Neural network-based double-wall cooling structure creep life prediction method

Abstract

The invention relates to a method, device and equipment for predicting the creep life of a double-wall cooling structure based on a neural network and a medium, and relates to the field of aero-engine double-wall turbine blade creep life prediction.The method comprises the steps that creep tests are conducted on a single-axis flat plate sample and a multi-axis flat plate sample under different stresses respectively; obtaining a uniaxial creep curve and a multi-axial creep life; fitting a uniaxial creep curve and a multiaxial creep life on the basis of finite element analysis results of the continuous damage model on the uniaxial flat plate sample finite element model and the multiaxial flat plate sample, and determining material parameters in the continuous damage model; performing finite element analysis on the finite element model of the double-wall cooling structure based on a continuous damage model of known material parameters to obtain bone point stress at a bone point on a fracture interface, and determining reference stress for predicting the creep life based on the bone point stress; and performing creep life prediction by taking the reference stress as the input of the creep life prediction model to obtain the predicted creep life of the double-wall cooling structure.

Description

Background Art

[0002] A double-wall cooling structure is a composite cooling structure typically composed of impingement holes, overflow holes, spoiler columns, and inner and outer wall panels. It offers high cooling efficiency. However, the structural characteristics of double-wall cooling structures pose certain strength risks and are prone to structural failure, with high-temperature creep being a major failure mode. Therefore, accurately predicting the creep life of double-wall cooling structures is crucial to ensuring the proper operation of mechanical components.

[0003] Conventional creep life prediction models based on damage mechanics are often used to predict the creep life of double-wall cooling structures. These models are often limited by complex damage mechanisms or sensitive parameter values, and fail to comprehensively consider the impact of multiple factors on creep life, resulting in limited prediction accuracy. Therefore, there is still room for improvement in the accuracy of creep life prediction for double-wall cooling structures.

[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute prior art known to ordinary technicians in the field. Summary of the Invention

[0005] The purpose of the present disclosure is to provide a method and device for predicting the creep life of a double-wall cooling structure based on a neural network, an electronic device, and a computer-readable storage medium, so as to improve the accuracy of the creep life prediction of the double-wall cooling structure at least to a certain extent.

[0006] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by practice of the present disclosure.

[0007] According to a first aspect of the present disclosure, a neural network-based creep life prediction method for a double-wall cooling structure is provided, comprising: performing creep tests on a uniaxial flat plate specimen and a multiaxial flat plate specimen under different stresses, respectively, to obtain a uniaxial creep curve and a multiaxial creep life; performing finite element analysis on a uniaxial flat plate specimen finite element model and a multiaxial flat plate specimen finite element model based on a continuous damage model, and fitting the uniaxial creep curve and the multiaxial creep life using the results of the finite element analysis to determine material parameters in the continuous damage model; performing finite element analysis on a finite element model of the double-wall cooling structure based on a continuous damage model with known material parameters to obtain a bone point stress at a bone point on a fracture interface, and determining a reference stress for creep life prediction based on the bone point stress; and performing creep life prediction using the reference stress as input to the creep life prediction model to obtain a predicted creep life of the double-wall cooling structure.

[0008] Optionally, before using the reference stress as the input of the creep life prediction model, the method further includes: training a neural network model based on creep test data corresponding to the target material in the material manual to obtain a creep life prediction model, where the target material is consistent with the material of the double-wall cooling structure.

[0009] Optionally, a neural network model is trained based on creep test data corresponding to the target material in the material manual to obtain a creep life prediction model, including: normalizing the creep test data composed of temperature, stress and creep life to obtain normalized creep test data; using the normalized creep test data as a training set, and using the Bayesian regularization function as a training function to train the back propagation neural network model to obtain a creep life prediction model.

[0010] Optionally, creep tests are performed on the uniaxial flat plate specimen and the multiaxial flat plate specimen under different stresses to obtain uniaxial creep curves and multiaxial creep life, including: using a flat plate specimen without a circular hole in the center as a uniaxial flat plate specimen, and performing creep tests on the uniaxial flat plate specimen under different stresses to obtain a uniaxial creep curve; using a flat plate specimen with a circular hole in the center as a multiaxial flat plate specimen, and performing creep tests on the multiaxial flat plate specimen under different stresses to obtain a multiaxial creep life.

[0011] Optionally, the materials of the uniaxial flat plate specimen, the multiaxial flat plate specimen, and the double-wall cooling structure are consistent.

[0012] Optionally, a finite element analysis is performed on the finite element model of the double-wall cooling structure based on a continuous damage model with known material parameters to obtain the bone point stress at the bone point on the fracture interface, and a reference stress for creep life prediction is determined based on the bone point stress, including: performing a finite element analysis on the finite element model of the double-wall cooling structure based on a continuous damage model with known material parameters to obtain the stress distribution on the fracture interface of the double-wall cooling structure; based on the stress distribution, determining the bone point stress corresponding to the bone point on the fracture interface where the stress does not change with time, and determining the reference stress for creep life prediction based on the bone point stress.

[0013] Optionally, the above-mentioned neural network-based double-wall cooling structure creep life prediction method also includes: performing a creep test on the double-wall cooling structure simulation part to obtain the actual creep life, wherein the structural dimensions and materials of the double-wall cooling structure simulation part and the double-wall cooling structure finite element model are consistent; comparing the actual creep life with the predicted creep life to obtain the prediction error of the creep life prediction model.

[0014] According to a second aspect of the present disclosure, a neural network-based creep life prediction device for a double-wall cooling structure is provided, comprising: a creep curve acquisition module, for performing creep tests on a uniaxial flat plate specimen and a multiaxial flat plate specimen under different stresses, respectively, to obtain a uniaxial creep curve and a multiaxial creep life; a material parameter determination module, for performing finite element analysis on a uniaxial flat plate specimen finite element model and a multiaxial flat plate specimen finite element model, respectively, based on a continuous damage model, and fitting the uniaxial creep curve and the multiaxial creep life using the results of the finite element analysis to determine the material parameters in the continuous damage model; a reference stress acquisition module, for performing finite element analysis on a finite element model of a double-wall cooling structure based on a continuous damage model with known material parameters, to obtain bone point stress at a bone point on a fracture interface, and to determine a reference stress for creep life prediction based on the bone point stress; and a creep life prediction module, for performing creep life prediction using the reference stress as input to the creep life prediction model, to obtain a predicted creep life of the double-wall cooling structure.

[0015] According to a third aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform any one of the above methods and possible implementations thereof by executing the executable instructions.

[0016] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, any one of the above methods and possible implementations thereof is implemented.

[0017] In the technical solution provided by the present disclosure, finite element analysis is performed on the finite element model of a uniaxial flat plate specimen and the finite element model of a multiaxial flat plate specimen based on a continuous damage model. The results of the finite element analysis are then used to fit the uniaxial creep curve and the multiaxial creep life to determine the material parameters in the continuous damage model. Then, based on the continuous damage model with known material parameters, a finite element analysis is performed on the finite element model of a double-wall cooling structure to obtain the bone point stress at the fracture interface. Based on the bone point stress, a reference stress for creep life prediction is determined. Finally, the reference stress is used as input to the creep life prediction model to perform creep life prediction. On the one hand, using the continuous damage model to obtain the reference stress of the double-wall cooling structure can improve the accuracy of stress data. On the other hand, fitting the results of the finite element analysis with the uniaxial creep curve and the multiaxial creep life to determine the material parameters in the continuous damage model enables the continuous damage model to better describe the damage process of the double-wall cooling structure, thereby improving the accuracy of creep life prediction for the double-wall cooling structure.

[0018] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification, are used to explain the principles of the present disclosure. Obviously, the drawings described below are only some embodiments of the present disclosure, and those skilled in the art can derive other drawings based on these drawings without inventive effort.

[0020] Figure 1 A flow chart of a method for predicting the creep life of a double-wall cooling structure based on a neural network according to an embodiment of the present disclosure is schematically shown.

[0021] Figure 2 A schematic diagram of a double-wall cooling structure simulation component in an embodiment of the present disclosure is shown.

[0022] Figure 3 The figure schematically shows the geometric dimensions of the impact plate simulation component in the embodiment of the present disclosure.

[0023] Figure 4 The schematic diagram of the geometric dimensions of the overflow plate simulation component in the embodiment of the present disclosure is schematically shown.

[0024] Figure 5 A schematic diagram of the fracture path of the double-wall cooling structure in an embodiment of the present disclosure is shown schematically.

[0025] Figure 6 A schematic diagram schematically illustrates the creep life prediction results in an embodiment of the present disclosure.

[0026] Figure 7 A schematic diagram schematically illustrating creep test data in an embodiment of the present disclosure.

[0027] Figure 8 A schematic diagram of a back propagation neural network in an embodiment of the present disclosure is schematically shown.

[0028] Figure 9 The block diagram of the neural network-based double-wall cooling structure creep life prediction device in an embodiment of the present disclosure is schematically shown.

[0029] Figure 10 A block diagram schematically illustrates an electronic device in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0030] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the present disclosure will be more comprehensive and complete and will fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or that other methods, components, devices, steps, etc. may be employed. In other cases, well-known technical solutions are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0031] In addition, the accompanying drawings are merely schematic illustrations of the present disclosure and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0032] Raising turbine inlet temperature is the most direct and effective means of improving aircraft engine thermal efficiency and operating performance. Therefore, current turbine inlet temperatures are designed to exceed the maximum temperature resistance of turbine blade materials. To enable turbine blades to operate normally at such high temperatures, advances in turbine blade cooling technology play a crucial role, in addition to advances in turbine blade materials and thermal barrier coatings. Current cooling structures in turbine blades primarily utilize film holes, which are machined onto the surface of the hollow blade. Cooling gas is ejected from the inside out to form a film, thereby preventing direct impact of high-temperature combustion gas on the blade. This traditional cooling structure effectively improves turbine blade cooling efficiency. However, as performance requirements for gas turbine engines continue to rise, current cooling structures will struggle to achieve higher cooling efficiencies to cope with rising operating temperatures. Therefore, a revolutionary cooling structure is urgently needed to further push the temperature limit of turbine inlet temperatures. Double-wall cooling structures, with their high cooling efficiency, are an effective solution to this technical challenge.

[0033] Specifically, the double-wall cooling structure is a composite cooling structure considered an advanced and promising turbine blade cooling technology. With advancements in casting and processing technology, the potential and possibilities of the double-wall cooling structure are further expanding. The double-wall cooling structure consists of impingement holes, overflow holes, spoiler columns, and inner and outer wall panels connected by the spoiler columns. This structure integrates impingement, convection, and film cooling. The internal air flow is turbulent, absorbing more heat and achieving higher cooling efficiency.

[0034] Although the double-wall cooling structure can bring more efficient cooling efficiency, its disadvantages are also prominent, the most serious of which is the challenge of strength. The impact holes, overflow holes, and spoiler columns connecting the wall panels not only destroy the structural integrity, but also easily produce local multi-axial stress states. Under the action of complex stress, cracks are easily generated, causing structural failure and shortening the service life of the structure. In actual service conditions, since the rotating blades are always subjected to centrifugal loads and are in a high-temperature environment, high-temperature creep is one of the main failure modes of the double-wall cooling structure. Therefore, accurately predicting the creep life of the double-wall cooling structure is crucial to ensuring the normal operation of mechanical components.

[0035] However, traditional creep life prediction models based on damage mechanics are based on numerous assumptions and are limited by complex damage mechanisms or sensitive parameter values. Consequently, they suffer from limited accuracy and cannot accurately predict the creep rupture life of structures, leading to their limited application in engineering. Furthermore, the creep failure of high-temperature structures is influenced by multiple factors. Commonly used empirical creep life prediction formulas based on the time-temperature parameter method, such as the Larson-Miller equation, the Manson-Succop equation, and the Orr-Sherby-Dorn equation, only relate creep life to applied stress and temperature, failing to consider the impact of various factors on the results or the nonlinear coupling relationships between these factors. Consequently, the prediction models have poor generalization capabilities. Furthermore, these empirical creep life prediction formulas are based on a large amount of uniaxial material-level test data (round bar specimens and flat plate specimens). For complex structures like double-walled structures, which are subjected to complex multiaxial stress states, obtaining the relationship between the creep life and time and stress requires a large number of structural-level tests under different temperatures and stress conditions, which is time-consuming and costly.

[0036] In order to solve some or all of the above-mentioned technical problems, the embodiments of the present disclosure provide a method for predicting the creep life of a double-wall cooling structure based on a neural network, which can fully consider the influence of various factors on the creep life prediction results and improve the accuracy of the creep life prediction of the double-wall cooling structure.

[0037] Next, refer to Figure 1As shown in , each step of the creep life prediction method of double-wall cooling structure based on neural network is explained in detail.

[0038] In step S110 , creep tests are performed on the uniaxial flat plate specimen and the multiaxial flat plate specimen under different stresses to obtain uniaxial creep curves and multiaxial creep life.

[0039] In the embodiment of the present disclosure, the double-wall cooling structure can be a cooling structure composed of an inner wall surface and an outer wall surface, which is used to improve the cooling efficiency of mechanical equipment. It can be applied to turbine blades in aircraft engines, and can also be applied to other mechanical equipment working under high-temperature conditions such as gas turbines, steam turbines, chemical reactors, etc.

[0040] In the disclosed embodiment, the uniaxial flat plate specimen may represent a specimen used to measure the uniaxial load characteristics of the same material as the double-wall cooling structure, and the multiaxial flat plate specimen may represent a specimen used to measure the multiaxial stress state of the flat plate made of the same material as the double-wall cooling structure in actual use, so as to determine the material parameters of the subsequent continuous damage model. By performing creep tests under different stresses on the uniaxial flat plate specimen, a uniaxial creep curve can be obtained, and by performing creep tests under different stresses on the multiaxial flat plate specimen, a multiaxial creep life can be obtained. Among them, the uniaxial creep curve can describe the relationship between the strain and creep life of the uniaxial flat plate specimen under different stresses and working temperatures, and the multiaxial creep life can represent the creep life of the multiaxial flat plate specimen under different conditions. For example, creep tests at 980 degrees Celsius can be performed on the uniaxial flat plate specimen and the multiaxial flat plate specimen.

[0041] In some embodiments, creep tests are performed on uniaxial flat plate specimens and multiaxial flat plate specimens under different stresses to obtain uniaxial creep curves and multiaxial creep life, specifically including the following steps: a flat plate specimen without a circular hole in the center is used as a uniaxial flat plate specimen, and creep tests are performed on the uniaxial flat plate specimen under different stresses to obtain a uniaxial creep curve; a flat plate specimen with a circular hole in the center is used as a multiaxial flat plate specimen, and creep tests are performed on the multiaxial flat plate specimen under different stresses to obtain a multiaxial creep life. Preferably, the diameter of the circular hole is 1 mm. Of course, in other embodiments of the present disclosure, the diameter of the circular hole can also be set to other suitable sizes such as 1.5 mm, 2 mm, 4 mm, etc. according to actual needs.

[0042] In step S120, finite element analysis is performed on the uniaxial flat plate specimen finite element model and the multiaxial flat plate specimen finite element model based on the continuous damage model, and the results of the finite element analysis are used to fit the uniaxial creep curve and the multiaxial creep life to determine the material parameters in the continuous damage model.

[0043] In the embodiment of the present disclosure, the continuous damage model can be used to describe the mathematical model of gradual damage and destruction of a double-wall cooling structure under continuous stress or environmental factors, which can more accurately predict the creep characteristics of materials under complex stress conditions. Among them, the continuous damage model can be the Kachanov-Rabotnov continuous damage model, which is mainly suitable for the study of creep damage of metal materials. Of course, in other exemplary embodiments of the present disclosure, the continuous damage model can also be other suitable models such as the Lemaitre model, the Mazars model, and the Chaboche model, which are not specifically limited in this exemplary embodiment.

[0044] In some embodiments, a uniaxial flat plate specimen finite element model and a multiaxial flat plate specimen finite element model are established using the commercial finite element software ABAQUS. Finite element analysis can be performed on each of the uniaxial flat plate specimen finite element model and the multiaxial flat plate specimen finite element model using the Kachanov-Rabotnov continuum damage model. Material parameters in the continuum damage model are obtained by simulating and fitting the uniaxial creep curve and the multiaxial creep life.

[0045] Specifically, the expression of the Kachanov-Rabotnov continuum damage model under uniaxial conditions is:

[0046]

[0047] in is the creep strain rate, σ is the stress, A and n are temperature-related material constants. ω is the damage variable, ranging from [0,1], where ω = 0 means no damage to the material and ω = 1 means fracture of the material. Its rate of change is:

[0048]

[0049] Among them, B, x, and φ are material constants.

[0050] Introducing the multiaxial factor into the uniaxial damage model and considering the combined effect of the maximum principal stress and the Mises equivalent stress, the multiaxial damage model is obtained:

[0051]

[0052] σ r =ασ1+(1-α)σ e (4)

[0053] Where, σ r , σ1, σ e are the fracture stress, maximum principal stress, and Mises equivalent stress, respectively. α is the material constant under multiaxial stress state, and its value range is [0,1].

[0054] Exemplarily, the model parameters of the Kachanov-Rabotnov continuous damage model are simulated and are shown in Table 1 below.

[0055] Table 1

[0056] A B n x φ <![CDATA[1.66875×10 -13 ]]> <![CDATA[6.41904×10 -18 ]]> 4.00087 5.48828 7.0

[0057] Next reference Figure 1 In step S130, a finite element analysis is performed on the finite element model of the double-wall cooling structure based on a continuous damage model with known material parameters to obtain the bone point stress at the bone point on the fracture interface, and a reference stress for creep life prediction is determined based on the bone point stress.

[0058] Specifically, a simulation of a double-wall cooling structure can be designed and established in advance, wherein the physical diagram of the simulation of the double-wall cooling structure is as follows: Figure 2 As shown, Figure 2 The left side is the impact plate simulation part, and the right side is the overflow plate simulation part. Figure 3 is a schematic diagram of the geometric dimensions of the impact plate simulation component. Figure 4 This is a schematic diagram of the geometric dimensions of the overflow plate simulation. A finite element model of the double-wall cooling structure was established based on the physical and dimensional drawings of the double-wall cooling structure simulation. This ensures that the structural dimensions and material parameters of the finite element model are consistent, allowing the finite element model to better reflect the stress characteristics of the double-wall cooling structure. Furthermore, bone points represent points on the load-bearing structure where stress does not change with creep time or the Norton exponent.

[0059] In some embodiments, a finite element analysis is performed on a finite element model of a double-wall cooling structure based on a continuous damage model with known material parameters to obtain the bone point stress at the bone point on the fracture interface, and a reference stress for creep life prediction is determined based on the bone point stress. Specifically, the following steps may be included: a finite element analysis is performed on a finite element model of a double-wall cooling structure based on a continuous damage model with known material parameters to obtain the stress distribution on the fracture interface of the double-wall cooling structure; based on the stress distribution, the bone point stress corresponding to the bone point on the fracture interface where the stress does not change with time is determined, and the reference stress for creep life prediction is determined based on the bone point stress.

[0060] Among them, the stress distribution can represent the distribution of the stress magnitude and direction of each point inside the double-wall cooling structure when it is subjected to external loads. The stress distribution is usually uneven, that is, the stress magnitude and direction at different positions may be different. Therefore, the bone point stress corresponding to the bone point on the fracture interface where the stress does not change with time can be selected to characterize the stress characteristics of the double-wall cooling structure. The bone point stress is substituted into the above multi-axial stress relationship (4) to calculate the reference stress corresponding to the finite element model of the double-wall cooling structure, thereby facilitating the accurate prediction of the creep life of the double-wall cooling structure.

[0061] For example, in actual applications, after performing finite element analysis on the finite element model of the double-wall cooling structure based on a continuous damage model with known material parameters, it was found that the results of the finite element analysis showed that cracks first initiated around the middle overflow hole, and the overflow plate was the first to fracture, causing the double-wall cooling structure to fail. Therefore, the middle fracture surface of the overflow plate can be selected as the dangerous section of the double-wall cooling structure. Figure 5 Schematic diagram of the fracture path of the double-wall cooling structure. Figure 5 As shown in Figure 2, along path 1 and path 2, the Mises stress of the corresponding nodes at different creep times can be obtained. Since path 2 is closer to the edge of the double-wall cooling structure, the stress is larger and the crack propagation distance is shorter, using the bone point stress at this location to calculate the equivalent stress of the double-wall cooling structure, that is, the reference stress, will make the creep life prediction too conservative. For this reason, the maximum principal stress and Mises stress at the bone point on path 1 can be selected, and the reference stress corresponding to the finite element model of the double-wall cooling structure can be calculated using the above multi-axial stress relationship (4).

[0062] Next reference Figure 1 In step S140, the reference stress is used as the input of the creep life prediction model to perform creep life prediction and obtain the predicted creep life of the double-wall cooling structure.

[0063] The creep life prediction model may be a prediction model that takes stress data and operating temperature as input and outputs the creep life of the double-wall cooling structure. Preferably, the creep life prediction model may be a neural network model, such as a back propagation (BP) neural network model, a feedforward neural network model, a radial basis function neural network model, a convolutional neural network model, or other suitable neural network models.

[0064] In some embodiments, before using the reference stress as the input of the creep life prediction model, the creep life prediction model can also be trained through the following steps: the neural network model is trained based on the creep test data corresponding to the target material in the material manual to obtain the creep life prediction model, and the target material is consistent with the material of the double-wall cooling structure.

[0065] Among them, the creep test data can represent the relationship between the operating temperature, stress data and creep life of the double-wall cooling structure.

[0066] Preferably, creep test data can be obtained from the material manual. Specifically, a target material commonly used for turbine blades can be selected as the research object, such as a material commonly used for turbine blades such as cast nickel-based alloys, titanium alloys, alloy steels, and aluminum alloys, and the selected target material is consistent with the material of the double-wall cooling structure. Then, multiple sets of creep test data under different temperature and stress conditions are extracted from the creep life curve of the material in the material manual to form a creep test data set, such as Figure 6 In addition, the creep test data can be divided into a training data set and a test data set. For example, 90% of the data can be classified as a training data set and 10% of the data can be classified as a test data set to train and test the creep life prediction model.

[0067] Furthermore, using neural networks to construct creep life prediction models can account for the impact of multiple factors on prediction results, as well as the nonlinear coupling relationships between these factors. This overcomes the limitations of traditional mechanical models, which are often limited by complex damage mechanisms or sensitive parameter values. Furthermore, model fitting and data mining can be performed based on existing test data, accelerating test design and saving time and costs.

[0068] In some embodiments, a neural network model is trained based on creep test data to obtain a creep life prediction model, which may specifically include the following steps: normalizing the creep test data composed of temperature, stress and creep life to obtain normalized creep test data; using the normalized creep test data as a training set, and using the Bayesian regularization function as a training function to train the back propagation neural network model to obtain a creep life prediction model.

[0069] The Bayesian regularization function is a regularization method used in machine learning and statistics to prevent model overfitting. The backpropagation neural network model is a multi-layer feedforward network that uses the backpropagation learning algorithm. Neurons in adjacent layers are fully connected, while neurons in each layer are disconnected. It is a typical nonlinear algorithm consisting of an input layer, one or more hidden layers, and an output layer, with each layer having one or more neurons.

[0070] Furthermore, in order to avoid the influence of the numerical value of the test data on the model training results, the creep test data can be normalized to control the input eigenvalues ​​between 0 and 1. For example, the Min-Max Normalization method can be used to perform a linear transformation on the original data and map it to the range of [0, 1], as shown in the following formula (5):

[0071] x=(xx min ) / (x ma" -x min ) (5)

[0072] Where x is the original data, x ma" 、x min are the maximum and minimum values ​​in the original data, respectively.

[0073] In some embodiments, a back propagation neural network can be used to construct a creep life prediction model. Specifically, the input layer of the back propagation neural network includes two inputs: operating temperature and stress, and the output of the output layer is creep life. The number of hidden layers and the number of neurons are determined by the optimal training results of the data. The structure of the back propagation neural network with two hidden layers can be as follows: Figure 7 As shown in Figure 2, the learning process consists of two processes: forward propagation of input and back propagation of error. Forward propagation is the process in which information enters the grid through the input layer and the output layer result is obtained through calculation of each layer. The calculation process can be shown in the following equations (6) to (8):

[0074]

[0075] Among them, w ％i , v i+ and u +m are the weights between two adjacent layers, θ i(1,.,1) is the threshold before the adjacent layer, f1 and f . is the activation function between adjacent layers. Backpropagation adjusts weights and thresholds by calculating the error between the output value and the expected value. Forward propagation and error backpropagation repeat over and over again, with weights continuously adjusted, until the error is reduced to an acceptable level, and training ends.

[0076] In some embodiments, since the training data set is relatively small, a Bayesian regularization function may be used as a training function for the creep life prediction model.

[0077] Furthermore, in some embodiments, the square of the Pearson correlation coefficient can be used to evaluate the trained back propagation neural network model to quantify the prediction performance of the model. The closer the square of the Pearson correlation coefficient is to 1, the higher the similarity between the predicted value and the observed value, and the higher the prediction accuracy of the model. The Pearson correlation coefficient is the quotient of the covariance and standard deviation between two features, which can be expressed as the following formula (9):

[0078]

[0079] Among them, y i,678 represents the expected value, y i,9r: Represents the predicted value.

[0080] Exemplarily, the trained back-propagation neural network model has three hidden layers, and the number of neurons in each hidden layer is 5, 4, and 3 respectively.

[0081] In some embodiments, the trained backpropagation neural network model can be tested using a test set to verify the model's fit and prediction capabilities. If the life prediction result is within twice the scatter band, the current backpropagation neural network model can be used as the creep life prediction model. If the error in the test result is greater than an error threshold, the learning algorithm, network structure, and initial weights of the current backpropagation neural network model can be modified, and model training and model testing can continue until the test result error is less than twice the scatter band.

[0082] In some embodiments, error analysis of creep test prediction results can also be performed through the following steps: a creep test is performed on a double-wall cooling structure simulation part to obtain an actual creep life, wherein the structural dimensions and materials of the double-wall cooling structure simulation part and the double-wall cooling structure finite element model are consistent; the actual creep life is compared with the predicted creep life to obtain the prediction error of the creep life prediction model.

[0083] Specifically, a double-wall cooling structure simulation part with the same structural dimensions and materials as the double-wall cooling structure finite element model can be designed and manufactured, and then a creep test can be carried out on the double-wall cooling structure simulation part to obtain the actual creep life of the double-wall cooling structure. At the same time, the actual creep life of the double-wall cooling structure can be compared with the corresponding predicted creep life to calculate the prediction error. The prediction results of creep life are as follows: Figure 8 As shown, it can be concluded that the maximum prediction error of the creep life prediction method in the present disclosure is 9%, and the prediction accuracy is relatively high.

[0084] The neural network-based creep life prediction method for a double-wall cooling structure in the above embodiment can utilize existing material-level creep test data to establish a neural network model between creep life and temperature and stress, without requiring a large amount of double-wall cooling structure creep test data to train the creep life prediction model, thereby saving time and cost; moreover, the equivalent stress of the double-wall cooling structure obtained by the Kachanov-Rabotnov continuous damage model and the bone point method can be used, and the equivalent stress can be brought into the trained network model to achieve accurate prediction of the creep life of the double-wall cooling structure; in addition, the creep life prediction method in the present disclosure can also be applied to the creep life prediction of other complex mechanical structures, such as gas turbines, steam turbines, pressure vessels and other applicable mechanical equipment.

[0085] In the technical solution provided by the present disclosure, finite element analysis is performed on the finite element model of a uniaxial flat plate specimen and the finite element model of a multiaxial flat plate specimen based on a continuous damage model. The results of the finite element analysis are then used to fit the uniaxial creep curve and the multiaxial creep life to determine the material parameters in the continuous damage model. Then, based on the continuous damage model with known material parameters, a finite element analysis is performed on the finite element model of a double-wall cooling structure to obtain the bone point stress at the fracture interface. Based on the bone point stress, a reference stress for creep life prediction is determined. Finally, the reference stress is used as the input of the creep life prediction model to perform creep life prediction. On the one hand, using the continuous damage model to obtain stress data for the double-wall cooling structure can improve the accuracy of the stress data. On the other hand, fitting the results of the finite element analysis with the uniaxial creep curve and the multiaxial creep life to determine the material parameters in the continuous damage model enables the continuous damage model to better describe the damage process of the double-wall cooling structure, thereby improving the accuracy of creep life prediction for the double-wall cooling structure.

[0086] Next, in the embodiment of the present disclosure, a device for predicting the creep life of a double-wall cooling structure based on a neural network is provided. Figure 9 As shown in , a neural network-based double-wall cooling structure creep life prediction device 900 includes: a creep curve acquisition module 901, a material parameter determination module 902, a reference stress acquisition module 903 and a creep life prediction module 904.

[0087] Among them, the creep curve acquisition module 901 can be used to perform creep tests on uniaxial flat plate specimens and multiaxial flat plate specimens under different stresses to obtain uniaxial creep curves and multiaxial creep lives; the material parameter determination module 902 can be used to perform finite element analysis on the uniaxial flat plate specimen finite element model and the multiaxial flat plate specimen finite element model based on the continuous damage model, and use the results of the finite element analysis to fit the uniaxial creep curve and multiaxial creep life to determine the material parameters in the continuous damage model; the reference stress acquisition module 903 can be used to perform finite element analysis on the finite element model of the double-wall cooling structure based on the continuous damage model with known material parameters to obtain the bone point stress at the bone point on the fracture interface, and determine the reference stress for creep life prediction based on the bone point stress; the creep life prediction module 904 can be used to use stress data as input to the creep life prediction model to perform creep life prediction and obtain the predicted creep life of the double-wall cooling structure.

[0088] In an exemplary embodiment of the present disclosure, the turbine cooling blade analysis data interpolation device may also include: a prediction model acquisition module, which can be used to train a neural network model based on creep test data corresponding to the target material in the material manual to obtain a creep life prediction model, and the target material is consistent with the material of the double-wall cooling structure.

[0089] In an exemplary embodiment of the present disclosure, the prediction model acquisition module may include: a normalization processing unit, which can be used to normalize creep test data composed of temperature, stress and creep life to obtain normalized creep test data; a model training unit, which can be used to use the normalized creep test data as a training set and use the Bayesian regularization function as a training function to train the back propagation neural network model to obtain a creep life prediction model.

[0090] In an exemplary embodiment of the present disclosure, the creep curve acquisition module may include: a uniaxial creep curve acquisition unit, which can be used to treat a flat plate specimen without a circular hole in the center as a uniaxial flat plate specimen, and perform creep tests on the uniaxial flat plate specimen under different stresses to obtain a uniaxial creep curve; a multiaxial creep life acquisition unit, which can be used to treat a flat plate specimen with a circular hole in the center as a multiaxial flat plate specimen, and perform creep tests on the multiaxial flat plate specimen under different stresses to obtain a multiaxial creep life.

[0091] In an exemplary embodiment of the present disclosure, the materials of the uniaxial flat plate specimen, the multiaxial flat plate specimen, and the double-wall cooling structure are consistent.

[0092] In an exemplary embodiment of the present disclosure, the reference stress acquisition module may include: a stress distribution acquisition unit, which can be used to perform finite element analysis on a finite element model of a double-wall cooling structure based on a continuous damage model with known material parameters to obtain the stress distribution on the fracture interface of the double-wall cooling structure; a bone point determination unit, which can be used to determine the bone point stress corresponding to the bone point on the fracture interface where the stress does not change with time based on the stress distribution, and determine the reference stress for creep life prediction based on the bone point stress.

[0093] In an exemplary embodiment of the present disclosure, the neural network-based double-wall cooling structure creep life prediction device also includes: a prediction error acquisition unit, which can be used to perform creep tests on the double-wall cooling structure simulation part to obtain the actual creep life, wherein the structural dimensions and materials of the double-wall cooling structure simulation part and the double-wall cooling structure finite element model are consistent; the actual creep life is compared with the predicted creep life to obtain the prediction error of the creep life prediction model.

[0094] It should be noted that the specific details of each part of the above-mentioned neural network-based double-wall cooling structure creep life prediction device have been described in detail in the implementation method of the double-wall cooling structure creep life prediction method based on neural network. The undisclosed details can be found in the implementation content of the method part, and therefore will not be repeated here.

[0095] In an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above method is also provided.

[0096] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods, or program products. Therefore, various aspects of the present disclosure may be implemented in the following forms: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software implementations, which may be collectively referred to herein as "circuits," "modules," or "systems."

[0097] Refer to the following Figure 10 1000 according to this embodiment of the present disclosure will be described. Figure 10 The electronic device 1000 shown is merely an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0098] like Figure 10 As shown, electronic device 1000 is implemented as a general-purpose computing device. Components of electronic device 1000 may include, but are not limited to, the aforementioned at least one processing unit 1010, the aforementioned at least one storage unit 1020, a bus 1030 connecting various system components (including storage unit 1020 and processing unit 1010), and a display unit 1040.

[0099] The storage unit stores program codes, which can be executed by the processing unit 1010, so that the processing unit 1010 performs the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification. For example, the processing unit 1010 can perform the following steps: Figure 1 Follow the steps shown in .

[0100] The storage unit 1020 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 10201 and / or a cache memory unit 10202 , and may further include a read-only memory unit (ROM) 10203 .

[0101] The storage unit 1020 may also include a program / utility 10204 having a set (at least one) of program modules 10205, such program modules 10205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0102] Bus 1030 may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0103] The electronic device 1000 can also communicate with one or more external devices 1100 (e.g., a keyboard, a pointing device, a Bluetooth device, etc.), one or more devices that enable a user to interact with the electronic device 1000, and / or any device that enables the electronic device 1000 to communicate with one or more other computing devices (e.g., a router, a modem, etc.). Such communication can occur via an input / output (I / O) interface 1050. Furthermore, the electronic device 1000 can also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network such as the Internet) via a network adapter 10100. As shown, the network adapter 10100 communicates with other modules of the electronic device 1000 via a bus 1030. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device 1000, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0104] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or an electronic device, etc.) to execute the method according to the embodiments of the present disclosure.

[0105] In exemplary embodiments of the present disclosure, a computer-readable storage medium is also provided, on which is stored a program product capable of implementing the aforementioned methods of this specification. In some possible implementations, various aspects of the present disclosure may also be implemented in the form of a program product comprising program code. When the program product is executed on a terminal device, the program code is configured to cause the terminal device to execute the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of the present disclosure.

[0106] According to an embodiment of the present disclosure, a program product for implementing the above-mentioned method can be a portable compact disc read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present disclosure is not limited thereto. In this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0107] The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0108] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0109] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0110] The program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, and the like, as well as conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a stand-alone software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0111] Furthermore, the above-mentioned figures are merely illustrative of the processes included in the methods according to exemplary embodiments of the present disclosure and are not intended to be limiting. It is readily understood that the processes illustrated in the above-mentioned figures do not indicate or limit the temporal order of these processes. Furthermore, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0112] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow from the general principles of the present disclosure and include common knowledge or customary techniques in the art not invented herein. The specification and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the claims.

[0113] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for predicting creep life of a double-wall cooling structure based on a neural network, characterized in that: include: The creep tests were carried out on uniaxial flat plate specimens and multiaxial flat plate specimens under different stresses, and the uniaxial creep curves and multiaxial creep life were obtained. Based on the continuous damage model, finite element analysis is performed on the uniaxial flat plate specimen finite element model and the multiaxial flat plate specimen finite element model respectively, and the uniaxial creep curve and the multiaxial creep life are fitted using the results of the finite element analysis to determine the material parameters in the continuous damage model; Based on the continuous damage model with known material parameters, a finite element analysis is performed on the finite element model of the double-wall cooling structure to obtain the bone point stress at the bone point on the fracture interface, and based on the bone point stress, a reference stress for creep life prediction is determined; The reference stress is used as the input of the creep life prediction model to perform creep life prediction and obtain the predicted creep life of the double-wall cooling structure.

2. The method according to claim 1, characterized in that Before using the reference stress as the input of the creep life prediction model, the method further includes: The neural network model is trained based on creep test data corresponding to the target material in the material manual to obtain the creep life prediction model, and the target material is consistent with the material of the double-wall cooling structure.

3. The method according to claim 2, characterized in that The method of training the neural network model based on creep test data corresponding to the target material in the material manual to obtain the creep life prediction model includes: Normalizing the creep test data consisting of temperature, stress and creep life to obtain normalized creep test data; The normalized creep test data is used as a training set, and the Bayesian regularization function is used as a training function to train the back propagation neural network model to obtain the creep life prediction model.

4. The method according to claim 1, characterized in that: The creep tests are performed on the uniaxial flat plate specimen and the multiaxial flat plate specimen under different stresses to obtain the uniaxial creep curve and the multiaxial creep life, including: A flat plate specimen without a circular hole in the center is used as the uniaxial flat plate specimen, and creep tests are performed on the uniaxial flat plate specimen under different stresses to obtain the uniaxial creep curve; A flat plate sample with a circular hole in the center is used as the multiaxial flat plate sample, and creep tests are performed on the multiaxial flat plate sample under different stresses to obtain the multiaxial creep life.

5. The method according to claim 1, characterized in that The creep life prediction model is constructed based on a back propagation neural network.

6. The method according to claim 1, characterized in that The continuous damage model based on known material parameters performs finite element analysis on the finite element model of the double-wall cooling structure to obtain the bone point stress at the bone point on the fracture interface, and determines the reference stress for creep life prediction based on the bone point stress, including: Performing finite element analysis on a finite element model of a double-wall cooling structure based on the continuous damage model with known material parameters to obtain stress distribution on a fracture interface of the double-wall cooling structure; Based on the stress distribution, the bone point stress corresponding to the bone point on the fracture interface where the stress does not change with time is determined, and based on the bone point stress, a reference stress for creep life prediction is determined.

7. The method according to claim 1, characterized in that The method further comprises: Conducting a creep test on a double-wall cooling structure simulation part to obtain an actual creep life, wherein the double-wall cooling structure simulation part and the double-wall cooling structure finite element model have the same structural dimensions and materials; The actual creep life is compared with the predicted creep life to obtain a prediction error of the creep life prediction model.

8. A double-wall cooling structure creep life prediction device based on neural network, characterized in that: include: The creep curve acquisition module is used to perform creep tests on uniaxial flat plate specimens and multiaxial flat plate specimens under different stresses to obtain uniaxial creep curves and multiaxial creep life; A material parameter determination module is used to perform finite element analysis on the uniaxial flat plate specimen finite element model and the multiaxial flat plate specimen finite element model respectively based on the continuous damage model, and use the results of the finite element analysis to fit the uniaxial creep curve and the multiaxial creep life to determine the material parameters in the continuous damage model; A reference stress acquisition module, used for performing finite element analysis on a double-wall cooling structure finite element model based on the continuous damage model with known material parameters, obtaining bone point stress at the bone point on the fracture interface, and determining a reference stress for creep life prediction based on the bone point stress; The creep life prediction module is used to use the reference stress as the input of the creep life prediction neural network model to perform creep life prediction and obtain the predicted creep life of the double-wall cooling structure.

9. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 7 by executing the executable instructions.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images