Training method of fatigue life prediction model for offshore wind power axial load grouting joint and fatigue life prediction method

Abstract

The application is suitable for the field of offshore wind power technology, and provides a training method and a fatigue life prediction method for a fatigue life prediction model of an offshore wind power axial bearing grouting joint. The training method comprises the following steps: obtaining a training data set, determining a predicted fatigue life and a predicted fatigue relationship coefficient by using a preset fatigue life prediction model according to node label data in the training data set, determining predicted loss data based on the predicted fatigue life and the predicted fatigue relationship coefficient, training the preset fatigue life prediction model according to the predicted loss data, and obtaining a trained fatigue life prediction model. The method can improve the prediction accuracy of the trained fatigue life prediction model in the case of insufficient training sample data, insufficient model input feature data and high data dispersion, and further realizes efficient and reliable prediction of the fatigue life of the grouting connection under the axial loading condition.

Description

Technical Field

[0001] This application belongs to the field of offshore wind power technology, and in particular relates to a training method and a fatigue life prediction method for a fatigue life prediction model of an axial load grouting node in offshore wind power. Background Technology

[0002] Grouting connection is a key force-transmitting component connecting the superstructure of offshore wind turbines to the foundation (such as...). Figure 1 As shown in the diagram, high-strength grouting material is used to fill the annular gap in the steel structure, forming a composite stress connection that firmly connects the wind turbine tower to the foundation. During the design life of offshore wind power structures (generally 20-25 years), grouting connections need to withstand the coupling effects of multi-source cyclic loads such as wind, waves, currents, and wind turbines for a long period of time, making fatigue failure the main potential failure mode.

[0003] Fatigue testing of grouting connections mainly relies on experiments and finite element analysis, using fatigue tests and test data to assess stress distribution and lifespan. However, full-scale tests are prohibitively expensive due to size limitations and the sheer number of tests (over a hundred million). Furthermore, empirical formulas and finite element analysis not only require enormous computational resources but also struggle to fully account for nonlinear behavior under complex working conditions.

[0004] Currently, predicting the fatigue life of grouting connections faces significant challenges, including insufficient data, limited input features, high data dispersion, low prediction accuracy, and limited efficiency. Therefore, an efficient and reliable fatigue life prediction method is needed. Summary of the Invention

[0005] This application provides a training method and a fatigue life prediction method for a fatigue life prediction model of grouted joints with axial load in offshore wind power, which can improve the accuracy of fatigue life prediction for grouted joints.

[0006] In a first aspect, embodiments of this application provide a training method for a fatigue life prediction model of grouted joints under axial load in offshore wind power, comprising:

[0007] Obtain a training dataset, which includes node label data corresponding to each of multiple axial bearing grouting nodes;

[0008] Based on the node label data, a preset fatigue life prediction model is used to predict the fatigue life and fatigue relationship coefficient of the axial bearing grouting node, and the predicted fatigue life and predicted fatigue relationship coefficient are obtained.

[0009] Based on the predicted fatigue life and the predicted fatigue relationship coefficient, the predicted loss data of the preset fatigue life prediction model is determined.

[0010] Based on the predicted loss data, the preset fatigue life prediction model is trained to obtain the trained fatigue life prediction model.

[0011] In some embodiments, the node label data includes geometric structure data, material characterization data, and environmental loading data. The step of predicting the fatigue life and fatigue relationship coefficient of the axially loaded grouting node using a preset fatigue life prediction model based on the node label data, to obtain the predicted fatigue life and predicted fatigue relationship coefficient, includes: performing a correlation analysis on the geometric structure data, the material characterization data, and the environmental loading data to obtain a correlation analysis result; filtering target node label data for the axially loaded grouting node from the node label data based on the correlation analysis result; and using the preset fatigue life prediction model to predict the fatigue life and fatigue relationship coefficient of the axially loaded grouting node based on the target node label data, to obtain the predicted fatigue life and predicted fatigue relationship coefficient.

[0012] In some embodiments, the preset fatigue life prediction model includes a preset fatigue life prediction network and a preset fatigue relationship prediction network. The step of predicting the fatigue life and fatigue relationship coefficient of the axially bearing grouting node using the preset fatigue life prediction model based on the target node label data to obtain the predicted fatigue life and predicted fatigue relationship coefficient includes: predicting the fatigue life of the axially bearing grouting node using the preset fatigue life prediction network based on the target node label data to obtain the predicted fatigue life; and predicting the corresponding fatigue relationship coefficient of the axially bearing grouting node in the high-cycle fatigue region using the preset fatigue relationship prediction network based on the target node label data to obtain the predicted fatigue relationship coefficient.

[0013] In some embodiments, determining the predicted loss data of the preset fatigue life prediction model based on the predicted fatigue life and the predicted fatigue relationship coefficient includes: extracting the actual fatigue life of the axially bearing grouting node from the training dataset, and comparing the predicted fatigue life with the actual fatigue life to obtain a fatigue life comparison result; determining the fatigue life loss data corresponding to the axially bearing grouting node based on the fatigue life comparison result; determining the residual loss data corresponding to the axially bearing grouting node based on the predicted fatigue life and the predicted fatigue relationship coefficient; and determining the predicted loss data of the preset fatigue life prediction model based on the fatigue life loss data and the residual loss data.

[0014] In some embodiments, determining the residual loss data corresponding to the axial load grouting node based on the predicted fatigue life and the predicted fatigue relationship coefficient includes: extracting node residual description data for constructing residual terms from the node label data of the axial load grouting node; determining first residual loss data corresponding to the axial load grouting node based on the predicted fatigue life and the node residual description data; determining second residual loss data corresponding to the axial load grouting node based on the predicted fatigue life, the node residual description data, and the predicted fatigue relationship coefficient; and determining the residual loss data corresponding to the axial load grouting node based on the first residual loss data and the second residual loss data.

[0015] Secondly, embodiments of this application provide a training device for a fatigue life prediction model of grouted joints under axial load in offshore wind power, comprising:

[0016] The first acquisition unit is used to acquire a training dataset, which includes node label data corresponding to each of multiple axial bearing grouting nodes.

[0017] The first prediction unit is used to predict the fatigue life and fatigue relationship coefficient of the axial bearing grouting node based on the node label data and a preset fatigue life prediction model, so as to obtain the predicted fatigue life and predicted fatigue relationship coefficient.

[0018] The loss determination unit is used to determine the predicted loss data of the preset fatigue life prediction model based on the predicted fatigue life and the predicted fatigue relationship coefficient.

[0019] The model training unit is used to train the preset fatigue life prediction model based on the predicted loss data to obtain the trained fatigue life prediction model.

[0020] Thirdly, embodiments of this application provide a fatigue life prediction method for grouted joints bearing axial loads in offshore wind power, including:

[0021] Obtain initial node description data for axial bearing grouting nodes, the initial node description data including geometric structure data, material characterization data and environmental loading data;

[0022] A correlation analysis was performed on the geometric structure data, the material characterization data, and the environmental loading data to obtain the correlation analysis results.

[0023] Based on the correlation analysis results, target node description data for the axial bearing grouting node is selected from the initial node description data.

[0024] Based on the target node description data, the fatigue life of the axial bearing grouting node is predicted using the aforementioned fatigue life prediction model, thereby obtaining the fatigue life of the axial bearing grouting node.

[0025] Fourthly, embodiments of this application provide a fatigue life prediction device for grouting joints with axial loads in offshore wind power projects, comprising:

[0026] The second acquisition unit is used to acquire the initial node description data of the axial bearing grouting node, the initial node description data including geometric structure data, material characterization data and environmental loading data;

[0027] The data analysis unit is used to perform correlation analysis on the geometric structure data, the material characterization data, and the environmental loading data to obtain the correlation analysis results.

[0028] A data filtering unit is used to filter out the target node description data of the axial bearing grouting node from the initial node description data based on the correlation analysis results.

[0029] The second prediction unit is used to predict the fatigue life of the axial bearing grouting node based on the target node description data and using a fatigue life prediction model, so as to obtain the fatigue life of the axial bearing grouting node.

[0030] Fifthly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the above-described training methods and fatigue life prediction methods.

[0031] Sixthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by an electronic device, causes the electronic device to implement the training method and fatigue life prediction method described above.

[0032] In a seventh aspect, embodiments of this application provide a computer program product that, when run on a terminal device, causes the terminal device to execute the training method and fatigue life prediction method described in any one of the first aspects.

[0033] It is understood that the beneficial effects of the second to seventh aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.

[0034] The beneficial effects of the embodiments in this application compared with the prior art are:

[0035] In the training process of the preset fatigue life prediction model, this application embodiment simultaneously predicts the fatigue life and fatigue relationship coefficient of the axially loaded grouting node, obtaining the predicted fatigue life and predicted fatigue relationship coefficient. The predicted fatigue life characterizes the mapping relationship between the node label data and the predicted fatigue life during the prediction process, while the predicted fatigue relationship coefficient characterizes the implicit constants corresponding to the physical constraints (including simple physical trends and implicit physical laws) during the prediction process. Based on this, the predicted fatigue life and predicted fatigue relationship coefficient are used together to construct the prediction loss data of the preset fatigue life prediction model. The preset fatigue life prediction model is then trained based on the prediction loss data, enabling the trained fatigue life prediction model to quickly capture the life change trend and improve the physical consistency and generalization ability of the prediction. This improves the prediction accuracy of the trained fatigue life prediction model even when training sample data is insufficient, model input feature data is limited, and data dispersion is high, thereby achieving efficient and reliable prediction of the fatigue life of grouting connections under axial loading conditions. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a schematic diagram of the basic structure of an offshore wind turbine;

[0038] Figure 2 This is a schematic diagram illustrating an application scenario of a training method for a fatigue life prediction model of an axial load grouting node in offshore wind power, as provided in an embodiment of this application.

[0039] Figure 3 This is a flowchart illustrating a training method for a fatigue life prediction model of an axial load grouting node for offshore wind power, as provided in an embodiment of this application.

[0040] Figure 4A This is a schematic diagram of the geometric structure of an axial load-bearing grouting node used in offshore wind power.

[0041] Figure 4B This is a schematic diagram of another geometric structure for an axial load-bearing grouting node used in offshore wind power.

[0042] Figure 5 This is another flowchart illustrating the training method for the fatigue life prediction model of the axial load grouting node of offshore wind power provided in the embodiments of this application.

[0043] Figure 6 This is a schematic diagram of the model structure of the preset fatigue life prediction model provided in the embodiments of this application;

[0044] Figure 7 These are schematic diagrams of SN curves corresponding to different industry standards;

[0045] Figure 8 This is a schematic diagram illustrating an application scenario of a fatigue life prediction method for axial load grouting nodes in offshore wind power, as provided in an embodiment of this application.

[0046] Figure 9 This is a flowchart illustrating a method for predicting the fatigue life of grouting nodes for axial loads in offshore wind power, as provided in an embodiment of this application.

[0047] Figure 10 This is a schematic diagram of the structure of the training device for the fatigue life prediction model of the axial load grouting node of offshore wind power provided in the embodiments of this application.

[0048] Figure 11 This is a schematic diagram of the fatigue life prediction device for axial load grouting nodes in offshore wind power provided in the embodiments of this application;

[0049] Figure 12 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0050] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0051] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0052] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0053] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0054] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0055] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0056] This application provides a training method, a fatigue life prediction method, and related equipment for a fatigue life prediction model of grouted joints under axial load in offshore wind turbines. The related equipment includes a training device for the fatigue life prediction model of grouted joints under axial load in offshore wind turbines, a fatigue life prediction device for grouted joints under axial load in offshore wind turbines, electronic equipment, a computer program product, and a computer-readable storage medium. The training device for the fatigue life prediction model of grouted joints under axial load in offshore wind turbines can be integrated into an electronic device, which can be a server or a terminal, etc.

[0057] The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud-preset databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery network (CDN), and big data and artificial intelligence platforms. The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these.

[0058] Figure 2 This illustration shows an application scenario diagram of a training method for a fatigue life prediction model of an axial load grouting node in offshore wind power, as provided in an embodiment of this application. For example... Figure 2 As shown, taking the training device for the fatigue life prediction model of axial load grouting nodes in offshore wind power as an example, the electronic device is integrated into an electronic device, which is a server. The server can obtain the training dataset, which includes node label data corresponding to each of the multiple axial load grouting nodes; based on the node label data, the preset fatigue life prediction model is used to predict the fatigue life and fatigue relationship coefficient of the axial load grouting nodes, obtaining the predicted fatigue life and predicted fatigue relationship coefficient; based on the predicted fatigue life and predicted fatigue relationship coefficient, the prediction loss data of the preset fatigue life prediction model is determined; based on the prediction loss data, the preset fatigue life prediction model is trained to obtain the trained fatigue life prediction model.

[0059] The following sections provide detailed descriptions of each example. It should be noted that the order in which the embodiments are described is not intended to limit the order of the embodiments.

[0060] Figure 3 This illustration shows a flowchart of a training method for a fatigue life prediction model of an axially loaded grouting node in offshore wind power, as provided in an embodiment of this application. Figure 3 As shown, the specific process of training the fatigue life prediction model for grouted joints with axial load in offshore wind power is as follows:

[0061] S101. Obtain the training dataset, which includes node label data corresponding to each of the multiple axial bearing grouting nodes.

[0062] In this embodiment, the number of node label data included in the training dataset can be determined based on the actual data collection situation. For example, the training dataset may include 50, 60, 70, or other numbers of samples, each sample corresponding to the node label data of an axial bearing grouting node.

[0063] For any axial load grouting node used in offshore wind power, the corresponding node label data may include geometric structure data, material characterization data, and environmental loading data.

[0064] Figure 4A A schematic diagram of the geometric structure of an axially loaded grouting joint for offshore wind power is shown. Figure 4B A schematic diagram of another geometric construction for an axially loaded grouting joint used in offshore wind power is shown. (See diagram below.) Figure 4A and Figure 4B As shown, the geometric data of the axial load-bearing grouting node may include the sleeve diameter (D). s), sleeve thickness (t) s ), pile diameter (D) p ), pile thickness (t) p Grouting thickness (t) g ), shear key width (w), grouting length (L) g ), the ratio of shear key height to spacing (h / s).

[0065] Material characterization data for axially loaded grouting joints may include the elastic modulus of the sleeve (E). s ), pile elastic modulus (E) p ), sleeve yield strength (σ Y_s Grouting compressive strength (UCD) g ), the yield strength of the pile (σ) Y_p ), grouting Poisson's ratio (v g Poisson's ratio of steel (v) s ).

[0066] The environmental loading data for axially loaded grouting nodes can be understood as the loading conditions of the axially loaded grouting nodes. Environmental loading data may include: loading frequency (f), minimum load (P)... min ), loading environment (env), ratio of maximum load to static interface shear strength (P) max / P static ).

[0067] S102. Based on the node label data, a preset fatigue life prediction model is used to predict the fatigue life and fatigue relationship coefficient of the axial bearing grouting node, and the predicted fatigue life and predicted fatigue relationship coefficient are obtained.

[0068] In this embodiment of the application, node label data can be input into a preset fatigue life prediction model so that the fatigue life and fatigue relationship coefficient of the axial bearing grouting node can be predicted through the preset fatigue life prediction model, and the predicted fatigue life and predicted fatigue relationship coefficient can be obtained.

[0069] Among them, the predicted fatigue life is the fatigue life of the axially loaded grouting node predicted by a preset fatigue life prediction model. The predicted fatigue life can characterize the mapping relationship between the node label data and the predicted fatigue life during the prediction process. The predicted fatigue relationship coefficient can be understood as an implicit constant that represents the physical constraints (including simple physical trends and implicit physical laws) during the prediction process, obtained by the preset fatigue life prediction model.

[0070] Specifically, based on the node label data, a preset fatigue life prediction model is used to predict the fatigue life and fatigue relationship coefficient of the axially loaded grouting node, obtaining the predicted fatigue life and predicted fatigue relationship coefficient. This process may include: performing correlation analysis on geometric structure data, material characterization data, and environmental loading data to obtain the correlation analysis results; based on the correlation analysis results, selecting target node label data for the axially loaded grouting node from the node label data; and based on the target node label data, using a preset fatigue life prediction model to predict the fatigue life and fatigue relationship coefficient of the axially loaded grouting node, obtaining the predicted fatigue life and predicted fatigue relationship coefficient.

[0071] In this embodiment of the application, correlation analysis can be performed on geometric structure data, material characterization data and environmental loading data based on Pearson correlation coefficient, and redundant features in node label data can be eliminated based on the correlation analysis results, thereby selecting the target node label data of axial bearing grouting nodes.

[0072] The Pearson correlation coefficient is used to measure the degree of linear correlation between two variables. In this embodiment, the Pearson correlation coefficient can be calculated between each pair of data covered by geometric construction data, material characterization data, and environmental loading data to obtain the degree of linear correlation between the data. Data with a linear correlation greater than a preset threshold is used as target node label data. At the same time, data with a linear correlation less than or equal to the preset threshold is removed.

[0073] It is understood that the preset threshold can be set according to the actual situation, and this application embodiment does not impose any restrictions on it.

[0074] Figure 5 This illustration shows another flowchart of the training method for a fatigue life prediction model of an axial load grouting node in offshore wind power, as provided in an embodiment of this application. Figure 5 As shown, based on the correlation analysis results, the target node label data corresponding to the axial bearing grouting nodes selected from the node label data can include: grouting thickness (t). g ), shear key height to spacing ratio (h / s), minimum load (P) min ), loading frequency (f), loading environment (env), and the ratio of maximum load to static interface shear strength (P) max / P static ).

[0075] After obtaining the target node label data, the target node label data can be stratified and randomly partitioned, and then normalized to obtain normalized target node label data. Specifically, the input features can be standardized using the z-score method, and the standardized data can then be input into a preset fatigue life prediction model.

[0076] Then, the normalized target node label data can be used as the model input data for the preset fatigue life prediction model. The preset fatigue life prediction model is then used to predict the fatigue life and fatigue relationship coefficient of the axially bearing grouting node, thus obtaining the predicted fatigue life and predicted fatigue relationship coefficient. The preset fatigue life prediction model is... Figure 5 In the physical information neural network, the normalized target node label data is... Figure 5 The normalized features are then analyzed. Furthermore, the data output by the preset fatigue life prediction model can be logarithmically transformed to obtain the predicted fatigue life and the coefficients relating it to fatigue prediction.

[0077] Figure 6 A schematic diagram of the model structure of the preset fatigue life prediction model provided in an embodiment of this application is shown. Figure 6 As shown, the preset fatigue life prediction model may include a preset fatigue life prediction network and a preset fatigue relationship prediction network.

[0078] The preset fatigue life prediction network can be understood as the main network of the preset fatigue life prediction model. It contains five hidden layers, each consisting of 128 neurons, used to map the six-dimensional input features to a single fatigue life (N) output. The activation function of the preset fatigue life prediction network can be Tanh, or other functions (such as Sigmoid, ReLU, LeakyReLU, pReLU, ELU, maxout, etc.).

[0079] The pre-defined fatigue relationship prediction network can be understood as an auxiliary network for the pre-defined fatigue life prediction model. It constrains the output of the network through two types of physical constraints (simple physical trends and implicit physical laws). The network consists of three hidden layers, each with 128 neurons, used to regress parameters A and B in the Basquin relation (used in fatigue mechanics to describe the behavior of materials in the high-cycle fatigue region). Parameters A and B are implicit constants in the Basquin relation. The activation function of the network can be Tanh, or other functions (such as Sigmoid, ReLU, LeakyReLU, pReLU, ELU, maxout, etc.).

[0080] Based on this, using the target node label data, a preset fatigue life prediction model is used to predict the fatigue life and fatigue relationship coefficient of the axially bearing grouting node, resulting in the predicted fatigue life and predicted fatigue relationship coefficient. This can include: using the target node label data, a preset fatigue life prediction network is used to predict the fatigue life of the axially bearing grouting node, resulting in the predicted fatigue life; using the target node label data, a preset fatigue relationship prediction network is used to predict the corresponding fatigue relationship coefficient of the axially bearing grouting node in the high-cycle fatigue region, resulting in the predicted fatigue relationship coefficient.

[0081] In this embodiment, both the preset fatigue life prediction network and the preset fatigue relationship prediction network use target node label data as input data. As mentioned above, the target node label data includes grouting thickness (t). g ), shear key height to spacing ratio (h / s), minimum load (P) min ), loading frequency (f), loading environment (env), and the ratio of maximum load to static interface shear strength (P) max / P static These six data points.

[0082] For a preset fatigue life prediction network, it can predict the fatigue life of axial bearing grouting nodes based on input data.

[0083] For a pre-defined fatigue relationship prediction network, the fatigue life of axially loaded grouting joints can be predicted based on simple physical trends, thus obtaining the predicted fatigue life. The simple physical trends are determined based on the SN curves provided in existing industry standards for fatigue life prediction of offshore wind turbine grouting connections (such as DNV-ST-0126 and DIN EN ISO 19902). The SN curve, also known as the stress-life curve, is a curve representing the relationship between the stress amplitude or load level (S) experienced by a material and the number of cycles (N) experienced at that stress level until fatigue fracture occurs. The number of cycles represents the fatigue life. Figure 7 This shows a schematic diagram of the SN curves corresponding to different industry standards. For example... Figure 7 As shown, fatigue life N decreases with increasing load level S, and the rate of decrease is faster at high load levels. This trend can be simplified by the derivative properties of the curve, i.e., the first derivative is less than zero, and the second derivative is greater than zero. The relevant expressions are as follows:

[0084]

[0085]

[0086] For the preset fatigue relationship prediction network, it can also predict the corresponding fatigue relationship coefficients of the axially loaded grouting nodes in the high-cycle fatigue region based on implicit physical laws, obtaining predicted fatigue relationship coefficients A and B. In this embodiment, the Basquin relationship can be embedded as an implicit physical law into the training process of the preset fatigue life prediction model. In each forward propagation, the prediction result of the preset fatigue relationship prediction network must satisfy the approximate constraint of the Basquin power function form. The expression of the Basquin power function form is as follows:

[0087]

[0088] Where A and B are implicit constants that depend on the geometry (data), material characterization (data), and environmental loading (data) of the axially loaded grouting node.

[0089] S103. Based on the predicted fatigue life and the predicted fatigue relationship coefficient, determine the predicted loss data of the preset fatigue life prediction model.

[0090] After obtaining the predicted fatigue life of the axially loaded grouting node using a preset fatigue life prediction network and the predicted fatigue relationship coefficient of the axially loaded grouting node in the high-cycle fatigue region using a preset fatigue relationship prediction network, the predicted loss data of the preset fatigue life prediction model can be determined based on the predicted fatigue life and the predicted fatigue relationship coefficient. Specifically, determining the predicted loss data of the preset fatigue life prediction model based on the predicted fatigue life and the predicted fatigue relationship coefficient can include: extracting the actual fatigue life of the axially loaded grouting node from the training dataset and comparing the predicted fatigue life with the actual fatigue life to obtain a fatigue life comparison result; determining the fatigue life loss data corresponding to the axially loaded grouting node based on the fatigue life comparison result; determining the residual loss data corresponding to the axially loaded grouting node based on the predicted fatigue life and the predicted fatigue relationship coefficient; and determining the predicted loss data of the preset fatigue life prediction model based on the fatigue life loss data and the residual loss data.

[0091] The training dataset also includes the actual fatigue lives of multiple axially loaded grouting nodes, allowing the extraction of these actual fatigue lives. Then, for any given axially loaded grouting node, its predicted fatigue life can be compared with its actual fatigue life to obtain a fatigue life comparison result. Based on this comparison result, the fatigue life loss data for that axially loaded grouting node can be determined.

[0092] Since the predicted fatigue life is obtained using a preset fatigue life prediction network, the fatigue life loss data can reflect the difference between the predicted value and the actual value of the preset fatigue life prediction network, thereby reflecting the prediction accuracy of the preset fatigue life prediction network. In this embodiment, the fatigue life loss data can be taken as the mean square error between the logarithm of the actual fatigue life and the logarithm of the predicted fatigue life, and the specific calculation formula is as follows:

[0093]

[0094] Where n represents the sample number (i.e., the number of the axial bearing grouting node), N true N represents the actual fatigue life. pred Representation and prediction of fatigue life, Loss mse This represents fatigue life loss data.

[0095] In this embodiment, residual loss data corresponding to the axial bearing grouting node can also be determined based on the predicted fatigue life and the predicted fatigue relationship coefficient. Residual loss data can be understood as loss data constructed after introducing a preset fatigue relationship prediction network, taking into account both simple physical trends and hidden physical laws. The residual loss data can assist the fatigue life loss data in constructing the predicted loss data for the entire preset fatigue life prediction model, so that the entire preset fatigue life prediction model can be updated in reverse based on the predicted loss data until the model converges, resulting in the trained fatigue life prediction model.

[0096] Among them, the residual loss data and fatigue life loss data can be added together to obtain the predicted loss data for determining the preset fatigue life prediction model.

[0097] In this embodiment of the application, determining the residual loss data corresponding to the axial load grouting node based on the predicted fatigue life and the predicted fatigue relationship coefficient may include: extracting node residual description data for constructing residual terms from the node label data of the axial load grouting node; determining the first residual loss data corresponding to the axial load grouting node based on the predicted fatigue life and the node residual description data; determining the second residual loss data corresponding to the axial load grouting node based on the predicted fatigue life, the node residual description data, and the predicted fatigue relationship coefficient; and determining the residual loss data corresponding to the axial load grouting node based on the first residual loss data and the second residual loss data.

[0098] As mentioned earlier, residual loss data is loss data constructed by comprehensively considering simple physical trends and hidden physical laws. Therefore, a first residual loss data can be constructed based on simple physical trends, a second residual loss data can be constructed based on hidden physical laws, and then the first and second residual loss data can be merged to obtain the final residual loss data.

[0099] In constructing the first residual loss data, nodal residual description data for constructing residual terms can be extracted from the nodal label data of the axially loaded grouting nodes; then, based on the predicted fatigue life and the nodal residual description data, the first residual loss data corresponding to the axially loaded grouting nodes is determined. The nodal residual description data can be any one of the target nodal label data, for example, it can be the ratio of the maximum load to the static interface shear strength (P...). max / P static Considering that fatigue life undergoes logarithmic transformation during data preprocessing and load levels are standardized using z-scores, the first residual loss data can be constructed based on the following formula:

[0100]

[0101] in, z-score normalization representing the load level, Loss physical This represents the first residual loss data.

[0102] In constructing the second residual loss data, the second residual loss data corresponding to the axially bearing grouting node can be determined based on the predicted fatigue life, nodal residual description data, and predicted fatigue relationship coefficients. The nodal residual description data can be any one of the target node label data; for example, it can be the ratio of the maximum load to the static interface shear strength (P...). max / P static Considering that fatigue life undergoes logarithmic transformation during data preprocessing and load levels are standardized using z-score, the second residual loss data can be constructed based on the following formula:

[0103] Loss hidden pyhsical =

[0104] Among them, Loss hidden physical This represents the second residual loss data.

[0105] Based on this, the predicted loss data of the preset fatigue life prediction model can be represented as follows:

[0106] Loss total =Loss mse +ωLoss physical +γLoss hidden physical

[0107] Among them, Loss total ω represents the predicted loss data, ω represents the weight coefficient corresponding to the first residual loss data constructed from simple physical trends, and γ represents the weight coefficient corresponding to the second residual loss data constructed from hidden physical laws.

[0108] S104. Based on the predicted loss data, train the preset fatigue life prediction model to obtain the trained fatigue life prediction model.

[0109] After obtaining the predicted loss data, the model parameters of the preset fatigue life prediction model can be updated in reverse based on the predicted loss data until the preset fatigue life prediction model converges, and finally the trained fatigue life prediction model is obtained.

[0110] In this embodiment, the hyperparameters of the preset fatigue life prediction model can be determined by combining five-fold cross-validation, and the Adam optimizer (an optimization algorithm) can be used for model training. The learning rate, number of training epochs, batch size, ω, γ, etc., can all be set according to actual conditions. For example, the learning rate can be set to 0.001, the number of training epochs can be set to 570, the batch size can be 3, ω can be set to 0.5, and γ can be set to 0.1.

[0111] In this embodiment of the application, while discussing the prediction results, the SHAP method (a machine learning model interpretation method based on game theory) and the physical consistency evaluation method can also be introduced to perform interpretability analysis on the preset fatigue life prediction model.

[0112] Since the preset fatigue life prediction model includes a preset fatigue life prediction network and a preset fatigue relationship prediction network, during the reverse update of the model parameters of the preset fatigue life prediction model, both the preset fatigue life prediction network and the preset fatigue relationship prediction network can be updated simultaneously. After training, the trained fatigue life prediction network and the trained fatigue relationship prediction network can be obtained. The trained fatigue life prediction model includes the trained fatigue life prediction network and the trained fatigue relationship prediction network. For ease of distinction, the trained fatigue life prediction model will be referred to as the fatigue life prediction model, the trained fatigue life prediction network as the fatigue life prediction network, and the trained fatigue relationship prediction network as the fatigue relationship prediction network.

[0113] In this embodiment, during the training of the preset fatigue life prediction model, the prediction loss data is determined through the synergistic effect of two types of physical constraints (simple physical trends and implicit physical laws). This enables the preset fatigue life prediction model to not only quickly capture the life change trend using simple physical trends, but also improve the physical consistency and generalization ability of the prediction by leveraging implicit physical laws. Thus, even with insufficient training sample data, limited model input feature data, and high data dispersion, the prediction results of the trained fatigue life prediction model can significantly improve the prediction accuracy while ensuring physical consistency. This achieves efficient and reliable prediction of the fatigue life of grouting connections under axial loading conditions, and has good scalability, allowing for the further introduction of complex working conditions or multi-physics field influences.

[0114] This application also provides a method, apparatus, electronic device, and computer-readable storage medium for predicting the fatigue life of grouted joints under axial load in offshore wind power. The fatigue life prediction apparatus can be integrated into an electronic device, which may be a server or a terminal, etc.

[0115] The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud-preset databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery network (CDN), and big data and artificial intelligence platforms. The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these.

[0116] Figure 8 This illustration shows an application scenario diagram of a fatigue life prediction method for axial load-bearing grouting nodes in offshore wind power, provided by an embodiment of this application. For example... Figure 8 As shown, a fatigue life prediction device for axial load grouting nodes in offshore wind power is integrated into an electronic device, which is a server, as an example. The server can acquire initial node description data of the axial load grouting node, which includes geometric structure data, material characterization data, and environmental loading data; perform correlation analysis on the geometric structure data, material characterization data, and environmental loading data to obtain the correlation analysis results; based on the correlation analysis results, select target node description data for the axial load grouting node from the initial node description data; and predict the fatigue life of the axial load grouting node using the aforementioned fatigue life prediction model based on the target node description data to obtain the fatigue life of the axial load grouting node.

[0117] The following sections provide detailed descriptions of each example. It should be noted that the order in which the embodiments are described is not intended to limit the order of the embodiments.

[0118] Figure 9 This illustration shows a flowchart of a fatigue life prediction method for grouted joints with axial loads in offshore wind power, provided by an embodiment of this application. Figure 9 As shown, the specific process of the fatigue life prediction method for grouted joints with axial load in offshore wind power is as follows:

[0119] S201. Obtain the initial node description data of the axial bearing grouting node. The initial node description data includes geometric structure data, material characterization data, and environmental loading data.

[0120] During the model application phase, for any axially loaded grouting node to be tested, the corresponding initial node description data can be obtained. The initial node description data may include geometric data, material characterization data, and environmental loading data.

[0121] The geometric data of the axial load-bearing grouting node may include the sleeve diameter (D). s ), sleeve thickness (t) s ), pile diameter (D) p ), pile thickness (t) p Grouting thickness (t) g ), shear key width (w), grouting length (L) g ), the ratio of shear key height to spacing (h / s).

[0122] Material characterization data for axially loaded grouting joints may include the elastic modulus of the sleeve (E). s ), pile elastic modulus (E) p ), sleeve yield strength (σ Y_s Grouting compressive strength (UCD) g ), the yield strength of the pile (σ) Y_p ), grouting Poisson's ratio (v g Poisson's ratio of steel (v) s ).

[0123] The environmental loading data for axially loaded grouting nodes can be understood as the loading conditions of the axially loaded grouting nodes. Environmental loading data may include: loading frequency (f), minimum load (P)... min ), loading environment (env), ratio of maximum load to static interface shear strength (P) max / P static ).

[0124] S202. Perform correlation analysis on geometric structure data, material characterization data and environmental loading data to obtain the correlation analysis results.

[0125] For the acquired geometric structure data, material characterization data, and environmental loading data, correlation analysis can be performed on the geometric structure data, material characterization data, and environmental loading data based on the Pearson correlation coefficient to obtain the correlation analysis results.

[0126] S203. Based on the correlation analysis results, select the target node description data of the axial bearing grouting node from the initial node description data.

[0127] In this embodiment, redundant features in the initial node description data can be removed based on the correlation analysis results, thereby filtering out the target node description data for axial bearing grouting nodes. For example, based on the correlation analysis results, the target node description data corresponding to the axial bearing grouting nodes filtered from the initial node description data may include: grouting thickness (t) g ), shear key height to spacing ratio (h / s), minimum load (P) min ), loading frequency (f), loading environment (env), and the ratio of maximum load to static interface shear strength (P) max / P static ).

[0128] S204. Based on the target node description data, the fatigue life of the axial bearing grouting node is predicted using a fatigue life prediction model to obtain the fatigue life of the axial bearing grouting node.

[0129] After obtaining the target node description data, it can be hierarchically and randomly partitioned and normalized to obtain normalized target node description data. Specifically, the input features can be standardized using the z-score method, and the standardized data can then be input into the trained fatigue life prediction model.

[0130] Then, the normalized target node description data can be used as the model input data for the fatigue life prediction model. The fatigue life of the axial bearing grouting node can be predicted by the fatigue life prediction model to obtain the fatigue life of the axial bearing grouting node.

[0131] It should be noted that although the post-trained fatigue life prediction model consists of two parts, the post-trained fatigue life prediction network and the post-trained fatigue relationship prediction network, in the process of model application, it is only necessary to use the post-trained fatigue life prediction network to process the target node description data to obtain the fatigue life of the axial bearing grouting node.

[0132] The fatigue life prediction model provided in this application was used to predict the fatigue life of grouted joints bearing axial loads in offshore wind power. The results show that almost all predicted values ​​fall within the three-fold error band, and the mean square error (MSE) is 0.2248. 2 With a value of 0.9078, it effectively overcomes the bottlenecks of traditional SN curve method, such as the inability to explain the prediction bias caused by the nonlinear effects of multi-physics fields, and the failure of extrapolation of small-sample training models under conditions of scarce data, few feature inputs, and low computational efficiency.

[0133] Therefore, the fatigue life prediction model provided in this application embodiment can be directly applied to the fatigue life prediction of grouted joints in the axial load of offshore wind turbines. It can quickly output fatigue life prediction results by inputting the geometric structure data, material characterization data, and environmental loading data of the grouted joint, providing support for structural design optimization and operation and maintenance decisions. Simultaneously, the fatigue life prediction model provided in this application embodiment can be combined with a structural health monitoring system to assess the remaining life of the grouted connection based on real-time recorded data, assisting in the formulation of scientific maintenance and replacement plans. Furthermore, the fatigue life prediction model provided in this application embodiment has good scalability, allowing for the introduction of more geometric, material, and load parameters within the existing framework to achieve higher accuracy in predicting fatigue failure modes.

[0134] Corresponding to the training method for the fatigue life prediction model of the axial load grouting node of offshore wind power described in the above embodiments, this application embodiment also provides a training device for the fatigue life prediction model of the axial load grouting node of offshore wind power. Figure 10 This illustration shows a schematic diagram of the training device for a fatigue life prediction model of an axial load-bearing grouting node in offshore wind power, as provided in an embodiment of this application. Figure 10 As shown, the training device for the fatigue life prediction model of the axial load grouting node in offshore wind power can include a first acquisition unit 301, a first prediction unit 302, a loss determination unit 303, and a model training unit 304, as follows:

[0135] (1) First acquisition unit 301;

[0136] The first acquisition unit 301 is used to acquire a training dataset, which includes node label data corresponding to each of the multiple axial bearing grouting nodes.

[0137] The node label data includes geometric structure data, material characterization data, and environmental loading data.

[0138] (2) First prediction unit 302;

[0139] The first prediction unit 302 is used to predict the fatigue life and fatigue relationship coefficient of the axial bearing grouting node based on the node label data and a preset fatigue life prediction model, so as to obtain the predicted fatigue life and predicted fatigue relationship coefficient.

[0140] For example, the first prediction unit 302 can be used to perform correlation analysis on geometric structure data, material characterization data and environmental loading data to obtain correlation analysis results; based on the correlation analysis results, target node label data of axial bearing grouting nodes is selected from the node label data; based on the target node label data, a preset fatigue life prediction model is used to predict the fatigue life and fatigue relationship coefficient of axial bearing grouting nodes to obtain predicted fatigue life and predicted fatigue relationship coefficient.

[0141] The preset fatigue life prediction model includes a preset fatigue life prediction network and a preset fatigue relationship prediction network.

[0142] For example, the first prediction unit 302 can be used to predict the fatigue life of the axial bearing grouting node based on the target node label data using a preset fatigue life prediction network, and obtain the predicted fatigue life; based on the target node label data, it can also predict the corresponding fatigue relationship coefficient of the axial bearing grouting node in the high-cycle fatigue region using a preset fatigue relationship prediction network, and obtain the predicted fatigue relationship coefficient.

[0143] (3) Loss determination unit 303;

[0144] The loss determination unit 303 is used to determine the predicted loss data of the preset fatigue life prediction model based on the predicted fatigue life and the predicted fatigue relationship coefficient.

[0145] For example, the loss determination unit 303 can be used to extract the actual fatigue life of the axially bearing grouting node from the training dataset, compare the predicted fatigue life with the actual fatigue life to obtain the fatigue life comparison result; determine the fatigue life loss data corresponding to the axially bearing grouting node based on the fatigue life comparison result; determine the residual loss data corresponding to the axially bearing grouting node based on the predicted fatigue life and the predicted fatigue relationship coefficient; and determine the predicted loss data of the preset fatigue life prediction model based on the fatigue life loss data and the residual loss data.

[0146] For example, the loss determination unit 303 can be specifically used to extract node residual description data for constructing residual terms from the node label data of the axial bearing grouting node; determine the first residual loss data corresponding to the axial bearing grouting node based on the predicted fatigue life and the node residual description data; determine the second residual loss data corresponding to the axial bearing grouting node based on the predicted fatigue life, the node residual description data and the predicted fatigue relationship coefficient; and determine the residual loss data corresponding to the axial bearing grouting node based on the first residual loss data and the second residual loss data.

[0147] (4) Model training unit 304

[0148] The model training unit 304 is used to train the preset fatigue life prediction model based on the prediction loss data to obtain the trained fatigue life prediction model.

[0149] In practice, each of the above units can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units, please refer to the previous method embodiments, which will not be repeated here.

[0150] Corresponding to the fatigue life prediction method for axial load grouting nodes of offshore wind power described in the above embodiments, this application also provides a fatigue life prediction device for axial load grouting nodes of offshore wind power. Figure 11 A schematic diagram of the fatigue life prediction device for axial load grouting nodes in offshore wind power, provided in an embodiment of this application, is shown. Figure 11 As shown, the fatigue life prediction device for grouting nodes with axial load in offshore wind power projects may include a second acquisition unit 401, a data analysis unit 402, a data filtering unit 403, and a second prediction unit 404, as follows:

[0151] (1) Second acquisition unit 401;

[0152] The second acquisition unit 401 is used to acquire the initial node description data of the axial bearing grouting node. The initial node description data includes geometric structure data, material characterization data and environmental loading data.

[0153] (2) Data analysis unit 402;

[0154] The data analysis unit 402 is used to perform correlation analysis on geometric structure data, material characterization data and environmental loading data to obtain correlation analysis results.

[0155] (3) Data filtering unit 403;

[0156] The data filtering unit 403 is used to filter the target node description data of the axial bearing grouting node from the initial node description data based on the correlation analysis results.

[0157] (4) Second prediction unit 404.

[0158] The second prediction unit 404 uses a fatigue life prediction model to predict the fatigue life of the axially bearing grouting node based on the target node description data, and obtains the fatigue life of the axially bearing grouting node.

[0159] In practice, each of the above units can be implemented as an independent entity or can be arbitrarily combined to be implemented as the same or several entities. For the specific implementation of each of the above units, please refer to the previous method embodiments, which will not be repeated here.

[0160] Figure 12 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 12 As shown, the electronic device 5 of this embodiment includes: at least one processor 50 ( Figure 12 The diagram shows only one of the components, a memory 51, and a computer program 52 stored in the memory 51 and executable on the at least one processor 50. When the processor 50 executes the computer program 52, it implements the steps in the above-described training method for any fatigue life prediction model for axial load grouting nodes in offshore wind power, as well as the steps in the fatigue life prediction method embodiments. Alternatively, when the processor 50 executes the computer program 52, it implements the steps in the above-described training method for any fatigue life prediction model for axial load grouting nodes in offshore wind power, as well as the steps in the fatigue life prediction method embodiments.

[0161] The electronic device 5 may include, but is not limited to, a processor 50 and a memory 51. Those skilled in the art will understand that... Figure 12 This is merely an example of electronic device 5 and does not constitute a limitation on electronic device 5. It may include more or fewer components than shown in the figure, or combine certain components, or different components. For example, it may also include input / output devices, network access devices, etc.

[0162] The processor 50 can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0163] In some embodiments, the memory 51 may be an internal storage unit of the electronic device 5, such as a hard disk or memory. In other embodiments, the memory 51 may be an external storage device of the electronic device 5, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the electronic device 5. Furthermore, the memory 51 may include both internal and external storage units of the electronic device 5. The memory 51 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory 51 can also be used to temporarily store data that has been output or will be output.

[0164] This application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by an electronic device, the electronic device implements the training method for the fatigue life prediction model of the axial bearing grouting node of offshore wind power and the steps in the fatigue life prediction method embodiments described above.

[0165] This application provides a computer program product, which includes a computer program. When the computer program is executed by an electronic device, the electronic device implements the training method for the fatigue life prediction model of the axial load grouting node of offshore wind power and the steps in the fatigue life prediction method embodiments described above.

[0166] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable storage medium can include at least: any entity or device capable of carrying computer program code to a device / electronic device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable storage media cannot be electrical carrier signals or telecommunication signals.

[0167] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0168] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0169] In the embodiments provided in this application, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings or direct couplings or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0170] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0171] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A training method for a fatigue life prediction model of an offshore wind power axial load grouted node, characterized in that, Comprise: Obtain a training data set, the training data set comprises a plurality of axial bearing grouting nodes respectively corresponding node label data; wherein, the node label data includes geometric configuration data, material characterization data and environmental loading data; According to the node label data, a preset fatigue life prediction model is used to predict the fatigue life and fatigue relationship coefficient of the axial bearing grouting node, and the predicted fatigue life and the predicted fatigue relationship coefficient are obtained; wherein, the predicted fatigue relationship coefficient is an implicit constant representing the physical constraint in the prediction process obtained by the preset fatigue life prediction model; Based on the predicted fatigue life and the predicted fatigue relationship coefficient, the prediction loss data of the preset fatigue life prediction model is determined; According to the prediction loss data, the preset fatigue life prediction model is trained to obtain a trained fatigue life prediction model; Wherein, the prediction loss data of the preset fatigue life prediction model is determined based on the predicted fatigue life and the predicted fatigue relationship coefficient, including: extracting the actual fatigue life of the axial bearing grouting node from the training data set, and comparing the predicted fatigue life with the actual fatigue life to obtain a fatigue life comparison result; according to the fatigue life comparison result, the fatigue life loss data corresponding to the axial bearing grouting node is determined; according to the predicted fatigue life and the predicted fatigue relationship coefficient, the residual loss data corresponding to the axial bearing grouting node is determined; based on the fatigue life loss data and the residual loss data, the prediction loss data of the preset fatigue life prediction model is determined.

2. The training method of the fatigue life prediction model of the offshore wind power axial bearing grouting node according to claim 1, wherein, According to the node label data, a preset fatigue life prediction model is used to predict the fatigue life and fatigue relationship coefficient of the axial bearing grouting node, and the predicted fatigue life and the predicted fatigue relationship coefficient are obtained; wherein, the predicted fatigue relationship coefficient is an implicit constant representing the physical constraint in the prediction process obtained by the preset fatigue life prediction model; The preset fatigue life prediction model comprises a preset fatigue life prediction network and a preset fatigue relationship prediction network, According to the target node label data, the preset fatigue life prediction network is used to predict the fatigue life of the axial bearing grouting node, and the predicted fatigue life is obtained; ​ 3. The training method of the fatigue life prediction model for offshore wind turbine axial load grouted nodes according to claim 2, characterized in that, ​ ​ ​ According to the target node label data, a preset fatigue relationship prediction network is used to predict a corresponding fatigue relationship coefficient of the axial load grouting node in a high-cycle fatigue region, to obtain the predicted fatigue relationship coefficient.

4. The training method of the fatigue life prediction model for offshore wind power axial load grouted nodes according to claim 1, characterized in that, The determining of the residual loss data corresponding to the axial load grouting node according to the predicted fatigue life and the predicted fatigue relationship coefficient comprises: extracting node residual description data for constructing a residual term from the node label data of the axial load grouting node; determining first residual loss data corresponding to the axial load grouting node according to the predicted fatigue life and the node residual description data; determining second residual loss data corresponding to the axial load grouting node according to the predicted fatigue life, the node residual description data and the predicted fatigue relationship coefficient; determining the residual loss data corresponding to the axial load grouting node based on the first residual loss data and the second residual loss data.

5. A method for fatigue life prediction of an offshore wind power axial load grouted node, characterized in that, comprises: obtaining initial node description data of an axial load grouting node, the initial node description data comprising geometric configuration data, material characterization data and environmental loading data; performing correlation analysis on the geometric configuration data, the material characterization data and the environmental loading data to obtain a correlation analysis result; according to the correlation analysis result, screening target node description data of the axial load grouting node from the initial node description data; according to the target node description data, using the fatigue life prediction model of claim 1 to predict the fatigue life of the axial load grouting node, to obtain the fatigue life of the axial load grouting node.

6. A training device for a fatigue life prediction model of an offshore wind power axial load grouted node, characterized in that, comprises: a first acquisition unit configured to acquire a training data set, the training data set comprising node label data corresponding to a plurality of axial load grouting nodes respectively; wherein the node label data comprises geometric configuration data, material characterization data and environmental loading data; a first prediction unit configured to, according to the node label data, use a preset fatigue life prediction model to predict the fatigue life and fatigue relationship coefficient of the axial load grouting node, to obtain a predicted fatigue life and a predicted fatigue relationship coefficient; wherein the predicted fatigue relationship coefficient is an implicit constant representing physical constraints in the prediction process obtained by the preset fatigue life prediction model; a loss determination unit configured to determine prediction loss data of the preset fatigue life prediction model based on the predicted fatigue life and the predicted fatigue relationship coefficient; a model training unit configured to train the preset fatigue life prediction model according to the prediction loss data, to obtain a trained fatigue life prediction model; The loss determination unit is specifically configured to: extract an actual fatigue life of the axial load bearing grouting node from the training data set, and compare the predicted fatigue life with the actual fatigue life to obtain a fatigue life comparison result; determine fatigue life loss data corresponding to the axial load bearing grouting node according to the fatigue life comparison result; determine residual loss data corresponding to the axial load bearing grouting node according to the predicted fatigue life and the predicted fatigue relationship coefficient; and determine prediction loss data of the preset fatigue life prediction model based on the fatigue life loss data and the residual loss data.

7. A device for fatigue life prediction of an offshore wind power axial load grouted node, characterized by, The method comprises: The second acquisition unit is configured to acquire initial node description data of the axial load bearing grouting node, the initial node description data comprising geometric configuration data, material characterization data and environmental loading data; The data analysis unit is configured to perform correlation analysis on the geometric configuration data, the material characterization data and the environmental loading data to obtain a correlation analysis result; The data screening unit is configured to screen target node description data of the axial load bearing grouting node from the initial node description data according to the correlation analysis result; The second prediction unit is configured to predict the fatigue life of the axial load bearing grouting node by using the fatigue life prediction model of claim 1 according to the target node description data, to obtain the fatigue life of the axial load bearing grouting node.

8. An electronic device, comprising: The computer program is executed by the processor to implement the training method of any one of claims 1-4 and the fatigue life prediction method of claim 5. The computer program is executed by the processor to implement the training method of any one of claims 1-4 and the fatigue life prediction method of claim 5.

9. A computer-readable storage medium storing a computer program, the computer program comprising instructions that, when executed by a computer, cause the computer to perform the method of any one of claims 1 to 8. ​

Images