigue strength output module is used to input the fatigue strength test data into a pre-trained material fatigue strength prediction model and output the fatigue strength corresponding to the fatigue strength test data; wherein, the fatigue strength is used to characterize the maximum alternating stress value that the target material can withstand under a preset load form and / or, the number of cycles that the target material can withstand without damage when the maximum alternating stress value is applied to the target material, and the material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer.

[0011] Thirdly, embodiments of the present invention also provide an electronic device, the electronic device comprising:

[0012] One or more processors;

[0013] Storage device for storing one or more programs.

[0014] When the one or more programs are executed by the one or more processors, the one or more processors implement the method for determining the fatigue strength of materials as described in any embodiment of the present invention.

[0015] Fourthly, embodiments of the present invention also provide a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform a method for determining the fatigue strength of a material as described in any of the embodiments of the present invention.

[0016] Fifthly, embodiments of the present invention also provide a computer program product, including a computer program, characterized in that, when executed by a processor, the computer program implements the method for determining the fatigue strength of a material as described in any embodiment of the present invention.

[0017] The technical solution of this disclosure determines the fatigue strength test data of a target material under at least one of the following indicators: heat treatment index, constituent element content index, physical performance index, and microstructure index. The fatigue strength test data is input into a pre-trained material fatigue strength prediction model, which outputs the fatigue strength corresponding to the test data. The fatigue strength characterizes the maximum alternating stress value that the target material can withstand under a preset load condition and / or the number of cycles that allow the target material to withstand the maximum alternating stress value without damage. The material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer. This solves the problem in the prior art where the fatigue strength of nickel alloy materials is determined based on the stress-life curve, which only reflects the change of fatigue strength with stress as a single dimension, providing only limited analysis and lacking flexibility and accuracy in fatigue strength determination. Based on fatigue strength test data under heat treatment indicators, constituent element content indicators, physical performance indicators, and microstructure indicators, this invention determines the fatigue strength corresponding to the fatigue strength test data based on a material fatigue strength prediction model. By comprehensively considering multiple material characteristic indicators, it can accurately model the complex relationship between the fatigue strength of the material and multiple material characteristic indicators. This not only improves the prediction accuracy of material fatigue strength but also takes into account the influence that a single factor cannot fully reflect, thereby improving the overall prediction capability and achieving the effect of improving the efficiency and accuracy of fatigue strength determination. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of exemplary embodiments of the present invention, the accompanying drawings used in describing the embodiments are briefly introduced below. Obviously, the accompanying drawings described are only a portion of the drawings of the embodiments to be described in this invention, and not all of the drawings. For those skilled in the art, other drawings can be obtained from these drawings without any creative effort.

[0019] Figure 1 This is a schematic flowchart of a method for determining the fatigue strength of a material provided in an embodiment of this disclosure;

[0020] Figure 2 This is a schematic diagram of a long short-term memory network layer provided in an embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of a material fatigue strength prediction model provided in an embodiment of the present invention;

[0022] Figure 4 This is a flowchart illustrating another method for determining the fatigue strength of materials provided in an embodiment of the present invention;

[0023] Figure 5This is a schematic diagram of the distribution of fatigue strength provided in an embodiment of the present invention;

[0024] Figure 6 This is a schematic diagram of a Pearson correlation coefficient provided in an embodiment of the present invention;

[0025] Figure 7 This is a schematic diagram of an optimized hyperparameter provided in an embodiment of the present invention;

[0026] Figure 8 This is a schematic diagram of the experimental and predicted values ​​of the fatigue strength of a material provided in an embodiment of the present invention;

[0027] Figure 9 This is a schematic diagram of the mean absolute error, root mean square error, coefficient of determination, and mean absolute percentage error of fatigue strength provided in an embodiment of the present invention.

[0028] Figure 10 This is a schematic diagram of a device for determining the fatigue strength of materials provided in an embodiment of the present invention;

[0029] Figure 11 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0030] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0031] Before introducing the technical solutions provided in the embodiments of this disclosure, the application scenarios can be illustrated first. The technical solutions provided in the embodiments of this disclosure can be applied to scenarios where the fatigue strength of any alloy material needs to be determined. For example, alloy materials can include nickel alloys, aluminum alloys, iron alloys, etc. In this embodiment, nickel alloys are used as an example. For nickel alloys, heat treatment indicators, constituent element content indicators, physical performance indicators, and microstructure indicators can be obtained. Based on the original data under at least one heat treatment indicator, constituent element content indicator, physical performance indicator, and microstructure indicator, the data is input into a pre-trained material fatigue strength prediction model to determine the fatigue strength of the nickel alloy. Based on the technical solutions of the embodiments of this disclosure, the fatigue strength of the target material can be determined based on the fatigue strength test data of the target material under the indicators and the material fatigue strength prediction model, thereby improving the efficiency and accuracy of fatigue strength determination.

[0032] Example 1

[0033] Figure 1 This is a flowchart illustrating a method for determining the fatigue strength of a material according to an embodiment of this disclosure. This embodiment is applicable to determining the fatigue strength of any alloy material. In this embodiment, nickel alloy material is used as an example for illustration. This method can be executed by a device for determining the fatigue strength of a material. This device can be implemented in the form of software and / or hardware. The hardware can be a mobile electronic device. This electronic device can execute the method for determining the fatigue strength of a material provided by this technical solution.

[0034] like Figure 1 As shown, the method includes:

[0035] S110. Determine the fatigue strength test data of the target material under at least one index.

[0036] In this embodiment, the target material can be any alloy material. For example, the target material can be a copper alloy, an aluminum alloy, or a nickel alloy. In this embodiment, the target material can be a nickel alloy. It should be noted that the target material is also a material whose fatigue strength is determined based on fatigue strength test data under at least one index.

[0037] Among them, at least one indicator includes at least one heat treatment indicator, constituent element content indicator, physical performance indicator, and microstructure indicator.

[0038] It should be noted that fatigue strength test data refers to the data of the target material under the conditions of heat treatment index, constituent element content index, physical property index, and microstructure index. Based on the fatigue strength test data of the target material, the fatigue strength of the target material can be determined.

[0039] In this embodiment, at least one heat treatment index includes: normalizing temperature index, quenching temperature index, quenching cooling rate index, carburizing time index, diffusion time index, quenching medium temperature index, tempering temperature index, and tempering time index; constituent element content index is the percentage content of each element constituting the target material; physical performance index includes target material reduction rate index; and microstructure index includes the area ratio of inclusions in plastic deformation and the area ratio of isolated inclusions.

[0040] The normalizing temperature index for nickel alloy materials refers to the temperature range to which the target material is heated during the normalizing process. In this embodiment, the normalizing temperature index for nickel alloy materials is typically between 900℃ and 1150℃, depending on the alloy composition and expected performance requirements. During normalizing, the nickel alloy material is heated to this temperature range and held for a certain period of time to improve the microstructure of the target material, making it more uniform and improving its mechanical properties. Excessive temperature may lead to grain coarsening in the nickel alloy material, affecting its toughness. The quenching temperature index for nickel alloy materials is typically between 900℃ and 1100℃, adjusted according to the alloy composition. The quenching process involves heating the material to this temperature and then rapidly cooling it to obtain high hardness and strength. Temperatures that are too high or too low will affect the final hardness and microstructure of the nickel alloy material. Temperatures that are too low may result in the inability to reach the appropriate phase transformation temperature, failing to obtain good hardness. In this embodiment, excessively high temperatures may increase the brittleness of the nickel alloy material. The quenching cooling rate index refers to the rate at which the material is rapidly cooled after being heated to the quenching temperature. In this embodiment, a relatively fast quenching cooling rate is required for the nickel alloy. Common quenching media include oil, water, or gas. The cooling rate must be fast enough to ensure the desired martensitic or other phase transformation structure is obtained, but overcooling should be avoided to prevent cracking or deformation. The specific quenching cooling rate is adjusted according to the alloy type, and is usually controlled between 10-100°C per second. Carburizing involves exposing the material surface to a carbon-rich atmosphere at high temperature, allowing carbon elements to diffuse into the nickel material surface, thereby increasing the surface hardness of the material. The carburizing time refers to the time the material is held at a high temperature during the carburizing process. In this embodiment, too short a carburizing time may result in insufficient surface hardening, while too long a time may result in excessively high carbon content on the surface of the nickel alloy, affecting mechanical properties. The diffusion time refers to the time required during heat treatment to ensure that carbon or other alloying elements diffuse uniformly into the interior of the material. During diffusion, elements are distributed in the material through atomic movement. Too short a diffusion time may result in uneven element distribution and unstable performance, while too long a diffusion time may lead to excessive changes in the material. In this embodiment, diffusion treatment (e.g., nitriding or fluorination) of nickel alloy materials is commonly used to improve surface hardness and corrosion resistance. The diffusion time is typically between 2 and 8 hours, depending on the atmosphere and temperature used. During diffusion, the nickel alloy material is heated to a certain high temperature and held for a certain period to promote the diffusion of alloying elements on the surface of the nickel alloy material. The quenching medium temperature refers to the temperature of the medium used to cool the material during the quenching process. The temperature of the quenching medium directly affects the cooling rate. Excessively high temperatures may lead to a decrease in the cooling rate, affecting the quenching effect; excessively low temperatures may lead to overcooling, easily causing cracks or other defects. In this embodiment, the quenching medium temperature of the nickel alloy is typically controlled between room temperature and 90°C. The cooling media used include oil, nitrogen, and air.The temperature of the quenching medium needs to be properly controlled to ensure effective cooling and avoid cracking or excessive stress. Tempering involves heating the material to an appropriate temperature for quenching, followed by tempering. Tempering is commonly used to improve the toughness and impact resistance of materials. In this embodiment, the tempering temperature for nickel alloys is generally between 550°C and 750°C, with the specific temperature depending on the alloy type and performance requirements. Too high a tempering temperature may cause the material to soften, while too low a temperature may not effectively improve toughness. For nickel alloys, the tempering time refers to the time the material is heated and held at the tempering temperature. In this embodiment, the tempering time for nickel alloys is typically between 1 and 4 hours, mainly depending on the required hardness and toughness. The tempering time directly affects the hardness and toughness of the nickel alloy; too short a time may result in insufficient tempering, failing to achieve the ideal overall performance, while too long a time may reduce the material's hardness and affect its tensile strength.

[0041] The constituent element content indicators include the percentage content of carbon, silicon, manganese, phosphorus, sulfur, nickel, chromium, copper, and molybdenum in the target material. The reduction rate indicator typically refers to the percentage of the reduced substance reduced by the reducing agent in a reduction reaction. The target material reduction rate indicator refers to the degree of transformation or recovery of certain components or phases in the material under specific conditions. In this embodiment, nickel metal is reduced from nickel oxide, so the reduction rate can be used to describe the proportion of nickel oxide reduced to nickel metal in this process. The inclusion area ratio indicator for plastic processing deformation refers to the ratio of the area occupied by inclusions in the material to the total area of ​​the material. Inclusions are independent phases formed by non-metallic substances in a metal matrix; they may originate from impurities in raw materials, reaction products during smelting, or introduced substances during processing. The presence of inclusions disrupts the continuity of the metal matrix, affecting the mechanical and processing properties of the material. Therefore, the inclusion area ratio indicator is an important parameter for measuring the purity of the material. By measuring and calculating the inclusion area ratio, the overall quality and processing performance of the material can be evaluated, providing an important basis for material selection and processing. The area ratio index for isolating inclusions refers to the ratio of the area required to effectively isolate inclusions from a material to the total area of ​​the material. This index focuses more on the distribution and morphology of inclusions within the material, and on how processing methods can effectively isolate inclusions to reduce their impact on material properties.

[0042] Optionally, the raw data of the target material under at least one index are obtained; for at least one index, the mean and standard deviation are determined based on the raw data under the index; for at least one index, each raw data under the index is normalized based on the mean and standard deviation to obtain the normalized data corresponding to each raw data under the index; and fatigue strength test data used for predicting the fatigue strength of the material are determined based on the normalized data under at least one index.

[0043] It should be noted that raw data refers to data on the target material obtained through experiments, measurements, etc. This data is unprocessed and directly reflects a certain characteristic or state of the target material. Raw data includes data for at least one heat treatment index, constituent element content index, physical performance index, and microstructure index. The mean refers to the average of the raw data for each index. The standard deviation refers to the standard deviation of the raw data for each index. For each raw data point, the normalization formula is:

[0044]

[0045] Where x represents each original data point of the target material under each indicator; μ represents the mean value corresponding to the corresponding indicator; σ represents the standard deviation corresponding to the corresponding indicator; and Z represents the normalized data corresponding to each original data point under each indicator after normalization processing.

[0046] It should be noted that for the raw data under a certain indicator, the normalized data for each raw data point is obtained by subtracting the mean of the raw data under that indicator from each raw data point and then dividing by the standard deviation of the raw data under that indicator. Normalizing the raw data under all indicators using the above method yields the normalized data for each raw data point under each indicator. The resulting normalized data for all indicators constitutes the fatigue strength test data used in predicting material fatigue strength.

[0047] Specifically, raw data for nickel alloy materials under heat treatment, constituent element content, physical properties, and microstructure parameters are obtained. The mean and standard deviation of each raw data point under each parameter are calculated. Based on each raw data point, its mean, standard deviation, and the normalization formula, normalized data for each raw data point are obtained. The obtained normalized data for each parameter are then used as fatigue strength test data for predicting the material's fatigue strength.

[0048] S120. Input the fatigue strength test data into the pre-trained material fatigue strength prediction model and output the fatigue strength corresponding to the fatigue strength test data.

[0049] Fatigue strength is used to characterize the maximum alternating stress value that the target material can withstand under a preset load form and / or the number of cycles that the target material can withstand without being damaged when the maximum alternating stress value is applied to the target material. The material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer.

[0050] It should be noted that convolutional neural network layers mainly include convolutional layers, pooling layers, and fully connected layers. Long Short-Term Memory (LSTM) networks are a special type of recurrent neural network (RNN) structure that addresses the vanishing or exploding gradient problems that traditional RNNs encounter when processing long sequences of data. LSM networks introduce memory units into the RNN structure, including forget gates, input gates, and output gates. See also... Figure 2 Where σ is the sigmoid function, As a candidate state, forget layer f t Its function is to determine which data to forget; input gate i t The Long Short-Term Memory (LSTM) network layer determines which data enters the current state. It uses the forget gate and input gate to decide which data to discard and which to retain, and the output gate... t The expressions for generating the output at the current moment are as follows:

[0051]

[0052] f t =σ(W f ·[h t-1, x t ]+b f );

[0053] i t =σ(W i ·[h t-1, x t ]+b i );

[0054] o t =σ(W o ·[h t-1 ,x t ]+b o );

[0055] Where W and b are the weight matrix and bias to be trained, respectively, and h t-1This is the output of the previous layer. Optionally, the input dimension of the Long Short-Term Memory (LSTM) network layer can be set to 20, the output dimension to 200, and the ReLU activation function can be used.

[0056] Convolutional neural network (CNN) layers possess the characteristics of local connectivity and weight sharing. In material fatigue strength prediction, CNN layers can be used to extract feature data during the fatigue process of nickel alloy materials, thereby achieving fatigue strength prediction. Long Short-Term Memory (LSTM) network layers can learn long-term dependencies, that is, they remember input information from a long time ago and use this information in the current output. In material fatigue strength prediction, LSM layers can be used to process fatigue strength test data under different indices during the material fatigue process, thereby achieving fatigue strength prediction.

[0057] In this embodiment, the material fatigue strength prediction model includes, in order of processing the input data, a model structure consisting of a convolutional layer, a pooling layer, a random deactivation layer, a long short-term memory network layer, a flattening layer, and a fully connected layer.

[0058] Among them, see Figure 3 Convolutional layers are primarily used to extract features from input fatigue strength test data, reducing the difficulty of network learning. Their convolution operation is... Where k and b represent the weights and biases of the i-th neuron in the l-th layer, and x is the l-th input of the j-th layer. Pooling layers are used to simplify computation and select and filter extracted features; their expression is: Where q represents l neurons in the i-th channel, and W is the pooling kernel size; the random deactivation layer is a technique used to prevent overfitting in neural networks. It reduces co-fitting among neurons by randomly setting the output of a subset of neurons to zero during training, thereby enhancing the model's generalization ability. The random deactivation layer can prevent overfitting, improve regularization, and enhance the robustness of material fatigue strength prediction models; the flattening layer is mainly used in convolutional neural networks. It converts the multidimensional feature maps generated by convolutional or pooling layers into one-dimensional vectors so that subsequent fully connected layers can process them. The flattening layer can convert data formats, maintain the order of features, and connect convolutional and fully connected layers, enabling material fatigue strength prediction models to handle more complex data; the fully connected layer integrates the information from the feature maps for the final regression task. Optionally, the input dimension of linear layer 1 can be set to 200, the output dimension to 8, and the activation function to ReLU; the input dimension of linear layer 2 can be set to 8, the output dimension to 1, and the activation function to Linear.

[0059] Specifically, a material fatigue strength prediction model is constructed based on convolutional neural network (CNN) layers and long short-term memory (LSTM) network layers. Following the order of input data processing, the model structure consists of convolutional layers, pooling layers, random deactivation layers, LSTM network layers, flattening layers, and fully connected layers. Convolutional layers are used to extract features from the input fatigue strength test data. The extracted fatigue strength test data is then input into the LSTM network layer to learn the sequential relationship between the input features. The output of the LSTM network layer is then used as the input to the fully connected layer. After the material fatigue strength prediction model is constructed and trained, fatigue strength test data is input into the pre-trained model, which outputs the fatigue strength corresponding to the test data.

[0060] The technical solution of this disclosure determines the fatigue strength test data of a target material under at least one of the following indicators: heat treatment index, constituent element content index, physical performance index, and microstructure index. The fatigue strength test data is input into a pre-trained material fatigue strength prediction model, which outputs the fatigue strength corresponding to the test data. The fatigue strength characterizes the maximum alternating stress value that the target material can withstand under a preset load condition and / or the number of cycles that allow the target material to withstand the maximum alternating stress value without damage. The material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer. This solves the problem in the prior art where the fatigue strength of nickel alloy materials is determined based on the stress-life curve, which only reflects the change of fatigue strength with stress as a single dimension, providing only limited analysis and lacking flexibility and accuracy in fatigue strength determination. Based on fatigue strength test data under heat treatment indicators, constituent element content indicators, physical performance indicators, and microstructure indicators, this invention determines the fatigue strength corresponding to the fatigue strength test data based on a material fatigue strength prediction model. By comprehensively considering multiple material characteristic indicators, it can accurately model the complex relationship between the fatigue strength of the material and multiple material characteristic indicators. This not only improves the prediction accuracy of material fatigue strength but also takes into account the influence that a single factor cannot fully reflect, thereby improving the overall prediction capability and achieving the effect of improving the efficiency and accuracy of fatigue strength determination.

[0061] Example 2

[0062] Figure 4 This is a flowchart illustrating another method for determining material fatigue strength according to an embodiment of the present invention. Based on the foregoing embodiments, the fatigue strength prediction model is explained in detail, and its specific implementation can be found in the technical solution of this embodiment. Technical terms that are the same as or corresponding to those in the above embodiments will not be repeated here.

[0063] like Figure 4 As shown, the method specifically includes the following steps:

[0064] S210. Obtain the training sample dataset.

[0065] The training sample dataset includes a test set and a training set, each containing multiple samples. Each sample includes experimental data to be used under at least one metric and the corresponding theoretical fatigue strength.

[0066] It should be noted that the data of the target material obtained through experiments, measurements, or directly from databases are normalized. The normalized data is the test data to be used. The theoretical fatigue strength refers to the actual fatigue strength corresponding to each test data point to be used.

[0067] It should be noted that, see Figure 5 By plotting kernel density lines and histograms, the distribution of theoretical fatigue intensity in the training sample dataset can be analyzed and visualized in depth. See also Figure 5 The horizontal axis represents the theoretical fatigue intensity in the training sample dataset, and the theoretical fatigue intensity data is divided into several continuous intervals, which are represented by bars on the graph. A smooth curve illustrates the distribution of theoretical fatigue intensity. This curve is generated by placing a smooth kernel (such as a Gaussian kernel) around each theoretical fatigue intensity data point and summing these kernels. The vertical axis represents the kernel density, that is, the probability density of theoretical fatigue intensity data near a certain fatigue intensity value. The area under the curve is usually 1, representing the total probability of the theoretical fatigue intensity data. Figure 5 As can be seen, the theoretical fatigue intensity values ​​in the training sample dataset span a wide range, from 200 MPa to 1200 MPa. However, most fatigue intensity data are concentrated between 400 MPa and 600 MPa. This indicates that the theoretical fatigue intensity of most samples in the training sample dataset is concentrated in this range. Histograms and kernel density curves not only provide a general understanding of the discrete distribution of fatigue intensity data but also allow for a more accurate grasp of the overall distribution trend and probability density of the fatigue intensity data.

[0068] Optionally, after obtaining the training sample dataset, for multiple training samples in the test set, the correlation between the experimental data to be used in the samples is calculated based on the Pearson correlation coefficient algorithm to obtain the first correlation strength between each pair of indicators; for multiple training samples in the test set, the correlation between the experimental data to be used in the samples and the theoretical fatigue strength is calculated based on the Pearson correlation coefficient algorithm to obtain the second correlation strength between each indicator and the theoretical fatigue strength.

[0069] The first correlation strength refers to the correlation strength between any two indicators, determined by the Pearson correlation coefficient algorithm. The second correlation strength refers to the correlation strength between any single indicator and the theoretical fatigue intensity, also determined by the Pearson correlation coefficient algorithm. The Pearson correlation coefficient algorithm is a statistical method used to measure the strength and direction of the linear relationship between two variables. See also... Figure 6 The strength of the correlation can be characterized by the Pearson correlation coefficient, also known as the linear correlation coefficient, which ranges from -1 to 1. A correlation coefficient of 1 indicates a perfectly positive linear correlation between each pair of indicators or each indicator and the theoretical fatigue intensity; that is, one indicator increases strictly with the increase of another, or the theoretical fatigue intensity increases strictly with the increase of a certain indicator. A correlation coefficient of -1 indicates a perfectly negative linear correlation between each pair of indicators or each indicator and the theoretical fatigue intensity; that is, one indicator decreases strictly with the increase of another, or the theoretical fatigue intensity decreases strictly with the increase of a certain indicator. A correlation coefficient of 0 indicates no linear correlation between each pair of indicators or each indicator and the theoretical fatigue intensity. The formula for calculating the Pearson correlation coefficient is:

[0070]

[0071] Where PC(x,y) is the Pearson correlation coefficient between variables, representing the correlation strength between any two indicators or the correlation strength between each indicator and the theoretical fatigue intensity. n is the total number of training samples in the test set. i and y i The characteristic data consists of the value of each indicator relative to the theoretical fatigue strength. and For feature data x i and y i The average value is the average value of each indicator and the theoretical fatigue strength.

[0072] Specifically, the normalized test data and the corresponding theoretical fatigue strengths are obtained. The test data and their corresponding theoretical fatigue strengths are divided into a test set and a training set in an 8:2 ratio. For multiple training samples in the test set, the correlation between the test data and the theoretical fatigue strengths is calculated using the Pearson correlation coefficient algorithm to obtain the first correlation strength between each pair of indicators. Based on the first correlation strength, it is possible to identify which indicators are highly correlated. Highly correlated indicators may provide repetitive or redundant information; therefore, when constructing a material fatigue strength prediction model, one representative indicator can be selected while other redundant indicators are removed, thereby simplifying the material fatigue strength prediction model, reducing computational load, and avoiding model overfitting. The correlation between the test data and the theoretical fatigue strengths in the samples is then calculated using the Pearson correlation coefficient algorithm to obtain the second correlation strength between each indicator and the theoretical fatigue strength. Based on the second correlation strength, it is possible to identify which indicators are significantly correlated with the theoretical fatigue strength. Indicators that are significantly correlated with the theoretical fatigue strength are crucial for predicting the fatigue strength of materials and can serve as the main features input to the material fatigue strength prediction model, thereby increasing the reliability of the model. Furthermore, when constructing a material fatigue strength prediction model, different weights can be assigned based on the correlation strength between each indicator and fatigue strength. Indicators with stronger correlations can be given greater weights to improve the prediction accuracy of the material fatigue strength prediction model.

[0073] S220. For multiple training samples in the training set, the experimental data to be used in the samples is processed based on the Bayesian optimization algorithm to determine the optimization hyperparameters, and multiple samples are processed under the optimization hyperparameters.

[0074] Bayesian optimization is a global optimization method based on a probabilistic model, suitable for high-dimensional optimization problems with high computational costs. When determining hyperparameters, Bayesian optimization can effectively reduce the number of trials and find the optimal combination of hyperparameters. In this embodiment of the invention, the optimized hyperparameters refer to the optimal set of model parameters determined by the Bayesian optimization algorithm, including the number of convolutional channels, kernel size, dropout rate, and learning rate.

[0075] Specifically, for the experimental data to be used and the corresponding theoretical fatigue strength under at least one metric from multiple samples in the test set of the training sample dataset, the optimal hyperparameters can be determined using the Bayesian optimization algorithm. After determining the optimal set of model parameters, including the number of convolution channels, the convolution kernel size, the dropout rate, and the learning rate, the optimized hyperparameters are input into the material fatigue strength prediction model to guide the subsequent model training process of the material fatigue strength prediction model.

[0076] S230. The processing results are evaluated based on the objective function corresponding to the Bayesian optimization algorithm. If the evaluation results meet the preset conditions, the obtained optimized hyperparameters are used as model parameters in the fatigue strength prediction model.

[0077] Generally, when determining hyperparameters using Bayesian optimization algorithms, a surrogate model can be selected based on methods such as Gaussian process regression. The surrogate model can model the distribution of the objective function and accurately represent its behavior. Based on the current surrogate model, Bayesian optimization can select the next combination of hyperparameters for evaluation. Here, the objective function is the validation accuracy or loss function of the material fatigue strength prediction model. The selected hyperparameters are used to train and evaluate the surrogate model, and the evaluation results can be used to update the surrogate model.

[0078] It should be noted that when selecting hyperparameter combinations, the selection can be based on the optimization range of the parameters. For example, see... Figure 7 The optimization ranges for the parameters can be as follows: the number of convolutional channels can be set from 8 to 128, the kernel size from 2 to 7, the dropout rate from 0.1 to 0.5, and the learning rate from 10 to 150. Once the optimization ranges are determined, the search space for each model parameter is defined in Bayesian optimization. The number of convolutional channels can be searched within a limited range of 8 to 128, the kernel size from 2 to 7, the dropout rate from 0.1 to 0.5, and the learning rate from 10 to 150. A surrogate model is trained using the selected set of hyperparameters, and its performance is evaluated. The evaluation results are used to update the surrogate model. Through iterative iteration, the surrogate model is updated each time, and the next optimal hyperparameter combination can be selected based on the optimization ranges.

[0079] Meeting the preset conditions in the evaluation results means that, as iterations and evaluations proceed, the Bayesian optimization algorithm eventually finds a set of optimized hyperparameter combinations, i.e., finds the optimized hyperparameters. After determining the optimized hyperparameters, the obtained optimized hyperparameters are used as model parameters in the fatigue strength prediction model to guide the subsequent model training process of the material fatigue strength prediction model.

[0080] Specifically, optimization ranges are defined for the number of convolutional channels, kernel size, dropout rate, and learning rate. After selecting a surrogate model, Bayesian optimization selects and evaluates hyperparameter combinations within the optimization range based on the current surrogate model. The surrogate model is trained using the selected hyperparameter combinations, and its performance is evaluated. The evaluation results are used to update the surrogate model. Through repeated iterations, the surrogate model is updated each time, and the next optimal hyperparameter combination is selected. As the iterations proceed, Bayesian optimization can more accurately estimate the optimal solution of the objective function. After multiple iterations and evaluations, Bayesian optimization finally finds a set of optimized hyperparameter combinations, i.e., optimized hyperparameters. The obtained optimized hyperparameters can be input into the fatigue intensity prediction model as model parameters to guide the subsequent training process.

[0081] S240. For multiple samples in the test set, determine the accuracy of the fatigue strength prediction model based on the multiple test samples.

[0082] Specifically, for fatigue strength prediction models whose parameters are optimized hyperparameters, multiple samples from the test set are input into the model. For each sample in the test set, the predicted fatigue strength corresponding to the input sample can be obtained based on the fatigue strength prediction model. Based on the predicted fatigue strength and theoretical fatigue strength of multiple samples in the test set, the accuracy of the fatigue strength prediction model can be determined.

[0083] For example, see Figure 8 For the predicted and theoretical fatigue strengths of multiple samples in the test set, a scatter plot of the predicted and theoretical fatigue strengths can quickly and intuitively reflect the prediction range and accuracy of the fatigue strength prediction model. Contour lines represent reference lines where the predicted fatigue strength is exactly equal to the theoretical fatigue strength. The closer the scatter plot formed by the predicted and theoretical fatigue strengths is to the contour lines, the higher the accuracy of the fatigue strength prediction model. A ±0.3 scatter line indicates that the predicted fatigue strength is 0.3 or 1.7 times the theoretical fatigue strength (i.e., deviating from the theoretical value by ±70%). The lower limit of the ±0.3 scatter line represents a predicted value far lower than the theoretical value, possibly indicating that the fatigue strength of the material or structure is severely underestimated, or that the prediction model has a significant error. The upper limit of the ±0.3 scatter line represents a predicted value far higher than the theoretical value, possibly indicating that the fatigue strength of the material or structure is overestimated, or that the prediction model is too conservative. A ±1.0 scatter line indicates that the predicted fatigue strength deviates completely from the theoretical fatigue strength, indicating a large deviation in the fatigue strength prediction model, which may require model optimization or structural improvement.

[0084] See Figure 9To comprehensively evaluate the predictive performance of the fatigue strength prediction model, the mean absolute error (MAE), root mean square error (RMSE), coefficient of determination (CDO), and mean absolute percentage error (MAS%) were used as metrics. The MAE directly reflects the average level of the prediction error of the fatigue strength prediction model; a smaller MAE indicates a more accurate prediction. The RMSE is sensitive to larger errors; a smaller RMSE indicates a more accurate prediction. The CDO ranges from 0 to 1; a value closer to 1 indicates a stronger interpretability and better fit of the model. The MAS expresses the prediction error as a percentage, facilitating comparison of errors across different datasets. A smaller MAS indicates a more accurate prediction. For example, based on the predicted fatigue strength and theoretical fatigue strength, the MAE is 0.1215, the RMSE is 0.1564, the CDO is 0.9808, and the MAS is 0.423. Compared with traditional fixed-parameter deep learning models, the fatigue strength prediction model of this disclosure effectively reduces prediction error, improves prediction accuracy and generalization ability, and fully verifies the superiority and practical value of the present invention in the field of fatigue strength prediction of nickel alloy materials.

[0085] S250. When the accuracy reaches the preset accuracy threshold, the fatigue strength prediction model is determined.

[0086] Specifically, as training progresses, when the number of training rounds exceeds 60, that is, the model has completely traversed the training set 60 times, the loss function tends to converge during the training process, the loss value decreases, the accuracy increases, the model performance tends to stabilize, and finally the fatigue intensity prediction model is determined.

[0087] The technical solution of this disclosure first obtains a training sample dataset. Then, for multiple training samples in the training set, the experimental data to be used in the samples is processed based on a Bayesian optimization algorithm to determine the optimized hyperparameters. Further, the processing results are evaluated based on the objective function corresponding to the Bayesian optimization algorithm. If the evaluation results meet preset conditions, the obtained optimized hyperparameters are used as model parameters in the fatigue strength prediction model. After determining the model parameters, for multiple samples in the test set, the accuracy of the fatigue strength prediction model is determined based on multiple test samples. When the accuracy reaches a preset accuracy threshold, the fatigue strength prediction model is determined. Bayesian optimization, by constructing a probabilistic model of the objective function, intelligently selects evaluation points and finds the optimal hyperparameter combination with fewer iterations. The Bayesian optimization algorithm achieves more accurate and efficient determination of the model parameters for the fatigue strength prediction model, thereby guiding the model to perform efficient and accurate training. This not only improves the model's prediction performance but also significantly optimizes the parameter tuning process. This achieves the effects of improving fatigue strength prediction accuracy, saving costs and time.

[0088] Example 3

[0089] Figure 10 This is a schematic diagram of a device for determining the fatigue strength of a material, provided in an embodiment of this disclosure. Figure 10 As shown, the device includes: a fatigue strength test data determination module 310 and a fatigue strength output module 320.

[0090] A fatigue strength test data determination module is used to determine the fatigue strength test data of the target material under at least one index; wherein, the at least one index includes at least one heat treatment index, constituent element content index, physical property index, and microstructure index; a fatigue strength output module is used to input the fatigue strength test data into a pre-trained material fatigue strength prediction model and output the fatigue strength corresponding to the fatigue strength test data; wherein, the fatigue strength is used to characterize the maximum alternating stress value that the target material can withstand under a preset load form and / or, the number of cycles that the target material can withstand without damage when the maximum alternating stress value is applied to the target material, and the material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer.

[0091] The technical solution of this disclosure determines the fatigue strength test data of a target material under at least one of the following indicators: heat treatment index, constituent element content index, physical performance index, and microstructure index. The fatigue strength test data is input into a pre-trained material fatigue strength prediction model, which outputs the fatigue strength corresponding to the test data. The fatigue strength characterizes the maximum alternating stress value that the target material can withstand under a preset load condition and / or the number of cycles that allow the target material to withstand the maximum alternating stress value without damage. The material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer. This solves the problem in the prior art where the fatigue strength of nickel alloy materials is determined based on the stress-life curve, which only reflects the change of fatigue strength with stress as a single dimension, providing only limited analysis and lacking flexibility and accuracy in fatigue strength determination. Based on fatigue strength test data under heat treatment indicators, constituent element content indicators, physical performance indicators, and microstructure indicators, this invention determines the fatigue strength corresponding to the fatigue strength test data based on a material fatigue strength prediction model. By comprehensively considering multiple material characteristic indicators, it can accurately model the complex relationship between the fatigue strength of the material and multiple material characteristic indicators. This not only improves the prediction accuracy of material fatigue strength but also takes into account the influence that a single factor cannot fully reflect, thereby improving the overall prediction capability and achieving the effect of improving the efficiency and accuracy of fatigue strength determination.

[0092] Based on the above technical solutions, the at least one heat treatment index includes: normalizing temperature index, quenching temperature index, quenching cooling rate index, carburizing time index, diffusion time index, quenching medium temperature index, tempering temperature index, and tempering time index; the constituent element content index is the percentage content of each element constituting the target material; the physical performance index includes the target material reduction rate index; and the microstructure index includes the area ratio of inclusions during plastic processing deformation and the area ratio of isolated inclusions.

[0093] Based on the above technical solutions, determining the fatigue strength test data of the target material includes: acquiring the raw data of the target material under the at least one index; for the at least one index, determining the mean and standard deviation based on the raw data under the index; for the at least one index, normalizing each raw data under the index based on the mean and standard deviation to obtain normalized data corresponding to each raw data under the index; and determining the fatigue strength test data used for predicting the fatigue strength of the material based on the normalized data under the at least one index.

[0094] Based on the above technical solutions, the material fatigue strength prediction model includes, according to the processing order of the input data, the model structure of the material fatigue strength prediction model is as follows: convolutional layer, pooling layer, random deactivation layer, long short-term memory network layer, flattening layer and fully connected layer.

[0095] Based on the above technical solutions, the device further includes: a training sample dataset acquisition module, an optimization hyperparameter determination module, a model parameter determination module, an accuracy determination module, and a fatigue strength prediction model determination module.

[0096] The training sample dataset acquisition module is used to acquire the training sample dataset, wherein the training sample dataset includes a test set and a training set, and both the test set and the training set include multiple samples, wherein the samples include experimental data to be used under at least one indicator and the corresponding theoretical fatigue strength;

[0097] The hyperparameter optimization determination module is used to process the experimental data to be used in the training samples based on the Bayesian optimization algorithm for multiple training samples in the training set, determine the optimization hyperparameters, and process multiple samples under the optimization hyperparameters.

[0098] The model parameter determination module is used to evaluate the processing results based on the objective function corresponding to the Bayesian optimization algorithm. If the evaluation results meet the preset conditions, the obtained optimization hyperparameters are used as model parameters in the fatigue strength prediction model.

[0099] An accuracy determination module is used to determine the accuracy of the fatigue strength prediction model based on multiple test samples in a test set.

[0100] The fatigue strength prediction model determination module is used to determine the fatigue strength prediction model when the accuracy reaches a preset accuracy threshold.

[0101] Based on the above technical solutions, the target material is a nickel alloy.

[0102] Based on the above technical solutions, after obtaining the training sample dataset, the device further includes: a correlation strength determination module and a correlation strength acquisition module.

[0103] The correlation strength determination module is used to calculate the correlation between the experimental data to be used in the test samples based on the Pearson correlation coefficient algorithm for multiple training samples in the test set, so as to obtain the correlation strength between every two original data under the index.

[0104] The correlation strength acquisition module is used to calculate the correlation between the experimental data to be used and the theoretical fatigue strength in the test samples based on the Pearson correlation coefficient algorithm for multiple training samples in the test set, so as to obtain the correlation strength between each original data and the theoretical fatigue strength under the index.

[0105] The apparatus for determining material fatigue strength provided in this disclosure can execute the method for determining material fatigue strength provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects for executing the method.

[0106] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this disclosure.

[0107] Example 4

[0108] Figure 11 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Refer to the following... Figure 11 It illustrates an electronic device suitable for implementing embodiments of the present disclosure (e.g., Figure 11 The diagram below shows the structure of the terminal device or server 500. The terminal device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and vehicle terminals (e.g., vehicle navigation terminals). Figure 11 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0109] like Figure 11 As shown, electronic device 500 may include a processing unit (e.g., central processing unit, graphics processor, etc.) 501, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 502 or a program loaded from storage device 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of electronic device 500. The processing unit 501, ROM 502, and RAM 503 are interconnected via bus 504. An edit / output (I / O) interface 505 is also connected to bus 504.

[0110] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic device 500 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 11 An electronic device 500 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0111] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, it performs the functions defined in the methods of embodiments of this disclosure.

[0112] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0113] The electronic device provided in this embodiment and the method for determining material fatigue strength provided in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0114] Example 5

[0115] This disclosure provides a computer storage medium storing a computer program that, when executed by a processor, implements the method for determining material fatigue strength provided in the above embodiments.

[0116] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0117] In some implementations, the server may communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and may interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and end-to-end networks (e.g., ad hoc end-to-end networks), as well as any currently known or future-developed networks.

[0118] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0119] The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to:

[0120] Determine the fatigue strength test data of the target material under at least one index; wherein, the at least one index includes at least one heat treatment index, constituent element content index, physical property index and microstructure index;

[0121] The fatigue strength test data is input into a pre-trained material fatigue strength prediction model, which outputs the fatigue strength corresponding to the fatigue strength test data. The fatigue strength is used to characterize the maximum alternating stress value that the target material can withstand under a preset load condition and / or the number of cycles that the target material can withstand without damage when the maximum alternating stress value is applied to the target material. The material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer.

[0122] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0123] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0124] The units described in the embodiments of this disclosure can be implemented in software or hardware. The names of the units are not, in some cases, intended to limit the specific unit.

[0125] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0126] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0127] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0128] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0129] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for determining the fatigue strength of a material, characterized in that, include: Determine the fatigue strength test data of the target material under at least one index; wherein, the at least one index includes at least one heat treatment index, constituent element content index, physical property index and microstructure index; The fatigue strength test data is input into a pre-trained material fatigue strength prediction model, which outputs the fatigue strength corresponding to the fatigue strength test data. The fatigue strength is used to characterize the maximum alternating stress value that the target material can withstand under a preset load condition and / or the number of cycles that the target material can withstand without damage when the maximum alternating stress value is applied to the target material. The material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer.

2. The method according to claim 1, characterized in that, The at least one heat treatment index includes: normalizing temperature index, quenching temperature index, quenching cooling rate index, carburizing time index, diffusion time index, quenching medium temperature index, tempering temperature index, and tempering time index; the constituent element content index is the percentage content of each element constituting the target material; the physical performance index includes the target material reduction rate index; and the microstructure index includes the area ratio of inclusions during plastic processing deformation and the area ratio of isolated inclusions.

3. The method according to claim 1, characterized in that, The fatigue strength test data for determining the target material includes: Obtain the raw data of the target material under at least one of the above indicators; For at least one indicator, the mean and standard deviation are determined based on the raw data under that indicator; For the at least one indicator, each original data point under the indicator is normalized based on the mean and standard deviation of the indicator to obtain the normalized data corresponding to each original data point under the indicator. Based on the normalized data under at least one of the aforementioned indicators, fatigue strength test data for use in predicting material fatigue strength are determined.

4. The method according to claim 1, characterized in that, The material fatigue strength prediction model includes: Based on the order of processing the input data, the model structure of the material fatigue strength prediction model is as follows: convolutional layer, pooling layer, random deactivation layer, long short-term memory network layer, flattening layer, and fully connected layer.

5. The method according to claim 1 or 4, characterized in that, Also includes: Obtain a training sample dataset, wherein the training sample dataset includes a test set and a training set, and both the test set and the training set include multiple samples, wherein the samples include experimental data to be used under at least one indicator and the corresponding theoretical fatigue strength; For multiple training samples in the training set, the experimental data to be used in the samples are processed based on the Bayesian optimization algorithm to determine the optimization hyperparameters, and multiple samples are processed under the optimization hyperparameters; The processing results are evaluated based on the objective function corresponding to the Bayesian optimization algorithm. If the evaluation results meet the preset conditions, the obtained optimized hyperparameters are used as model parameters in the fatigue strength prediction model. For multiple samples in the test set, the accuracy of the fatigue strength prediction model is determined based on the multiple test samples; When the accuracy reaches a preset accuracy threshold, the fatigue strength prediction model is determined.

6. The method according to claim 1, characterized in that, The target material is a nickel alloy.

7. An apparatus for determining the fatigue strength of a material, characterized in that, include: The fatigue strength test data determination module is used to determine the fatigue strength test data of the target material under at least one index; wherein, the at least one index includes at least one heat treatment index, constituent element content index, physical property index and microstructure index; The fatigue strength output module is used to input the fatigue strength test data into a pre-trained material fatigue strength prediction model and output the fatigue strength corresponding to the fatigue strength test data; wherein, the fatigue strength is used to characterize the maximum alternating stress value that the target material can withstand under a preset load form and / or, the number of cycles that the target material can withstand without damage when the maximum alternating stress value is applied to the target material, and the material fatigue strength prediction model is determined by at least a convolutional neural network layer and a long short-term memory network layer.

8. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs. When one or more programs are executed by one or more processors, the one or more processors implement the method for determining the fatigue strength of a material as described in any one of claims 1-6.

9. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the method for determining the fatigue strength of a material as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method for determining the fatigue strength of a material according to any one of claims 1-6.

Images