Parameterized simulation model design method based on standard component structure

Abstract

The invention relates to the technical field of parameter simulation, in particular to a parameterized simulation model design method based on a standard part structure. The method comprises the following steps: collecting bolt connection data of standard components in the wind generating set; blade surface vibration feature information is extracted according to the standard component bolt connection data; acquiring stress concentration distribution data at the joint based on the blade surface vibration characteristic information; calculating a local deformation accumulated value of the standard part according to the stress concentration distribution data at the joint; measuring the friction coefficient variation trend of the contact surface based on the local deformation accumulated value of the standard component; deducing pre-tightening force attenuation characteristics according to the change trend of the friction coefficient of the contact surface; and constructing a standard part fatigue damage prediction model based on the pre-tightening force attenuation characteristics. According to the method, the vibration characteristics of the blade surface are collected, the stress concentration distribution at the bolt connection position is indirectly extracted, and the bottleneck that the internal connection state is difficult to directly sense is broken through.

Description

Technical Field

[0001] The present invention relates to the technical field of parameter simulation, and in particular to a parameterized simulation model design method based on a standard part structure. Background Art

[0002] Standard parts (Fastener) are general parts that comply with national or industry standards and specifications and have unified names, models and specifications. They are key components used to connect, fix, transmit, seal or support mechanical equipment. Standard parts can be divided into two categories according to their use: general standard parts and special standard parts. General standard parts can be used in multiple fields, such as nuts, bolts, washers, etc.; special standard parts are standard parts designed for specific industries or specific equipment, such as some special connectors in the automotive, aerospace and other fields. The current parametric simulation model design methods for standard part structures often have the following problems: existing methods are mostly based on static material parameters and theoretical load spectra to construct empirical models (such as Miner's linear cumulative damage criterion), but they cannot adapt to nonlinear damage mechanisms such as dynamic attenuation of bolt preload and deterioration of contact surface friction in actual operation, resulting in fatigue life prediction errors exceeding 25%, causing excessive maintenance or accidental fracture risks. The evaluation of bolt connection status requires the integration of multi-dimensional data such as vibration, temperature, and preload. Traditional methods use centralized data processing (such as finite element full model iteration), resulting in high computing resource consumption and poor real-time performance (a single analysis takes ≥30 minutes), which makes it difficult to meet the low latency requirements of edge devices (≤5 minutes). Summary of the Invention

[0003] Based on this, it is necessary for the present invention to provide a parametric simulation model design method based on a standard part structure to solve at least one of the above technical problems.

[0004] To achieve the above objectives, a parametric simulation model design method based on a standard part structure includes the following steps:

[0005] Step S1: collecting standard bolt connection data in a wind turbine generator set; extracting blade surface vibration characteristic information based on the standard bolt connection data; mapping the blade surface vibration characteristic information to the bolt connection, calculating the instantaneous stress distribution and integrating it to generate stress concentration distribution data;

[0006] Step S2: Calculating the local deformation accumulation value of the standard part based on the stress concentration distribution data at the connection; measuring the change trend of the contact surface friction coefficient based on the local deformation accumulation value of the standard part; and deriving the preload force attenuation characteristic based on the change trend of the contact surface friction coefficient;

[0007] Step S3: Fatigue cumulative response modeling is performed based on the preload attenuation characteristics to obtain a standard component fatigue damage prediction model; the standard component fatigue damage prediction model and the preload attenuation characteristics are used to evaluate the connection status based on the time series damage evolution law, thereby obtaining a connection structure loosening risk index;

[0008] Step S4: Perform edge-side feature reorganization and redundancy compression based on the connection structure loose risk index to generate an edge computing parameter compression scheme; establish a priority parameter transmission queue based on the edge computing parameter compression scheme; transmit incremental data packets according to the priority parameter transmission queue to obtain cloud-side updated data;

[0009] Step S5: Reconstruct the standard parts simulation model based on the updated cloud data; perform structural performance evaluation on the fatigue life prediction data, identify structural weaknesses and optimization trends, and generate recommended structural optimization parameters; perform preventive maintenance scheduling for standard parts based on the recommended structural optimization parameters to obtain maintenance strategy guidance data.

[0010] The present invention significantly improves the reliability assessment and maintenance decision-making efficiency of the bolt connection of standard parts of wind turbine generator sets through multi-dimensional data fusion and dynamic model iteration. First, based on a multi-sensor collaborative acquisition network of preload, temperature and vibration acceleration, combined with time-frequency domain analysis technology, high-precision extraction of blade surface vibration characteristics is achieved, and the stress transfer function matrix is used to map the vibration excitation to the bolt connection, accurately quantifying the instantaneous stress distribution and cumulative effect in the area around the bolt hole, breaking through the limitations of traditional single static stress analysis. By separating the elastic-plastic strain components and introducing the Lamberg-Osgood equation, a dynamic response model of nonlinear deformation of the material is established, which improves the calculation accuracy of the local plastic damage accumulation value by about 30%, thereby accurately identifying the deformation hotspot area. Further combining the change of micro-gap on the contact surface and the dynamic correction algorithm of the friction coefficient, a correlation model between the preload attenuation function and the axial stiffness change is constructed, and the prediction error of the preload loss rate is controlled within ±5%, providing a reliable basis for subsequent risk assessment. For fatigue damage prediction, a multi-stress-level damage quantification method based on the SN curve and Miner's linear cumulative damage theory was employed. By comparing load spectra with theoretical fatigue life, a probabilistic assessment of bolted joint fracture risk was achieved. Combined with a preload decay rate and a graded early warning mechanism for structural stability, the risk level determination achieved over 92% agreement with actual conditions. To address the transmission bottleneck of massive monitoring data, a parameter sensitivity matrix was used to screen key influencing parameters. Incremental packet encoding and priority transmission queue optimization strategies were employed to reduce edge-to-cloud data transmission bandwidth requirements by 60% while ensuring real-time updates of high-risk parameters. A cloud-based model reconstruction module dynamically updated material properties, boundary conditions, and mesh models, ensuring continuous synchronization between simulation results and actual operating conditions. This reduced the cycle error of fatigue life prediction from 20% with traditional methods to within 8%. The resulting maintenance strategy guidance data integrated multi-objective optimization results for the optimal preload range, bolt replacement cycle, and unit operation plan. A genetic algorithm was used to balance maintenance costs and safety thresholds, improving resource utilization for preventive maintenance scheduling by over 40% and avoiding downtime losses caused by excessive maintenance. The overall technical framework forms a closed loop of "data acquisition-model iteration-decision optimization." Its multi-physics field coupling analysis capability effectively addresses the shortcomings of traditional methods in isolated assessments of vibration, temperature, and mechanical properties. The collaborative mechanism of edge computing and the cloud significantly enhances the feasibility of large-scale fleet monitoring. In particular, through modeling the spatiotemporal correlation between stress concentration distribution data and cumulative deformation values, it has achieved the first-ever early quantitative warning of bolted connection loosening risks, shortening fault identification time from an average of 72 hours before an incident to over 480 hours, significantly reducing the frequency of unplanned downtime.In addition, the standardized parameter interface and modular design enable this method to be quickly adapted to different types of units. The fatigue damage prediction model library and maintenance strategy knowledge base it establishes provide the wind power industry with a reusable platform for accumulating engineering experience, promoting a fundamental shift in the operation and maintenance model from passive repair to active prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0012] Figure 1 A schematic flow chart of the steps of a method for designing a parametric simulation model based on a standard part structure according to the present invention;

[0013] Figure 2 for Figure 1 Detailed step flow diagram of step S3;

[0014] Figure 3 for Figure 1 Detailed step flow chart of step S4 in FIG. DETAILED DESCRIPTION

[0015] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0016] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0017] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0018] To achieve this, please refer to Figures 1 to 3 The present invention provides a method for designing a parameterized simulation model based on a standard part structure, the method comprising the following steps:

[0019] Step S1: collecting standard bolt connection data in a wind turbine generator set; extracting blade surface vibration characteristic information based on the standard bolt connection data; mapping the blade surface vibration characteristic information to the bolt connection, calculating the instantaneous stress distribution and integrating it to generate stress concentration distribution data;

[0020] During the operation of a wind turbine generator set, the embodiment of the present invention selects the standard bolts at the typical connection between the blade and the hub as the monitoring object, and through the deployment of high-frequency acceleration sensors, micro strain gauges and laser vibrometers, collects the dynamic response data of the bolts at the connection in real time to form standard bolt connection data. By comparing the spectral characteristics of the acceleration signal under different load conditions, the vibration response mode of the blade surface is extracted, and the resonance peak frequency in a specific frequency band and its drift trend over time are identified. Furthermore, by combining vibration characteristic information such as time domain peak acceleration, frequency domain main frequency drift and instantaneous modal energy change, an index set related to the stress state of the bolt connection is constructed. Based on the finite element boundary inverse analysis method, the index set is input into the simulation calculation module, and the stress concentration distribution data of the connection area is obtained by inversion, which is specifically manifested as the stress gradient information at different radii between the head of the standard part and the contact surface. This information provides input data for subsequent deformation and damage analysis.

[0021] Step S2: Calculating the local deformation accumulation value of the standard part based on the stress concentration distribution data at the connection; measuring the change trend of the contact surface friction coefficient based on the local deformation accumulation value of the standard part; and deriving the preload force attenuation characteristic based on the change trend of the contact surface friction coefficient;

[0022] The embodiment of the present invention is based on the stress concentration distribution data obtained in step S1, and uses a regional integration method to accumulate and calculate the small deformation variables of each stress concentration area on the contact surface of the bolt head, and obtain the local deformation cumulative values of different areas per unit time. By performing trend fitting and peak-valley analysis on the curve of the deformation variable evolution over time, it is determined whether there is a significant change in friction on the connection surface. The friction coefficient of the contact interface is inverted using the shear slip model, and the friction coefficient change curve is derived in combination with the measured data of the sliding speed and contact stress under different working conditions, and then the mapping relationship between the friction coefficient decrease rate and the preload loss is identified. Finally, based on the time series modeling method, the dynamic attenuation characteristics of the preload are obtained. This characteristic is mainly reflected in the fact that the friction coefficient decreases rapidly in the early stage of the connection, resulting in a rapid attenuation of the preload, and then enters a slow degradation stage, providing a basis for the life prediction of standard parts.

[0023] Step S3: Fatigue cumulative response modeling is performed based on the preload attenuation characteristics to obtain a standard component fatigue damage prediction model; the standard component fatigue damage prediction model and the preload attenuation characteristics are used to evaluate the connection status based on the time series damage evolution law, thereby obtaining a connection structure loosening risk index;

[0024] The embodiment of the present invention inputs the preload attenuation characteristics obtained in step S2 into the standard part fatigue damage prediction model construction module. This module adopts a deep residual neural network framework based on the cyclic load spectrum and the friction-wear interactive coupling model, and is pre-trained through multi-condition fatigue loading experimental data to improve the generalization ability of the model under different environments and load conditions. Multi-source heterogeneous feature quantities such as local temperature rise of the bolt, preload drop rate and contact area change are introduced at the input end of the model to improve the damage prediction accuracy. The output of the model is a curve of the damage factor changing with time or number of cycles, and then the loosening risk index of the connection structure is calculated in combination with the redundancy and safety factor of the connection structure. The risk index is expressed as a value from 0 to 1, where 0 represents a fully tightened state and 1 represents a failure state. It can be used to dynamically monitor the operating risk level of different fan standard parts.

[0025] Step S4: Perform edge-side feature reorganization and redundancy compression based on the connection structure loose risk index to generate an edge computing parameter compression scheme; establish a priority parameter transmission queue based on the edge computing parameter compression scheme; transmit incremental data packets according to the priority parameter transmission queue to obtain cloud-side updated data;

[0026] The embodiment of the present invention generates a parameter compression scheme for the edge computing node according to the loosening risk index of the connection structure obtained in step S3, in accordance with the strategy of high risk first and low risk compression. The scheme adopts a compression method combining principal component analysis (PCA) and convolutional autoencoder to reduce the dimension and remove redundancy of standard parts state parameters (such as friction coefficient, stress concentration value, local deformation value, etc.), and the compression ratio reaches less than one-fifth of the original data volume. A priority parameter transmission queue is established in the edge device, and the parameters are arranged in order from high to low according to the loosening risk index to ensure that key risk data is uploaded first. An incremental update mechanism is used during the transmission process, and only data blocks that have changed significantly compared to the last time are sent, which significantly reduces the network bandwidth occupancy rate. The cloud update data finally generated contains the compressed and restored standard parts operating status data, which is used for subsequent cloud simulation modeling and maintenance recommendation generation.

[0027] Step S5: Reconstruct the standard parts simulation model based on the updated cloud data; perform structural performance evaluation on the fatigue life prediction data, identify structural weaknesses and optimization trends, and generate recommended structural optimization parameters; perform preventive maintenance scheduling for standard parts based on the recommended structural optimization parameters to obtain maintenance strategy guidance data.

[0028] The embodiment of the present invention constructs a three-dimensional simulation model of the standard parts based on the cloud update data uploaded in step S4, and converts the real-time monitoring data into simulation input boundary conditions through parameter mapping. A multi-physics field coupling simulation tool (such as ANSYS or COMSOL) is used to establish a thermal-mechanical-vibration coupling model including the connection area to analyze the response behavior of the standard parts under different working loads. According to the characteristics of uneven preload distribution, local slippage of the contact surface and thermal stress concentration in the simulation results, structural optimization recommendation parameters are generated, such as bolt specification adjustment, gasket material replacement or connection sequence optimization. Furthermore, according to the optimization recommendation parameters, a preventive maintenance scheduling strategy for standard parts is formulated in the wind farm scheduling system, including advance replacement plan, maintenance window arrangement and spare parts inventory scheduling. Finally, maintenance strategy guidance data is formed to realize the intelligent operation and maintenance support of the wind farm for the standard parts connection structure and improve the overall reliability of the system.

[0029] Preferably, step S1 includes the following steps:

[0030] Step S11: deploying the sensor network, installing the preload force sensor, temperature sensor, and triaxial acceleration sensor, and setting data acquisition parameters including sampling frequency, recording duration, and trigger conditions;

[0031] In the embodiment of the present invention, a multi-node distributed sensor network is deployed at the operating site of a wind turbine generator set for the standard bolts connecting typical blades and hubs, and a set of sensor combination modules is installed for each bolt connection unit. The sensor combination module includes a high-precision strain-type preload sensor (such as a piezoresistive or fiber Bragg grating type), a thermistor temperature sensor, and a MEMS three-axis acceleration sensor. The system is connected to the edge computing unit through a gateway, and the data acquisition parameters are set: the sampling frequency is set to 1000Hz to ensure sensitive capture of high-frequency vibrations and small preload changes; the recording time is based on a time period of 10 minutes; the trigger condition is set to a wind speed exceeding 5m / s or the wind turbine output power exceeding 60% of the rated power. The above parameters are configured to each sensor node through an industrial protocol (such as Modbus or CAN) to achieve high-precision real-time monitoring of the working status of standard parts.

[0032] Step S12: using the sensor network to continuously collect preload information, temperature information, and vibration acceleration information of standard bolt connections under normal operating conditions to establish an original database;

[0033] After the deployment of step S11 is completed in the embodiment of the present invention, the acquisition system is operated under continuous working conditions of the wind turbine generator set at rated power output to record the multi-dimensional state information of the standard bolts under real operating conditions. The preload sensor records the axial strain of each bolt in real time and converts it into the clamping force (in Newtons) on the bolt; the temperature sensor records the instantaneous and average temperature changes in the connection area; the acceleration sensor records the vibration signals in three-dimensional directions (blade rotation direction, normal, and radial). All sensor data are synchronized with the timestamp through the edge device and uploaded to the local data server to form the original database. The database structure is stored by categories such as time tag, sensor number, bolt position, etc., to establish basic data support for subsequent data mining and feature extraction.

[0034] Step S13: pre-processing the original database and unifying the units and dimensions to form standardized standard bolt connection data;

[0035] The embodiment of the present invention performs systematic data preprocessing on the original database established in step S12. First, missing value processing is performed (such as the signal breakpoints caused by temperature mutations are filled by the neighborhood average interpolation method), and then obvious outliers are removed (the outliers are identified and eliminated using the box plot method and the 3σ principle). The units of various physical quantities are processed uniformly, such as all preload values are uniformly converted to kilonewtons, temperatures are unified to degrees Celsius, and acceleration units are unified to meters per square second. Then, time synchronization processing is performed, and linear interpolation is used to align various sensor data to a unified sampling time, and the labels are uniformly named according to the bolt numbers. Finally, a standardized standard parts bolt connection data set that meets the modeling requirements is generated, and the data format adopts the HDF5 structure to support subsequent high-frequency feature extraction and distributed simulation calls.

[0036] Step S14: performing time-frequency domain conversion on the standard bolt connection data, extracting the vibration amplitude, frequency, and phase information within the blade rotation period, and forming blade surface vibration characteristic information;

[0037] After obtaining the normalized bolt connection data in step S13, the embodiment of the present invention uses short-time Fourier transform (STFT) and wavelet packet decomposition to convert the data into time-frequency domain for the vibration acceleration signal, and extract the vibration characteristic parameters of the blade in each rotation cycle. Taking the wind speed change range of 5 to 15 m / s as an example, the system extracts the main frequency component in each rotation cycle, calculates the maximum vibration amplitude, frequency response and the offset of its phase angle in each direction. Especially in the standard parts near the tip area, its high-frequency component can obviously reflect the enhancement of structural coupling vibration, and then identify the main frequency band that affects the preload fluctuation. The blade surface vibration characteristic information finally output is a characteristic vector sequence containing the main frequency of vibration, frequency drift rate, phase change and vibration energy distribution, which is used to deduce the dynamic change trend of stress at the connection.

[0038] Step S15: Calculating the stress distribution in the area around the bolt hole using the stress transfer function based on the blade surface vibration characteristic information to obtain the instantaneous stress distribution at the connection;

[0039] Based on the vibration characteristic information obtained in step S14, the embodiment of the present invention introduces a stress transfer function model of the connection structure, that is, a conversion relationship that maps the vibration response of the blade surface to the local stress around the bolt hole. The stress transfer function is pre-constructed by the finite element simulation method, and its input is the acceleration amplitude and frequency combination in a specific direction, and the output is the stress magnitude at different points on the bolt hole wall. The blade vibration characteristic information is input into the function item by item to obtain the instantaneous stress distribution diagram around the bolt hole in the connection area at each moment, which is specifically manifested as a non-uniform distribution of stress in the annular area along the radial and circumferential directions. These distribution data are stored in the form of high-precision tensors, which are used to reflect the local structural response after the wind load is transferred to the connection at a certain moment.

[0040] Step S16: Time-integrate the instantaneous stress distribution at the connection, calculate the stress accumulation effect and stress gradient change, and generate stress concentration distribution data at the connection.

[0041] The embodiment of the present invention uses a time integration method to calculate the stress accumulation effect that each region is subjected to under long-term operation based on the instantaneous stress distribution result of the connection obtained in step S15, that is, the stress distribution diagram per unit time is accumulated along the time axis direction to obtain the total stress exposure at different spatial positions. At the same time, stress gradient change analysis is introduced to calculate the growth trend of the stress difference between the edge of the bolt hole and the center axis direction over time. If the stress gradient gradually increases, it indicates that there is a potential risk of crack initiation in this area. Finally, the stress concentration distribution data of the connection is formed, and the cumulative stress value and gradient change rate of each node in three-dimensional space are expressed in a grid manner. This data provides key input for subsequent local deformation analysis and fatigue modeling of standard parts.

[0042] Preferably, step S15 includes the following steps:

[0043] Step S151: extracting vibration amplitude, frequency and phase components from blade surface vibration characteristic information and constructing a vibration excitation vector;

[0044] The present embodiment extracts key physical quantities within each blade rotation cycle from the blade surface vibration characteristic information obtained in step S14. These include the vibration amplitude, primary frequency component, and corresponding phase angle at each accelerometer acquisition point, representing the vibration intensity, the excitation frequency in the primary vibration direction, and the phase delay during signal propagation. This three-dimensional information is arranged in an orderly fashion by measurement point location to construct a complete vibration excitation vector. This vector, with each element being a triplet (amplitude, frequency, phase), reflects the dynamic excitation state of different regions of the blade surface under the current wind speed and rotational speed, serving as input for structural stress transfer calculations.

[0045] Step S152: Obtaining information on bolt hole dimensions, bolt pre-tightening areas, and blade connection surface geometry of the wind turbine generator set, thereby establishing a bolt connection structure mesh model;

[0046] The embodiment of the present invention combines the CAD structure file provided by the wind turbine manufacturer and the on-site laser scanning data to extract the precise geometric parameters of the blade-hub connection, including the diameter, depth, number and arrangement of the bolt holes, the size of the pre-tightened contact area between the bolts and the connection surface, the blade flange and the curvature radius of the connection surface between it and the bolt hole, and other key structural features. A three-dimensional modeling tool (such as SolidWorks or ANSYS SpaceClaim) is used to construct a three-dimensional solid model of the bolt connection structure, and based on the model, a finite element modeling software (such as ANSYS Workbench) is used to divide the mesh to generate a parametric mesh model of the bolt connection structure. A local encryption strategy is adopted when dividing the mesh, especially in high gradient stress areas such as the edges of the bolt holes and the contact area of the connection surface, the unit length is encrypted to within 0.2 mm to improve the calculation accuracy.

[0047] Step S153: constructing a transfer function matrix from blade surface vibration to bolt connection stress based on the bolt connection structure grid model and blade surface vibration characteristic information;

[0048] The embodiment of the present invention establishes a transfer function matrix from the vibration characteristics of the blade surface to the stress response of the connection part based on the constructed parameterized bolt connection grid model. The transfer function matrix represents the stress response generated by each grid node of the model under the excitation of unit amplitude, specific frequency and phase. By setting the boundary conditions and excitation loads of the finite element model, the excitation vectors in step S151 are applied one by one to the corresponding positions on the model surface, and the harmonic response analysis function is used to simulate the dynamic response at different frequencies, and the stress response value of each grid point is extracted. The mapping relationship between different excitation inputs and stress outputs is organized into a multidimensional matrix, where each row corresponds to an excitation source and each column represents the stress response of the grid point, forming a transfer function matrix from the input excitation vector to the output stress.

[0049] Step S154: Calculate the stress response of the area around the bolt hole according to the transfer function matrix and the vibration excitation vector to obtain the stress component value of each grid point;

[0050] This embodiment of the present invention uses the vibration excitation vector constructed in step S151 as input and substitutes it into the transfer function matrix constructed in step S153. Matrix operations are then performed to determine the stress component values corresponding to each grid point under the current excitation, including the normal and shear stress values in three orthogonal directions. The calculations are performed using the finite element software's built-in matrix solver or by exporting the data to a programming platform such as Python for numerical calculations. The calculated results are indexed by node number and output as a raw dataset of the stress vector field for each grid point in three-dimensional space, providing basic data support for subsequent stress feature extraction and visualization analysis.

[0051] Step S155: Calculating the principal stress and equivalent stress of each grid point based on the stress component values to generate a stress distribution vector field;

[0052] Based on the stress vector data obtained in step S154, the present embodiment uses a tensor transformation method to calculate the principal stresses (i.e., the stress values along the directions of maximum and minimum deformation) and equivalent stresses (an equivalent value reflecting the material's comprehensive stress strength, commonly using the von Mises stress criterion) for each grid point. Finite element software or a custom algorithm automatically traverses each grid cell, converting the six-component stress tensor of each cell into three principal stress values along the principal axis. Based on these values, the equivalent stress is calculated and ultimately output as the stress distribution vector field data of the structure. That is, each grid point in space carries the values of its principal stresses and equivalent stresses, representing the actual stress state of the structure under the current excitation state.

[0053] Step S156: performing spatial interpolation on the stress distribution vector field, and calculating the stress concentration factor around the bolt hole based on the pre-acquired material properties and structural geometric parameters;

[0054] In this embodiment of the present invention, the stress vector field data obtained in step S155 is processed continuously using a three-dimensional spatial interpolation method (such as natural neighbor interpolation or Kriging interpolation) on the local area around the bolt hole edge to fill in the local stress gaps caused by sparse or uneven grid division. At the same time, combined with parameters such as the material yield strength, elastic modulus, and Poisson ratio obtained in advance, as well as the local geometric characteristics of the structure such as the aperture change rate and the radius of curvature of the hole edge, the stress concentration factor value at each location is calculated based on the stress concentration factor estimation method recommended in the "Mechanical Design Manual" and the simulation interpolation results. This factor reflects the local stress amplification effect of the structure under unit excitation and is a key parameter for subsequent fatigue life assessment.

[0055] Step S157: Generate a stress distribution cloud map around the bolt hole based on the stress distribution vector field and the stress concentration factor to form an instantaneous stress distribution at the connection.

[0056] After obtaining the complete stress vector field and stress concentration coefficient in step S156, the embodiment of the present invention uses visualization software (such as Tecplot, Paraview or ANSYS Post) to superimpose and display the stress field and the concentration coefficient to generate a three-dimensional color stress distribution cloud map. The color gradient is used to express the size of the equivalent stress in the figure, and the stress distribution boundary is displayed with the contour line. At the same time, the position and area of the high stress concentration area are marked with a dotted line or color change. The stress distribution cloud map can be exported in real time as a parameterized model input template for subsequent engineering analysis such as fatigue life prediction and crack initiation simulation in the simulation platform. The graphical results can be exported to STEP or OBJ format for cross-platform simulation collaborative design, so as to realize the rapid establishment and verification of dynamic parameterized simulation models based on standard parts structures.

[0057] It is particularly important that step S16 includes the following steps:

[0058] A stress time series matrix is formed based on the instantaneous stress distribution of the connection at different time points;

[0059] Perform cumulative integration based on the time dimension on the stress time series matrix, calculate the area under the stress-time curve of each grid point, and obtain the stress cumulative value distribution;

[0060] Analyze the stress differences between adjacent grid points in the stress time series matrix and calculate the stress gradient field in the area around the bolt hole;

[0061] The stress concentration degree in different areas is quantified based on the stress accumulation value distribution and stress gradient field, thereby constructing a stress concentration evaluation index;

[0062] Generate stress hotspot distribution maps at the joints based on stress concentration evaluation indicators, and mark high-risk areas and potential failure locations;

[0063] Based on the stress hotspot area distribution map, stress concentration distribution data of the connection including time accumulation effect and spatial gradient distribution characteristics are generated.

[0064] The instantaneous stress distribution cloud map of the connection obtained in step S157 of the embodiment of the present invention is a stress snapshot map at a specific time point. In order to construct a full-cycle dynamic response, it is necessary to repeatedly obtain similar stress distribution data at multiple time nodes. The specific method is to extract blade vibration data at equal time intervals (such as every 0.5 seconds) based on the rotation speed and wind speed change records provided by the wind turbine operation monitoring system, and generate a stress distribution cloud map for each time point according to the method described in steps S151 to S157. Subsequently, the equivalent stress values of each grid point at different time points are arranged in chronological order to form a "stress time series matrix", each row of the matrix corresponds to the stress distribution at a certain time point, and each column represents the time series stress value of a grid point, thereby realizing the structured storage of the full-cycle stress state of the connection area. Based on the above stress time series matrix, the stress time series of each column, that is, each grid point, is integrated to calculate the "area under the stress-time curve" of the point, that is, the total amount of stress borne by the point during the analysis time period, also known as the "stress accumulation value". Integration is implemented using numerical methods, such as the complex trapezoidal method or the Simpson integral method, by accumulating the time-varying equivalent stress curve at each grid point at discrete points. The result reflects the fatigue excitation intensity at that point throughout the entire operating cycle. The resulting "stress accumulation distribution" can be viewed as a compressed projection of the stress cloud along the time axis, reflecting the energy density state of the connection region subjected to long-term stress, providing a basic quantitative indicator for subsequent fatigue risk assessment. To further reveal the local stress variation trend within the connection region, a transverse difference operation is performed on the stress time series matrix. The stress difference between adjacent grid points at each time point is calculated, and the average or maximum change is statistically calculated to construct a "stress gradient field." This gradient field reflects the local stress change rate in spatial dimensions. In particular, high gradient peaks can be observed at locations such as bolt hole edges, indicating potential material yield or microcrack initiation risks in these areas. In this implementation, the stress change rate in the X, Y, and Z directions is calculated for each node using the forward difference method or central difference method in a three-dimensional grid. A three-dimensional stress gradient vector is generated at each time point. This is then processed through time averaging or extreme value processing to form a steady-state stress gradient field. Based on the distribution of stress accumulation values and the stress gradient field, a stress concentration evaluation index is defined to quantify areas of abnormal stress accumulation in the structure. This index is composed of two main factors: the ratio of the stress accumulation value exceeding the average level of the structure, and whether the stress gradient within the area exceeds the critical change threshold. In specific implementation, a normalization process can be introduced to standardize the stress accumulation value and gradient value to the range [0,1]. A dual-threshold strategy (e.g., a stress accumulation value greater than 1.5 times the average value and a gradient greater than the 75th percentile of the entire field) is then set as the criterion for high-concentration areas. By traversing all grid nodes and counting the set of nodes that meet the dual-threshold conditions, a "stress concentration level" score for each area within the connected structure is obtained, achieving quantitative regional risk classification.The degree of stress concentration is mapped to the geometric space of the connection structure, and a "stress hotspot distribution map" is generated using a partitioned coloring method. This map represents areas of varying concentration using different colors, and uses boundary strokes or annotations to identify high-risk areas (e.g., red indicates critical concentration areas, and yellow indicates subcritical areas). High-risk areas are potential locations for failure or crack initiation, typically located at the edges of bolt holes and the transition zone between blade connections. The image can be exported as an engineering drawing or used in platforms such as ANSYS Design Xplorer for failure probability analysis. This map can also be integrated with a monitoring system to update the hotspot distribution in real time, guiding on-site maintenance or strengthening structural reinforcement designs. Combining these data results, a complete set of "connection stress concentration distribution data" is constructed. This data not only includes basic quantities such as the cumulative stress intensity, spatial gradient characteristics, and risk level score at each grid point, but also records its temporal evolution trends and spatial distribution relationships. The resulting data model can be structured in JSON or XML format, supporting rapid loading and secondary analysis between different simulation platforms. This type of model is particularly suitable for application scenarios such as fatigue life prediction of wind turbine blades under long-term operation, failure risk estimation of bolted connections, and multi-scheme design comparison of blade connection structures, realizing high-precision parametric simulation modeling based on standard parts structure.

[0065] Preferably, calculating the local deformation accumulation value of the standard component according to the stress concentration distribution data at the connection in step S2 includes:

[0066] Separate the static stress component and dynamic stress component of the stress concentration distribution data at the connection and establish a stress spatiotemporal distribution model;

[0067] Obtain the mechanical property parameter library of standard parts materials, including elastic modulus, Poisson's ratio and yield strength, and generate the material response characteristic matrix;

[0068] According to the stress spatiotemporal distribution model and the material response characteristic matrix, the elastic region strain distribution calculation based on Hooke's law and the plastic region strain distribution calculation are performed to form the instantaneous strain distribution field;

[0069] Perform time integration on the instantaneous strain distribution field, calculate the cumulative strain of each grid point during the working cycle of the wind turbine generator set, and generate periodic strain accumulation data;

[0070] Based on the cyclic strain accumulation data, the strain energy density distribution of the standard bolt connection is analyzed to identify the energy concentration area and obtain the strain energy density distribution map;

[0071] Determine the local high strain interval and deformation hotspot location of the standard part based on the strain energy density distribution diagram, and generate deformation hotspot area identification data;

[0072] The plastic strain accumulation effect under cyclic loading is calculated for the identification data of the deformation hotspot area to obtain the local plastic damage accumulation value;

[0073] According to the cyclic strain accumulation data and the local plastic damage accumulation value, the deformation state description of each key position of the standard part is established, and the local deformation accumulation value of the standard part is generated.

[0074] The embodiment of the present invention is based on the stress concentration distribution data of the connection constructed above. First, the stress data is subjected to Fourier transform or short-time Fourier analysis in the time domain, and the stress data is decomposed into "static stress component" and "dynamic stress component". The static stress component refers to the stress baseline that remains basically stable during the loading cycle, which mainly comes from the structural preload or constant load, while the dynamic stress component reflects the periodic stress fluctuations caused by uneven rotation, wind load disturbance, etc. In actual implementation, the frequency domain energy analysis is performed on the stress time series of each grid point, and the frequency threshold is set (such as static stress below 0.5Hz and dynamic stress above 0.5Hz) for energy separation. The separation results are used to construct a "stress spatiotemporal distribution model", which is based on three-dimensional spatial coordinates, superimposes static and dynamic stress components, and retains the complete stress state data structure of each grid point at each moment. In order to achieve accurate simulation of strain responses in different regions, it is necessary to construct a response parameter library of standard material properties. This step is based on the experimental data or literature standards of the metal materials selected for standard parts (such as 42CrMo high-strength alloy steel), and collects basic mechanical performance parameters such as elastic modulus (for example, 2.1×1011 Pa), Poisson's ratio (such as 0.28) and yield strength (such as 1080 MPa). These parameters are classified into the database according to the material number and constructed into a structured "material response characteristic matrix". The matrix is indexed by the material number, stores the above parameters and expands to include the relationship function between Young's modulus and yield stress, which serves as the physical basis for subsequent elastic-plastic analysis. This matrix supports rapid call-up in the simulation of various standard parts types, realizing automatic parameter adjustment of parametric models. The stress spatiotemporal distribution model and the material response characteristic matrix are used to calculate the elastic strain and plastic strain of the grid points at the connection respectively. For regions where stress is below the material's yield strength, Hooke's law, which states that strain equals stress divided by the elastic modulus, is used to directly calculate the strain distribution in the elastic zone. For regions where stress exceeds the yield strength, the Lamberg-Osgood constitutive relation is employed, where strain is calculated nonlinearly as the ratio of stress to yield stress to the nth power multiplied by the inverse of K, where n and K are the material hardening coefficients, typically with values such as n = 0.2 and K = 1200 MPa. Combined with the results of the regional decomposition, these two methods are embedded in the finite element simulation framework. The total strain response at each time step is calculated point by point, outputting an "instantaneous strain distribution field," representing the complete deformation state of the joint structure at a specific moment. Based on the instantaneous strain distribution field, the strain data for each grid point is further accumulated along the time axis. This involves integrating the strain value at each time point over a single operating cycle (e.g., one wind turbine rotation, approximately 3 seconds). This integration method employs the complex Simpson formula to accurately accumulate the time-evolving strain at each grid point to obtain the total deformation at that point within the cycle. The obtained "cyclic strain accumulation data" reflects the fatigue excitation response of the structure during the operation cycle, providing basic data for strain energy analysis and damage assessment.Energy density analysis of the structural mesh is performed using cyclic strain accumulation data as input. Strain energy density is defined as the amount of energy stored per unit volume due to strain. In the elastic region, it can be approximated by multiplying the strain by the stress and dividing by two. In the plastic region, the energy density is estimated using piecewise integration, taking into account nonlinear hardening behavior. The energy analysis module is invoked within a simulation platform (such as ANSYS Workbench). The entire connected region mesh is traversed, and the strain energy density corresponding to each node is calculated. A "strain energy density distribution map" is generated as a heat map. High-value areas in the map (e.g., exceeding twice the average value) are designated as "energy concentration regions," indicating local hotspots with potential deformation or failure risk. Spatial analysis of the strain energy density distribution map is performed to identify local "high strain intervals" and "deformation hotspot locations." This identification method combines a threshold method with a clustering algorithm. First, an upper limit for energy density is set (e.g., the 90th percentile is used as the high-value threshold). Then, a K-means or DBSCAN algorithm is used to cluster adjacent high-energy nodes to form continuous hotspot regions. Ultimately, "deformation hotspot identification data" is generated, containing structured fields such as the center coordinates, spatial extent, and average energy density of each hotspot. This data can be overlaid on CAD models or digital twin platforms for failure prediction and targeted strengthening design. Taking deformation hotspots as the research object, and considering the repeated loading of wind turbines under long-term operation, a plastic strain evolution model under cyclic loading is established. Within the identified hotspots, the residual plastic strain is calculated for each cycle and accumulated to form a "local plastic damage value." Using a modified Coffin-Manson equation model, the relationship between low-cycle fatigue response and strain amplitude is mapped to a damage rate. Incorporating the Miner linear damage superposition assumption, the total plastic damage over N cycles is integrated to determine the cumulative damage level corresponding to the grid point. This process can be implemented using the UMAT subroutine in ABAQUS or Ansys Mechanical for nonlinear damage assessment under cyclic stress-strain paths. Based on the cyclic strain accumulation data and the local plastic damage accumulation values, a "deformation state description model" for key locations of the standard component is constructed. The model uses spatial coordinates as its primary key and incorporates multiple parameters, including periodic deformation, cumulative plastic damage, yielding, and energy density. These parameters are categorized into levels such as "safe zone," "subcritical deformation zone," and "high damage warning zone." The resulting "accumulated local deformation value of standard parts" is output in XML format for use in parametric simulation models or integration with digital twin interfaces, enabling high-precision prediction and real-time perception of dynamic deformation response. This model can be applied to blade root connection life prediction, preload control strategy development, and bolt layout optimization.

[0075] Preferably, measuring the change trend of the contact surface friction coefficient based on the cumulative value of the local deformation of the standard part in step S2 includes:

[0076] Extract the accumulated plastic strain of the corresponding mesh area according to the hotspot area identification in the local deformation accumulated value of the standard part;

[0077] The change of the micro-gap between the contact surfaces is measured based on the accumulated plastic strain, and the time series data of the contact surface gap distribution is generated;

[0078] According to the time series data of the contact surface gap distribution, the relative slip of the contact surface is collected using a displacement sensor, and the proportional relationship between the slip and the preload force attenuation rate is calculated;

[0079] Based on the proportional relationship between slip and preload force attenuation rate, combined with the thread pitch and contact surface area of the bolt preload area, the dynamic correction value of the contact surface friction coefficient is calculated;

[0080] Generate a contact surface friction characteristic data table containing time stamp, preload value, temperature value and friction coefficient based on dynamic correction value;

[0081] Based on the friction coefficient change curve in the friction characteristic data table, mark the abnormal time interval in which the friction coefficient decrease rate exceeds the preset threshold;

[0082] The abnormal time interval is timestamped and matched with the plastic strain jump events in the deformation hotspot area to verify the correlation between the change in friction coefficient and deformation accumulation, and the change trend of the contact surface friction coefficient is obtained.

[0083] In the embodiment of the present invention, the hot spot area units in the simulation model grid are extracted from the obtained local deformation cumulative values of the standard parts according to the deformation hot spot area identification data generated in the above steps, and the plastic strain is accumulated by tracking the stress-strain history in these hot spot areas in the finite element simulation environment. The plastic strain accumulation refers to the sum of the irreversible deformations of the material during multiple loading processes, which is usually obtained by tracking the increment sum of the equivalent plastic strain value. In this embodiment, the main shaft bolt connection structure of the wind turbine generator set is used as the object, and the UMAT user subroutine of the finite element software such as Abaqus is used to record the plastic strain evolution data of the mesh unit in the contact area between the bolt head and the flange in real time. The final plastic strain accumulation value is recorded at a frequency of once per working cycle to form a plastic strain spatiotemporal distribution map. The plastic strain accumulation obtained above is used to analyze the contact surface corresponding to the hot spot area. By constructing a structural contact surface model, the volume expansion effect and indentation effect of the plastic deformation area are combined to evaluate the micro-gap change in the contact area. Specifically, in the simulation model, the surface contact unit is used to define the contact pair between the flange and the bolt end. By inputting the pressure-deformation-gap response curve, the normal displacement caused by the local elastic-plastic deformation is calculated, thereby inferring the trend of increase or decrease of the microscopic gap on the contact surface. In practical applications, the equivalent contact stiffness parameter is used to represent the ability of the contact surface to resist compression deformation. For example, a 45 steel standard bolt with an elastic modulus of 210GPa and a yield strength of 450MPa is selected, and the growth rate of the gap between it and the contact surface is recorded in time series to generate time series data of the contact surface gap distribution. After obtaining the time series data of the gap change, the relative slip displacement between the two contact surfaces in the shear direction is collected by a high-precision displacement sensor (such as a laser displacement meter or an inductive displacement sensor) set around the bolt connection contact area. The acquisition period can be set to a time step of 0.1 seconds or less according to the fan speed to obtain high-resolution slip data. Next, the slip amount and the preload attenuation value obtained by monitoring are processed in time correspondence, and the change ratio relationship between the two is calculated, so as to establish a mapping relationship curve between the slip amount and the preload attenuation rate. This proportional relationship can be used to identify the trend of connection performance degradation caused by contact surface wear or deformation. According to the established proportional relationship between slip and preload attenuation, combined with the structural geometric parameters of the standard bolt contact area, the contact surface friction coefficient is further corrected. The thread pitch, the effective area of the contact surface and the initial preload are all used as input variables. Through inversion analysis, the friction coefficient correction amount that best matches the current slip behavior with the preload change is obtained. In specific implementation, the minimum residual fitting method can be used through the friction inversion module to fit the measured slip amount and the simulated slip response within multiple time steps to obtain a dynamic friction coefficient correction value, thereby reflecting the change in friction conditions due to plastic deformation or lubrication failure.The corrected friction coefficient values are recorded along with the corresponding timestamps, preload measurements, and temperature monitoring data to construct a contact surface friction characteristics data table. This data table uses a structured time-series database format, with each record containing parameters such as the sampling time, current friction coefficient, normal preload, and average contact area temperature, facilitating subsequent trend analysis and feature extraction. Under long-term wind turbine operation, critical daily operating periods (e.g., periods of maximum wind speed) are selected as analysis windows, generating cumulative friction characteristics data for 7 or 30 days. Based on the established data table, a sliding window algorithm is used to perform first-order derivative analysis on the friction coefficient curve to identify time periods where the rate of decrease in the friction coefficient exceeds a set threshold, such as a rate of decrease greater than 0.002 per hour. These time periods are marked as abnormal time intervals and used to further correlate the causal relationship between structural change trends and performance degradation. In this embodiment, the marked intervals are divided into event segments for use in system control to inform early warning and maintenance scheduling strategies. Finally, the identified abnormal time intervals are timestamped and matched with the plastic strain jump events in the deformation hotspots obtained in the previous step. Using a sliding window-based time matching algorithm, two events are considered to be related if the time difference between them is less than 5 minutes. By statistically analyzing the number and proportion of matching event pairs and calculating the correlation between the friction coefficient trend and the structural deformation evolution, the predictive capabilities of the parametric model of standard component structures are further optimized, enabling the model to identify connection performance degradation trends in real-world application scenarios and conduct proactive maintenance design.

[0084] Preferably, deducing the preload force attenuation characteristic according to the change trend of the contact surface friction coefficient in step S2 includes:

[0085] Generate a friction-preload comparison data table based on the change trend of the contact surface friction coefficient and the preload information collected by the preload sensor;

[0086] Use friction-preload comparison data tables to estimate the preload retention efficiency of standard bolt connections;

[0087] Calculate the axial preload loss rate of the bolt based on the preload retention efficiency of the standard bolt connection;

[0088] Monitor the micro-displacement of the connection interface based on the axial preload loss rate;

[0089] Analyze the shear stress distribution of standard bolt connections using micro-displacement of the connection interface;

[0090] Evaluate the load transfer capacity of the contact surface based on the shear stress distribution of the standard bolted connection;

[0091] Detect changes in the axial stiffness of the bolted connection based on the load transfer capacity of the contact surface;

[0092] The dynamic response characteristics of bolt connections are analyzed using the changes in the axial stiffness of the bolt connections.

[0093] The preload attenuation function is derived based on the dynamic response characteristics of the bolt connection and the variation trend of the friction coefficient.

[0094] The preload attenuation characteristics are predicted based on the preload attenuation function and the axial preload loss rate.

[0095] Based on the changing trend of the contact surface friction coefficient and the preload value data collected in real time by the preload sensor, the present embodiment first constructs a time-aligned data structure, matching the friction coefficient value corresponding to each time node with the preload value at the same moment to form a friction-preload comparison data table. In this embodiment, based on the wind turbine main shaft connection bolts, a high-frequency data sampling module is used to acquire preload and friction coefficient data at a frequency of once per second. Data matching is performed using the timestamp field in the database, and an interpolation algorithm is used to handle situations where the sampling time intervals vary slightly. This generates a complete friction-preload comparison table, providing basic data support for subsequent analysis. Using this friction-preload comparison data table, the preload retention efficiency of standard bolt connections during long-term service is evaluated. In this embodiment, preload retention efficiency is defined as the ratio of the actual preload maintained within a specific time period to the initially set preload. This efficiency reflects the load retention capacity of the connection structure after being affected by friction changes. In the specific operation, the simulation data of the bolt connection with an operating cycle of one month is selected, and the average preload value within the initial 10 minutes is used as the benchmark value. The average preload value of each day is calculated and compared with the benchmark value to obtain the preload retention rate in the time series, forming a preload retention efficiency curve. The curve can be used to monitor the long-term stability of the connection system performance. After obtaining the preload retention efficiency curve of the bolt connection, its axial preload loss rate is further calculated. In this embodiment, the loss rate is defined as the rate at which the preload decreases per unit time, which reflects the trend of weakening of the axial tension after the bolt is loaded. The specific method is to perform numerical differentiation on the preload retention efficiency curve and extract the daily or hourly preload change rate using the five-point difference or sliding average method. In the application of wind power main shaft connection, taking the initial preload of the bolt as 250kN as an example, if the preload on the 10th day drops to 230kN, the average loss rate can be calculated to be 0.083kN per hour, which is used to determine whether to trigger the maintenance strategy. According to the change of the above-mentioned axial preload loss rate, the micro-displacement of the connection interface is monitored. Micro-displacement refers to the microscopic relative slip or compression deformation of the contact interface under the interaction of shear and tension and compression, which is usually obtained by placing resistance strain gauges or laser displacement sensors on the edge of the contact surface. During the implementation process, key displacement measurements are performed in different time intervals according to the loss rate. The displacement signal collected by the sensor is noise-filtered and converted into the corresponding micro-displacement curve. The change data is recorded with a resolution of 0.01mm to determine the actual response of the contact surface at different loss stages. The above micro-displacement data is used to construct a shear stress distribution map of the standard bolt connection. Shear stress refers to the force state of the interface perpendicular to the axis of the bolt. Its distribution can be inferred through finite element simulation combined with measured micro-displacement.In the specific operation, the bolt-flange joint surface was analyzed. Finite element load distribution boundary conditions were established based on micro-displacement differences in different regions. The shear stress distribution field was restored using equivalent stress reconstruction technology, thereby obtaining the shear stress gradient at each location in the connection area. In wind turbine equipment, it has been observed that the edge areas of the joint surface often carry greater shear forces, which can be used to assess potential slip risk. Based on this shear stress distribution, the load transfer capacity of the contact surface was further evaluated, namely, the maximum shear force that can be transferred per unit area of the contact interface was determined. In implementation, high-value areas of the shear stress distribution were integrated and summed, and the equivalent shear strength was calculated by combining the contact area and the friction coefficient. This was then compared with the shear yield strength of the material to determine whether there were any overloaded areas. If the shear stress at multiple unit nodes approached or exceeded the material limit, it indicates a decrease in the load transfer capacity of the contact surface, and structural optimization or repair measures should be considered. This analysis helps to establish a model for the direct impact of friction attenuation on load-carrying capacity. Based on the changing trends in load transfer capacity, the axial stiffness of the bolted joint was monitored. Axial stiffness is defined as the inverse ratio of the axial deformation caused by a unit axial load and can be obtained by recording the slope of the load-displacement curve during simulated loading tests. In this embodiment, the preload and displacement data collected in the early stage are used to perform fitting analysis at different time points to obtain a stiffness curve that changes hour by hour, and identify the stiffness reduction trend caused by bolt loosening, wear or material degradation. If the stiffness decreases by more than 10% in a certain time period, it is determined that the connection performance has significantly degraded. According to the change in axial stiffness, the dynamic response characteristics of the bolt connection structure are analyzed, including its displacement amplification effect and resonance behavior under periodic loads. The dynamic response characteristics are dynamically analyzed by loading a typical operating cycle waveform (such as a sine or wind load disturbance) in the simulation and combining it with the current stiffness model. In implementation, a sweep frequency load in the range of 1Hz to 10Hz is loaded on the main shaft model, and the response peak frequency and amplitude are identified by the frequency response function, and the correlation between it and the stiffness change is observed to determine whether the connection system has an instability trend or resonance risk. Based on the dynamic response characteristics and the trend of friction coefficient change, the preload attenuation function is derived. This function is used to characterize the preload change process over time, and usually adopts an empirical function fitting method, such as a power function or an exponential decay model. During implementation, a nonlinear least-squares fit was performed using a previously constructed dataset, with time as the independent variable and friction coefficient and stiffness as weighting factors. This attenuation model, which comprehensively reflects the effects of load transfer and slip, was established. This validated function can be used in structural residual life prediction and intelligent operation and maintenance systems. Based on the constructed preload attenuation function and axial preload loss rate, the preload attenuation characteristics of bolted connections in standard component structures were predicted, including key indicators such as the future trend of preload values and the estimated time to reach the critical maintenance point.During the implementation process, the actual working load curve of the current system and environmental parameters such as temperature and humidity are input, and the preload change curve for the next 30 days and 60 days is extrapolated through the attenuation function. The maintenance threshold is combined to determine the next preload adjustment or bolt replacement time, providing data support for the safe operation of the structure.

[0096] Preferably, step S3 includes the following steps:

[0097] Step S31: constructing a fatigue damage prediction model for a standard component based on the preload force attenuation characteristics;

[0098] The embodiment of the present invention is based on the preload attenuation function obtained in the above steps and its attenuation characteristics over time, and constructs a mathematical model for fatigue damage prediction of standard parts. The specific operation is as follows: first, the SN curve (stress-life curve) of the standard bolt material is selected as the basic data of fatigue performance, and then the number of equivalent load cycles experienced by the bolt in a specific time interval is calculated by combining the historical preload change curve obtained by the attenuation function under actual loading conditions. Then, the Miner linear cumulative damage theory is used to accumulate the stress amplitude and the corresponding number of cycles in each small time interval to obtain the cumulative fatigue damage value, and finally a standard part fatigue life prediction model is constructed. During the implementation process, in order to ensure the parameterization flexibility of the simulation model, an automatic iteration module is written in Python to update the fatigue life output corresponding to different preload attenuation rates, which is convenient for rapid deployment in multiple working conditions. For example, in the simulation of the foundation bolts of a certain wind power equipment, when the initial preload is 120kN and the annual attenuation rate is 10kN, the model predicts that its fatigue life is significantly lower than the design life, suggesting that the structural optimization design needs to be strengthened.

[0099] Step S32: Evaluate the risk probability of bolt connection fracture based on the standard part fatigue damage prediction model;

[0100] The embodiment of the present invention is based on the cumulative value of bolt damage output by the fatigue damage prediction model constructed in the previous step, and further combines the statistical reliability theory to evaluate its fracture risk probability. The specific implementation method includes: first, standardizing the cumulative fatigue damage value, and constructing a probability density function (usually normal or log-normal distribution) with reference to the fatigue limit distribution characteristics of the standard parts, and then calculating the possibility that the current bolt connection is at the fracture critical value based on the cumulative distribution probability of the actual cumulative damage value under the distribution function. In order to improve the accuracy of the model, the Monte Carlo method is introduced into the simulation model, and multiple simulation samplings are performed for different initial preloads, bolt material properties and loading conditions to obtain the fracture risk probability distribution under various conditions. For example, in the bolt connection structure of a wind turbine tower, the predicted fracture probability of a certain model of M30 bolt is 0.12 under the conditions of 1000 hours of annual operation and 7kN of annual preload attenuation, indicating that there is a medium risk level and a special attention mechanism needs to be set in the monitoring plan.

[0101] Step S33: calculating the overall strength of the connection structure based on the preload attenuation characteristics and the risk probability of bolt connection fracture;

[0102] The embodiment of the present invention utilizes the above-mentioned fracture risk probability and the known preload attenuation characteristics to comprehensively evaluate the overall strength of the connection structure. In terms of operation, the standard parts connection structure is first analyzed by a multi-condition static load and fatigue load combination through a simulation platform (such as Abaqus combined with MATLAB interface), the bearing capacity of the structure under different damage degrees is extracted, and the strength output of each section is subjected to risk-weighted averaging with the fracture probability as the weight, to obtain a comprehensive index representing the current overall strength of the connection structure. The unit of this indicator is kilonewton (kN) or megapascal (MPa), and the higher the value, the greater the residual strength. For example, in the ship deck connector, the strength loss of the standard M24 bolt group is evaluated by a three-dimensional parametric model. Under the conditions of an average preload attenuation of 18% and a local damage index of 0.3, the overall connection strength is reduced from the initial 180MPa to 134MPa.

[0103] Step S34: analyzing the stability of the connection structure according to the overall strength of the connection structure, wherein the stability of the connection structure includes a stable state, a critical state, and an unstable state;

[0104] The embodiment of the present invention analyzes the structural stability status of the connection structure based on the calculation of the overall strength index. The specific judgment method includes: comparing the overall strength with the design ultimate bearing value. If the current strength is greater than or equal to 85% of the original design value, it is judged as a "stable state"; if it is between 65% and 85%, it is a "critical state"; if it is lower than 65%, it is judged as an "unstable state". In the actual simulation system, the structural status label is automatically output through the preset stability threshold function module. At the same time, the number of bolts, distribution density and local micro-displacement trend chart are combined to realize visual risk identification. For example, in the elevated bridge pier connection model, when it is detected that the connection strength is 71% of the design value and the cumulative value of bolt damage in the local strain area is high, the simulation system automatically determines it as a "critical state" and uses a yellow highlighted area on the graphical interface to remind structural maintenance personnel to conduct a focused inspection.

[0105] Step S35: generating risk level data based on a predefined risk rating standard according to the stability status of the connection structure and the preload attenuation characteristics; wherein when the stability status of the connection structure is shown as a stable state and the attenuation rate is ≤2kN / month, the risk level data is low; when the stability status of the connection structure is shown as a critical state or the attenuation rate is 2-5kN / month, the risk level data is medium; when the stability status of the connection structure is shown as an unstable state or the attenuation rate is greater than 5kN / month, the risk level data is high;

[0106] After identifying the structural stability status and the corresponding preload decay rate, the embodiment of the present invention generates standard part connection risk level data according to the set risk rating rules. The operation process is: first, the preload decay rate is input into the rating module together with the structural status in units of kN / month, and the risk level label (low, medium, high) is generated according to the rule classification, and then the level result is bound to the corresponding connection node number (such as the bolt number) to generate a risk level database. In the system implementation, the SQL database structure is combined with the front-end automatic screening script to achieve batch processing and rapid call of risk level data. For example, in the high-speed rail sleeper connection system, the monthly monitoring data analysis found that the decay rate of a certain connection section was 3.8kN / month and was in a "critical state". The system automatically assessed the section as a "medium level" risk and marked it as a yellow warning node.

[0107] Step S36: Bind the risk level data with the bolt grid number and the preload loss rate to generate a connection structure loosening risk index.

[0108] The embodiment of the present invention binds the risk level data generated in the previous step to the standard part connection grid number (i.e., the unique identifier of each bolt in the simulation model) and its corresponding preload loss rate to form a loosening risk index for the connection structure. Operationally, by defining the "loosening risk index" as a comprehensive scoring function, the function takes the risk level value, the displacement increment corresponding to the bolt number, and the preload loss ratio as input factors, and outputs a risk index value between 0 and 1. The closer the index is to 1, the more likely the connection is to loosen or fail. The calculation module is integrated into the simulation software (such as ANSYS or a custom Python simulation framework), and the results are displayed in a three-dimensional visualization interface through a color gradient to achieve an intuitive risk situation map for engineering personnel. For example, in a certain aircraft engine bracket connection assembly, through loosening risk index analysis, it was found that the risk index of the bolt numbered N45 was as high as 0.83, and the system automatically prompted that the connection node needed to be replaced or re-reinforced.

[0109] Preferably, step S31 includes the following steps:

[0110] Step S311: extracting the preload force attenuation rate and attenuation period from the preload force attenuation characteristics as fatigue analysis parameters;

[0111] The embodiment of the present invention extracts preload attenuation characteristic data from the preload monitoring system, wherein the preload attenuation rate is defined as the amount of decrease in preload per unit time (for example, in kN / month), and the attenuation period represents the complete time interval in which the preload significantly decreases (for example, 3 months, 6 months, etc.). These two parameters together determine the variable load characteristics that the bolt bears during use, and therefore serve as key inputs in fatigue analysis. The specific operation is: by collecting the preload historical data of the bolt online monitoring system, the least squares method is used to perform linear or exponential fitting on the preload variation curve over time, and the average attenuation rate and the corresponding observation period are extracted. For example, in a bolt simulation model of a wind power tower, the monitoring data shows that the average preload attenuation in the first quarter is 9kN, then the corresponding attenuation rate is 3kN / month, and the attenuation period is 3 months, which serves as the input basis for subsequent load spectrum generation.

[0112] Step S312: establishing a basic material property library for standard parts, including elastic modulus, yield strength, and fatigue limit;

[0113] The embodiment of the present invention establishes a basic property library of standard parts materials, which is used for stress response and fatigue limit calculation in subsequent fatigue life prediction models. The library is composed of material performance parameters of different types of standard bolts (such as 8.8-grade, 10.9-grade carbon steel or stainless steel bolts), mainly including elastic modulus (reflecting the material's ability to resist deformation, in GPa), yield strength (the minimum stress for plastic deformation of the material, in MPa) and fatigue limit (the maximum stress amplitude that the material can withstand under high-cycle fatigue conditions). Parameter data can be extracted from GB / T 3098, ISO 898 or technical manuals provided by various manufacturers and entered into a database system (such as SQLite or Excel file). Taking a certain M24 bolt as an example, its material is 42CrMo steel, with an elastic modulus of 210GPa, a yield strength of 930MPa and a fatigue limit of 450MPa, which serves as the boundary conditions for subsequent stress level evaluation.

[0114] Step S313: Calculate the cyclic load history of each bolt based on the fatigue analysis parameters and generate a load spectrum;

[0115] The embodiment of the present invention generates the cyclic load history of each bolt based on the fatigue analysis parameters (preload attenuation rate and attenuation cycle) extracted above, and constructs a load spectrum accordingly. The load spectrum is a set of sequences that describe the load fluctuations over time in fatigue analysis, which can reflect the actual working conditions of variable loads. In the parameterized simulation model, the attenuation rate is converted into equivalent load amplitude fluctuations, and a time series load input is constructed in combination with typical working conditions (such as vibration frequency, external periodic load). For example, in the analysis of the M24 bolts at the tower connection, if the monthly attenuation rate is 3kN and it is subject to 5Hz periodic fluctuations caused by wind, a load sequence with an initial preload of 120kN and a decrease of 0.05kN per cycle can be generated, forming load spectrum data of 4,320,000 cycles in three months for simulation analysis.

[0116] Step S314: Counting the number of cycles at different stress levels in the load spectrum to form a stress-cycle number distribution;

[0117] The embodiment of the present invention performs statistics on the above-mentioned load spectrum, extracts the number of cycles under different stress amplitude intervals, and obtains a stress-cycle number distribution. The operation method is: the load data in the load spectrum is converted into stress values through the bolt force area, and then the stress intervals are divided into intervals of 10MPa, such as 50-60MPa, 60-70MPa, etc., and the number of load cycles in each interval is counted. This step can use Python scripts in conjunction with the NumPy library to implement data processing and group statistics to generate a distribution list. For example, in a certain tower simulation model, statistics show that the number of cycles in the stress range of 80-90MPa is the largest, which is 150,000 times, followed by 90-100MPa, which is 90,000 times. The distribution results will be used for subsequent life calculations.

[0118] Step S315: Calculate the number of cycles allowed at each stress level based on the SN curve according to the material basic property library to obtain the theoretical fatigue life;

[0119] The embodiment of the present invention uses the fatigue limit and SN curve model in the basic material property library to calculate the number of cycles allowed at each stress level, thereby obtaining the theoretical fatigue life. The SN curve reflects the logarithmic relationship between stress amplitude and fatigue life, and can usually be derived by standard formula or obtained by fitting material test data. In the simulation model, by calling the curve function, the stress amplitude corresponding to each stress level is input to calculate the theoretically allowed number of cycles. For example, the fatigue life corresponding to a stress amplitude of 80MPa under the SN curve of a certain high-strength steel is 300,000 times, and 90MPa corresponds to 150,000 times. This result constitutes a theoretical life table, which provides basic data for subsequent fatigue damage calculations.

[0120] Step S316: Divide the actual number of cycles in the stress-cycle number distribution by the theoretical fatigue life to calculate the fatigue damage degree; linearly accumulate the fatigue damage degrees of all stress levels to obtain the overall fatigue damage value;

[0121] The embodiment of the present invention calculates the fatigue damage degree based on the stress-cycle number distribution and the theoretical fatigue life, and accumulates the total fatigue damage value. The calculation of the fatigue damage degree is based on the Miner linear damage accumulation criterion, that is, the actual number of cycles at each stress level is divided by the allowable number of cycles at that level to obtain the single damage degree. The damage degrees at all stress levels are then summed up to form the final fatigue damage value. In the simulation script, the data tables are compared and summarized item by item. For example, in the 80MPa range, the actual number of cycles is 150,000 times and the allowable number is 300,000 times, so the damage degree is 0.5; in the 90MPa range, the actual number is 90,000 times and the allowable number is 150,000 times, the damage degree is 0.6, and the total fatigue damage after accumulation is 1.1, indicating that the bolt has exceeded the fatigue life limit.

[0122] Step S317: establishing a functional relationship between the overall fatigue damage value and time to form a standard component fatigue damage prediction model.

[0123] The embodiment of the present invention fits the fatigue damage value obtained in the previous step with the corresponding time data point, establishes a functional relationship of fatigue damage growth over time, and finally constructs a fatigue damage prediction model. The model can adopt linear, exponential or hyperbolic function form, and the optimal fitting method is determined according to the damage growth trend. In the model implementation, the curve_fit function in the SciPy library is used to perform curve fitting on the fatigue damage value-time point series, and output the model function and its parameters. Taking the aforementioned three-month simulation period as an example, the corresponding cumulative fatigue damage values are 0.3, 0.7, and 1.1 respectively. After fitting, an exponential fatigue damage growth function is obtained, which is used as the core model for fatigue life monitoring and remaining life prediction of standard parts. The simulation platform is embedded in engineering applications to realize visual early warning and life prediction functions.

[0124] Preferably, step S4 includes the following steps:

[0125] Step S41: establishing a parameter sensitivity analysis matrix according to the connection structure loosening risk index;

[0126] The embodiment of the present invention lists all possible parameters that affect the stability and loosening risk of bolted connections, including preload, decay rate, material elastic modulus, bolt type, ambient temperature and humidity, etc. Then, a sensitivity analysis method is used to calculate the degree of influence of each parameter on the loosening risk index of the connection structure. Common methods include local sensitivity analysis (such as perturbation method) or global sensitivity analysis (such as variance analysis method). By performing a systematic analysis of all parameters, a sensitivity analysis matrix is generated, in which each row in the matrix represents a parameter, and each column represents the contribution of the parameter to the loosening risk index under different conditions. For example, if the decay rate contributes more to the risk index, its correlation value in the matrix will be higher; conversely, when the influence of the preload is smaller, the correlation value is lower. This matrix provides a basis for screening key parameters in subsequent steps.

[0127] Step S42: Based on the parameter sensitivity analysis matrix, the parameters with an absolute correlation value of ≥0.5 are screened and marked as key influencing parameters, and the rest are marked as secondary influencing parameters;

[0128] The embodiment of the present invention evaluates the correlation coefficient between each parameter and the loosening risk index based on the sensitivity analysis matrix obtained in step S41. The specific operation is to obtain the influence of the parameter by calculating the Pearson correlation coefficient between each parameter and the risk index. If the absolute value of the correlation coefficient between a certain parameter and the loosening risk index is greater than or equal to 0.5, it is considered that the parameter has a strong influence on the risk index and is marked as a key influencing parameter; if the absolute value of the correlation coefficient is less than 0.5, it is marked as a secondary influencing parameter. For example, in a certain practical application, the correlation coefficient between the preload attenuation rate and the loosening risk index may be 0.7, so it is marked as a key influencing parameter; and the fatigue limit of the material may have a correlation coefficient of 0.4 with the risk index, which is a secondary influencing parameter.

[0129] Step S43: formulating a data compression priority strategy based on the key impact parameters and the secondary impact parameters;

[0130] The embodiment of the present invention formulates a data compression strategy based on the key influencing parameters and secondary influencing parameters screened out in step S42. For key influencing parameters, since they have a greater impact on the loosening risk index, more data should be retained as a priority for higher-precision monitoring; while for secondary influencing parameters, a higher compression ratio can be adopted to reduce the amount of data transmitted. The specific operation is to set the data compression ratio of each parameter according to the criticality of the parameter. For example, for the preload attenuation rate and ambient temperature and humidity (if marked as key influencing parameters), a low compression ratio (such as keeping 70% of the original data) is adopted, while for the fatigue limit of the material and the bolt type (such as being marked as secondary influencing parameters), a higher compression ratio (such as only retaining 30% of the original data) can be adopted. Through this strategy, the data transmission and storage costs can be optimized to the maximum extent while ensuring the calculation accuracy of the loosening risk index.

[0131] Step S44: using the data compression priority strategy to perform monitoring data importance classification;

[0132] The embodiment of the present invention grades the importance of all data actually monitored according to the data compression priority strategy in step S43. First, all monitoring data are grouped by parameter type, and each group of data is scored according to its importance in the loosening risk prediction model. The monitoring data corresponding to key parameters will be assigned higher weights, while secondary parameters will correspond to lower weights. For example, if the weight of the monitoring data of the preload decay rate in the model is 0.8, the priority of this data is higher; and the priority of the data of the ambient temperature and humidity is 0.4. Based on this classification, the system will decide which data requires more storage resources and transmission bandwidth, and which data can be properly compressed and optimized for transmission.

[0133] Step S45: Generate an edge computing parameter compression scheme based on the importance classification of the monitoring data;

[0134] The embodiment of the present invention uses the result of the importance classification of the monitoring data in S44 to generate an edge computing parameter compression scheme. Edge computing refers to preliminary data processing and analysis at the device end or close to the data source to reduce data transmission and delay. When generating a compression scheme, the system will select appropriate compression algorithms and compression ratios to process different categories of data based on the importance priority of the data. Specifically, for data with high importance, a low-loss compression algorithm (such as lossless compression or mild compression) is used to ensure the accuracy of the data; for data with lower importance, a more aggressive compression algorithm (such as lossy compression) can be used to reduce data storage and transmission requirements. For example, for the bolt preload attenuation rate and temperature change data (high importance), lossless compression is used, while for ambient humidity data (low importance), a lossy compression algorithm is used.

[0135] Step S46: Establish a priority parameter transmission queue based on the edge computing parameter compression scheme; transmit incremental data packets according to the priority parameter transmission queue to obtain updated data in the cloud.

[0136] According to the compression scheme generated in step S45 of the embodiment of the present invention, the system will establish a parameter transmission queue in order of priority. Parameters with high priority (such as key parameters such as preload decay rate and decay period) will be placed at the front of the transmission queue, while parameters with low priority (such as secondary parameters such as temperature, humidity, and air pressure) will be placed at the back. During data transmission, the system will transmit incremental data packets in the order of the queue, and include the incremental part of the new round of monitoring results in each data transmission. These incremental data will be sent to the cloud server for processing and updating. The specific operation is that after the edge computing device analyzes and generates incremental data packets locally, it sends them to the cloud through the network in order of priority. The cloud system processes the received data packets in real time and updates the loosening risk index. For example, if the preload decay rate changes significantly at a certain moment, the data will be sent as an incremental packet first, and if the temperature data changes less, it can be sent later.

[0137] It is particularly important that step S46 includes the following steps:

[0138] Design a data transmission power consumption optimization strategy based on packet size and transmission distance according to the edge computing parameter compression scheme;

[0139] Calculate the transmission priority score of each parameter using the data transmission power consumption optimization strategy;

[0140] Create a priority parameter transmission queue based on the transmission priority score of each parameter;

[0141] Packet grouping and encoding based on priority parameter transmission queue;

[0142] Incremental data packet transmission is implemented based on data packet grouping and encoding to obtain updated data in the cloud.

[0143] The present embodiment considers that packet size and transmission distance directly affect communication power consumption and designs an optimization strategy to balance these two factors. In practical applications, packet size and transmission power consumption are positively correlated: larger packets consume more power during transmission. The relationship between transmission distance and power consumption is that the longer the distance, the greater the power consumption. Therefore, when designing the optimization strategy, it is necessary to first measure the packet size corresponding to each parameter and calculate the power consumption under different transmission distances based on the wireless communication technology (such as Wi-Fi, Bluetooth, Zigbee, etc.) and the actual operating environment of the device. For longer-distance transmissions, the packet size should be minimized as much as possible; for shorter-distance transmissions, the packet size can be appropriately increased while ensuring data accuracy. Furthermore, based on the importance of each parameter and actual application requirements, the transmission duration and frequency are determined to further optimize power consumption. For example, when transmitting important parameters such as preload decay rate, smaller packets and shorter transmission distances are used to reduce power consumption. Based on the data transmission power consumption optimization strategy designed in step S51, a power consumption optimization score is assigned to each parameter. First, based on the relationship between packet size, transmission distance, and power consumption, the power consumption of each parameter under different transmission conditions is calculated. Then, power consumption and parameter importance are weighted to derive a transmission priority score for each parameter. Specifically, key parameters with a greater impact on loosening risk (such as preload decay rate and decay period) are prioritized for transmission, and power consumption of these parameters is optimized to reduce energy consumption during transmission. Within the framework of power optimization, the transmission of less important parameters can be appropriately delayed to reduce power consumption. For example, for parameters with less impact on risk, such as bolt type, the transmission distance can be relatively long, and the data packet size can be appropriately increased to improve transmission efficiency and reduce the power consumption overhead of frequent transmission. All monitoring data parameters are sorted by priority to form a priority parameter transmission queue. First, ensure that the most critical parameters in the risk prediction model (such as preload decay rate and bolt loosening risk index) are at the front of the queue. These parameters will be processed and transmitted first during the data transmission process. Second, less important parameters (such as ambient temperature and humidity, material fatigue limit, etc.) will be placed at the back of the queue and their transmission will be delayed based on network bandwidth and power optimization strategies. This transmission queue is based on a comprehensive ranking of importance and power consumption, ensuring that data with a greater impact on structural safety and risk assessment is transmitted first within limited transmission resources, while also reducing power consumption. For example, among high-priority parameters, the preload decay rate parameter may be prioritized for transmission along with the actual load status of the bolt. The data in the priority parameter transmission queue is grouped, with each group containing several parameter values, which are allocated based on priority. The grouping and encoding of data packets not only takes into account the order in which the data is transmitted, but is also designed based on power consumption optimization requirements.Data packets for high-priority parameters are grouped using smaller packet sizes and simplified encoding methods to facilitate faster transmission and reduce power consumption. For lower-priority parameters, the packet size can be appropriately increased, while compression coding techniques can be used to reduce transmission bandwidth usage. For example, bolt preload decay rate and load history data are grouped into small packets for priority transmission, while temperature and humidity data can be packaged into larger packets and compressed to conserve bandwidth. Coding techniques such as Huffman coding or the LZ77 compression algorithm may be used to improve transmission efficiency. Based on the packet grouping and encoding results in step S54, incremental data packets are transmitted. Incremental transmission refers to transmitting only the data that has changed since the last data update, rather than the entire data set. Each packet contains new or changed monitoring data within a specific time period, optimizing bandwidth and power consumption during transmission. Data packets are sent according to transmission priority, ensuring that high-priority data is sent first. For example, if the preload decay rate changes significantly, the corresponding packet is sent to the cloud for update first. After receiving the incremental data, the cloud decompresses and decodes it and updates the parameter values in the risk prediction model, enabling real-time data updates and risk assessments. Once the cloud-based update is complete, a new loosening risk index or fatigue damage prediction result is generated for subsequent decision-making.

[0144] Preferably, step 5 comprises the following steps:

[0145] Step S51: extracting the latest status parameters of the standard bolt connection from the cloud updated data to obtain the latest status data;

[0146] In the embodiment of the present invention, the cloud updates the relevant parameters of the bolt connection through the received incremental data packets, including the preload decay rate, load history, ambient temperature and humidity, and actual load data of the bolt connection. These data reflect the changes in the working state of the bolt connection in different time periods, such as the decay trend of the preload and the possibility of bolt loosening. Based on the transmitted data, the system will extract the state parameters of each bolt connection, such as actual preload, stress distribution, temperature change, etc. The extraction process includes data cleaning, outlier processing and parameter alignment to ensure that all data accurately reflects the actual situation. For example, if the preload decay rate changes too much, it is necessary to focus on monitoring the connection point and extract its detailed parameters of the change as the basic data for subsequent simulation and analysis.

[0147] Step S52: updating the basic parameters of the pre-acquired basic simulation model of the standard component bolt connection according to the latest status data, adjusting the boundary conditions and material properties, regenerating the mesh model, and forming a standard component simulation model;

[0148] The embodiment of the present invention updates the basic parameters of the pre-acquired basic simulation model of the standard parts bolt connection according to the latest status data, adjusts the boundary conditions and material properties, regenerates the mesh model, and forms a standard parts simulation model. Based on the latest status data extracted from the cloud, the system will first update the basic simulation model of the standard parts bolt connection. This includes adjusting the preload at the bolt connection, updating the boundary conditions to reflect the current stress state, and correcting the material properties (such as elastic modulus, yield strength, etc.) according to the latest temperature and humidity data. The adjusted material properties and boundary conditions will affect the calculation of stress and deformation in the simulation model. Then, the system regenerates the mesh model of the standard parts, takes into account the geometry and load distribution of the bolt connection, and refines the mesh division so that the simulation results can more accurately reflect the actual situation. A new standard parts simulation model is formed through the updated mesh model and modified parameters, which will be used for subsequent dynamic execution and fatigue life prediction.

[0149] Step S53: dynamically executing the standard component simulation model, calculating the standard component structural response, and generating fatigue life prediction data;

[0150] After the standard parts simulation model is updated in the embodiment of the present invention, the system will perform dynamic simulation to simulate the dynamic response of the bolt connection in the actual working environment. This step takes into account the influence of various environmental factors such as periodic changes in load, wind speed fluctuations, temperature changes, etc. on the bolt connection, and performs calculations through finite element analysis (FEA) technology. Dynamic simulation will output structural response data such as stress, strain, and displacement, and generate fatigue life predictions based on these data. Fatigue life prediction is based on the accumulated stress-strain history, combined with the SN curve and historical load spectrum of the material, to predict the possible fatigue damage level and remaining service life of the standard parts in the future working process. The generated fatigue life data will provide a fatigue damage distribution map for each bolt connection, helping to identify which bolt connections are more prone to fatigue failure.

[0151] Step S54: performing structural performance evaluation on the fatigue life prediction data, identifying structural weaknesses and optimization trends, and generating structural optimization recommendation parameters;

[0152] In an embodiment of the present invention, the system performs structural performance evaluation based on fatigue life prediction data. First, the system analyzes the fatigue damage degree of each bolt connection and identifies those bolt connections with more serious damage and shorter fatigue life. These connection points identified as weak links need to be optimized or reinforced as a priority. Through statistical analysis of fatigue life prediction data, the system can also discover some trends, such as frequent failures of certain bolt connections due to material degradation or improper operation, or accelerated damage due to excessive loads. Based on these trends, the system will generate a series of optimization suggestions, such as increasing the preload of a specific bolt connection, or selecting a more fatigue-resistant material to replace the current material to extend the service life of the bolt connection. In addition, the system may also recommend regular inspection or replacement of bolt connections to avoid the risk of failure.

[0153] Step S55: Calculate the optimal preload range, bolt replacement cycle, and inspection frequency based on the recommended structural optimization parameters, and generate maintenance strategy guidance data including maintenance time, specific measures, and required resources in combination with the wind turbine operation plan.

[0154] In the embodiment of the present invention, based on the structural optimization recommendation parameters generated in step S54, the system will calculate the optimal preload range to ensure that the bolt connection maintains the optimal stress state during operation and avoids damage caused by over-tightening or loosening. In addition, the system will also calculate the replacement cycle and inspection frequency of the bolts. These parameters are derived based on fatigue life prediction data, the health status of the bolt connection, and changes in material properties. Combined with the operation plan of the wind turbine, the system generates a detailed maintenance strategy guidance data, which includes the maintenance time of each bolt connection, specific maintenance measures (such as bolt tightening, material replacement, etc.) and the required resources (such as tools, materials and labor). For example, if certain bolt connections have high fatigue damage, the system may recommend arranging maintenance in advance or even replacing the bolts to avoid failure of the wind turbine.

[0155] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0156] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A parametric simulation model design method based on standard parts structure, characterized in that: The following steps are involved: Step S1: collecting standard bolt connection data in a wind turbine generator set; extracting blade surface vibration characteristic information based on the standard bolt connection data; mapping the blade surface vibration characteristic information to the bolt connection, calculating the instantaneous stress distribution and integrating it to generate stress concentration distribution data; Step S2: Calculating the local deformation accumulation value of the standard part based on the stress concentration distribution data at the connection; measuring the change trend of the contact surface friction coefficient based on the local deformation accumulation value of the standard part; and deriving the preload force attenuation characteristic based on the change trend of the contact surface friction coefficient; Step S3: Fatigue cumulative response modeling is performed based on the preload attenuation characteristics to obtain a standard component fatigue damage prediction model; the standard component fatigue damage prediction model and the preload attenuation characteristics are used to evaluate the connection status based on the time series damage evolution law, thereby obtaining a connection structure loosening risk index; Step S4: Perform edge-side feature reorganization and redundancy compression based on the connection structure loose risk index to generate an edge computing parameter compression scheme; Establish a priority parameter transmission queue based on the edge computing parameter compression scheme; Incremental data packet transmission is performed according to the priority parameter transmission queue to obtain updated data from the cloud; Step S5: Reconstruct the standard parts simulation model based on the updated cloud data; perform structural performance evaluation on the fatigue life prediction data, identify structural weaknesses and optimization trends, and generate recommended structural optimization parameters; perform preventive maintenance scheduling for standard parts based on the recommended structural optimization parameters to obtain maintenance strategy guidance data.

2. The method for designing a parameterized simulation model based on a standard component structure according to claim 1, wherein: Step S1 includes the following steps: Step S11: deploying the sensor network, installing the preload force sensor, temperature sensor, and triaxial acceleration sensor, and setting data acquisition parameters including sampling frequency, recording duration, and trigger conditions; Step S12: using the sensor network to continuously collect preload information, temperature information, and vibration acceleration information of standard bolt connections under normal operating conditions to establish an original database; Step S13: pre-processing the original database and unifying the units and dimensions to form standardized standard bolt connection data; Step S14: performing time-frequency domain conversion on the standard bolt connection data, extracting the vibration amplitude, frequency, and phase information within the blade rotation period, and forming blade surface vibration characteristic information; Step S15: Calculating the stress distribution in the area around the bolt hole using the stress transfer function based on the blade surface vibration characteristic information to obtain the instantaneous stress distribution at the connection; Step S16: Time-integrate the instantaneous stress distribution at the connection, calculate the stress accumulation effect and stress gradient change, and generate stress concentration distribution data at the connection.

3. The method for designing a parameterized simulation model based on a standard component structure according to claim 2, wherein: Step S15 includes the following steps: Step S151: extracting vibration amplitude, frequency and phase components from blade surface vibration characteristic information and constructing a vibration excitation vector; Step S152: Obtaining information on bolt hole dimensions, bolt pre-tightening areas, and blade connection surface geometry of the wind turbine generator set, thereby establishing a bolt connection structure mesh model; Step S153: constructing a transfer function matrix from blade surface vibration to bolt connection stress based on the bolt connection structure grid model and blade surface vibration characteristic information; Step S154: Calculate the stress response of the area around the bolt hole according to the transfer function matrix and the vibration excitation vector to obtain the stress component value of each grid point; Step S155: Calculating the principal stress and equivalent stress of each grid point based on the stress component values to generate a stress distribution vector field; Step S156: performing spatial interpolation on the stress distribution vector field, and calculating the stress concentration factor around the bolt hole based on the pre-acquired material properties and structural geometric parameters; Step S157: Generate a stress distribution cloud map around the bolt hole based on the stress distribution vector field and the stress concentration factor to form an instantaneous stress distribution at the connection.

4. The method for designing a parameterized simulation model based on a standard component structure according to claim 3, wherein: Calculating the local deformation accumulation value of the standard component based on the stress concentration distribution data at the connection in step S2 includes: Separate the static stress component and dynamic stress component of the stress concentration distribution data at the connection and establish a stress spatiotemporal distribution model; Obtain the mechanical property parameter library of standard parts materials, including elastic modulus, Poisson's ratio and yield strength, and generate the material response characteristic matrix; According to the stress spatiotemporal distribution model and the material response characteristic matrix, the elastic region strain distribution calculation based on Hooke's law and the plastic region strain distribution calculation are performed to form the instantaneous strain distribution field; Perform time integration on the instantaneous strain distribution field, calculate the cumulative strain of each grid point during the working cycle of the wind turbine generator set, and generate periodic strain accumulation data; Based on the cyclic strain accumulation data, the strain energy density distribution of the standard bolt connection is analyzed to identify the energy concentration area and obtain the strain energy density distribution map; Determine the local high strain interval and deformation hotspot location of the standard part based on the strain energy density distribution diagram, and generate deformation hotspot area identification data; The plastic strain accumulation effect under cyclic loading is calculated for the identification data of the deformation hotspot area to obtain the local plastic damage accumulation value; According to the cyclic strain accumulation data and the local plastic damage accumulation value, the deformation state description of each key position of the standard part is established, and the local deformation accumulation value of the standard part is generated.

5. The method for designing a parameterized simulation model based on a standard component structure according to claim 4, wherein: The measurement of the change trend of the contact surface friction coefficient based on the accumulated value of the local deformation of the standard part in step S2 includes: Extract the accumulated plastic strain of the corresponding mesh area according to the mark of the deformation hotspot area in the local deformation accumulated value of the standard part; The change of the micro-gap between the contact surfaces is measured based on the accumulated plastic strain, and the time series data of the contact surface gap distribution is generated; According to the time series data of the contact surface gap distribution, the relative slip of the contact surface is collected using a displacement sensor, and the proportional relationship between the slip and the preload force attenuation rate is calculated; Based on the proportional relationship between slip and preload force attenuation rate, combined with the thread pitch and contact surface area of the bolt preload area, the dynamic correction value of the contact surface friction coefficient is calculated; Generate a contact surface friction characteristic data table containing time stamp, preload value, temperature value and friction coefficient based on dynamic correction value; Based on the friction coefficient change curve in the friction characteristic data table, mark the abnormal time interval in which the friction coefficient decrease rate exceeds the preset threshold; The abnormal time interval is timestamped and matched with the plastic strain jump events in the deformation hotspot area to verify the correlation between the change in friction coefficient and deformation accumulation, and the change trend of the contact surface friction coefficient is obtained.

6. The method for designing a parameterized simulation model based on a standard component structure according to claim 5, characterized in that: Derivation of the preload force attenuation characteristics based on the contact surface friction coefficient variation trend in step S2 includes: Generate a friction-preload comparison data table based on the change trend of the contact surface friction coefficient and the preload information collected by the preload sensor; Use friction-preload comparison data tables to estimate the preload retention efficiency of standard bolt connections; Calculate the axial preload loss rate of the bolt based on the preload retention efficiency of the standard bolt connection; Monitor the micro-displacement of the connection interface based on the axial preload loss rate; Analyze the shear stress distribution of standard bolt connections using micro-displacement of the connection interface; Evaluate the load transfer capacity of the contact surface based on the shear stress distribution of the standard bolted connection; Detect changes in the axial stiffness of the bolted connection based on the load transfer capacity of the contact surface; The dynamic response characteristics of bolt connections are analyzed using the changes in the axial stiffness of the bolt connections. The preload attenuation function is derived based on the dynamic response characteristics of the bolt connection and the variation trend of the friction coefficient. The preload attenuation characteristics are predicted based on the preload attenuation function and the axial preload loss rate.

7. The method for designing a parameterized simulation model based on a standard component structure according to claim 6, wherein: Step S3 includes the following steps: Step S31: constructing a fatigue damage prediction model for a standard component based on the preload force attenuation characteristics; Step S32: Evaluate the risk probability of bolt connection fracture based on the standard part fatigue damage prediction model; Step S33: calculating the overall strength of the connection structure based on the preload attenuation characteristics and the risk probability of bolt connection fracture; Step S34: analyzing the stability of the connection structure according to the overall strength of the connection structure, wherein the stability of the connection structure includes a stable state, a critical state, and an unstable state; Step S35: generating risk level data based on a predefined risk rating standard according to the stability status of the connection structure and the preload attenuation characteristics; wherein when the stability status of the connection structure is shown as a stable state and the attenuation rate is ≤2kN / month, the risk level data is low; when the stability status of the connection structure is shown as a critical state or the attenuation rate is 2-5kN / month, the risk level data is medium; when the stability status of the connection structure is shown as an unstable state or the attenuation rate is greater than 5kN / month, the risk level data is high; Step S36: Bind the risk level data with the bolt grid number and the preload loss rate to generate a connection structure loosening risk index.

8. The method for designing a parameterized simulation model based on a standard component structure according to claim 7, wherein: Step S31 includes the following steps: Step S311: extracting the preload force attenuation rate and attenuation period from the preload force attenuation characteristics as fatigue analysis parameters; Step S312: establishing a basic material property library for standard parts, including elastic modulus, yield strength, and fatigue limit; Step S313: Calculate the cyclic load history of each bolt based on the fatigue analysis parameters and generate a load spectrum; Step S314: Counting the number of cycles at different stress levels in the load spectrum to form a stress-cycle number distribution; Step S315: Calculate the number of cycles allowed at each stress level based on the SN curve according to the material basic property library to obtain the theoretical fatigue life; Step S316: Divide the actual number of cycles in the stress-cycle number distribution by the theoretical fatigue life to calculate the fatigue damage degree; linearly accumulate the fatigue damage degrees of all stress levels to obtain the overall fatigue damage value; Step S317: establishing a functional relationship between the overall fatigue damage value and time to form a standard component fatigue damage prediction model.

9. The method for designing a parameterized simulation model based on a standard component structure according to claim 8, wherein: Step S4 includes the following steps: Step S41: establishing a parameter sensitivity analysis matrix according to the connection structure loosening risk index; Step S42: Based on the parameter sensitivity analysis matrix, the parameters with an absolute correlation value of ≥0.5 are screened and marked as key influencing parameters, and the rest are marked as secondary influencing parameters; Step S43: formulating a data compression priority strategy based on the key impact parameters and the secondary impact parameters; Step S44: using the data compression priority strategy to perform monitoring data importance classification; Step S45: Generate an edge computing parameter compression scheme based on the importance classification of the monitoring data; Step S46: Establish a priority parameter transmission queue based on the edge computing parameter compression scheme; transmit incremental data packets according to the priority parameter transmission queue to obtain updated data in the cloud.

10. The method for designing a parameterized simulation model based on a standard component structure according to claim 9, wherein: Step 5 includes the following steps: Step S51: extracting the latest status parameters of the standard bolt connection from the cloud updated data to obtain the latest status data; Step S52: updating the basic parameters of the pre-acquired basic simulation model of the standard component bolt connection according to the latest status data, adjusting the boundary conditions and material properties, regenerating the mesh model, and forming a standard component simulation model; Step S53: dynamically executing the standard component simulation model, calculating the standard component structural response, and generating fatigue life prediction data; Step S54: performing structural performance evaluation on the fatigue life prediction data, identifying structural weaknesses and optimization trends, and generating structural optimization recommendation parameters; Step S55: Calculate the optimal preload range, bolt replacement cycle, and inspection frequency based on the recommended structural optimization parameters, and generate maintenance strategy guidance data including maintenance time, specific measures, and required resources in combination with the wind turbine operation plan.

Images