A structure error tolerance model construction method for van body connection positioning

Abstract

The application relates to the technical field of structural design and optimization, and particularly discloses a structural error tolerance model construction method for compartment connection positioning, which comprises the following steps: acquiring real point cloud data of a bolt connection pair through three-dimensional laser scanning, realizing parameterized digital geometric modeling in combination with a deep neural network, and giving material properties and adaptive mesh division; generating a dynamic load sequence based on a vehicle driving condition, applying the dynamic load sequence to the model, and extracting full-field displacement vector data by using a digital image correlation technology; further identifying a micro-slip area of a contact interface, and calculating an energy dissipation index of the micro-slip area; establishing a quantitative mapping relationship between the energy dissipation and fatigue life through experiment calibration, taking a target life as a constraint, iteratively optimizing key design parameters by using a pattern search method, and finally outputting a connection structure scheme meeting the durability requirement.

Description

Technical Field

[0001] This invention relates to the field of structural design and optimization technology, specifically to a method for constructing a structural error tolerance model for compartment connection and positioning. Background Technology

[0002] The connection between the body and the base is typically achieved using bolted joints to allow for detachable assembly and positioning. These connection structures endure complex dynamic loads during service, including inertial forces from vehicle movement, vibrations caused by road surface excitation, and alternating stresses resulting from changing operating conditions. Under these loads, the connection interface is highly susceptible to minute relative slippage (microslip), leading to interfacial friction, energy dissipation, and localized stress concentration, ultimately inducing fretting fatigue damage and becoming a major cause of connection structure failure. Traditional connection structure designs often rely on empirical formulas or static strength checks, which struggle to accurately reflect the nonlinear contact behavior and dynamic response characteristics under actual operating conditions. Although digital image correlation (DIC) technology has been used in recent years for experimental measurement of deformation fields, it remains costly, difficult to replicate, and hard to integrate with the design process. Furthermore, existing simulation methods still have shortcomings in modeling accuracy, load realism, and life prediction accuracy, particularly lacking quantitative analysis models that systematically correlate actual geometric errors, dynamic loads, microslip behavior, and fatigue life. Therefore, there is an urgent need for a connection structure design optimization method that can comprehensively consider manufacturing deviations, real working conditions and mechanical response, so as to improve connection reliability and achieve error-tolerant design.

[0003] The technical problem solved by this invention is:

[0004] In the existing technology, the design of bolted connection structures makes it difficult to accurately predict their microslippage behavior and fatigue life under complex dynamic loads. This leads to premature failure of the connection pair in actual use due to fretting wear and energy accumulation and dissipation, affecting the safety and durability of the overall structure. Summary of the Invention

[0005] The purpose of this invention is to provide a method for constructing a structural error tolerance model for compartment connection and positioning, so as to solve the problems mentioned above.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A method for constructing a structural error tolerance model for compartment connection and positioning includes the following steps:

[0008] S1: Construct a parametric digital geometric model that includes bolts, connecting plates and the base, and assign corresponding material properties to each component;

[0009] S2: Generate a benchmark digital speckle image with random speckle characteristics based on a parametric digital geometric model, and simultaneously generate a dynamic load sequence simulating vehicle driving conditions.

[0010] S3: Apply the dynamic load sequence step by step to the parametric digital geometric model for mechanical solution. After each load step, calculate the deformed digital speckle image on the surface of the parametric digital geometric model. Compare the deformed digital speckle image with the reference digital speckle image in a non-contact full-field comparison and extract the full-field displacement vector data of the connecting pair under each load step.

[0011] S4: Identify the micro-slip regions on the contact interface of the connecting pair based on the full-field displacement vector data, and calculate the energy dissipation index corresponding to the micro-slip regions in each load cycle;

[0012] S5: Establish a quantitative mapping relationship between energy dissipation index and fatigue life. With the preset target life as a constraint, iteratively optimize the design parameters of the parameterized digital geometric model to ensure that the energy dissipation index of the key area meets the life requirements, and output the final optimized connection structure scheme.

[0013] As a further aspect of the present invention: the construction process of the parameterized digital geometric model is as follows:

[0014] The actual point cloud data of the connector is obtained by 3D laser scanning. The point cloud data is then denoised and simplified to establish an initial parametric digital geometric model that reflects the actual manufacturing tolerances and surface morphology.

[0015] Based on deep neural networks, feature recognition and parametric reconstruction of the initial parametric digital geometric model are performed. The key dimensions in the initial parametric digital geometric model are defined as adjustable design variables, where the input is the feature vector of point cloud data and the output is the structural topology and dimensional constraint relationship of the parametric digital geometric model.

[0016] The parametric digital geometric model is adaptively meshed using multi-scale mesh generation technology. A dense surface mesh is generated in the contact area, while a sparse mesh is used in non-critical areas, thus establishing a parametric digital geometric model suitable for high-precision contact analysis.

[0017] As a further aspect of the present invention: the generation of the dynamic load sequence simulating vehicle driving conditions specifically includes:

[0018] Based on the geometric and material properties of the parametric digital geometric model, the mass distribution characteristics and structural stiffness characteristics are automatically extracted, and the overall mass matrix and stiffness matrix are calculated.

[0019] Based on the kinematic parameters of the vehicle under typical operating conditions, and combined with the mass matrix and stiffness matrix, the basic inertial load components at the connection interface are calculated using the Newton-Euler equations.

[0020] Using road spectrum excitation data as input, load transfer path analysis is performed through stiffness matrix to calculate the dynamic additional load components caused by road excitation.

[0021] The basic inertial load components and the dynamic additional load components are superimposed and synthesized in the time domain to generate a long-term dynamic load sequence that matches the characteristics of the parametric digital geometric model.

[0022] As a further aspect of the present invention: the stepwise application of dynamic load sequences to a parametric digital geometric model for mechanical solution specifically includes:

[0023] In the mechanical solution process of each incremental load step, the complete Newton-Rafaelsen iteration method is used to solve the equilibrium equations considering geometric nonlinearity and contact nonlinearity to ensure the convergence and accuracy of the solution results.

[0024] After the solution converges in each load step, the spatial coordinate changes of all surface nodes of the parameterized digital geometric model are extracted, the discrete node displacements are reconstructed into a continuous model surface deformation field, and a deformed digital speckle image is generated accordingly.

[0025] The initial value prediction of the speckle images before and after deformation is performed using the phase correlation method, and then fine matching is performed by combining the zero-mean normalized cross-correlation function to calculate the two-dimensional displacement component of each pixel.

[0026] By integrating the displacement increments of all load steps, time-series full-field displacement vector data of the connection pair throughout the entire dynamic load history is constructed.

[0027] As a further aspect of the present invention: the non-contact full-field comparison of the deformed digital speckle image and the reference digital speckle image specifically includes:

[0028] Define a regular grid of dots on the reference digital speckle image as the center of the computational sub-region to be tracked;

[0029] For the deformed digital speckle image under each load step, the inverse combination Gauss-Newton algorithm based on gray-level gradient optimization is used to iteratively calculate each calculation sub-region and solve the displacement field that minimizes the gray-level difference between the sub-regions in the two images.

[0030] A displacement field smoothing constraint based on a quadratic B-spline function is introduced to regularize the solved initial displacement field in order to suppress non-physical fluctuations caused by noise.

[0031] By calculating the spatial gradient of the displacement field in adjacent computational sub-regions, the continuous strain tensor distribution of the entire field is obtained, and high-precision full-field displacement vector data is output by combining the displacement field.

[0032] As a further aspect of the present invention: the identification of micro-slippage areas on the contact interface of the connection pair specifically includes:

[0033] Extract the tangential displacement time history data of the contact interface nodes within a complete load cycle, and calculate the displacement amplitude and phase angle distribution of each node;

[0034] The displacement characteristics of nodes are automatically classified, and nodes with similar amplitude and phase characteristics are grouped into the same slip mode region;

[0035] Calculate the displacement direction consistency coefficient within each cluster region, and preliminarily determine the regions with a direction consistency coefficient lower than a first set threshold as micro-slip regions;

[0036] Morphological closing operations are performed on the initially identified microslip regions to eliminate discrete noise points and fill the voids inside the regions, generating continuous and complete microslip region boundaries.

[0037] As a further aspect of the present invention: the calculation of the energy dissipation index corresponding to the micro-slip region within each load cycle specifically includes:

[0038] Within the identified microslip region, the tangential stress time history curve and the relative slip velocity time history curve of each node are extracted throughout the entire load cycle.

[0039] The tangential stress time history curve of each node is multiplied point by point with the relative slip velocity time history curve to obtain the instantaneous friction power time history curve of the corresponding node.

[0040] Numerical integration is performed on the instantaneous friction power time history curve of each node to calculate the energy dissipation value of a single node in one load cycle.

[0041] The energy dissipation values ​​of all nodes within the microslip region are weighted and accumulated, with the weights allocated according to the contact area represented by each node, to obtain the total energy dissipation index of the corresponding microslip region.

[0042] As a further aspect of the present invention: the establishment of a quantitative mapping relationship with fatigue life specifically includes:

[0043] Experiments were conducted on standard fretting fatigue specimens of various materials to obtain data sets of failure cycle counts corresponding to different energy dissipation index levels. Nonlinear regression analysis was used to fit the experimental data sets, establishing a damage accumulation model with energy dissipation index as the independent variable and failure cycle count as the dependent variable. Based on the damage equivalence principle, the complex cycles in the actual load spectrum were simplified into equivalent constant amplitude cycles, and the critical energy dissipation index threshold was determined. A complete quantitative mapping relationship curve between energy dissipation and lifetime was constructed using interpolation methods, serving as a criterion for lifetime prediction.

[0044] As a further aspect of the present invention: the iterative optimization of the design parameters of the parametric digital geometric model specifically includes:

[0045] The thickness of the connecting plate, the bolt preload, and the radius of curvature of the contact surface are determined as key design variables, and their feasible value ranges are set. A pattern search method is used to generate several candidate parameter combinations in the neighborhood of the current design point. By re-executing the mechanical solution and energy dissipation calculation process, the energy dissipation index values ​​corresponding to each candidate combination are directly obtained. Based on the critical energy dissipation threshold corresponding to the target life, the candidate schemes that meet the requirements are selected, and the one with the best index is used as the new benchmark design point for the next round of iteration. When no better solution can be found in several consecutive iterations, the calculation is terminated, and the current optimal parameter combination and the corresponding connection structure scheme are output.

[0046] The beneficial effects of this invention are:

[0047] (1) By constructing precise parametric digital geometric models, simulating complex mechanical behavior, and predicting fatigue life based on energy dissipation indices, this invention can effectively identify and optimize the design parameters of key areas. This not only improves the anti-microslip capability of the connection structure under dynamic loads but also ensures its stability and safety throughout its service life. Therefore, connection structures designed and optimized using the method of this invention can significantly improve their durability and long-term reliability.

[0048] (2) Traditional design methods rely on numerous physical experiments to verify and optimize design schemes, which is both time-consuming and expensive. This invention utilizes advanced numerical simulation technology to complete the entire process from modeling to optimization in a virtual environment, significantly reducing the need for physical prototypes. Furthermore, the application of optimization strategies such as pattern search allows design parameters to be explored efficiently over a wider range, quickly finding the optimal design scheme that meets the target lifespan requirements. This method not only accelerates the product development cycle but also reduces R&D costs, saving valuable resources for enterprises. Attached Figure Description

[0049] The invention will now be further described with reference to the accompanying drawings.

[0050] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Please see Figure 1 As shown, this invention provides a method for constructing a structural error tolerance model for compartment connection and positioning, comprising the following steps:

[0053] S1: Construct a parametric digital geometric model that includes bolts, connecting plates and the base, and assign corresponding material properties to each component;

[0054] S2: Generate a benchmark digital speckle image with random speckle characteristics based on a parametric digital geometric model, and simultaneously generate a dynamic load sequence simulating vehicle driving conditions.

[0055] S3: Apply the dynamic load sequence step by step to the parametric digital geometric model for mechanical solution. After each load step, calculate the deformed digital speckle image on the surface of the parametric digital geometric model. Compare the deformed digital speckle image with the reference digital speckle image in a non-contact full-field comparison and extract the full-field displacement vector data of the connecting pair under each load step.

[0056] S4: Identify the micro-slip regions on the contact interface of the connecting pair based on the full-field displacement vector data, and calculate the energy dissipation index corresponding to the micro-slip regions in each load cycle;

[0057] S5: Establish a quantitative mapping relationship between energy dissipation index and fatigue life. With the preset target life as a constraint, iteratively optimize the design parameters of the parameterized digital geometric model to ensure that the energy dissipation index of the key area meets the life requirements, and output the final optimized connection structure scheme.

[0058] In S1, a parametric digital geometric model including bolts, connecting plates, and the base is constructed, and corresponding material properties are assigned to each component, specifically including:

[0059] Accurate point cloud data of bolted connections was acquired using 3D laser scanning technology. A 3D laser scanner was used to perform omnidirectional scanning of the actual bolts, connecting plates, and base components, maintaining appropriate scanning distance and angular overlap to ensure complete and high-precision surface point cloud data. The acquired raw point cloud data underwent preprocessing, including outlier removal, noise filtering, and point cloud simplification. Outlier removal employed a statistical filtering algorithm, calculating the average distance between each point and its neighbors, removing points whose distance exceeded three times the standard deviation. Noise filtering used a Gaussian filtering algorithm, smoothing random noise by applying a weighted average to the point cloud. Point cloud simplification employed a curvature-based method, reducing point density in flat areas and maintaining higher density in feature-rich areas, improving processing efficiency while maintaining accuracy. The processed point cloud data preserved the actual manufacturing tolerances and microscopic morphological features of the part surface, providing accurate input data for subsequent modeling.

[0060] Feature recognition and parametric reconstruction of preprocessed point cloud data are performed using a deep neural network. A deep neural network with an encoder-decoder structure is constructed. The encoder consists of five convolutional layers to extract global and local features of the point cloud. The decoder contains three fully connected layers that map the extracted features to control parameters of the parametric model. The network input consists of the 3D coordinates and normal vector information of each point in the point cloud, and the output consists of the structural topology and dimensional constraints of the parametric digital geometric model. The training process uses a point cloud dataset containing various bolt connection types, with each sample labeled with key dimensions and topological relationships. By learning the mapping relationship between point cloud features and the parametric model, the network can automatically identify bolt thread features, hole locations in connection plates, and geometric features of contact surfaces, and transform these features into adjustable design variables.

[0061] Adaptive mesh generation of the parametric digital geometric model is performed using multi-scale mesh generation technology. Different mesh density strategies are adopted based on the contact analysis requirements of different regions of the model. In the contact areas between bolts and connecting plates, and between connecting plates and the substrate, a surface adaptive refinement algorithm is used to generate high-density meshes, with mesh sizes controlled within 0.1 mm to accurately capture the contact stress distribution. In non-critical areas far from the contact region, a sparser mesh is used, with mesh sizes increased to 1 mm to improve computational efficiency. During mesh generation, the geometric model is first subjected to surface triangulation to ensure that the mesh quality meets the analysis requirements; the side length ratio of triangular elements is controlled within 1.5, and the maximum interior angle does not exceed 120 degrees. Then, according to the needs of contact analysis, multi-layer mesh refinement is performed in the refinement areas to ensure sufficient mesh nodes at the contact interface to accurately describe the contact behavior.

[0062] The final parametric digital geometric model contains complete geometric features, material properties, and mesh information, accurately reflecting the geometric characteristics of the actual bolted connection. This model not only preserves the manufacturing tolerances and surface morphology of the actual parts but also allows for flexible adjustment of key dimensions through parametric design variables, providing a reliable digital foundation for subsequent contact mechanics analysis. The model's mesh generation scheme ensures both the accuracy of the contact area analysis and overall computational efficiency, making it suitable for numerical simulation analysis under various working conditions. The digital model established using this method provides effective technical support for the design optimization and performance prediction of bolted connections.

[0063] In S2, a baseline digital speckle image with random speckle characteristics is generated based on a parametric digital geometric model, and a dynamic load sequence simulating vehicle driving conditions is also generated, specifically including:

[0064] Based on the geometric and material properties of a parametric digital geometric model, mass distribution characteristics and structural stiffness characteristics are automatically extracted, and the overall mass and stiffness matrices are calculated. Geometric properties include the three-dimensional dimensions, volume, and surface area of ​​each component, directly obtained through the parametric model. Material properties include parameters such as elastic modulus, Poisson's ratio, and density, which are set according to the actual material type. The calculation of mass distribution characteristics is based on the volume and material density of each component, obtaining the mass contribution of each mesh node through numerical integration, and finally assembling it into the overall mass matrix. The calculation of structural stiffness characteristics uses the finite element method. Based on the mesh generation results and material elastic parameters, the stiffness matrix of each element is calculated, and then the overall stiffness matrix is ​​formed through an element assembly process. Throughout this process, the dimensions of the mass and stiffness matrices are consistent with the number of degrees of freedom of the model, ensuring the accuracy of the dynamic analysis.

[0065] Based on the kinematic parameters of a vehicle under typical operating conditions, and combining the mass and stiffness matrices, the basic inertial load components at the connection interface are calculated using the Newton-Euler equations. Typical vehicle operating conditions include basic motion states such as uniform speed driving, acceleration, braking, and turning. Each operating condition corresponds to different kinematic parameters, such as acceleration, angular velocity, and radius of curvature. The Newton-Euler equations describe the relationship between rigid body motion and forces. By treating the entire vehicle as a multi-rigid-body system, the system's motion equations are established. During the calculation, firstly, the acceleration and angular acceleration of the center of mass of each component are calculated based on the kinematic parameters. Then, combining the mass matrix and rotational inertia parameters, the inertial forces and moments generated by each component are calculated. Finally, through the force transmission relationships, the components of these inertial loads at the bolted connection interface are calculated, forming a time series of the basic inertial loads.

[0066] Using road spectrum excitation data as input, load transfer path analysis is performed through the stiffness matrix to calculate the dynamic additional load components caused by pavement excitation. The road spectrum excitation data originates from actual road measurements and includes vibration excitation information caused by pavement unevenness, typically expressed as power spectral density. Load transfer path analysis is based on structural dynamics theory, describing the relationship between excitation and response points through a frequency response function. In the calculation process, the road spectrum excitation data is first converted to the time domain to generate a pavement excitation time series. Then, the frequency response characteristics of the system are calculated using the global stiffness matrix and mass matrix. Finally, the dynamic response generated by pavement excitation at the interface is calculated through convolution operations to obtain the dynamic additional load components.

[0067] The basic inertial load components and the dynamic additional load components are superimposed and synthesized in the time domain to generate a long-term dynamic load sequence that matches the characteristics of the parametric digital geometric model. The superposition and synthesis process is performed in the time domain to ensure the synchronization of the two load components in time. The basic inertial load components reflect the quasi-static load generated by changes in vehicle motion, while the dynamic additional load components reflect the dynamic load generated by road surface excitation. During the synthesis process, the time steps of the two load components are first aligned to ensure consistent sampling frequencies. Then, the two load components at the same moment are synthesized by vector addition to obtain the total load value at that moment. By traversing the entire time series, a long-term dynamic load sequence is generated. The generated load sequence contains information on the magnitude, direction, and point of application of the load, perfectly matching the characteristics of the parametric digital geometric model, providing accurate boundary conditions for subsequent contact analysis.

[0068] In S3, a dynamic load sequence is applied stepwise to the parametric digital geometric model for mechanical solution. After each load step, a deformed digital speckle image of the parametric digital geometric model surface is calculated. The deformed digital speckle image is then compared with a reference digital speckle image in a non-contact full-field manner to extract the full-field displacement vector data of the connecting joint under each load step. Specifically, this includes:

[0069] In the mechanical solution process of each incremental load step, the complete Newton-Rafaelsen iterative method is used to solve the equilibrium equations considering geometric and contact nonlinearities. The dynamic load sequence is discretized into multiple incremental steps, with each incremental step representing 1% to 5% of the total load. In each iteration, the tangent stiffness matrix and residual force vector of the structure are first calculated based on the current displacement state. The tangent stiffness matrix considers both geometric stiffness effects and contact state changes. Geometric nonlinearity is represented by including stress stiffening effects, and contact nonlinearity is handled by a contact algorithm based on the penalty function method. During the iteration process, the displacement increment is updated by solving a system of linear equations until the norm of the residual force is less than a set tolerance or the maximum number of iterations is reached. This iterative method can effectively handle strongly nonlinear problems, ensuring that the solution process converges within a reasonable number of iterations and obtains accurate mechanical response results.

[0070] After convergence at each load step, the spatial coordinate changes of all surface nodes in the parameterized digital geometric model are extracted, reconstructing the discrete node displacements into a continuous deformation field on the model surface. The node displacement data is directly output from the finite element solver, containing displacement components for each node in three coordinate directions. To obtain the continuous deformation field, a shape function-based interpolation method is used to convert the discrete node displacements into displacement values ​​at any point on the model surface. For triangular or quadrilateral surface elements, corresponding linear or bilinear shape functions are used for interpolation calculations. Based on the reconstructed continuous displacement field and combined with the initial baseline digital speckle image, a corresponding deformed digital speckle image is generated using an image deformation algorithm. The image deformation process considers the continuity and smoothness of the displacement field to ensure that the generated deformed speckle image accurately reflects the deformation state of the model surface, providing accurate input data for subsequent digital image correlation analysis.

[0071] Initial value estimation of speckle images before and after deformation is performed using the phase correlation method, followed by fine matching using a zero-mean normalized cross-correlation function. The phase correlation method first performs a Fast Fourier Transform on both the reference and deformed images, obtaining initial displacement values ​​at the integer pixel level by calculating the cross-power spectrum and inverse transform. This step enables rapid estimation of the overall displacement over a large area, providing good initial conditions for subsequent fine matching. After obtaining the initial values, sub-region matching is performed using the zero-mean normalized cross-correlation function. This algorithm exhibits good robustness to illumination changes and noise. During the calculation, a sub-region of a certain size is selected centered on the point to be measured, and the position in the deformed image that maximizes the cross-correlation function is searched, thereby obtaining sub-pixel level displacement estimates.

[0072] By integrating the displacement increments across all load steps, a time-series full-field displacement vector dataset of the connection joint throughout the entire dynamic load history is constructed. The displacement field calculated for each load step is the incremental displacement relative to the initial state. To obtain the cumulative displacement, the displacement increments from each load step are vector-superimposed. The superposition process is based on the consistency of node numbering in the finite element model, ensuring that the displacement history of each node is correctly recorded. The final dataset contains the displacement time series of each surface node under all load steps, with the data format being a complete record of the three-dimensional vector field changing over time. This dataset fully describes the deformation history of the connection joint under dynamic loading, providing fundamental data for subsequent mechanical performance evaluation and fatigue analysis.

[0073] A regular grid of dots is defined on the reference digital speckle image, serving as the center of the computational sub-region to be tracked. The grid spacing is set to five to fifteen pixels, depending on the measurement accuracy requirements and computational resources. Each grid dot corresponds to the center position of a computational sub-region, the size of which is determined by the speckle feature size. The grid arrangement covers the entire region of interest, ensuring complete deformation field information is captured. When arranging the grid dots, areas with indistinct features should be avoided, ensuring that each sub-region contains sufficient speckle features for reliable matching.

[0074] For the deformed digital speckle image under each load step, an inverse combination Gauss-Newton algorithm based on gray-level gradient optimization is used for iterative calculation. This algorithm solves for the displacement field by minimizing the gray-level difference between the reference image sub-region and the deformed image sub-region. In each iteration, the gray-level gradient matrix of the reference image sub-region is first calculated, then the Hessian matrix is ​​constructed and the displacement correction is solved. The inverse combination method is characterized by fixing the gray-level gradient calculation on the reference image, avoiding recalculating the gradient of the deformed image in each iteration, thus improving computational efficiency. The iterative process continues until the norm of the displacement correction is less than a set threshold (usually 0.01 pixels) or the maximum number of iterations is reached.

[0075] A displacement field smoothing constraint based on a quadratic B-spline function is introduced to regularize the solved initial displacement field. Due to image noise and speckle quality, the directly calculated displacement field may contain non-physical fluctuations. To suppress this noise, a parameterized representation of the displacement field is established using a quadratic B-spline function. The control point spacing of the B-spline function is set according to the smoothness requirements of the displacement field, typically five to ten calculation points apart. Discrete displacement measurements are fitted to the B-spline function using the least squares method, thus achieving smoothing of the displacement field.

[0076] The continuous strain tensor distribution across the entire field is obtained by calculating the spatial gradient of the displacement field in adjacent computational sub-regions. The spatial gradient of the displacement field is calculated using the central difference method. For each internal point on a regular grid, the strain components are calculated using the displacement values ​​of its neighboring points. The strain tensor under plane strain conditions includes three independent components: normal strain in the x-direction, normal strain in the y-direction, and shear strain. During strain calculation, the continuity and differential consistency of the displacement field are considered, and the least squares method is used to improve the accuracy of strain calculation. The final strain field is represented in tensor form, fully describing the strain distribution on the joint surface. Combined with the displacement field data, a complete mechanical response dataset containing displacement and strain information is output, providing a comprehensive basis for the mechanical performance analysis of the joint.

[0077] In S4, micro-slip regions on the contact interface of the connecting pair are identified based on full-field displacement vector data, and the energy dissipation index corresponding to the micro-slip regions in each load cycle is calculated, specifically including:

[0078] Tangential displacement time history data of the contact interface nodes within a complete load cycle is extracted, and the displacement amplitude and phase angle distribution of each node are calculated. The tangential displacement time history data is obtained from the full-field displacement vector dataset, containing the tangential displacement value of each node at each time step within the load cycle. The displacement time history data is analyzed using Fast Fourier Transform (FFT) to extract the displacement amplitude and phase angle corresponding to the fundamental frequency component. The displacement amplitude represents the range of motion of the node in the tangential plane, and the phase angle reflects the time delay of the node's motion relative to the reference signal. During the calculation, the displacement time history of each node is detrended to eliminate the influence of rigid body displacement, ensuring that the analyzed displacement data accurately reflects the relative sliding motion. This step obtains the motion characteristic parameters of all nodes on the contact interface, providing a data foundation for subsequent region identification.

[0079] The displacement characteristics of nodes are automatically classified, grouping nodes with similar amplitude and phase characteristics into the same slip pattern region. A density-based clustering algorithm is used to analyze the amplitude and phase characteristics of nodes. The algorithm first constructs a feature space, where each node is represented as a two-dimensional data point using its displacement amplitude and phase angle. By calculating the Euclidean distance between data points, regions with higher density are identified as cluster centers. During the clustering process, the neighborhood radius parameter is set to 0.1, and the minimum sample size parameter is set to five to ensure that only regions with significant statistical characteristics are identified as valid clusters. Each cluster region represents a specific slip pattern with similar displacement amplitude and phase characteristics. This automatic classification process can effectively distinguish different slip behavior regions on the contact interface.

[0080] The displacement direction consistency coefficient is calculated for each cluster region. Regions with a consistency coefficient below a first set threshold are initially identified as micro-slip regions. The displacement direction consistency coefficient is evaluated by calculating the average angle between the displacement direction vectors of the nodes within the region. First, the displacement direction of each node is represented as a unit vector, and then the average value of all node direction vectors is calculated. The displacement direction consistency coefficient is defined as the magnitude of the average vector, ranging from 0 to 1. The closer the coefficient is to 1, the more consistent the movement directions of the nodes within the region; the closer the coefficient is to zero, the more dispersed the movement directions. A first set threshold of 0.6 is set. When the displacement direction consistency coefficient of a region is below this threshold, it indicates that the movement directions of the nodes within that region are inconsistent, and complex relative slip behavior exists. Therefore, it is initially identified as a micro-slip region.

[0081] Morphological closing operations are performed on the initially identified micro-slip regions to eliminate discrete noise points and fill internal voids. The morphological closing operation consists of a sequential combination of dilation and erosion operations. First, a structuring element is used to dilate the region; this element is a circular template with a radius of three nodes. Dilation connects adjacent micro-regions and fills small voids. Then, an erosion operation is performed, using the same structuring element to restore the original boundary of the region while maintaining its connectivity. This process effectively eliminates isolated points caused by measurement noise while preserving the integrity of the main micro-slip regions. After morphological processing, a continuous and complete boundary of the micro-slip region is obtained, providing an accurate spatial definition for subsequent energy dissipation calculations.

[0082] Within the identified micro-slip region, the tangential stress time history curves and relative slip velocity time history curves of each node were extracted throughout the entire load cycle. The tangential stress time history data were obtained from the finite element analysis results, containing the tangential stress value of each node at each time step within the load cycle. The relative slip velocity time history data were obtained by numerically differentiating the tangential displacement time history, and the instantaneous velocity at each time step was calculated using the central difference method. During the numerical differentiation process, the time step size was determined based on the load cycle period and sampling frequency to ensure the accuracy of the velocity calculation. The data extraction scope was limited to the identified micro-slip region, with a focus on areas exhibiting significant relative slip.

[0083] The tangential stress time history curve and the relative slip velocity time history curve at each node are multiplied point by point to obtain the instantaneous friction power time history curve for the corresponding node. The multiplication is performed in the time domain; the stress and velocity values ​​at the same time point are multiplied to obtain the instantaneous friction power value at that time point. Instantaneous friction power represents the energy dissipation rate per unit contact area, reflecting the instantaneous energy conversion efficiency during the micro-slip process. During the calculation, it is ensured that the stress and velocity data have the same time step and are time-aligned to avoid calculation errors caused by time asynchrony. This operation converts mechanical parameters into energy parameters, preparing for subsequent energy integration calculations.

[0084] Numerical integration is performed on the instantaneous frictional power time history curve of each node to calculate the energy dissipation of a single node within one load cycle. The numerical integration employs the composite Simpson's law, dividing the load cycle period into an even number of equally spaced sub-intervals. An approximate integral value is calculated using Simpson's formula for each sub-interval, and the results from all sub-intervals are then summed to obtain the total integral value. The integration step size is determined based on the frequency characteristics of the load cycle, typically set to 1% of the cycle length. The numerical integration process considers the nonlinear characteristics of the power time history curve, enabling accurate calculation of the total energy dissipation within each load cycle. The calculated energy dissipation value represents the total energy consumed by the node through friction in a complete load cycle.

[0085] The energy dissipation values ​​of all nodes within the micro-slip region are weighted and accumulated, with weights allocated based on the contact area represented by each node. The weight of each node is determined by the contact area it controls, calculated through the area allocation of the element to which the node belongs. For triangular elements, the node weight is one-third of the element area; for quadrilateral elements, a bilinear shape function is used for area allocation. The weighted accumulation process first multiplies the energy dissipation value of each node by its corresponding weight, then sums the weighted energy dissipation values ​​of all nodes to obtain the total energy dissipation index of the micro-slip region. This index quantifies the total frictional energy loss in this region over a single load cycle, providing crucial information for fatigue life assessment and thermal effect analysis of the connection pair.

[0086] In S5, a quantitative mapping relationship between energy dissipation indices and fatigue life is established. With a preset target life as a constraint, the design parameters of the parameterized digital geometric model are iteratively optimized to ensure that the energy dissipation indices of key areas meet life requirements. The final optimized connection structure scheme is then output, specifically including:

[0087] Experiments were conducted on standard fretting fatigue specimens made of various materials to obtain a set of failure cycle counts corresponding to different energy dissipation index levels. Standard fretting fatigue specimens were used, with materials including structural steel and aluminum alloys commonly used in joint pairs. During the experiments, a constant normal load and cyclic tangential displacement were applied on a dedicated fretting fatigue testing machine, while the energy dissipation index at the specimen contact interface was measured. By changing the displacement amplitude and normal load level, a series of fatigue experimental data under different energy dissipation index levels were obtained. Each experimental condition was repeated 3 to 5 times to ensure the statistical reliability of the data. The experiments continued until obvious cracks appeared in the specimen or the stiffness decreased significantly; the number of cycles at this point was recorded as the failure cycle count. Finally, a complete set of experimental data containing energy dissipation indexes and corresponding failure cycle counts was obtained, providing an experimental basis for establishing a life prediction model.

[0088] A damage accumulation model was established by fitting the experimental data using nonlinear regression analysis, with energy dissipation index as the independent variable and the number of failure cycles as the dependent variable. The nonlinear regression analysis employed the least squares method, and the model parameters were determined through iterative optimization. The damage accumulation model adopted a power function form, where the number of failure cycles is proportional to the negative power of the energy dissipation index. During the regression analysis, the experimental data were first logarithmically transformed into a linear relationship for preliminary fitting, and then the Gauss-Newton method was used for precise parameter estimation. The goodness of fit of the model was evaluated using the coefficient of determination, requiring a value of 0.9 or higher to ensure the predictive accuracy of the model. The established damage accumulation model can quantitatively describe the relationship between energy dissipation index and fatigue life, providing a mathematical model basis for life prediction.

[0089] Based on the damage equivalence principle, the complex cycles in the actual load spectrum are simplified into equivalent constant-amplitude cycles, and the critical energy dissipation index threshold is determined. The damage equivalence principle employs Miner's linear cumulative damage theory, equating fatigue damage under variable-amplitude loads to damage under constant-amplitude loads. In practice, rainflow counting analysis is first performed on the actual load spectrum to identify all complete load cycles and their amplitude distributions. Then, based on the damage accumulation model, the damage degree generated by each load cycle is calculated, and all damage degrees are linearly superimposed to obtain the total damage. Finally, based on the target life requirement, the allowable equivalent constant-amplitude energy dissipation index value is calculated as the critical threshold. This threshold represents the maximum allowable energy dissipation level that the connecting pair can safely withstand within the target life.

[0090] A complete quantitative mapping curve of energy dissipation and lifetime is constructed using interpolation methods, serving as a criterion for lifetime prediction. Within the scope of the experimental data, a continuous mapping curve is constructed using cubic spline interpolation. For regions outside the experimental data range, logarithmic linear extrapolation is used for appropriate extension. During interpolation, the monotonically decreasing characteristic of the curve is ensured, conforming to the fundamental principle that greater energy dissipation corresponds to a shorter lifetime. The final mapping curve covers a wide range from long lifetime regions with low energy dissipation to short lifetime regions with high energy dissipation, providing comprehensive criteria for the design of connectors with different lifetime requirements. This curve is stored in graphical form for easy retrieval and application during the optimization process.

[0091] The thickness of the connecting plate, bolt preload, and radius of curvature of the contact surface were identified as key design variables, and their engineering-feasible value ranges were defined. The connecting plate thickness was set to vary from 5 mm to 20 mm based on structural space constraints and process requirements. The bolt preload was set to vary from 20,000 N to 100,000 N based on bolt specifications and material strength. The radius of curvature of the contact surface was set to vary from 50 mm to 200 mm based on fit requirements and machining capabilities. The value range of each design variable was verified for engineering feasibility to ensure the optimization results are highly achievable. These design variables have a significant impact on the stiffness characteristics and contact state of the connection pair, and are key factors affecting energy dissipation indicators.

[0092] A pattern search method is used to generate several candidate parameter combinations in the neighborhood of the current design point. By re-executing the mechanical solution and energy dissipation calculation process, the energy dissipation index values ​​corresponding to each candidate combination are directly obtained. The pattern search method generates test points around the current design point according to a specific pattern, which is usually represented by coordinate direction or simplex vertex direction. In each iteration step, 5 to 10 candidate parameter combinations are generated, each representing a specific set of design variable values. For each candidate combination, the complete finite element analysis process is re-executed, including load application, mechanical solution, microslip identification, and energy dissipation calculation, to directly obtain the energy dissipation index of the key area corresponding to the design. Although this direct calculation method has a higher computational cost, it can accurately reflect the impact of changes in design variables on energy dissipation.

[0093] Based on the critical energy dissipation threshold corresponding to the target lifetime, candidate solutions that meet the requirements are selected, and the one with the best performance is used as the new baseline design point for the next iteration. The selection process first eliminates candidate solutions whose energy dissipation performance exceeds the critical threshold to ensure that the lifetime requirement is met. Among the qualified solutions, the one with the lowest energy dissipation performance is selected as the current optimal solution. If the energy dissipation performance of all candidate solutions exceeds the threshold, the solution with the lowest performance is selected as the new baseline design point. This new baseline design point will serve as the center for the next round of pattern search, continuing to generate new candidate parameter combinations within its neighborhood for further exploration. In this way, the optimization process gradually progresses towards a design that meets the lifetime requirement and minimizes energy dissipation.

[0094] The calculation terminates when no better solution is found after multiple iterations, outputting the current optimal parameter combination and the corresponding connection structure scheme. The termination condition is set as follows: the improvement of the optimal objective function value is less than 1% after three consecutive iterations, indicating that the optimization process has converged to a local optimum. Before terminating the calculation, the current optimal design is finally verified, including a complete mechanical performance evaluation and manufacturing process check. The final optimization results include specific design parameter values, corresponding predicted energy dissipation indices, and a reliability assessment report that meets the target life. Detailed three-dimensional geometric models and process parameter requirements are also provided to ensure that the optimization results can be directly used for engineering implementation.

[0095] The working principle of this invention is as follows: A parametric digital geometric model including bolts, connecting plates, and a substrate is constructed, and material properties are assigned to each component. Based on this model, a reference digital speckle image with random speckle characteristics and a dynamic load sequence simulating vehicle driving conditions are generated. The dynamic load is gradually applied to the model for mechanical solution, the deformed digital speckle image is calculated and compared with the reference image, and the full-field displacement vector data is extracted. Based on the displacement vector data, micro-slip regions are identified and energy dissipation indexes are calculated. According to the relationship between energy dissipation indexes and fatigue life, the design parameters are iteratively optimized to meet the preset target life requirements, and the optimized connection structure scheme is output.

[0096] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for constructing a structural error tolerance model for compartment connection and positioning, characterized in that, Includes the following steps: S1: Construct a parametric digital geometric model that includes bolts, connecting plates and the base, and assign corresponding material properties to each component; S2: Generate a benchmark digital speckle image with random speckle characteristics based on a parametric digital geometric model, and simultaneously generate a dynamic load sequence simulating vehicle driving conditions. S3: Apply the dynamic load sequence step by step to the parametric digital geometric model for mechanical solution. After each load step, calculate the deformed digital speckle image on the surface of the parametric digital geometric model. Compare the deformed digital speckle image with the reference digital speckle image in a non-contact full-field comparison and extract the full-field displacement vector data of the connecting pair under each load step. Specifically, it includes: Define a regular grid of dots on the reference digital speckle image as the center of the computational sub-region to be tracked; For the deformed digital speckle image under each load step, the inverse combination Gauss-Newton algorithm based on gray-level gradient optimization is used to iteratively calculate each calculation sub-region and solve the displacement field that minimizes the gray-level difference between the sub-regions in the two images. A displacement field smoothing constraint based on a quadratic B-spline function is introduced to regularize the solved initial displacement field in order to suppress non-physical fluctuations caused by noise. By calculating the spatial gradient of the displacement field in adjacent computational sub-regions, the continuous strain tensor distribution across the entire field is obtained, and high-precision full-field displacement vector data is output by combining the displacement field. S4: Identify micro-slip regions on the contact interface of the connecting pair based on full-field displacement vector data, and calculate the energy dissipation index corresponding to the micro-slip regions in each load cycle, specifically including: Within the identified microslip region, the tangential stress time history curve and the relative slip velocity time history curve of each node are extracted throughout the entire load cycle. The tangential stress time history curve of each node is multiplied point by point with the relative slip velocity time history curve to obtain the instantaneous friction power time history curve of the corresponding node. Numerical integration is performed on the instantaneous friction power time history curve of each node to calculate the energy dissipation value of a single node in one load cycle. The energy dissipation values ​​of all nodes within the microslip region are weighted and summed, with the weights allocated according to the contact area represented by each node, to obtain the total energy dissipation index for the corresponding microslip region. S5: Establish a quantitative mapping relationship between energy dissipation index and fatigue life. With the preset target life as a constraint, iteratively optimize the design parameters of the parameterized digital geometric model to ensure that the energy dissipation index of the key area meets the life requirements, and output the final optimized connection structure scheme.

2. The method for constructing a structural error tolerance model for compartment connection and positioning according to claim 1, characterized in that, The process of constructing the parametric digital geometric model is as follows: The actual point cloud data of the connector is obtained by 3D laser scanning. The point cloud data is then denoised and simplified to establish an initial parametric digital geometric model that reflects the actual manufacturing tolerances and surface morphology. Based on deep neural networks, feature recognition and parametric reconstruction of the initial parametric digital geometric model are performed. The key dimensions in the initial parametric digital geometric model are defined as adjustable design variables, where the input is the feature vector of point cloud data and the output is the structural topology and dimensional constraint relationship of the parametric digital geometric model. The parametric digital geometric model is adaptively meshed using multi-scale mesh generation technology. A dense surface mesh is generated in the contact area, while a sparse mesh is used in non-critical areas, thus establishing a parametric digital geometric model suitable for high-precision contact analysis.

3. The method for constructing a structural error tolerance model for compartment connection and positioning according to claim 1, characterized in that, The generation of the dynamic load sequence simulating vehicle driving conditions specifically includes: Based on the geometric and material properties of the parametric digital geometric model, the mass distribution characteristics and structural stiffness characteristics are automatically extracted, and the overall mass matrix and stiffness matrix are calculated. Based on the kinematic parameters of the vehicle under typical operating conditions, and combined with the mass matrix and stiffness matrix, the basic inertial load components at the connection interface are calculated using the Newton-Euler equations. Using road spectrum excitation data as input, load transfer path analysis is performed through stiffness matrix to calculate the dynamic additional load components caused by road excitation. The basic inertial load components and the dynamic additional load components are superimposed and synthesized in the time domain to generate a long-term dynamic load sequence that matches the characteristics of the parametric digital geometric model.

4. The method for constructing a structural error tolerance model for compartment connection and positioning according to claim 1, characterized in that, The stepwise application of dynamic load sequences to a parametric digital geometric model for mechanical solution specifically includes: In the mechanical solution process of each incremental load step, the complete Newton-Rafaelsen iteration method is used to solve the equilibrium equations considering geometric nonlinearity and contact nonlinearity to ensure the convergence and accuracy of the solution results. After the solution converges in each load step, the spatial coordinate changes of all surface nodes of the parameterized digital geometric model are extracted, the discrete node displacements are reconstructed into a continuous model surface deformation field, and a deformed digital speckle image is generated accordingly. The phase correlation method is used to estimate the initial value of the speckle images before and after deformation, and then the zero-mean normalized cross-correlation function is combined to perform fine matching to calculate the two-dimensional displacement component of each pixel. By integrating the displacement increments of all load steps, a time-series full-field displacement vector dataset of the connection pair throughout the entire dynamic load history is constructed.

5. The method for constructing a structural error tolerance model for compartment connection and positioning according to claim 1, characterized in that, The identification of micro-slippage areas on the contact interface of the connector specifically includes: Extract the tangential displacement time history data of the contact interface nodes within a complete load cycle, and calculate the displacement amplitude and phase angle distribution of each node; The displacement characteristics of nodes are automatically classified, and nodes with similar amplitude and phase characteristics are grouped into the same slip mode region; Calculate the displacement direction consistency coefficient within each cluster region, and preliminarily determine the regions with a direction consistency coefficient lower than a first set threshold as micro-slip regions; Morphological closing operations are performed on the initially identified microslip regions to eliminate discrete noise points and fill the voids inside the regions, generating continuous and complete microslip region boundaries.

6. The method for constructing a structural error tolerance model for compartment connection and positioning according to claim 1, characterized in that, The establishment of a quantitative mapping relationship between fatigue life and fatigue life specifically includes: Experiments were conducted on standard fretting fatigue specimens of various materials to obtain data sets of failure cycle counts corresponding to different energy dissipation index levels. Nonlinear regression analysis was used to fit the experimental data sets, establishing a damage accumulation model with energy dissipation index as the independent variable and failure cycle count as the dependent variable. Based on the damage equivalence principle, the complex cycles in the actual load spectrum were simplified into equivalent constant amplitude cycles, and the critical energy dissipation index threshold was determined. A complete quantitative mapping relationship curve between energy dissipation and lifetime was constructed using interpolation methods, serving as a criterion for lifetime prediction.

7. The method for constructing a structural error tolerance model for compartment connection and positioning according to claim 1, characterized in that, The iterative optimization of the design parameters of the parametric digital geometric model specifically includes: The thickness of the connecting plate, the bolt preload, and the radius of curvature of the contact surface are determined as key design variables, and their feasible value ranges are set. A pattern search method is used to generate several candidate parameter combinations in the neighborhood of the current design point. By re-executing the mechanical solution and energy dissipation calculation process, the energy dissipation index values ​​corresponding to each candidate combination are directly obtained. Based on the critical energy dissipation threshold corresponding to the target life, the candidate schemes that meet the requirements are selected, and the one with the best index is used as the new benchmark design point for the next round of iteration. When no better solution can be found in several consecutive iterations, the calculation is terminated, and the current optimal parameter combination and the corresponding connection structure scheme are output.

Images