A fatigue failure assessment method for fan duct nodes based on linear damage criteria

By using linear damage criteria and dynamic stiffness monitoring, the problem of balancing computational efficiency and accuracy in fatigue assessment of offshore wind turbine pipe nodes was solved, achieving efficient and accurate full-life fatigue damage assessment, eliminating numerical errors and mesh interference, and improving the reliability of assessment results.

CN123116677APending Publication Date: 2026-11-17HUANENG GUANGDONG SHANTOU OFFSHORE WIND POWER CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611246797.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-17
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies struggle to balance the efficiency of long-term multi-field coupled calculations with the accuracy of local stress extraction when assessing fatigue failure of offshore wind turbine pipe nodes. The degradation of nonlinear stiffness in the pile and soil leads to numerical step abrupt changes, local stress extraction is affected by mesh singularity interference, and the fatigue resistance reduction effect caused by the enlargement of thick plate components is ignored, resulting in unreliable assessments.

Method used

An evaluation method based on linear damage criteria is adopted. By constructing a linear elastic spatial influence matrix and a pile-soil stiffness-impedance matrix, stiffness degradation parameters are monitored. Overlapping time windows are used for smooth fusion and algebraic mapping operations to avoid grid singular interference, eliminate ineffective stress cycles, introduce a thickness correction index, and combine the SN fatigue curve to evaluate the fatigue damage degree throughout the entire life cycle.

Benefits of technology

It improves the computational efficiency and accuracy of fatigue assessment of wind turbine pipe nodes, eliminates numerical truncation errors and high-frequency spurious oscillations, accurately assesses fatigue damage throughout the entire lifespan, and improves the reliability of assessment results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123116677A_ABST
    Figure CN123116677A_ABST
Patent Text Reader

Abstract

This invention relates to the field of offshore wind power generation and structural engineering technology, and discloses a fatigue failure assessment method for wind turbine pipe nodes based on a linear damage criterion. The method includes: constructing a finite element sub-model, applying a unit load to generate a linear elastic spatial influence matrix and a compensation matrix; constructing a multi-field coupling model, configuring a nonlinear spring to form a pile-soil stiffness-resistance matrix; inputting a load sequence for solution, monitoring stiffness degradation parameters, and performing smooth fusion using an overlapping time window when a trigger threshold is reached to generate end load and mud surface angular displacement time histories; inputting the above time histories into a preset matrix, and generating hot spot stress time histories through algebraic mapping and stress extrapolation; performing rainflow counting on the hot spot stress time histories, and outputting the full-life fatigue damage degree by combining the S-N curve. This invention uses algebraic mapping to replace the underlying time-domain integration to reduce computational overhead and eliminate truncation errors during stiffness updates, thereby improving the efficiency and accuracy of fatigue assessment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of offshore wind power generation and structural engineering technology, specifically to a fatigue failure assessment method for wind turbine pipe nodes based on linear damage criteria. Background Technology

[0002] Pipe nodes of offshore wind power support structures are subjected to cyclic alternating loads such as wind and waves for a long time, resulting in stress concentration and fatigue failure. Accurate assessment of fatigue damage throughout the life of pipe nodes is of great significance.

[0003] Existing evaluation methods require long-term multi-field coupled dynamic time-domain analysis. If a refined three-dimensional finite element model is directly used for long-period global integration or stepwise local solution, the amount of iterative computation is large, making it difficult to balance long-term solution efficiency and local stress extraction accuracy in engineering practice.

[0004] Meanwhile, the soil around the piles under the foundation will undergo plastic cumulative deformation and nonlinear stiffness degradation under alternating loads. When dealing with this dynamic support boundary, the existing time-domain continuous integration method will cause numerical step jumps if the foundation stiffness is directly updated, resulting in truncation errors and high-frequency spurious oscillations in the multi-field coupled solution, which will distort the extracted load time history. If a constant initial stiffness is used, it cannot reflect the structural displacement amplification effect caused by soil degradation.

[0005] Furthermore, the weld toe area at the intersection of pipe nodes exhibits geometric discontinuities, which can lead to mesh singularities and stress noise interference in finite element analysis. Directly extracting stress from this area makes it difficult to accurately characterize the macroscopic stress amplitude. In addition, conventional methods often neglect the fatigue resistance reduction effect caused by the enlargement of the thick plate components of large wind turbines, resulting in deviations in the calculation of the allowable number of cycles. These defects together restrict the reliability of the full-life fatigue assessment of wind turbine pipe nodes. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a fatigue failure assessment method for wind turbine pipe nodes based on linear damage criteria. This method solves the problems of difficulty in balancing the efficiency of long-term multi-field coupling calculations and the accuracy of local stress extraction in existing wind turbine pipe node fatigue assessments, the numerical truncation error caused by directly updating the nonlinear degradation stiffness of pile and soil in time-domain integration, and the unreliability of full-life fatigue assessment due to mesh singular interference in local stress extraction.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a fatigue failure assessment method for wind turbine pipe nodes based on a linear damage criterion, the method comprising the following steps: A three-dimensional finite element sub-model of the target wind turbine pipe node is constructed and a target extrapolation node set is defined. A unit load is applied to the three-dimensional finite element sub-model to extract the stress tensor to generate a linear elastic spatial influence matrix, and a unit gravity stress response vector and mud surface angular displacement compensation matrix are generated. An overall aerodynamic, hydrodynamic, and elastic coupling model was constructed, and nonlinear spring elements were configured in the pile foundation penetration section to form a pile-soil stiffness-resistance matrix. The input time history load sequence is used to perform the solution, and the stiffness degradation parameter of the pile-soil stiffness-impedance matrix is ​​monitored. If the stiffness degradation parameter reaches the set trigger threshold, an overlapping time window is set to perform smooth fusion and generate a global end load time history matrix and mud surface angle displacement time history. The global end load time history matrix and the mud surface angular displacement time history are input into the linear elastic spatial influence matrix, the unit gravity stress response vector and the mud surface angular displacement compensation matrix to perform algebraic mapping operation and spatial stress extrapolation operation to generate hot spot stress time history; Rainflow counting is performed on the hot spot stress time history, and the full-life fatigue damage degree is output by combining the matched SN fatigue curve and the linear cumulative damage criterion.

[0008] By reducing the dimensionality of the local mesh time-domain integral to a pre-calculated static matrix algebraic mapping, the algebraic multiplication of the spatial dimensionality reduction matrix and the time-domain load vector is used to replace the finite element bottom mesh solution at each time step, thus reducing computational overhead. At the same time, a dynamic stiffness monitoring and overlapping time window smooth reconstruction mechanism is introduced to eliminate the numerical truncation error introduced when updating the nonlinear decay of the support boundary, thereby achieving synchronous mapping between soil plastic degradation and the dynamic response of the superstructure.

[0009] Further, the step of applying a unit load to the three-dimensional finite element sub-model to extract the stress tensor and generate the linear elastic spatial influence matrix includes: establishing control reference points at the centroid positions of the main pipe end section and the branch pipe end section of the three-dimensional finite element sub-model; sequentially and independently applying unit force components along the spatial coordinate axes and unit moment components around the spatial coordinate axes to the control reference points, eliminating the rigid body displacement of the three-dimensional finite element sub-model, and then extracting the three-dimensional stress tensor output by the target extrapolation node set; arranging the stress components of all physical nodes in the target extrapolation node set into column vectors, and horizontally splicing them according to a fixed order of pipe end number, degree of freedom number, and stress component number to form the linear elastic spatial influence matrix in a two-dimensional matrix data format.

[0010] The above steps transform the multi-degree-of-freedom load response of the multi-pipe end into a two-dimensional matrix, establish a linear mapping benchmark from the end load to the local stress field, and avoid repeated finite element iterations for complex intersection regions.

[0011] Furthermore, the steps of generating the unit gravity stress response vector and the mud surface angular displacement compensation matrix include: applying a uniformly distributed load of a unit gravitational field acceleration value in the vertically downward direction of global gravity, forcing the three-dimensional finite element sub-model to generate a unit overturning angular displacement of a fixed radian value around the horizontal axis; extracting the principal stress change values ​​corresponding to the target extrapolation node set under the action of the unit gravitational field acceleration value, and organizing them into the unit gravity stress response vector; and extracting the principal stress change values ​​corresponding to the target extrapolation node set under the action of the unit overturning angular displacement, and organizing them into the mud surface angular displacement compensation matrix.

[0012] The above steps are used to compensate for the nonlinear additional bending moment caused by gravity in the wind turbine support structure under large deformation conditions, as well as the secondary stress introduced by the tilt of the bottom foundation, thereby improving the accuracy of stress extraction for structures with large length-to-diameter ratios.

[0013] Furthermore, the step of configuring nonlinear spring elements in the pile foundation penetration section to form a pile-soil stiffness-impedance matrix includes: configuring py springs representing horizontal lateral constraints, tz springs representing axial lateral friction, and Qz springs representing pile end bearing capacity at discrete structural nodes along the pile foundation depth direction, respectively; calculating and generating nonlinear load-displacement relationship curves; extracting the initial tangent stiffness values ​​of the nonlinear load-displacement relationship curves at the zero displacement point to construct a local foundation stiffness matrix; mapping the local foundation stiffness matrix to the global coordinate system through a coordinate transformation matrix and performing finite element stiffness integration to generate the pile-soil stiffness-impedance matrix in the initial state.

[0014] The above steps characterize the spatial nonlinear interaction between the pile and soil interface by discretizing spring elements, transforming the complex soil constitutive relationship into the foundation stiffness matrix, and providing accurate initial mechanical boundary conditions for the overall dynamic solution.

[0015] Further, the step of monitoring the stiffness degradation parameter of the pile-soil stiffness-impedance matrix includes: extracting the horizontal displacement value and lateral resistance value of discrete structural nodes at the mud surface elevation, dividing the lateral resistance value by the horizontal displacement value to calculate the equivalent secant stiffness; calculating the stiffness degradation parameter based on the equivalent secant stiffness and the initial equivalent stiffness of the pile-soil stiffness-impedance matrix at the mud surface elevation in the initial state; when the stiffness degradation parameter reaches the set trigger threshold, sending an interrupt signal to pause the multi-field coupled dynamics solution process, and saving the global displacement vector, velocity vector, and acceleration vector of the structure at the interruption time as local state boundary conditions.

[0016] The above steps use equivalent secant stiffness to dynamically characterize the attenuation of support boundary impedance caused by soil plastic deformation, providing a triggering basis for the adaptive adjustment of time-domain integral boundary conditions and avoiding the problem of small displacement response caused by using the initial stiffness throughout the process.

[0017] Furthermore, the step of setting an overlapping time window to perform smooth fusion and generate a global end load time history matrix and mud surface angular displacement time history includes: setting the overlapping time window with the interruption time as the starting time; retaining the unupdated pile-soil stiffness-impedance matrix, so that the dynamic integral operation under the unupdated stiffness reference continues to be executed until the end of the overlapping time window, and extracting and generating the preceding data sequence. The pile-soil stiffness-impedance matrix is ​​updated using the ratio of the equivalent secant stiffness to the initial equivalent stiffness as a stiffness reduction factor. The integration operation is restarted with the local state boundary conditions until the end of the overlapping time window, and a subsequent data sequence is extracted and generated. The cosine-weighted smoothing function is applied to mathematically reconstruct and smooth the preceding and subsequent data sequences. The smoothing and fusion results are aligned and spliced ​​along the overall time axis. The time series data of the six degrees of freedom internal forces at each pipe end are extracted to generate the global end load time history matrix. The spatial rotational angular displacement sequence is extracted to generate the mud surface angular displacement time history.

[0018] The above steps, by retaining the overlapping calculations of the old state and introducing the new state, and using the cosine weighting function for transition, can avoid high-frequency spurious oscillations of the structure caused by a numerical step change in the stiffness matrix.

[0019] Further, the steps of performing algebraic mapping operations and spatial stress extrapolation operations include: extracting the multi-pipe end load vector after coordinate transformation and degree of freedom rearrangement from the global end load time history matrix; multiplying the multi-pipe end load vector with the linear elastic spatial influence matrix to calculate the nominal dynamic stress response of the structure; extracting the gravity proportionality coefficient and multiplying the unit gravity stress response vector with the gravity proportionality coefficient; extracting the mud surface angular displacement vector from the mud surface angular displacement time history and multiplying the mud surface angular displacement vector with the mud surface angular displacement compensation matrix to calculate the additional dynamic stress response; and performing algebraic superposition of the nominal dynamic stress response of the structure and the additional dynamic stress response in the time domain to generate a dynamic stress time history matrix.

[0020] The above steps algebraically decouple the overall nominal response of the structure from the additional response generated by local large deformation. By using the time-domain linear superposition mechanism, the local dynamic stress under multi-load coupling can be reconstructed without re-performing the finite element integration.

[0021] Furthermore, the step of performing spatial stress extrapolation calculations to generate hotspot stress time histories includes: Multiple extrapolation paths are divided along the direction perpendicular to the weld toe normal of the intersection line. On each extrapolation path, a physical node is located at a first reference distance from the outer edge of the weld toe of the intersection line and a physical node is located at a second reference distance, forming the target extrapolation node set. The first reference distance and the second reference distance are determined based on the reference wall thickness parameter of the pipe node. Extract the first stress time series data corresponding to the physical node at the first reference distance position from the dynamic stress time history matrix, and extract the second stress time series data corresponding to the physical node at the second reference distance position. The first stress time series data is multiplied by the first spatial stress extrapolation coefficient, and the second stress time series data is multiplied by the second spatial stress extrapolation coefficient and summed to generate the hot spot stress time history at the intersection line of the pipe node.

[0022] The above steps select a specific wall thickness multiple location for linear geometric extrapolation to avoid local stress concentration and mesh singularity interference caused by micro-geometric discontinuities at the weld toe root, thereby obtaining accurate structural macro-geometric stress.

[0023] Further, the step of performing rainflow counting on the hotspot stress time history includes: identifying local maxima and local minima in the hotspot stress time history, eliminating invalid stress cycles where the adjacent difference between the local maxima and the local minima is less than a set filtering threshold, and generating an alternating stress extremum sequence; extracting the stress amplitudes that constitute closed hysteresis loops in the alternating stress extremum sequence, obtaining the actual number of cycles corresponding to each stress amplitude according to a preset stress level division interval; introducing a thickness correction index to reduce and correct the reference parameters of the SN fatigue curve, and calculating the allowable number of cycles corresponding to each stress amplitude.

[0024] The above steps eliminate stress noise that contributes little to fatigue damage by setting a filtering threshold, and introduce a thickness correction index to eliminate the fatigue resistance reduction effect caused by the enlargement of thick plate size, thereby improving the accuracy of the allowable cycle count calculation.

[0025] In a preferred embodiment of the present invention, the step of combining the matched SN fatigue curves and the linear cumulative damage criterion to output the full-life fatigue damage degree includes: dividing the actual number of cycles corresponding to each stress amplitude by the allowable number of cycles corresponding to each stress amplitude and summing the results to generate the short-term fatigue damage degree; obtaining the probability distribution data of occurrence of each discrete environmental condition within the design life cycle; multiplying the short-term fatigue damage degree of each discrete environmental condition by the corresponding probability of occurrence and by the total number of time blocks corresponding to the design life cycle; and performing linear superposition to generate the full-life fatigue damage degree.

[0026] The above steps are based on the linear cumulative damage criterion, which extends the short-term fatigue characteristics under discrete sea states according to the statistical probability of occurrence, so as to realize the quantitative assessment of fatigue damage throughout the entire life cycle of the structure.

[0027] This invention provides a fatigue failure assessment method for wind turbine pipe joints based on a linear damage criterion. It has the following beneficial effects: 1. This invention constructs a linear elastic spatial influence matrix to reduce the dimensionality of complex local mesh time-domain integrals to a pre-calculated static matrix algebraic mapping. This design utilizes the algebraic multiplication of the spatial dimensionality reduction matrix and the time-domain load vector to replace the finite element bottom-level solution at each time step in the traditional method. While ensuring the accuracy of local stress extraction, it reduces the resource consumption of long-term multi-field coupled calculations.

[0028] 2. This invention introduces a dynamic stiffness monitoring and overlapping time window smooth reconstruction mechanism. By monitoring the pile-soil stiffness degradation parameters and using a cosine weighting function to smooth the transition between the preceding and following data sequences when a threshold is triggered, this mechanism eliminates the numerical truncation error generated when updating the stiffness of the support boundary, avoids high-frequency spurious oscillations of the structure caused by a step change in the stiffness matrix, and improves the calculation accuracy of the dynamic response of the structure under nonlinear support boundaries.

[0029] 3. This invention uses linear geometric extrapolation at a specific wall thickness multiple when generating hot spot stress time history, and introduces stress filtering threshold and thickness correction index in the rainflow counting stage. The combination of these features not only avoids mesh singularity interference caused by micro-geometric discontinuities at the weld toe root, but also eliminates stress noise that does not contribute to fatigue damage and corrects the resistance reduction effect of thick plates, thereby improving the reliability of the whole life fatigue damage assessment results. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the system architecture of the present invention; Figure 2 This is a schematic diagram of the method flow of the present invention; Figure 3 This is a schematic diagram illustrating the principle of finite element sub-model construction and two-order space matrix extraction of the present invention; Figure 4 This is a schematic diagram illustrating the principle of constructing the multi-field coupling model and initializing pile-soil impedance in this invention. Figure 5 This is a schematic diagram of the process of stiffness degradation monitoring and temporal overlap reconstruction fusion of the present invention; Figure 6 This is a schematic diagram illustrating the principle of algebraic mapping solution and hotspot stress extrapolation of the present invention; Figure 7 This is a schematic diagram of the rainflow counting and full-life fatigue damage calculation process of the present invention; Figure 8This is a schematic diagram of comparative test data for the fatigue damage assessment method of the present invention.

[0031] Among them, 110 is the processor; 120 is the main memory; 130 is the persistent storage medium; 131 is the mapping preprocessing module; 132 is the coupled dynamics module; 133 is the data reconstruction module; and 134 is the fatigue assessment module. Detailed Implementation

[0032] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] See attached document Figure 1 The present invention provides a fatigue failure assessment system for wind turbine pipe nodes based on linear damage criteria, including hardware devices, which provide a physical support environment for the operation of the fatigue failure assessment method for wind turbine pipe nodes based on linear damage criteria.

[0034] The hardware device is equipped with a processor 110, a main memory 120, and a persistent storage medium 130. The persistent storage medium 130 is used to store computer program instructions. The processor 110 reads and executes the computer program instructions to realize the data processing logic and calculation logic for fatigue failure assessment of wind turbine pipe nodes. The main memory 120 serves as a data cache carrier during the data calculation process and is used to temporarily store time history matrix data and spatial influence matrix data during the multi-field coupling calculation process.

[0035] The persistent storage medium 130 stores computer program instructions that constitute the evaluation control logic. The computer program instructions are divided into multiple data processing modules that work together. The persistent storage medium 130 includes: a mapping preprocessing module 131, a coupled dynamics module 132, a data reconstruction module 133, and a fatigue evaluation module 134. The processor 110 performs corresponding operations by reading and running the above-mentioned data processing modules.

[0036] The mapping preprocessing module 131 is used to establish a three-dimensional finite element sub-model of the target wind turbine pipe node. The mapping preprocessing module 131 discretizes and locates the target extrapolation node set on the three-dimensional finite element sub-model, applies unit loads in multiple degrees of freedom directions to the pipe end in sequence, extracts the stress tensor corresponding to the target extrapolation node set and assembles it to generate a linear elastic spatial influence matrix. At the same time, the mapping preprocessing module 131 calculates the additional stress response by applying gravitational field acceleration and unit overturning angle displacement of mud surface, and assembles it to generate a spatial deflection compensation matrix.

[0037] The Coupled Dynamics Module 132 is used to construct the overall aerodynamic, hydrodynamic, and elastic coupling model of the wind turbine generator. At the boundary between the pile foundation entry section and the soil in the overall aerodynamic, hydrodynamic, and elastic coupling model, the Coupled Dynamics Module 132 configures nonlinear spring elements along the discrete structural nodes in the pile foundation depth direction to form the initial pile-soil stiffness and impedance matrix. The Coupled Dynamics Module 132 receives the wind field environmental load sequence and the wave and ocean current environmental load sequence input from the outside and performs the time-domain dimension structural and geotechnical dynamic coupling solution operation.

[0038] The data reconstruction module 133 establishes a real-time data interaction channel with the coupled dynamics module 132. During the time-domain solution process, the data reconstruction module 133 monitors the stiffness degradation parameters of the pile-soil stiffness-impedance matrix. When the stiffness attenuation reaches the set trigger threshold, the data reconstruction module 133 interrupts the current solution step of the coupled dynamics module 132 and executes the boundary stiffness update command. The data reconstruction module 133 extracts the preceding and following sequences within the overlapping time window, executes the weighted smoothing fusion algorithm, and outputs the global end load time history matrix and mud surface angular displacement time history after eliminating numerical truncation errors.

[0039] The fatigue assessment module 134 receives the linear elastic spatial influence matrix, unit gravity stress response vector, and mud surface angular displacement compensation matrix generated by the mapping preprocessing module 131, as well as the global end load time history matrix and mud surface angular displacement vector time history output by the data reconstruction module 133.

[0040] The fatigue assessment module 134 performs pipe end numbering, degree of freedom order and coordinate direction transformation on the global end load time history matrix to make the column vector order of the global end load time history matrix consistent with that of the linear elastic space influence matrix. Then, it performs algebraic mapping operation and introduces extrapolation coefficients to perform hot spot stress extrapolation operation to generate hot spot stress time history at the intersection line position. The fatigue assessment module 134 performs rainflow statistics and cumulative damage calculation based on the hot spot stress time history to obtain the full life fatigue damage result.

[0041] In the data flow logic, the mapping preprocessing module 131 completes the matrix-based dimensionality reduction of spatial geometric features, the coupled dynamics module 132 performs the forward dynamic solution under long-period environmental load input, the data reconstruction module 133 coordinates the synchronous mapping between the nonlinear attenuation law of pile and soil and the dynamic solution process of the superstructure, and the fatigue assessment module 134 integrates the spatial matrix formed in the preprocessing stage with the time-domain vector set generated in the operation stage and directly outputs the fatigue damage results.

[0042] See attached document Figure 2 This invention provides a fatigue failure assessment method for wind turbine pipe nodes based on a linear damage criterion, the method comprising the following steps: S100, the mapping preprocessing module 131 constructs a three-dimensional finite element sub-model of the target wind turbine pipe node. The mapping preprocessing module 131 discretizes and defines the target extrapolation node set on the three-dimensional finite element sub-model of the target wind turbine pipe node. The mapping preprocessing module 131 applies unit loads to the main pipe end and branch pipe end of the three-dimensional finite element sub-model of the target wind turbine pipe node in sequence according to the six degrees of freedom of the triaxial force and triaxial moment of each pipe end. The total number of unit loads is equal to the product of the number of pipe ends participating in the mapping and the number of degrees of freedom of each pipe end. The mapping preprocessing module 131 extracts the stress tensor corresponding to the target extrapolation node set and assembles it to generate a linear elastic spatial influence matrix.

[0043] Simultaneously, the mapping preprocessing module 131 applies unit gravitational field acceleration and unit overturning angle displacement of mud surface to the three-dimensional finite element sub-model of the target wind turbine pipe node, respectively. The mapping preprocessing module 131 calculates the additional stress response generated by the target extrapolated node set, and assembles and generates the unit gravitational stress response vector and mud surface angle displacement compensation matrix according to the additional stress response.

[0044] S200, the coupled dynamics module 132 constructs an overall aerodynamic, hydrodynamic, and elastic coupling model including wind turbine blades, nacelle, tower, and jacket. At the boundary of the interaction between the pile foundation entry section and the soil in the overall aerodynamic, hydrodynamic, and elastic coupling model, the coupled dynamics module 132 configures nonlinear spring elements along the discrete structural nodes in the direction of pile foundation depth. The coupled dynamics module 132 forms the pile-soil stiffness-resistance matrix in the initial state through the nonlinear spring elements.

[0045] S300, the coupled dynamics module 132 inputs the time history load sequence of the external environment to perform multi-field coupled dynamics solution. During the multi-field coupled dynamics solution process, the data reconstruction module 133 continuously monitors the stiffness degradation parameter of the pile-soil stiffness-impedance matrix. The stiffness degradation parameter is calculated based on the initial equivalent stiffness and the current equivalent secant stiffness in the lateral direction of the target at the mud surface elevation. When the stiffness degradation parameter reaches the set trigger threshold, the data reconstruction module 133 performs time-domain load segmentation based on the set trigger threshold.

[0046] The data reconstruction module 133 sets an overlapping time window with the stiffness update trigger time as the starting time. The data reconstruction module 133 performs smooth fusion on the local time history data through the overlapping window algorithm. Based on the smooth fusion result, the data reconstruction module 133 generates the reconstructed global end load time history matrix and mud surface angular displacement time history.

[0047] S400, the fatigue assessment module 134 receives the linear elastic spatial influence matrix, the unit gravity stress response vector, the mud surface angular displacement compensation matrix, the global end load time history matrix, and the mud surface angular displacement time history. The fatigue assessment module 134 inputs the global end load time history matrix and the mud surface angular displacement time history after coordinate transformation and degree of freedom rearrangement into the linear elastic spatial influence matrix, the unit gravity stress response vector, and the mud surface angular displacement compensation matrix to perform algebraic mapping operations.

[0048] After the algebraic mapping operation is completed, the fatigue assessment module 134 performs a spatial stress extrapolation operation on the result of the algebraic mapping operation in conjunction with the hot spot stress extrapolation criterion. The fatigue assessment module 134 generates the hot spot stress time history at the intersection line of the pipe node through the spatial stress extrapolation operation.

[0049] S500, fatigue assessment module 134 performs rainflow counting on hot spot stress time history. Fatigue assessment module 134 extracts stress amplitude and cycle count statistics through rainflow counting. Fatigue assessment module 134 matches the SN fatigue curve corresponding to the pipe node configuration based on the welding detail category, pipe node configuration, service environment, corrosion protection status and wall thickness parameters of the target wind turbine pipe node.

[0050] The fatigue assessment module 134 calculates the working condition damage degree of each working condition in combination with the Miner linear cumulative damage criterion. The fatigue assessment module 134 linearly superimposes the working condition damage degree of all working conditions and outputs the full life fatigue damage degree of the target fan pipe node.

[0051] See attached document Figure 3 Step S100 specifically includes the following sub-steps: S110, the mapping preprocessing module 131 constructs a three-dimensional finite element sub-model of the target wind turbine pipe node and performs parameter optimization closed loop. The mapping preprocessing module 131 obtains the structural parameters and material mechanical property parameters in the wind turbine design drawings. The structural parameters include the outer diameter of the main pipe, the wall thickness of the main pipe, the outer diameter of the branch pipe, the wall thickness of the branch pipe, and the intersection angle parameters of the target wind turbine pipe node. The material mechanical property parameters include the elastic modulus and Poisson's ratio of the steel.

[0052] The mapping preprocessing module 131 uses structural parameters to generate a three-dimensional geometric entity through Boolean intersection operation in a three-dimensional coordinate system, and uses material mechanical property parameters to assign physical reference properties to the three-dimensional geometric entity. In the specific finite element modeling process, the mapping preprocessing module 131 uses three-dimensional solid elements to mesh the intersection weld seam area where the main pipe and branch pipe intersect, while using shell elements to mesh the straight pipe extension section far away from the intersection weld seam area.

[0053] The mapping preprocessing module 131 establishes a kinematic multi-point constraint connection at the interface flat surface between the three-dimensional solid element and the shell element, and couples the translational degree of freedom of the boundary node of the three-dimensional solid element with the translational and rotational degree of freedom of the boundary node of the shell element, so as to realize the continuous transmission of spatial node displacement components and rotational components.

[0054] To ensure the convergence of stress calculation results and the accuracy of physical mapping in the three-dimensional finite element sub-model, the mapping preprocessing module 131 performs mesh convergence iteration and benchmark verification on the three-dimensional finite element sub-model. The above mesh convergence iteration and benchmark verification process constitutes the parameter iteration and state verification mechanism of the finite element analysis model.

[0055] The mapping preprocessing module 131 sets the initial mesh size to perform preliminary element division of the three-dimensional finite element sub-model and performs static calculations. It extracts the initial maximum principal stress at the weld toe position of the intersection line. The mapping preprocessing module 131 gradually reduces the mesh size of the three-dimensional solid element according to the set scaling factor of 0.5 to 0.8, and the mesh size reduction range is controlled between 20 mm and 5 mm. It also repeatedly calls the static solver of the mainstream large-scale commercial finite element analysis software ANSYS to generate a series of principal stress values ​​under different mesh refinement levels.

[0056] The mapping preprocessing module 131 calculates the relative change rate of principal stress between two adjacent iteration steps. When the relative change rate of principal stress is less than the set convergence tolerance, the mapping preprocessing module 131 stops the mesh refinement iteration. The set convergence tolerance is usually set to within 2%.

[0057] Subsequently, the mapping preprocessing module 131 compares the converged principal stress values ​​with the theoretical stress concentration coefficients calculated based on the empirical formulas of the International Institute of Welding. If the relative error is within the set allowable range of 3% to 5%, the mesh state and stiffness matrix parameters of the current three-dimensional finite element sub-model are locked, thereby completing the closed loop of three-dimensional finite element sub-model production and parameter tuning.

[0058] S120, the mapping preprocessing module 131 discretizes the extrapolation node set and assembles it to generate the linear elastic spatial influence matrix. The core principle of extracting the spatial influence matrix is ​​to use the linear elastic superposition principle to reduce the dimensionality of the complex local mesh solution to a pre-calculated matrix algebraic mapping. In order to accurately capture the stress concentration effect, the mapping preprocessing module 131 discretizes and defines the target extrapolation node set on the three-dimensional finite element sub-model. Based on the geometric extrapolation technology requirements for the assessment of hot spot stress in pipe nodes, the mapping preprocessing module 131 takes the intersection weld toe as the structural starting point and extracts the mesh node coordinates along the pipe wall surface and the path perpendicular to the normal direction of the intersection weld toe.

[0059] The mapping preprocessing module 131 uses the intersection line as a reference and divides multiple extrapolation paths along the circumferential angle. On each extrapolation path, it locates a physical node at a first reference distance from the outer edge of the weld toe of the intersection line and a physical node at a second reference distance from the outer edge of the weld toe of the intersection line. The first reference distance and the second reference distance are determined based on the reference wall thickness parameter of the pipe node. In a specific embodiment of the present invention, the first reference distance is specifically taken as 0.4 times the reference wall thickness, and the second reference distance is specifically taken as 1.0 times the reference wall thickness.

[0060] The 0.4 times reference wall thickness and 1.0 times reference wall thickness are standard value ranges determined by avoiding the singular region of notch stress at the weld toe root and capturing the nominal stress gradient of the structure. The mapping preprocessing module 131 sets and merges the physical nodes at the 0.4 times reference wall thickness position and the physical nodes at the 1.0 times reference wall thickness position to form the target extrapolation node set.

[0061] The reference wall thickness is the main pipe wall thickness or branch pipe wall thickness of the solid mesh where the target extrapolated node set is located in the target fan pipe node. The target extrapolated node set can completely enclose the saddle, crown and transition stress area between the saddle and crown of the pipe node structure in physical space where stress concentration occurs.

[0062] The mapping preprocessing module 131 applies unit loads to the main pipe end and the branch pipe end of the three-dimensional finite element sub-model respectively. The mapping preprocessing module 131 establishes control reference points at the centroid positions of the main pipe end section and the branch pipe end section of the three-dimensional finite element sub-model respectively. The mapping preprocessing module 131 establishes a rigid coupling constraint system between the edge nodes of the main pipe end section and the edge nodes of the branch pipe end section and the control reference points established at the centroid positions of the main pipe end section and the branch pipe end section respectively.

[0063] The mapping preprocessing module 131 sequentially and independently applies unit loads in six spatial degrees of freedom directions to the control reference point. The six spatial degrees of freedom directions include unit force components along three orthogonal spatial coordinate axes and unit torque components around three orthogonal spatial coordinate axes. In the set unit load, the unit force amplitude can be 1N and the unit torque amplitude can be 1N·m.

[0064] When a unit load is applied to any degree of freedom direction of any pipe end control reference point, the mapping preprocessing module 131 applies boundary constraints or equivalent elastic boundary constraints consistent with the overall aerodynamic and hydrodynamic and elastic coupling model to the other pipe end control reference points. By applying inertial release boundaries to achieve static equilibrium of the structure or configuring weak spring elements with minimal stiffness, the rigid body motion degrees of freedom are constrained to eliminate rigid body displacements of the three-dimensional finite element sub-model and ensure the uniqueness of the static solution results. The mapping preprocessing module 131 calls the underlying static solution program to perform finite element analysis and extracts the three-dimensional stress tensor output by the target extrapolation node set under each unit load.

[0065] The mapping preprocessing module 131 uses the three-dimensional stress tensor output by the target extrapolation node set under each unit load to assemble and generate a linear elastic spatial influence matrix. The mapping preprocessing module 131 arranges the stress components of all physical nodes in the target extrapolation node set under each single degree of freedom unit load into column vectors according to their spatial location numbers. The mapping preprocessing module 131 horizontally splices all the column vectors generated by all control reference points and all spatial degrees of freedom in a fixed order of pipe end number, degree of freedom number and stress component number to form a two-dimensional matrix data format.

[0066] The number of columns in the linear elastic space influence matrix is ​​equal to the product of the number of pipe ends participating in the mapping and the number of degrees of freedom for each pipe end. When the pipe ends participating in the mapping include main pipe ends and branch pipe ends, the number of columns in the linear elastic space influence matrix is ​​12. The mathematical expression for matrix assembly executed by the mapping preprocessing module 131 is: ; In the formula, The linear elastic spatial influence matrix; The stress response column vector output by the target extrapolation node set when a unit load in the first degree of freedom direction is applied to the pipe end of the three-dimensional finite element sub-model. The stress response column vector output by the target extrapolation node set when a unit load in the second degree of freedom direction is applied to the pipe end of the three-dimensional finite element sub-model; To apply the first... The stress response column vector output by the target extrapolation node set under a unit load in each degree of freedom direction; To apply the first... The stress response column vector output by the target extrapolation node set under a unit load in each degree of freedom direction; The degree of freedom number is the variable; The total number of degrees of freedom under a unit load.

[0067] S130, the mapping preprocessing module 131 calculates the additional stress response and generates a unit gravity stress response vector and a mud surface angular displacement compensation matrix. The mapping preprocessing module 131 applies a unit gravitational field acceleration and a unit overturning angular displacement of the mud surface to the three-dimensional finite element sub-model. To compensate for the nonlinear additional bending moment caused by gravity under large deformation of the wind turbine support structure and the secondary stress introduced by the tilting of the bottom foundation, the mapping preprocessing module 131 assigns material density properties to the discrete mesh of the three-dimensional finite element sub-model. The mapping preprocessing module 131 applies a uniformly distributed load with a magnitude equal to the unit gravitational acceleration value, for example, 1 m / s², in the direction perpendicularly downwards along the global gravity. 2 Or the unit value under the equivalent unit system.

[0068] The mapping preprocessing module 131 extracts the coordinates of the spatial nodes of the bottom mud surface of the overall jacket structure as the rotation center of the structure. Based on the spatial position of the target fan pipe node in the overall jacket coordinate system, it establishes the coordinate mapping relationship between the rotation center of the mud surface and the control reference point of the three-dimensional finite element sub-model. The unit overturning angle displacement of the mud surface is converted into the equivalent displacement boundary or equivalent additional load boundary of the pipe end of the three-dimensional finite element sub-model. The mapping preprocessing module 131 forces the three-dimensional finite element sub-model to generate a unit overturning angle displacement with a fixed radian value around the horizontal axis. The fixed radian value is set to 1 rad or a small rotation angle that meets the small deformation assumption. The specific value range of this small rotation angle is set to 0.01 rad to 0.05 rad.

[0069] The mapping preprocessing module 131 calculates the additional stress response generated by the target extrapolation node set, and assembles and generates the unit gravity stress response vector and mud surface angular displacement compensation matrix based on the additional stress response. The mapping preprocessing module 131 extracts the principal stress change values ​​of the target extrapolation node set at each node position under the action of unit gravity field acceleration and under the action of unit overturning angular displacement of mud surface through the linear static solver.

[0070] The mapping preprocessing module 131 organizes the principal stress change values ​​corresponding to the unit gravitational field acceleration into a unit gravitational stress response vector, and organizes the principal stress change values ​​corresponding to the unit overturning angular displacement of the mud surface into a mud surface angular displacement compensation matrix. When only a single horizontal overturning direction is considered, the mud surface angular displacement compensation matrix is ​​a single column vector; when multiple rotational degrees of freedom are considered, the number of columns in the mud surface angular displacement compensation matrix is ​​consistent with the dimension of the mud surface angular displacement vector.

[0071] The unit gravity stress response vector and the mud surface angular displacement compensation matrix independently play a compensation role in subsequent algebraic mapping operations, among which, This is the unit gravity stress response vector of the target extrapolation node set of the three-dimensional finite element sub-model under the action of unit gravity field acceleration; The mud surface angular displacement compensation matrix of the target extrapolation node set in the three-dimensional finite element submodel under the action of a unit overturning angular displacement of the mud surface. See attached document Figure 4 Step S200 specifically includes the following sub-steps: S210, the coupled dynamics module 132 constructs the overall aerodynamic, hydrodynamic and elastic coupling model and completes parameter optimization iteration. The coupled dynamics module 132 obtains the whole machine design data table of the wind turbine generator set. The whole machine design data table includes the aerodynamic airfoil coordinates of each section of the wind turbine blade, the outer diameter and wall thickness sequence of the tower along the elevation, the total mass and center of gravity coordinates of the nacelle, and the spatial topological node coordinates of the jacket.

[0072] The coupled dynamics module 132 uses the overall design data table to generate a three-dimensional spatial beam element and mass point matrix of the wind turbine blades, nacelle, tower and jacket, and constructs the initial topology of the overall aerodynamic and hydrodynamic and elastic coupling model. In this embodiment, the coupled dynamics module 132 uses beam elements to simulate the main load-bearing structure of the wind turbine blades, tower and jacket, and simplifies the nacelle into a rigid body node with concentrated mass and rotational inertia.

[0073] Before performing time-domain coupled calculations, in order to ensure that the overall aerodynamic, hydrodynamic and elastic coupled model can truly reflect the dynamic characteristics of the actual physical prototype and avoid modal distortion in high-frequency dynamic calculations, the coupled dynamics module 132 performs modal calibration and parameter tuning iterations on the overall aerodynamic, hydrodynamic and elastic coupled model. The above modal calibration and parameter tuning iteration process constitutes the parameter training and verification closed loop of the dynamic calculation model.

[0074] The coupled dynamics module 132 calls the eigenvalue solving algorithm to calculate the initial natural frequencies and mode shapes of the overall aerodynamic, hydrodynamic and elastic coupling model, focusing on extracting the first-order tower forward and backward swaying frequency and the first-order tower left and right swaying frequency of the overall model.

[0075] The coupled dynamics module 132 compares the first-order tower's forward and backward swaying frequency and its left and right swaying frequency with the first and third harmonic excitation ranges of the wind turbine's rotation frequency. If the natural frequency falls into the resonance excitation range or the error with the factory-calibrated frequency of the physical prototype exceeds the set limit of 2% to 3%, the coupled dynamics module 132 iteratively fine-tunes the local stiffness multipliers of the tower beam unit and the eccentric inertia parameters of the nacelle's concentrated mass point based on the internally established sensitivity analysis matrix. The above-mentioned coupled dynamics time-domain solution process can be realized by internally calling the dynamic calculation engine of the international open-source wind turbine aeroelastic coupling analysis tool OpenFAST.

[0076] The coupled dynamics module 132 repeatedly executes the eigenvalue solving and parameter fine-tuning steps until the core natural frequency of the overall aerodynamic, hydrodynamic and elastic coupling model avoids the resonance range and meets the calibration error requirements, thereby locking the global stiffness and mass matrix state of the dynamic model.

[0077] S220, the coupled dynamics module 132 configures the input conditions of the multi-field environmental load sequence and generates the pile-soil stiffness-impedance matrix in the initial state. In order to drive the subsequent multi-field coupled dynamics solution, the coupled dynamics module 132 configures the input conditions of the multi-field environmental load sequence. The coupled dynamics module 132 obtains the input average wind speed, wind shear index and turbulence intensity parameters, and calls the three-dimensional turbulence wind field generation algorithm to generate the aerodynamic wind speed time series acting on the wind turbine blades. The turbulence spectrum calculation method used here can be specifically based on the conventional Kaimal turbulence spectrum or von Karman turbulence spectrum in this field.

[0078] The coupled dynamics module 132 acquires the effective wave height and spectral peak period parameters, and calculates the wave particle water velocity and water acceleration by combining the wave spectrum function and wave kinematics analytical theory. The wave spectrum function can be selected from the JONSWAP spectrum or the Pierson-Moskowitz spectrum, which are suitable for the hydrological characteristics of the target sea area. At the same time, the coupled dynamics module 132 acquires the surface current velocity parameters to generate the current profile distribution that varies with water depth. The coupled dynamics module 132 converts the wave particle water velocity, wave particle water acceleration and current profile distribution into a hydrodynamic time series load acting on the surface of the underwater structure of the jacket according to the Morison equation.

[0079] When constructing the foundation boundary constraints of the overall aerodynamic, hydrodynamic and elastic coupling model, the coupled dynamics module 132 configures nonlinear spring elements at each discrete structural node along the depth direction of the pile foundation at the interaction boundary between the pile foundation entry section and the soil in the overall aerodynamic, hydrodynamic and elastic coupling model. Considering that the offshore wind turbine foundation will bear the reciprocating lateral load brought by the environment for a long time, the soil around the pile will exhibit nonlinear yielding characteristics.

[0080] The coupled dynamics module 132 configures a py spring representing horizontal lateral constraint, a tz spring representing axial lateral friction, and a Qz spring representing pile end bearing capacity at each discrete structural node along the pile foundation penetration depth. The coupled dynamics module 132 obtains the soil physical and mechanical parameters from the geological survey report of the target sea area. The soil physical and mechanical parameters include the undrained shear strength, effective unit weight, and internal friction angle of each soil layer. The coupled dynamics module 132 uses the soil physical and mechanical parameters and combines them with the empirical formulas in the API or DNV specifications known in the field of marine engineering to calculate and generate the nonlinear load-displacement relationship curves of the nonlinear spring element at each soil layer depth.

[0081] The coupled dynamics module 132 forms the pile-soil stiffness-resistance matrix in the initial state through nonlinear spring elements. From the perspective of physical state evaluation, at the initial moment of dynamic calculation, the wind turbine foundation is in a static equilibrium state without being subjected to external dynamic environmental loads. The coupled dynamics module 132 extracts the initial tangent stiffness value of the nonlinear load-displacement relationship curve at the zero displacement point and constructs the local foundation stiffness matrix of each discrete structural node in the local coordinate system.

[0082] The coupled dynamics module 132 maps the local foundation stiffness matrices at all discrete structural nodes to the global coordinate system through coordinate transformation matrices and performs finite element stiffness integration operations. The coupled dynamics module 132 generates the calculation expression for the pile-soil stiffness-resistance matrix in the initial state as follows: ; In the formula, This represents the pile-soil stiffness-resistance matrix under initial conditions. The sequence number variable for the configured nonlinear spring unit; The total number of discrete structural nodes for configuring nonlinear spring units; For the first The coordinate transformation matrix of the node containing a nonlinear spring element from the local coordinate system to the global coordinate system; For the first The transpose of the coordinate transformation matrix corresponding to each nonlinear spring element; For the first The local foundation stiffness matrix of a nonlinear spring element at the initial displacement zero point.

[0083] See attached document Figure 5 Step S300 specifically includes the following sub-steps: S310, when the coupled dynamics module 132 performs the solution, the data reconstruction module 133 monitors the stiffness degradation parameters of the pile-soil stiffness-impedance matrix and performs time-domain load segmentation. The coupled dynamics module 132 inputs the time-history load sequence of the external environment to perform multi-field coupled dynamics solution. During the multi-field coupled dynamics solution process, as the wind and wave loads continue to be applied, the soil around the pile will undergo irreversible plastic deformation. At the structural dynamics level, the plastic deformation of the soil is directly reflected as the attenuation of the nonlinear impedance of the support boundary.

[0084] To accurately capture the long-term impact of weakened foundation stiffness on the dynamic response of the upper wind turbine pipe nodes, the data reconstruction module 133 continuously monitors the stiffness degradation parameters of the pile-soil stiffness-impedance matrix. Within a specific dynamic solution time step, the data reconstruction module 133 extracts in real time the horizontal displacement values ​​of discrete structural nodes at the mud surface elevation and the lateral resistance values ​​output by the nonlinear spring elements configured at the discrete structural nodes at the mud surface elevation.

[0085] When the absolute value of the horizontal displacement is greater than the preset lower displacement limit of 0.001m to 0.005m, the data reconstruction module 133 divides the lateral resistance value by the horizontal displacement value to calculate the equivalent secant stiffness at the current time step. When the absolute value of the horizontal displacement is less than or equal to the preset lower displacement limit of 0.001m to 0.005m, the data reconstruction module 133 uses the initial tangent stiffness or the equivalent secant stiffness of the previous effective time step as the equivalent secant stiffness at the current time step. The calculation of the equivalent secant stiffness provides a quantitative physical basis for the dynamic adjustment of the time-domain integral boundary conditions of structural dynamics.

[0086] The data reconstruction module 133 calculates the stiffness degradation parameters based on the equivalent secant stiffness and the initial reference stiffness, and compares the stiffness degradation parameters with the set trigger threshold. The mathematical expression for the stiffness degradation parameters is as follows: ; In the formula, This is the stiffness degradation parameter; This represents the initial equivalent stiffness of the pile-soil stiffness-impedance matrix at the mud surface elevation. This represents the equivalent secant stiffness at the mud surface elevation at the current solution time step.

[0087] When the stiffness degradation parameter reaches the set trigger threshold, the data reconstruction module 133 performs time-domain load segmentation based on the set trigger threshold. In order to avoid unnecessary consumption of computing resources caused by excessively frequent matrix updates, and to prevent structural response distortion caused by stiffness update lag, the set trigger threshold is set to a range of 5% to 15%. The set trigger threshold is a conventional empirical range that achieves a balance between engineering calculation accuracy and calculation efficiency. In this embodiment, the data reconstruction module 133 sets the set trigger threshold to 10%.

[0088] Once the stiffness degradation parameter is greater than or equal to 10%, the data reconstruction module 133 immediately sends an interrupt signal to pause the current multi-field coupled dynamics solution process of the coupled dynamics module 132. The data reconstruction module 133 saves the global displacement vector, velocity vector and acceleration vector of the structure at the moment of interruption of the current multi-field coupled dynamics solution process as local state boundary conditions.

[0089] Subsequently, the data reconstruction module 133 calls the current equivalent secant stiffness to reconstruct the nonlinear spring element and updates the pile-soil stiffness-impedance matrix. When reconstructing the nonlinear spring element, the data reconstruction module 133 uses the ratio of the current equivalent secant stiffness to the initial equivalent stiffness as the stiffness reduction factor to proportionally correct the load-displacement relationship curves of the py spring, tz spring, or Qz spring at the corresponding depth, and inherits the spring displacement, spring internal force, and structural state variables from the previous stage. The data reconstruction module 133 restarts the dynamic time-domain integration operation of the next stage using the saved local state boundary conditions as initial conditions.

[0090] S320, the data reconstruction module 133 sets an overlapping time window to perform smooth fusion, generating the reconstructed global end load time history matrix and mud surface angular displacement time history. Since the stiffness matrix of the nonlinear spring element undergoes a numerical mutation at the segmented connection point, directly splicing the solution results of two adjacent segments will introduce numerical truncation error, resulting in high-frequency false oscillation of the structure. The high-frequency false oscillation of the structure caused by the numerical truncation error will interfere with the accuracy of the fatigue assessment of the pipe node.

[0091] Therefore, the data reconstruction module 133 sets an overlapping time window and performs smooth fusion of local time history data through the overlapping window algorithm. During the continuous calculation period after the interruption time of the current multi-field coupled dynamics solution process, the data reconstruction module 133 extracts an overlapping time window with the interruption time as the start time and the 1s to 3s after the interruption time as the end time.

[0092] To achieve effective data overlap, the data reconstruction module 133 controls the dynamic solver to generate a calculation branch at the time of interruption: on the one hand, it retains the old pile-soil stiffness-impedance matrix that has not been updated, so that the dynamic integral operation under the unupdated stiffness reference continues to be executed from the time of interruption to the end of the overlap time window, and extracts and generates the preceding data sequence. On the other hand, using the updated pile-soil stiffness-impedance matrix and the saved local state boundary conditions, the integration operation is restarted from the interruption time to the end of the overlapping time window to extract and generate the subsequent data sequence. The preceding data sequence and the subsequent data sequence have the same time start, time end, and time step. The data reconstruction module 133 applies a cosine-weighted smoothing function to mathematically reconstruct the preceding and subsequent data sequences. The mathematical equation for smooth fusion is as follows: ; ; In the formula, For smoothed and fused time history data; The current time variable within the overlapping time window; For the current time variable The corresponding smoothing weight coefficients; For the preceding data sequence at the current time variable The response value below; For the subsequent data sequence at the current time variable The response value below; Pi is a constant. This represents the start time of the overlapping time window; This represents the end time of the overlapping time window.

[0093] After completing the smooth fusion of all time-domain load segments, the data reconstruction module 133 aligns and stitches the smooth fusion results along the overall time axis. From the stitched global response data, the data reconstruction module 133 extracts the time series data of the six degrees of freedom internal forces at each pipe end of the target wind turbine pipe node at the centroid position of the main pipe end and branch pipe end section, which correspond one-to-one with the control reference point position of the three-dimensional finite element sub-model. This generates the reconstructed global end load time history matrix. The pipe end number, degree of freedom order, and positive moment direction in the global end load time history matrix are consistent with the arrangement order of the unit load case in the linear elastic space influence matrix.

[0094] Meanwhile, the data reconstruction module 133 extracts the spatial rotational angular displacement sequence of the nodes at the mud surface elevation of the overall jacket structure, and generates the reconstructed mud surface angular displacement time history. The global end load time history matrix and the mud surface angular displacement time history are persistently stored in a two-dimensional matrix format with equal time steps, providing a continuous time-domain load boundary for subsequent stress mapping.

[0095] See attached document Figure 6 Step S400 specifically includes the following sub-steps: S410, the fatigue assessment module 134 receives the preceding data matrix and performs algebraic mapping operations to generate a dynamic stress time history matrix. The fatigue assessment module 134 receives the linear elastic spatial influence matrix, the unit gravity stress response vector, the mud surface angular displacement compensation matrix, the global end load time history matrix, and the mud surface angular displacement time history. In the fatigue analysis system of a linear structure, the dynamic stress response of the structure can be equivalently replaced by a linear combination of the static stress field and the dynamic time history load. The fatigue assessment module 134 inputs the global end load time history matrix and the mud surface angular displacement time history into the linear elastic spatial influence matrix, the unit gravity stress response vector, and the mud surface angular displacement compensation matrix to perform algebraic mapping operations.

[0096] In the specific algebraic mapping operation, the fatigue assessment module 134 extracts the six-degree-of-freedom end load vector corresponding to each discrete time step in the global end load time history matrix, and the fatigue assessment module 134 extracts the spatial rotation angular displacement value corresponding to each discrete time step in the mud surface angular displacement time history.

[0097] To ensure the mathematical validity of algebraic mapping operations in vector space, the fatigue assessment module 134 establishes a column dimension matching mechanism between the end load vectors of the multi-pipe end and the linear elastic space influence matrix. This ensures that the triaxial force and triaxial moment components of each pipe end in the global end load time history matrix correspond one-to-one with the static unit load conditions of each pipe end in each degree of freedom direction in the linear elastic space influence matrix, in the same coordinate system order. Before the algebraic mapping operation, the fatigue assessment module 134 performs direction transformation and component rearrangement of the end load vectors based on the coordinate transformation matrix between the global coordinate system of the overall model and the local coordinate system of the target pipe node.

[0098] The fatigue assessment module 134 performs matrix multiplication between the end load vector of the multi-pipe after coordinate transformation and degree of freedom rearrangement and the linear elastic space influence matrix to calculate the nominal dynamic stress response of the structure caused by the end dynamic load of each pipe at each discrete time step in the global end load time history matrix.

[0099] Simultaneously, the fatigue assessment module 134 extracts the preset gravity proportionality coefficient, multiplies the unit gravity stress response vector with the gravity proportionality coefficient, and multiplies the mud surface angular displacement vector corresponding to the mud surface angular displacement time history at the current discrete time step with the mud surface angular displacement compensation matrix to calculate the additional dynamic stress response introduced by the gravity field and the tilt of the supporting foundation; wherein, the gravity proportionality coefficient is the ratio of the actual gravitational acceleration value of the physical environment where the target wind turbine is located to the unit gravitational field acceleration value applied during preprocessing.

[0100] The fatigue assessment module 134 algebraically superimposes the nominal dynamic stress response and the additional dynamic stress response of the structure in the time domain to generate the dynamic stress time history matrix of the target extrapolated node set on the global time axis. The mathematical expression for the algebraic mapping operation performed by the fatigue assessment module 134 is as follows: ; In the formula, Extrapolating the target node set in time variable The dynamic stress response vector at the location; The time variable at the current discrete time step; The linear elastic spatial influence matrix; The global end load time history matrix in time variable The six-degree-of-freedom end load vector at the location; This is the unit gravity stress response vector of the target extrapolation node set of the three-dimensional finite element sub-model under the action of unit gravity field acceleration; This is the gravity proportionality coefficient; The mud surface angular displacement compensation matrix is ​​the target extrapolation node set of the three-dimensional finite element sub-model under the action of unit overturning angular displacement of the mud surface; The time history of mud surface angular displacement in time variable The spatial rotational angular displacement value at that location.

[0101] Algebraic mapping operations, based on a dimension reduction calculation method of matrix multiplication and algebraic superposition, avoid repeatedly calling the underlying finite element solver in each time step, thus improving the processing efficiency of long-term multi-field coupled data.

[0102] S420, fatigue assessment module 134 performs spatial stress extrapolation calculation in conjunction with hot spot stress extrapolation criterion to generate hot spot stress time history. After the algebraic mapping calculation is completed, fatigue assessment module 134 performs spatial stress extrapolation calculation on the result of algebraic mapping calculation in conjunction with hot spot stress extrapolation criterion. In order to accurately assess the real fatigue damage at the weld toe position of the intersection line of the pipe node in the target wind turbine pipe node, it is necessary to remove the local stress concentration and mesh singular interference caused by geometric discontinuity at the root of the weld toe.

[0103] The fatigue assessment module 134 extracts the first stress time series data corresponding to the physical node at a position of 0.4 times the reference wall thickness and the second stress time series data corresponding to the physical node at a position of 1.0 times the reference wall thickness from the dynamic stress time history matrix. The first stress time series data and the second stress time series data are the time history data of the preset stress evaluation components. The preset stress evaluation components are the maximum principal stress components or the structural stress components perpendicular to the direction of the weld toe of the intersection line, and they are kept consistent in the same fatigue assessment process.

[0104] The fatigue assessment module 134, based on the linear geometric extrapolation criterion, calculates the hot spot stress time history at the intersection line of pipe nodes using the first and second stress time series data. Based on the multiple extrapolation paths divided when the target extrapolation node set is generated by the mapping preprocessing module 131, each extrapolation path includes a physical node at a position of 0.4 times the reference wall thickness and a physical node at a position of 1.0 times the reference wall thickness. The fatigue assessment module 134 performs hot spot stress extrapolation calculations for each extrapolation path. The mathematical formula for the spatial stress extrapolation calculation performed by the fatigue assessment module 134 is: ; In the formula, The position of the pipe node intersection line in time variable Numerical values ​​of hot spot stress response at the location; The time variable at the current discrete time step; This is the extrapolation coefficient for the first spatial stress; For the first stress time series data in time variable The stress value at the location; This is the extrapolation coefficient for the second spatial stress; For the second stress time series data in time variable The stress value at that location.

[0105] For determining the spatial stress extrapolation coefficient, the fatigue assessment module 134 takes the value according to the geometric interpolation ratio given by the DNV or CIDECT specifications known in the marine engineering field. In one embodiment of the present invention, based on the linear extrapolation path of 0.4 times the reference wall thickness and 1.0 times the reference wall thickness, and based on the mathematical interpolation algorithm that determines a straight line from two points and solves the ordinate intercept at the x-coordinate of 0, the fatigue assessment module 134 determines the value of the spatial stress extrapolation coefficient corresponding to the first stress time series data as 1.67 and the value of the spatial stress extrapolation coefficient corresponding to the second stress time series data as -0.67.

[0106] The fatigue assessment module 134 generates the hot spot stress time history at the intersection line of the pipe node through spatial stress extrapolation calculation. When the target extrapolation node set includes multiple extrapolation paths, the fatigue assessment module 134 generates the hot spot stress time history corresponding to each extrapolation path, and calculates the fatigue damage degree of each extrapolation path in the subsequent fatigue damage calculation. The maximum fatigue damage degree or damage envelope value is taken as the fatigue damage degree result of the target wind turbine pipe node. The hot spot stress time history reflects the structural stress state of the pipe node intersection line position dynamically evolving with multiple external environmental loads.

[0107] See attached document Figure 7 Step S500 specifically includes the following sub-steps: S510, the fatigue assessment module 134 performs extreme point screening and rainflow counting processing on the hot spot stress time history. The fatigue assessment module 134 receives the hot spot stress time history at the intersection line of the pipe node. In order to improve the calculation efficiency of fatigue counting and filter out high-frequency numerical noise, the fatigue assessment module 134 performs extreme point screening on the hot spot stress time history. The fatigue assessment module 134 identifies the local maximum and local minimum values ​​in the hot spot stress time history, and removes invalid stress cycles where the adjacent difference between the local maximum and local minimum values ​​is less than the set filtering threshold. The specific value range of the set filtering threshold is 1MPa to 2MPa, and an alternating stress extreme value sequence is generated.

[0108] According to the principle of cumulative damage due to material fatigue, small-amplitude alternating stress contributes little to the overall fatigue damage of the structure, and retaining small-amplitude alternating stress data will increase the computational cost of subsequent rainflow counting. In one embodiment of the present invention, the filtering threshold can be set to 2% to 5% of the fatigue limit of the corresponding pipe node material. The basic algorithm for stress sequence extreme point identification and invalid fluctuation elimination can be based on existing signal processing technology.

[0109] The fatigue assessment module 134 performs four-point rainflow counting processing on the alternating stress extreme value sequence. The fatigue assessment module 134 extracts the stress amplitude that constitutes a closed hysteresis loop in the alternating stress extreme value sequence, and performs statistics on the extracted stress amplitude according to the preset stress level division interval to obtain each stress amplitude and the actual number of cycles corresponding to each stress amplitude.

[0110] S520, fatigue assessment module 134 matches SN fatigue curve and calculates the allowable number of cycles. Fatigue assessment module 134 obtains the reference SN curve parameters of the material corresponding to the target wind turbine pipe node, and determines the corresponding SN fatigue curve according to the welding detail category, pipe node configuration, service environment, corrosion protection status and wall thickness parameters of the target wind turbine pipe node. The wall thickness of the offshore wind turbine jacket pipe node is greater than the thickness of the standard test specimen. The thickness effect will reduce the fatigue life of the target wind turbine pipe node.

[0111] To make theoretical calculations more closely reflect the fatigue attenuation characteristics of actual engineering structures, the fatigue assessment module 134 introduces a thickness correction index to reduce the stress amplitude of the benchmark SN curve. Based on the SN curve equation, the fatigue assessment module 134 calculates the permissible number of cycles corresponding to each stress amplitude. The stress amplitude or stress range output by the rainflow count is consistent with the stress parameter definition in the SN curve equation. When the determined SN fatigue curve is a bilinear or multilinear form, the fatigue assessment module 134 adaptively calls the SN curve intercept and slope parameters corresponding to the interval representing the stress amplitude or number of cycles. The mathematical equation for fatigue life calculation executed by the fatigue assessment module 134 is as follows: ; In the formula, In the first Each stress level division interval represents the permissible number of cycles a material can withstand under stress amplitude. The index variable is used to divide the stress level intervals; This is the intercept parameter of the SN curve; For the first The representative stress amplitude of each stress level division interval; The actual wall thickness of the target fan pipe node; Used as a reference wall thickness; For thickness correction index; This is the slope parameter of the SN curve.

[0112] In practice, the specific values ​​of the reference wall thickness and the thickness correction index can be obtained from the well-known marine engineering design specifications in the field. For example, the reference wall thickness is often taken as 16mm or 32mm, and the value range of the thickness correction index can be set between 0.15 and 0.25.

[0113] S530, the fatigue assessment module 134 outputs the full-life fatigue damage degree based on the linear fatigue cumulative damage theory. The fatigue assessment module 134 calculates the short-term fatigue damage degree under the current environmental conditions based on the linear fatigue cumulative damage theory. The short-term fatigue damage degree corresponds to the fatigue damage within a single standard simulation duration under the current environmental conditions. The single standard simulation duration is a preset time history calculation time block. The fatigue assessment module 134 divides the actual number of cycles corresponding to each stress amplitude by the allowable number of cycles corresponding to each stress amplitude, and accumulates the damage contribution of all stress amplitude levels to generate the short-term fatigue damage degree. The mathematical expression for the short-term fatigue damage degree calculation performed by the fatigue assessment module 134 is as follows: ; In the formula, This represents the short-term fatigue damage under current environmental conditions. The index variable is used to divide the stress level intervals; This represents the total number of stress level intervals. For the first The actual number of cycles corresponding to each stress amplitude; For the first Each stress level division interval represents the permissible number of cycles a material can withstand under stress amplitude.

[0114] The fatigue assessment module 134 calculates the full-life fatigue damage degree by combining the sea state scatter map. The fatigue assessment module 134 obtains the probability distribution data of various environmental conditions at the wind farm site within the design life cycle. The fatigue assessment module 134 multiplies the short-term fatigue damage degree of all discrete environmental conditions by the probability of occurrence corresponding to each discrete environmental condition, and then multiplies it by the total number of time blocks corresponding to the design life cycle, summing the results to generate the full-life fatigue damage degree. The mathematical expression for calculating the full-life fatigue damage degree by the fatigue assessment module 134 is as follows: ; In the formula, The fatigue damage degree over the entire lifespan; The total number of time blocks within the design lifecycle; This is a variable representing the sequence number of the environmental operating condition; The total number of environmental operating conditions; For the first The probability of an environmental condition occurring throughout its entire life cycle; For the first The short-term fatigue damage degree corresponding to each environmental working condition.

[0115] The fatigue assessment module 134 compares the full-life fatigue damage with the set fatigue safety threshold and outputs the structural safety assessment result of the target wind turbine pipe node. Under normal circumstances, the theoretical upper limit of the set fatigue safety threshold used as the basis for structural safety judgment is 1. The set fatigue safety threshold is preset according to the safety level, design specifications and safety factor requirements adopted by the target project. When the safety redundancy requirements of specific projects need to be considered, the set fatigue safety threshold can be conservatively set to a value less than 1, specifically a safety tolerance value between 0.8 and 0.95. The fatigue calculation parameters of the relevant target wind turbine pipe node material can be directly retrieved from the standard data in the DNV-RP-C203 specification library of Det Norske Veritas.

[0116] To aid in understanding the technical solution of this invention, a specific application example of fatigue assessment for fan pipe nodes is provided below.

[0117] This embodiment performs fatigue assessment on a K-type intersecting pipe node at the bottom layer of the jacket foundation of a 10MW offshore wind turbine. The structural parameters are as follows: the outer diameter of the main pipe is 1500mm, the wall thickness of the main pipe is 40mm; the outer diameter of the branch pipe is 800mm, the wall thickness of the branch pipe is 25mm, the material is marine engineering structural steel, and the elastic modulus is set to 206GPa.

[0118] In the mapping preprocessing stage, the mapping preprocessing module 131 establishes a local three-dimensional finite element sub-model of the K-type pipe node. The mapping preprocessing module 131 uses the main pipe wall thickness of 40mm as the reference parameter and extracts solid mesh nodes at distances of 16mm and 40mm from the weld toe along the normal direction perpendicular to the weld toe of the intersection line, and combines them into a target extrapolation node set.

[0119] The mapping preprocessing module 131 independently applies unit loads in six degrees of freedom directions to the ends of the main branch pipe, extracts the stress tensor of the target extrapolation node set and splices them to generate a linear elastic spatial influence matrix, while simultaneously applying a 1 m / s² load. 2 The downward gravitational field and the mud surface overturning angle displacement of 0.02 rad generate a unit gravity stress response vector and a mud surface angle displacement compensation matrix.

[0120] In the multi-field coupled dynamics solution stage, the coupled dynamics module 132 loads the overall structural parameters of the wind turbine generator set, configures nonlinear spring elements characterizing the shear strength of the soil layer at the nodes of the pile foundation entry section, assembles the initial pile-soil stiffness impedance matrix, and loads a sequence of random turbulent wind field and irregular wave environmental loads with a total time domain length of 600 seconds.

[0121] During the time-domain calculation, the data reconstruction module 133 calculates the equivalent secant stiffness of the pile and soil at the mud surface in real time. When the calculation reaches the 213.5th second, the spring stiffness degradation parameter at the mud surface elevation reaches the set 10% trigger threshold. The data reconstruction module 133 pauses the current time-domain integration step, updates the nonlinear spring element with the current equivalent secant stiffness, and then extracts the pre-sequence and post-sequence of the period from the 213.5th second to the 215.5th second obtained before and after the update. It performs a smooth fusion operation based on cosine weights and splices them to generate a global end load time history matrix covering a complete 600-second cycle.

[0122] During the fatigue assessment phase, the fatigue assessment module 134 performs coordinate transformation on the global end load time history matrix, performs matrix multiplication on the global end load time history matrix after coordinate transformation and the linear elastic spatial influence matrix, and superimposes gravity and mud surface angular displacement compensation terms to generate a dynamic stress response vector. The fatigue assessment module 134 extracts the stress time series data at 16mm and 40mm, substitutes them with extrapolation coefficients of 1.67 and -0.67 respectively, and generates the hot spot stress time history at the pipe node intersection line position.

[0123] After removing invalid stress fluctuations by setting a filter threshold of 1.2 MPa, four-point rainflow counting is performed. The fatigue assessment module 134 calls the corresponding T-node SN fatigue curve, introduces a thickness correction index of 0.2 to calculate the allowable number of cycles for each stress amplitude, accumulates according to the Miner criterion to obtain the short-term fatigue damage degree of the environmental condition, and then sums and outputs the full-life fatigue damage degree by combining the full-life sea state scatter map.

[0124] To verify the technical effectiveness of this scheme, the full-size transient finite element method and the constant stiffness nominal stress method were selected for benchmark comparison. The full-size transient finite element method established a single overall model of the complete wind turbine and fine pipe node for nonlinear transient solution; the constant stiffness nominal stress method kept the initial pile-soil stiffness unchanged throughout the entire calculation cycle, and the calculation input conditions were the same 600-second test sea state.

[0125] Table 1: Comparative Experimental Data of Fatigue Damage Assessment Methods

[0126] According to Table 1 and Figure 8 It can be seen that the full-size transient finite element method takes 487215.4 seconds, while the technical solution of the present invention takes 1423.6 seconds. The difference in time consumption indicates that the present solution achieves dimensionality reduction by mapping the linear elastic spatial influence matrix constructed by the preprocessing module 131. In the time domain calculation, the processor 110 uses matrix algebra mapping to replace the stepwise iterative decomposition of the underlying stiffness matrix of the finite element, thereby reducing the computational resource consumption of long-term multi-field coupled simulation.

[0127] The peak hotspot stress output by the constant stiffness nominal stress method is 114.61 MPa, and the full-life fatigue damage degree is 0.5419. The peak hotspot stress of the technical solution of this invention is 134.82 MPa, and the damage degree is 0.7952, which is close to the full-size benchmark method's 136.27 MPa and 0.8103. This indicates that the data reconstruction module 133 corrects the deficiency of the constant stiffness method in underestimating structural stress and fatigue damage by dynamically monitoring and updating the nonlinear spring stiffness and taking into account the weakening effect of the support foundation caused by soil yielding.

[0128] The truncation error of both the full-size transient finite element method and the constant stiffness nominal stress method has a high-frequency cycle number of 0. The full-size transient finite element method performs continuous nonlinear iterations during the solution process, while the constant stiffness nominal stress method does not update the stiffness throughout the entire calculation cycle. Neither of them has experienced any interventionist numerical abrupt changes in the stiffness matrix, so no truncation error is generated.

[0129] The technical solution of this invention performs piecewise updates of the stiffness matrix based on a set trigger threshold. The number of high-frequency cycles of the recorded truncation error is 3. The data reconstruction module 133 performs smooth fusion of the preceding and following sequences by setting an overlapping time window, controls the dynamic response step phenomenon caused by the piecewise update of stiffness, and the smoothing process suppresses most of the false high-frequency numerical noise. This avoids the fatigue assessment module 134 from including a large number of truncation errors in the stress cycle during the rainflow counting stage, thus maintaining the effectiveness of the cumulative damage calculation while taking into account the computational efficiency.

[0130] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A fatigue failure assessment method for fan pipe joints based on linear damage criteria, characterized in that, The method includes the following steps: A three-dimensional finite element sub-model of the target wind turbine pipe node is constructed and a target extrapolation node set is defined. A unit load is applied to the three-dimensional finite element sub-model to extract the stress tensor to generate a linear elastic spatial influence matrix, and a unit gravity stress response vector and mud surface angular displacement compensation matrix are generated. An overall aerodynamic, hydrodynamic, and elastic coupling model was constructed, and nonlinear spring elements were configured in the pile foundation penetration section to form a pile-soil stiffness-resistance matrix. The input time history load sequence is used to perform the solution, and the stiffness degradation parameter of the pile-soil stiffness-impedance matrix is ​​monitored. If the stiffness degradation parameter reaches the set trigger threshold, an overlapping time window is set to perform smooth fusion and generate a global end load time history matrix and mud surface angle displacement time history. The global end load time history matrix and the mud surface angular displacement time history are input into the linear elastic spatial influence matrix, the unit gravity stress response vector and the mud surface angular displacement compensation matrix to perform algebraic mapping operation and spatial stress extrapolation operation to generate hot spot stress time history; Rainflow counting is performed on the hot spot stress time history, and the full-life fatigue damage degree is output by combining the matched SN fatigue curve and the linear cumulative damage criterion.

2. The fatigue failure assessment method for wind turbine pipe nodes based on linear damage criteria according to claim 1, characterized in that, The step of applying a unit load to the three-dimensional finite element sub-model to extract the stress tensor and generate the linear elastic spatial influence matrix includes: Establish control reference points at the centroid positions of the main pipe end section and the branch pipe end section of the three-dimensional finite element sub-model; Unit force components along the spatial coordinate axes and unit torque components around the spatial coordinate axes are sequentially and independently applied to the control reference point. After eliminating the rigid body displacement of the three-dimensional finite element sub-model, the three-dimensional stress tensor output by the target extrapolation node set is extracted. The stress components of all physical nodes in the target extrapolation node set are arranged into column vectors and horizontally spliced ​​according to a fixed order of pipe end number, degree of freedom number and stress component number to form the linear elastic spatial influence matrix in a two-dimensional matrix data format.

3. The fatigue failure assessment method for wind turbine pipe nodes based on linear damage criteria according to claim 1, characterized in that, The steps for generating the unit gravity stress response vector and the mud surface angular displacement compensation matrix include: A uniformly distributed load of unit gravitational field acceleration is applied in the direction perpendicular to the downward direction of global gravity, forcing the three-dimensional finite element sub-model to generate a unit overturning angle displacement of a fixed radian value around the horizontal axis; Extract the principal stress change values ​​corresponding to the target extrapolation node set under the action of the unit gravitational field acceleration value, and organize them into the unit gravitational stress response vector; Extract the principal stress change values ​​corresponding to the target extrapolation node set under the unit overturning angle displacement, and organize them into the mud surface angle displacement compensation matrix.

4. The fatigue failure assessment method for fan pipe nodes based on linear damage criteria according to claim 1, characterized in that, The step of configuring nonlinear spring elements in the pile foundation penetration section to form a pile-soil stiffness-resistance matrix includes: At discrete structural nodes along the depth direction of the pile foundation, a py spring representing horizontal lateral constraint, a tz spring representing axial lateral friction, and a Qz spring representing pile end bearing capacity are respectively configured, and a nonlinear load-displacement relationship curve is generated. Extract the initial tangent stiffness value of the nonlinear load-displacement relationship curve at the zero displacement point, and construct a local foundation stiffness matrix; The local foundation stiffness matrix is ​​mapped to the global coordinate system through a coordinate transformation matrix and finite element stiffness integration is performed to generate the pile-soil stiffness-resistance matrix in the initial state.

5. The fatigue failure assessment method for wind turbine pipe nodes based on linear damage criteria according to claim 1, characterized in that, The steps for monitoring the stiffness degradation parameters of the pile-soil stiffness-resistance matrix include: Extract the horizontal displacement and lateral resistance values ​​of discrete structural nodes at the mud surface elevation, and calculate the equivalent secant stiffness by dividing the lateral resistance value by the horizontal displacement value. The stiffness degradation parameter is calculated based on the equivalent secant stiffness and the initial equivalent stiffness of the pile-soil stiffness-impedance matrix at the mud surface elevation. When the stiffness degradation parameter reaches the set trigger threshold, an interrupt signal is sent to pause the multi-field coupled dynamics solution process, and the global displacement vector, velocity vector and acceleration vector of the structure at the time of interruption are saved as local state boundary conditions.

6. The fatigue failure assessment method for wind turbine pipe nodes based on linear damage criteria according to claim 5, characterized in that, The steps of setting an overlapping time window to perform smooth fusion and generate the global end load time history matrix and mud surface angular displacement time history include: Set the overlapping time window with the interruption time as the starting time; The unupdated pile-soil stiffness-impedance matrix is ​​retained, and the dynamic integration operation under the unupdated stiffness reference continues to be executed until the end of the overlapping time window, thereby extracting and generating the preceding data sequence. The pile-soil stiffness-impedance matrix is ​​updated using the ratio of the equivalent secant stiffness to the initial equivalent stiffness as the stiffness reduction factor. The integration operation is restarted with the local state boundary conditions until the end of the overlapping time window, and the subsequent data sequence is extracted and generated. The cosine-weighted smoothing function is applied to mathematically reconstruct and smooth the preceding and following data sequences. The smoothing and fusion results are aligned and spliced ​​along the overall time axis. The time series data of the six degrees of freedom internal forces at each pipe end are extracted to generate the global end load time history matrix. The spatial rotational angular displacement sequence is extracted to generate the mud surface angular displacement time history.

7. The fatigue failure assessment method for wind turbine pipe nodes based on linear damage criteria according to claim 1, characterized in that, The steps for performing algebraic mapping operations and spatial stress extrapolation operations include: Extract the multi-pipe end load vector from the global end load time history matrix after coordinate transformation and degree of freedom rearrangement, multiply the multi-pipe end load vector with the linear elastic spatial influence matrix, and calculate the nominal dynamic stress response of the structure. Extract the gravity proportionality coefficient and multiply the unit gravity stress response vector with the gravity proportionality coefficient; extract the mud surface angular displacement vector in the mud surface angular displacement time history and multiply the mud surface angular displacement vector with the mud surface angular displacement compensation matrix; calculate the additional dynamic stress response. The nominal dynamic stress response of the structure and the additional dynamic stress response are algebraically superimposed in the time domain to generate a dynamic stress time history matrix.

8. The fatigue failure assessment method for fan pipe nodes based on linear damage criteria according to claim 7, characterized in that, The step of performing spatial stress extrapolation calculation to generate hotspot stress time history includes: Multiple extrapolation paths are divided along the direction perpendicular to the weld toe normal of the intersection line. On each extrapolation path, a physical node is located at a first reference distance from the outer edge of the weld toe of the intersection line and a physical node is located at a second reference distance, forming the target extrapolation node set. The first reference distance and the second reference distance are determined based on the reference wall thickness parameter of the pipe node. Extract the first stress time series data corresponding to the physical node at the first reference distance position from the dynamic stress time history matrix, and extract the second stress time series data corresponding to the physical node at the second reference distance position. The first stress time series data is multiplied by the first spatial stress extrapolation coefficient, and the second stress time series data is multiplied by the second spatial stress extrapolation coefficient and summed to generate the hot spot stress time history at the intersection line of the pipe node.

9. The fatigue failure assessment method for wind turbine pipe nodes based on linear damage criteria according to claim 1, characterized in that, The step of performing rainflow counting on the hotspot stress time history includes: Identify the local maximum and local minimum values ​​in the hotspot stress time history, remove invalid stress cycles where the adjacent difference between the local maximum and local minimum is less than a set filtering threshold, and generate an alternating stress extreme value sequence. Extract the stress amplitudes that form closed hysteresis loops from the alternating stress extreme value sequence, and obtain the actual number of cycles corresponding to each stress amplitude according to the preset stress level division interval; A thickness correction index is introduced to reduce and correct the baseline parameters of the SN fatigue curve, and the allowable number of cycles corresponding to each stress amplitude is calculated.

10. The fatigue failure assessment method for wind turbine pipe nodes based on linear damage criteria according to claim 9, characterized in that, The step of combining the matched SN fatigue curve and the linear cumulative damage criterion to output the full-life fatigue damage degree includes: The actual number of cycles corresponding to each stress amplitude is divided by the allowable number of cycles corresponding to each stress amplitude and then summed to generate the short-term fatigue damage degree. Obtain the probability distribution data of occurrence of each discrete environmental condition within the design life cycle, multiply the short-term fatigue damage degree of each discrete environmental condition by the corresponding probability of occurrence, multiply by the total number of time blocks corresponding to the design life cycle, and perform linear superposition and summation to generate the full life fatigue damage degree.