Part creep strain prediction method and system based on grid nodes

Abstract

The invention discloses a part creep strain prediction method and system based on grid nodes. The method comprises the steps of constructing a first creep finite element model of a part under a harsh working condition to obtain full-life-cycle creep process information of each grid node in the model, and extracting a training data set from the full-life-cycle creep process information to train a creep strain prediction model, and constructing a second creep finite element model of the part under the current working condition to obtain previous creep process information for prediction of each grid node in the model, and inputting the previous creep process information for prediction into the trained creep strain prediction model to predict future creep strain information of the part.

Description

Technical Field

[0001] The invention relates to the field of engine component creep strain prediction, and in particular to a component creep strain prediction method and system based on grid nodes. Background Art

[0002] With the development of computer technology, finite element technology has become an important tool for checking the performance, strength, and life of engine components, improving the efficiency of engine design. For engine components under high temperature and heavy load, in addition to strength verification, creep strain analysis is also required in engineering. In order to obtain more accurate creep information, the calculation process often only uses the stress relaxation method. However, this method will greatly increase the calculation time when dealing with long-term creep problems, reducing the efficiency of creep strain analysis. With the continuous development of neural networks and deep learning algorithms, existing researchers have applied deep learning methods to the prediction of creep fatigue life of engine combustion chambers based on existing creep experimental data, using experimental parameters such as temperature, average strain, strain ratio, full load time, and loading speed as input. However, this method requires a large number of experimental samples, and when the geometric characteristic parameters of the research object change, its prediction accuracy will decrease, which is not conducive to the replacement of engine component products in the design stage.

[0003] Therefore, in order to meet the demand for high-efficiency and high-precision creep strain prediction of engine components, it is necessary to propose a creep strain prediction method that can be applied to components with different working conditions and different geometric parameters to improve the efficiency of engine product design. Summary of the invention

[0004] This Summary is provided to introduce some concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.

[0005] In order to overcome the inefficiency of existing creep strain prediction methods for engine parts due to the long time consumption and the requirement of many experimental samples, the present invention provides a creep strain prediction method based on grid nodes. The method uses the historical creep information of grid nodes as input and uses a neural network algorithm to achieve rapid prediction of creep strain of parts.

[0006] Specifically, in one embodiment of the present invention, a method for predicting creep strain of a component based on a grid node is provided, the method comprising:

[0007] Constructing a first creep finite element model of a component under harsh working conditions to obtain full life cycle creep history information of all mesh nodes in the first creep finite element model;

[0008] Extracting a training data set from the full life cycle creep history information to train a creep strain prediction model;

[0009] Constructing a second creep finite element model of the component under the current working condition to obtain past creep history information for prediction of one or more grid nodes to be predicted in the second creep finite element model; and

[0010] The predicted past creep history information is input into the trained creep strain prediction model to predict the future creep strain information of the one or more to-be-predicted mesh nodes of the component.

[0011] In one embodiment of the present invention, the severe working condition and the current working condition include loads and geometric characteristic parameters to be changed, and the full life cycle creep history information and the predicted past creep history information include creep strain size and rate, stress, and temperature.

[0012] In one embodiment of the present invention, the first creep finite element model and the second creep finite element model are constructed using a stress relaxation method.

[0013] In one embodiment of the present invention, extracting a training data set from the full life cycle creep history information further includes: extracting past creep history information for training and a time to be predicted from the full life cycle creep history information; converting the past creep history information for training into standard creep history information for training with a series of time points by an interpolation method; using the standard creep history information for training and the time to be predicted as prediction input; and extracting future creep strain information at the time to be predicted as prediction output, the prediction input and the prediction output constituting the training data set.

[0014] In the above-mentioned embodiment of the present invention, training the creep strain prediction model further includes converting the strain information in the training standard creep history information in the training data set into the logarithmic domain for standardization processing and applying a neural network algorithm to train the creep strain prediction model, wherein the creep strain prediction model is a multi-layer perceptron model.

[0015] In the above-mentioned embodiment of the present invention, inputting the past creep history information for prediction into the trained creep strain prediction model further includes: using an interpolation method to convert the past creep history information for prediction into standard creep history information for prediction having a series of time points consistent with the training standard creep history information; converting the strain information in the standard creep history information for prediction into a logarithmic domain for standardization; and inputting the standard creep history information for prediction together with the time to be predicted as prediction input into the trained creep strain prediction model.

[0016] In another embodiment of the present invention, a system for predicting creep strain of a component based on a mesh node is provided, the system comprising:

[0017] Creep finite element model building module, which is configured to:

[0018] Constructing a first creep finite element model of a component under harsh working conditions to obtain full life cycle creep history information of all mesh nodes in the first creep finite element model; and

[0019] Constructing a second creep finite element model of the component under the current working condition to obtain past creep history information for prediction of one or more grid nodes to be predicted in the second creep finite element model;

[0020] A creep strain prediction model training module, configured to extract a training data set from the full life cycle creep history information to train a creep strain prediction model; and

[0021] The creep strain prediction module is configured to input the predicted past creep history information into a trained creep strain prediction model to predict future creep strain information of the one or more to-be-predicted mesh nodes of the component.

[0022] In one embodiment of the present invention, the creep strain prediction model training module is further configured to extract a training data set from the full life cycle creep history information by the following operations: extracting past creep history information for training and the time to be predicted from the full life cycle creep history information; converting the past creep history information for training into standard creep history information for training with a series of time points by an interpolation method; using the standard creep history information for training and the time to be predicted as prediction input; and extracting future creep strain information at the time to be predicted as prediction output, and the prediction input and the prediction output constitute the training data set.

[0023] In the above-mentioned embodiment of the present invention, the creep strain prediction model training module is further configured to train the creep strain prediction model by converting the strain information in the training standard creep history information in the training data set into the logarithmic domain for standardization processing and applying a neural network algorithm to train the creep strain prediction model, wherein the creep strain prediction model is a multi-layer perceptron model.

[0024] In the above-mentioned embodiment of the present invention, the creep strain prediction module is further configured to input the past creep history information for prediction into the trained creep strain prediction model through the following operations: using an interpolation method to convert the past creep history information for prediction into standard creep history information for prediction having a series of time points consistent with the training standard creep history information; the strain information in the standard creep history information for prediction is converted into a logarithmic domain for standardization; and the standard creep history information for prediction is used together with the time to be predicted as the prediction input of the trained creep strain prediction model.

[0025] In yet another embodiment of the present invention, a computer readable medium storing computer executable instructions is provided, the instructions comprising:

[0026] Instructions for constructing a first creep finite element model of a component under harsh working conditions to obtain full life cycle creep history information of all mesh nodes in the first creep finite element model;

[0027] Instructions for extracting a training data set from the full life cycle creep history information to train a creep strain prediction model;

[0028] An instruction for constructing a second creep finite element model of the component under the current working condition to obtain past creep history information for prediction of one or more grid nodes to be predicted in the second creep finite element model; and

[0029] The instructions are used to input the predicted past creep history information into the trained creep strain prediction model to predict the future creep strain information of the one or more to-be-predicted mesh nodes of the component.

[0030] After studying the description of the specific exemplary embodiments of the present invention below in conjunction with the accompanying drawings, other aspects, features and embodiments of the present invention will be apparent to those of ordinary skill in the art. Although features of the present invention may be discussed below with respect to certain embodiments and accompanying drawings, all embodiments of the present invention may include one or more of the advantageous features discussed herein. In other words, although one or more embodiments may be discussed to have certain advantageous features, one or more of such features may also be used according to the various embodiments of the present invention discussed herein. In a similar manner, although exemplary embodiments may be discussed below as device, system or method embodiments, it should be appreciated that such exemplary embodiments may be implemented in various devices, systems, and methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to understand in detail the manner in which the above-stated features of the present disclosure are used, a more particular description of the content briefly summarized above may be made with reference to various aspects, some of which are illustrated in the accompanying drawings. It should be noted, however, that the accompanying drawings illustrate only certain typical aspects of the present disclosure and are not to be considered limiting of its scope, as the description may admit to other equally effective aspects.

[0032] Figure 1 A schematic block diagram of a system for predicting creep strain of a component based on mesh nodes according to an embodiment of the present disclosure is shown.

[0033] Figure 2 A graph showing a unique assumption of a mesh node creep strain law according to one embodiment of the present disclosure.

[0034] Figure 3 A schematic diagram of constructing a training data set according to an embodiment of the present disclosure is shown.

[0035] Figure 4 A diagram showing a comparison between the predicted value of creep strain corresponding to the model training under severe working conditions and the model prediction under current working conditions and the true value according to an embodiment of the present disclosure is shown.

[0036] Figure 5 A schematic diagram of a creep strain prediction scheme for engine parts based on mesh nodes according to an embodiment of the present disclosure is shown.

[0037] Figure 6 A flow chart of a method for predicting creep strain of a component based on mesh nodes according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0038] The various embodiments will be described in more detail below with reference to the accompanying drawings that form part of the present invention and illustrate various specific exemplary embodiments. However, the various embodiments can be implemented in many different forms and should not be interpreted as limiting the various embodiments set forth herein; on the contrary, these embodiments are provided to make the present disclosure thorough and complete, and to fully convey the scope of these embodiments to those of ordinary skill in the art. The various embodiments may be implemented in accordance with methods, systems, or devices. Therefore, these embodiments may be implemented in hardware, in full software, or in a combination of software and hardware. Therefore, the following specific embodiments are not limiting.

[0039] The steps in each flow chart may be performed by hardware (e.g., processor, engine, memory, circuit), software (e.g., operating system, application, driver, machine / processor executable instructions), or a combination thereof. As will be appreciated by those skilled in the art, the methods involved in each embodiment may include more or fewer steps than shown.

[0040] When analyzing the sensitivity of creep strain of engine parts to geometric parameters and reliability analysis, the stress relaxation method can give the creep strain of each part, but when the calculated creep time is long, it will take a lot of computing time and reduce product design efficiency. Although deep learning or neural network algorithms based on experimental samples can provide good accuracy in predicting creep life, they can often only give the creep strain of a single point and cannot obtain creep information of different parts of the entire part. In addition, this method requires a large number of experimental samples, so there are also efficiency issues.

[0041] To this end, the present invention uses the grid nodes of the finite element model of engine parts as the analysis object to realize the creep strain prediction of the entire parts. The technical problems that need to be solved in this process include:

[0042] (1) The problem of establishing a data set for creep strain prediction of finite element mesh nodes of components;

[0043] (2) The problem of predicting creep strain of components.

[0044] Specifically, in view of the defects in the prior art, the method proposed by the present invention includes using a stress relaxation method to construct a first creep finite element model of a component under harsh working conditions to obtain the full life cycle creep history information of each grid node in the model, extracting a training data set from the full life cycle creep history information to train a creep strain prediction model, constructing a second creep finite element model of the component under other working conditions to obtain the past creep history information for prediction of each grid node in the model, and inputting the past creep history information for prediction into the trained creep strain prediction model to predict the future creep strain information of the component. In addition, the present invention also discloses a system including a device / module for executing each step in the above method.

[0045] The various aspects of the present disclosure will be described in more detail and more comprehensively through block diagrams and method flow charts below.

[0046] Figure 1 FIG. 1 is a schematic block diagram of a system 100 for predicting creep strain of a component based on mesh nodes according to an embodiment of the present disclosure. Figure 1 Based on the combination Figures 2 to 5 Detailed description will be given of various aspects of the system 100 for predicting creep strain of a component based on mesh nodes according to the present invention.

[0047] like Figure 1 As shown, the system 100 includes a creep finite element model building module 102 , a creep strain prediction model training module 104 , and a creep strain prediction module 106 .

[0048] In one embodiment of the present invention, first, the creep finite element model building module 102 may be configured to build a first creep finite element model of a component under harsh working conditions to obtain full life cycle creep history information of all mesh nodes in the first creep finite element model.

[0049] Specifically, combined Figure 5 , Figure 5 A schematic diagram of a creep strain prediction scheme for engine components based on mesh nodes according to an embodiment of the present disclosure is shown. First, at 502, a creep finite element model (i.e., a first creep finite element model) of a component to be analyzed under harsh working conditions is constructed by the creep finite element model construction module 102, and at 504, all mesh nodes in the creep finite element model are calculated at time t 0 ~t n In one embodiment of the present invention, the full life cycle creep history information may include creep strain magnitude and rate, stress, temperature, etc., and in other embodiments of the present invention, it may also include any other suitable creep history information.

[0050] As can be understood by those skilled in the art, in the process of constructing a creep finite element model and determining the creep history information of all mesh nodes in the model over the entire life cycle, the unique creep assumption can be followed, that is: for a certain engine component to be analyzed made of the same material and shape, when the stress, strain, and temperature of the early creep of a certain mesh node in its finite element model under any operating condition are determined, the change law of its later creep strain is uniquely determined.

[0051] Specifically, Figure 2 A graph showing a unique assumption of the creep strain law of a mesh node according to an embodiment of the present disclosure is shown. Figure 2 For an engine component under a certain load, the strain and stress information of any mesh node in the finite element model during the early creep process is as follows: Figure 2 The solid line shows the temperature T 0 When , the strain in the later creep process is unique and obeys Figure 2 The dotted line rule is shown.

[0052] Back to 502, the reason why the creep finite element model construction module 102 can construct the creep finite element model of the component under the harsh working condition is that considering that the creep history information formed by all mesh nodes of the finite element model under the harsh working condition can include the creep history information of the mesh nodes to be predicted under other working conditions (such as the current working condition or the actual working condition), so a relatively harsh working condition is selected here, and in one embodiment of the present invention, the working condition may include the load and the geometric characteristic parameters to be changed. In one embodiment of the present invention, the first creep finite element model and can be constructed by the stress relaxation method, but in other embodiments, any other suitable method can also be used to construct the creep finite element model.

[0053] The creep strain prediction model training module 104 may be configured to extract a training data set from the full life cycle creep history information to train a creep strain prediction model. Figure 5 , continuing to 506, the creep strain prediction model training module 104 can establish a data set of creep history of network nodes.

[0054] Specifically, Figure 3 FIG. 1 shows a schematic diagram of a training data set constructed according to an embodiment of the present disclosure. In one embodiment of the present invention, following the creep uniqueness assumption, as Figure 3 As shown, for the grid node to be predicted, the creep information of the whole life cycle under the harsh working condition is known, and the creep strain prediction model training module 104 can be further configured to extract the past t for training from the creep history information of the whole life cycle 0 ~t i-1 The creep history information at time t (i.e., the past creep history information for training, which may include the creep strain magnitude and creep strain rate) and the time to be predicted t j (i≤j≤n), the past training creep history information is converted into the standard training creep history information with a series of time points through interpolation method, the standard training creep history information and the time to be predicted are taken as input, and the corresponding t j The future creep strain information at the moment is used as output to form a set of data sets with a sample size of n-j+1. Since the time nodes of the creep history information output by the finite element are affected by the creep calculation method, it is necessary to obtain the creep information at a series of moments through an interpolation method to form creep history information that is convenient for training and prediction. As can be understood by those skilled in the art, the series of time points or moments can be set according to actual conditions or needs, and are not limited to any specific time points or moments.

[0055] The same method is used to generate data sets for all other grid nodes and integrate them together. The number of samples in this data set will be the product of the number of grid nodes and the number of moments to be predicted, which will provide a sufficient data set for the neural network algorithm used in the creep strain prediction model.

[0056] In one embodiment of the present invention, in order to improve the prediction accuracy, the creep strain prediction model training module 104 may be further configured to convert the strain information in the training standard creep history information in the training data set into the logarithmic domain for standardization processing (e.g. Figure 5 508 in the figure) and applying a neural network algorithm to train the creep strain prediction model to train the creep strain prediction model (as shown in FIG. Figure 5 In this embodiment of the present invention, the creep strain prediction model (such as Figure 5 512 in the figure) is a multilayer perceptron model, and in other embodiments of the present invention, it may be any other suitable neural network model, without being limited to any specific model. In another embodiment of the present invention, the data set may be divided into a training set and a test set according to a certain ratio, and a neural network algorithm is applied to train the model to obtain a neural network model that can be used for creep strain prediction. As can be understood by those skilled in the art, the normalization process performed on the strain information in the standard creep history information for prediction may include any suitable normalization process, without being limited to any specific processing method.

[0057] Figure 4 A graph showing the comparison between the predicted creep strain values ​​and the true values ​​corresponding to the model training under the harsh working condition and the model prediction under the current working condition according to an embodiment of the present disclosure is shown. In one embodiment of the present invention, the mean square error of the training set and the test set is 5.046×10 -8 ,like Figure 4 As shown in the left figure.

[0058] After obtaining the trained creep strain prediction model, the creep finite element model building module 102 may be configured to build a second creep finite element model (e.g., Figure 5 Specifically, in one embodiment of the present invention, the creep finite element model building module 102 can calculate the creep finite element model of the same component in the past t under the current working condition (other working conditions). 0 ~t i-1 The creep finite element model at time t is used to obtain the creep finite element model of each mesh node in the model at the past t 0 ~t i-1The past creep history information used for prediction at the moment includes creep strain, stress, temperature, etc. In one embodiment of the present invention, the second creep finite element model can be constructed by using a stress relaxation method. In other embodiments, any other suitable method can be used to construct the creep finite element model.

[0059] Then, at 516, the creep finite element model construction module 102 may extract the past creep history information for prediction at past moments of a certain (certain) grid node to be predicted in the second creep finite element model. Specifically, in one embodiment of the present invention, the creep finite element model construction module 102 may construct a second creep finite element model of the same component under other working conditions (such as when it is necessary to predict the future creep strain information of a component under the current working condition, the other working condition is the current working condition) and obtain the past creep history information of each grid node of the model at past moments. 0 ~t i-1 Then, the creep finite element model building module 102 can extract the creep history information of a certain (certain) grid node to be predicted in the past t from the second creep finite element model. 0 -t i-1 The prediction of time uses past creep history information, including creep strain size, creep strain rate, etc.

[0060] At 516 , the creep strain prediction module 106 may use an interpolation method to transform the predicted past creep history information into predicted standard creep history information having a series of time points consistent with the training standard creep history information, so as to be input into the trained standard creep history information.

[0061] At 518 , as at 508 , the creep strain prediction module 106 may also convert the strain information in the predicted standard creep history information into a logarithmic domain for standardization (normalization) for input into the trained creep strain prediction model 512 .

[0062] In one embodiment of the present invention, the creep strain prediction module 106 may be configured to input the past creep history information into the trained creep strain prediction model to predict the future creep strain information of the one or more to-be-predicted mesh nodes of the component. In this embodiment, in addition to converting the past creep history information for prediction into standard creep history information for prediction and the creep strain prediction module 106 in the standard creep history information for prediction may be further configured to input the past creep history information into the trained creep strain prediction model by further using the standard creep history information for prediction together with the to-be-predicted time as the prediction input of the trained creep strain prediction model.

[0063] Combination Figure 5At 520, the creep strain prediction model 512 can be used to predict and output the creep strain information of the grid node to be predicted at a certain time in the future (the time to be predicted). In one embodiment of the present invention, the output predicted by the creep strain prediction model can be converted into real strain information according to the normalized parameters to complete the prediction of creep strain. The final effect is as follows Figure 4 As shown in the right figure, the mean square error between the true value and the -7 , although it is larger than the error when training the model, it still meets the engineering requirements.

[0064] In another embodiment of the present invention, after the creep strain prediction model is trained at 510 , the prediction accuracy may be output at 522 .

[0065] Figure 6 A flowchart of a method 600 for predicting creep strain of a component based on mesh nodes according to an embodiment of the present disclosure is shown.

[0066] like Figure 6 As shown, method 600 starts at step 602, constructing a first creep finite element model of a component under a harsh working condition to obtain the full life cycle creep history information of all mesh nodes in the first creep finite element model. In one embodiment of the present invention, the harsh working condition and the current working condition include loads and geometric characteristic parameters to be changed, and the full life cycle creep history information includes creep strain magnitude and rate, stress, and temperature. In one embodiment of the present invention, the first creep finite element model and the second creep finite element model are constructed using a stress relaxation method.

[0067] Next, method 600 proceeds to step 604, extracting a training data set from the full life cycle creep history information to train a creep strain prediction model. In one embodiment of the present invention, extracting a training data set from the full life cycle creep history information further includes extracting past training creep history information and a time to be predicted from the full life cycle creep history information; converting the past training creep history information into standard training creep history information with a series of time points by an interpolation method; using the standard training creep history information and the time to be predicted as prediction input; and extracting future creep strain information at the time to be predicted as a prediction output, wherein the prediction input and the prediction output constitute the training data set. In one embodiment of the present invention, training the creep strain prediction model further includes converting strain information in the standard training creep history information in the training data set into a logarithmic domain for standardization and applying a neural network algorithm to train the creep strain prediction model, wherein the creep strain prediction model is a multilayer perceptron model.

[0068] Then, the method 600 proceeds to step 606 , where a second creep finite element model of the component under the current working condition is constructed to obtain past creep history information for prediction of one or more to-be-predicted mesh nodes in the second creep finite element model.

[0069] Finally, the method 600 continues to step 608, inputting the past creep history information for prediction into the trained creep strain prediction model to predict the future creep strain information of the one or more grid nodes to be predicted of the component. In one embodiment of the present invention, inputting the past creep history information for prediction into the trained creep strain prediction model further includes: using an interpolation method to convert the past creep history information for prediction into standard creep history information for prediction with a series of time points consistent with the training standard creep history information; converting the strain information in the standard creep history information for prediction into a logarithmic domain for standardization; and inputting the standard creep history information for prediction together with the time to be predicted as prediction input into the trained creep strain prediction model.

[0070] After step 608 , method 600 ends.

[0071] In summary, for the problem of creep strain prediction of engine parts, the existing method is to directly use the working condition as the input parameter, the maximum creep strain or the overall creep performance as the output, and use a neural network or deep learning method for prediction. This method has a large sample size requirement, a long experimental cycle, and high costs. The present invention establishes a creep strain prediction model based on the creep strain history information of the mesh nodes of the finite element model of the parts. Compared with the above method, it has the following innovations:

[0072] (1) The model is trained based on the creep history information of grid nodes, which has stronger generalization ability;

[0073] (2) The sample size of the data set is the product of the number of mesh nodes of the finite element model of the component to be predicted and the number of moments to be predicted. A sufficient training set can be obtained after one creep finite element analysis, which greatly shortens the time for obtaining creep samples.

[0074] (3) By extracting the creep strain history information of any grid node in the finite element creep calculation model of components under other working conditions, the creep strain of the node at a certain moment in the future can be quickly predicted.

[0075] Embodiments of the present invention have been described above with reference to block diagrams and / or the operating instructions of methods, systems, and computer program products according to embodiments of the present invention. Each function / action noted in the frame may occur in an order different from that shown in any flow chart. For example, depending on the function / action involved, two frames shown in succession may actually be performed substantially simultaneously, or these frames may sometimes be performed in an opposite order.

[0076] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for predicting creep strain of a component based on a mesh node, the method comprising: Constructing a first creep finite element model of a component under harsh working conditions to obtain full life cycle creep history information of all mesh nodes in the first creep finite element model; Extracting a training data set from the full life cycle creep history information to train a creep strain prediction model; Constructing a second creep finite element model of the component under the current working condition to obtain past creep history information for prediction of one or more grid nodes to be predicted in the second creep finite element model; as well as The predicted past creep history information is input into a trained creep strain prediction model to predict future creep strain information of the one or more to-be-predicted mesh nodes of the component.

2. The method of claim 1, wherein the severe working condition and the current working condition include loads and geometric characteristic parameters to be changed, and the creep history information includes creep strain magnitude and rate, stress, and temperature. 3 . The method according to claim 1 , wherein the first creep finite element model and the second creep finite element model are constructed using a stress relaxation method.

4. The method of claim 1, wherein extracting a training data set from the full life cycle creep history information further comprises: Extracting past creep history information for training and the time to be predicted from the full life cycle creep history information; Converting the past creep history information for training into standard creep history information for training with a series of time points by an interpolation method; Using the training standard creep history information and the time to be predicted as prediction input; as well as extracting the future creep strain information at the time to be predicted as the prediction output, The prediction input and the prediction output constitute the training data set.

5. The method of claim 4, wherein training the creep strain prediction model further comprises converting strain information in the training standard creep history information in the training data set into a logarithmic domain for standardization and applying a neural network algorithm to train the creep strain prediction model, wherein the creep strain prediction model is a multi-layer perceptron model.

6. The method of claim 4, wherein inputting the predicted past creep history information into the trained creep strain prediction model further comprises: Using an interpolation method to convert the past creep history information for prediction into standard creep history information for prediction having time points consistent with the series of time points in the standard creep history information for training; Converting the strain information in the predicted standard creep history information into a logarithmic domain for standardization; as well as The prediction standard creep history information and the time to be predicted are input as prediction inputs into the trained creep strain prediction model.

7. A system for predicting creep strain of a component based on a mesh node, the system comprising: Creep finite element model building module, which is configured to: Constructing a first creep finite element model of a component under harsh working conditions to obtain full life cycle creep history information of all mesh nodes in the first creep finite element model; as well as Constructing a second creep finite element model of the component under the current working condition to obtain past creep history information for prediction of one or more grid nodes to be predicted in the second creep finite element model; A creep strain prediction model training module, configured to extract a training data set from the full life cycle creep history information to train a creep strain prediction model; as well as The creep strain prediction module is configured to input the predicted past creep history information into a trained creep strain prediction model to predict future creep strain information of the one or more to-be-predicted mesh nodes of the component.

8. The system of claim 7, wherein the creep strain prediction model training module is further configured to extract a training data set from the full life cycle creep history information by: Extracting past creep history information for training and the time to be predicted from the full life cycle creep history information; Converting the past creep history information for training into standard creep history information for training with a series of time points by an interpolation method; Using the training standard creep history information and the time to be predicted as prediction input; and extracting the future creep strain information at the time to be predicted as the prediction output, The prediction input and the prediction output constitute the training data set.

9. The system of claim 8, wherein the creep strain prediction model training module is further configured to train the creep strain prediction model by converting strain information in the training standard creep history information in the training data set into a logarithmic domain for standardization and applying a neural network algorithm to train the creep strain prediction model, wherein the creep strain prediction model is a multi-layer perceptron model.

10. The system of claim 8, wherein the creep strain prediction module is further configured to input the predicted past creep history information into the trained creep strain prediction model by: Using an interpolation method to convert the past creep history information for prediction into standard creep history information for prediction having time points consistent with the series of time points in the standard creep history information for training; The strain information in the prediction standard creep history information is converted into a logarithmic domain for standardization; and The prediction standard creep history information and the time to be predicted are used as prediction inputs of the trained creep strain prediction model.

Images