Fine modeling method and system based on digital design of steel truss beam

Abstract

This invention discloses a refined modeling method and system based on the digital design of steel trusses, belonging to the field of digital design and data processing of civil engineering structures. It acquires the design parameter set of steel truss components and measured point cloud data at nodes, calculates a multi-scale geometric feature matrix through a morphological correction model, constructs a node spatial coordinate system based on this matrix, and forms a three-dimensional constraint set. The three-dimensional constraint set is input into a multi-source data fusion model to generate an associated constraint probability matrix. Constraint folding operations are performed on the multi-scale geometric feature matrix based on this matrix to obtain a stress sensitivity prediction curve. The geometric deviation influence factors of each component position are calculated, and adaptive feature replacement of the digital component morphology is performed based on the influence factors, thereby generating an optimized refined digital model of the steel truss. This invention can improve the accuracy of the digital model of steel truss components, realize the quantitative expression of actual structural deviations, and accurately reflect the influence of stress paths.

Description

Technical Field

[0001] This invention relates to the field of digital design and data processing of civil engineering structures, specifically to a refined modeling method and system based on the digital design of steel trusses. Background Technology

[0002] Existing steel truss structures typically rely on 2D drawings, parametric component libraries, and finite element analysis software for modeling during the engineering design phase. However, this process has significant shortcomings: First, a large number of implicit component constraints in 2D design information cannot be automatically identified, such as complex node geometry between members, setback relationships, plate openings, and structural accessories, leading to discrepancies between the digital model and the actual structure. Second, existing parametric modeling often uses regular nodes as templates, making it difficult to accurately represent minute dimensional deviations caused by transportation, welding, and temperature deformation on-site, thus affecting stress path analysis, fatigue assessment, and prediction of overall truss stability. Third, complex spatial nodes are prone to constraint conflicts under multi-source data fusion, such as the lack of a unified feature mapping mechanism between scanned point clouds, BIM models, and design parameters, making it impossible to construct a realistic and reliable high-precision digital steel truss model.

[0003] Especially in structures such as long-span steel truss bridges and heavy-duty stadium steel trusses, the joint structures often exhibit complex forms such as curved surfaces, three-dimensional weld groups, and irregular stiffening ribs. Traditional modeling methods cannot depict the actual stress-sensitive parts, which can easily lead to local fatigue damage prediction deviations exceeding engineering tolerances. In addition, due to the large number of steel truss components and the complex connection relationships between members, existing digital processes are unable to automatically identify and quantify small structural units (such as 1–3 mm weld toe transition fillets, local indentations, and discretized geometric features of nodes), which seriously restricts the precision and traceability of digital design for steel trusses. Summary of the Invention

[0004] The purpose of this invention is to provide a refined modeling method and system based on the digital design of steel trusses, so as to overcome the shortcomings of the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a refined modeling method based on digital design of steel trusses, comprising:

[0006] The design parameter set of steel truss components and the measured point cloud data at the nodes are obtained, and a multi-scale geometric feature matrix containing the actual geometric deviation of the components is calculated through the morphological correction model.

[0007] Based on the multi-scale geometric feature matrix, a node spatial coordinate system is constructed, and a node construction topology network is established. The scattered nodes in the point cloud are reconstructed into component constraint boundaries, forming a three-dimensional constraint set of rods, plates and welds at the node.

[0008] The three-dimensional constraint set is input into the multi-source data fusion model to simulate the transmission path of the design and construction logic, and outputs the correlation constraint probability matrix between nodes.

[0009] Based on the associated constraint probability matrix, constraint folding operation is performed on the multi-scale geometric feature matrix to establish the mapping relationship between the actual force path direction of the component and the actual geometric deviation of the node, and to generate a force sensitivity prediction curve.

[0010] Based on the stress sensitivity prediction curve, the geometric deviation influence factor of each component position is calculated, and the influence factor is adaptively replaced with the component model to generate an optimized fine digital model of the steel truss beam.

[0011] Preferably, the step of calculating the multi-scale geometric feature matrix containing the true geometric deviations of the component through the morphological correction model includes:

[0012] Spatial reference alignment is performed between the design parameter set of the steel truss girder and the measured point cloud data at the nodes. A constraint-driven layer-by-layer matching algorithm is used to extract the initial geometric deviation distribution of the girder shape and form a basic geometric error field.

[0013] Based on the aforementioned basic geometric error field, a multi-scale decomposition operator is constructed. Through a scale-progressive morphological filtering model, geometric feature information of weld toe fillet, local indentation, and overall component posture change is separated to generate a multi-scale morphological deviation set.

[0014] The geometric deviation correlation weight matrix is ​​constructed using the multi-scale morphological deviation set, and node-level interpolation correction is performed on the three-dimensional discrete surface of the component to obtain a refined feature point cloud describing the coupling relationship between the local and overall morphology of the component.

[0015] The refined feature point cloud is input into the morphological correction model to calculate the cumulative deviation and orientation consistency index of the component at each scale, and the multi-scale geometric feature matrix is ​​generated accordingly.

[0016] Preferably, the step of forming the three-dimensional constraint set of the member-plate-weld at the node includes:

[0017] Based on the principal direction component of the consistency index of deviation direction of each scale in the multi-scale geometric feature matrix, the eigenvector decomposition method is used to determine the principal axis direction of the node, and a node spatial coordinate system is constructed so that the node coordinate axes correspond to the actual force path of the component, the direction of local geometric disturbance and the normal direction of the plate, respectively.

[0018] The node spatial coordinate system is mapped to the node scatter area in the point cloud data. The node scatter is classified by a point-by-point clustering algorithm based on distance constraints, generating a primary boundary set representing weld boundaries, plate boundaries and rod end boundaries.

[0019] Based on the axial relationship between the primary boundary set and the node spatial coordinate system, the local boundary at the node is reconstructed by a joint solution method of arc fitting and plane constraint, so as to obtain the constraint boundary surface that satisfies the geometric continuity of the component connection.

[0020] The constrained boundary surfaces are combined according to the connection topology of rod-plate-weld to form a three-dimensional constraint set of components with node spatial orientation, local structural scale and stress correlation.

[0021] Preferably, the step of establishing the correlation constraint probability matrix between the output nodes includes:

[0022] The spatial boundary parameters of the weld surface, plate surface and rod end surface, which are characterized by the three-dimensional constraints of the components, are respectively subjected to feature quantization to form a multi-source feature vector set that includes geometric continuity index, curvature change rate and force direction matching degree.

[0023] Based on the multi-source feature vector set, a node construction association graph is constructed. By assigning different connection weights to geometric adjacency relationships, force orientation relationships, and surface contact conditions, a weighted topology structure for simulating the design construction logic transmission path is generated.

[0024] The weighted topology is input into the multi-source data fusion model. By calculating the similarity of node features and the path weight decay strategy, the constraint propagation strength between each node is iteratively obtained to obtain the initial association score matrix describing the geometric dependence and force correlation of the nodes.

[0025] The initial association scoring matrix is ​​normalized, and valid association items are screened based on the construction logic consistency threshold. Finally, a node association constraint probability matrix that satisfies the requirements of node geometric continuity and force connection is output.

[0026] Preferably, the step of generating the force sensitivity prediction curve includes:

[0027] The node constraint strength in the correlation constraint probability matrix is ​​paired with the cumulative deviation at each scale of the multi-scale geometric feature matrix. A weighted folding coefficient is applied to the deviation based on the probability value to obtain a multi-scale deviation reconstruction sequence that reflects the degree of force path constraint.

[0028] Based on the multi-scale deviation reconstruction sequence, the direction is expanded according to the principal axis component of the force path direction in the node spatial coordinate system, so that the deviation components form analyzable directional deviation sets in the force path direction, the local disturbance direction and the plate normal direction, respectively.

[0029] The directional deviation set is subjected to scale fusion processing. By taking the trend of deviation change of the force path direction as the dominant quantity, an increasing or decreasing sequence of deviation changes with scale is constructed to obtain a trend function characterizing the sensitivity of node geometric deviation to force path response.

[0030] The trend function is continuously interpolated and fitted within the scale domain to form a force sensitivity prediction curve.

[0031] Preferably, the step of adaptively replacing the influencing factors with the component model to generate an optimized refined digital model of the steel truss girder includes:

[0032] Based on the stress sensitivity prediction curve, the deviation sensitivity values ​​of the components at each scale are extracted. The sensitivity values ​​are compared with the directional deviation set at each component location item by item, and the geometric deviation influence factor, which characterizes the contribution of geometric deviation to the stress path, is calculated.

[0033] The geometric deviation influence factor is mapped to the three-dimensional geometric domain of the component. The local morphological parameters of the component surface are adaptively adjusted according to the magnitude of the influence factor, so that the adjustment weight of the morphological parameters of the region corresponding to the force path direction is higher than the adjustment weight of the local disturbance direction and the plate normal direction.

[0034] Based on the adjusted morphological parameters, feature replacement operations are performed on the original digital model of the component, so that the replacement results simultaneously meet the consistency requirements of the node spatial coordinate system and the geometric continuity requirements of the three-dimensional constraint set, thereby forming a local fine model of the component with deviation compensation capability.

[0035] The local fine model is reassembled according to the component connection relationship to generate an optimized fine digital model of the steel truss beam that is globally consistent and reflects the stress-sensitive characteristics.

[0036] This invention also provides a sophisticated modeling system based on the digital design of steel trusses, comprising:

[0037] Component data acquisition module: acquires the design parameter set of steel truss components and the measured point cloud data at the nodes, and calculates a multi-scale geometric feature matrix containing the actual geometric deviation of the components through the morphological correction model;

[0038] The constraint generation module constructs a node spatial coordinate system based on the multi-scale geometric feature matrix and establishes a node construction topology network. It reconstructs the scattered nodes in the point cloud into component constraint boundaries, forming a three-dimensional constraint set of rods, plates and welds at the nodes.

[0039] Construction logic association calculation module: Input the three-dimensional constraint set into the multi-source data fusion model, simulate the transmission path of the construction logic, and output the association constraint probability matrix between nodes;

[0040] Sensitivity calculation module: Based on the associated constraint probability matrix, performs constraint folding operation on the multi-scale geometric feature matrix to establish the mapping relationship between the actual force path direction of the component and the actual geometric deviation of the node, and generates a force sensitivity prediction curve;

[0041] Model Adaptive Update Module: Based on the stress sensitivity prediction curve, calculate the geometric deviation influence factor of each component position, and adaptively replace the influence factor with the component model to generate an optimized fine digital model of the steel truss beam.

[0042] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0043] 1. This invention establishes a complete technology chain, from multi-scale geometric deviation extraction, nodal spatial coordinate system construction, 3D constraint set reconstruction, multi-source data fusion to calculate nodal relationships, constraint folding calculations, to adaptive optimization of component digital models, achieving a deep representation of the true form of steel truss components. Compared with traditional methods relying on idealized parameter modeling, this invention can capture local weld toe fillet errors, plate normal micro-deviations, and member end assembly deviations at the millimeter scale. By probabilistically quantifying the construction logic through inter-node correlation constraints, the transmission effect of geometric deviations is accurately characterized, thus establishing a high-fidelity mapping relationship between the actual structure of the steel truss and the force path. This capability significantly improves the consistency between digital modeling and actual engineering construction, providing higher-precision input data for structural safety assessment.

[0044] 2. This invention, by introducing stress sensitivity prediction curves and geometric deviation influence factors, achieves adaptive feature replacement of component morphology. This enables the digital model to actively compensate for the disturbance of the stress path caused by structural errors, generating a refined digital model of steel truss girders with structural consistency, stress rationality, and global geometric continuity. Compared with existing technologies that rely solely on simple error superposition or manual correction methods, the model optimization method provided by this invention can achieve automated and quantitative adjustments in complex structural areas, effectively reducing the prediction error of local stress concentration caused by geometric deviations. This makes structural stress analysis, fatigue life calculation, and health monitoring more reliable, providing a digital analysis foundation with engineering value for scenarios such as long-span bridges and high-rise steel structures. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0046] Figure 1 This is a flowchart of the method of the present invention.

[0047] Figure 2 This is a flowchart of the system modules of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0049] Example 1, please refer to Figure 1 As shown in this embodiment, the refined modeling method based on the digital design of steel trusses includes:

[0050] The design parameter set of steel truss components and the measured point cloud data at the nodes are obtained, and a multi-scale geometric feature matrix containing the actual geometric deviation of the components is calculated through the morphological correction model.

[0051] In implementing the method of this invention, the design parameter set of the steel truss components is first extracted from the structural modeling file during the steel truss design phase. This component design parameter set includes the theoretical length of the component, theoretical cross-sectional dimensions, bevel angle at the end of the member, gusset plate thickness, theoretical fillet radius of the weld toe, and the component installation posture vector. To ensure the accuracy of the model input information, measured point cloud data at the nodes is acquired using a 3D laser scanning device. This point cloud must cover the entire member connection area of ​​the steel truss node, with a point density of no less than 200 measurement points per square centimeter to meet the accuracy requirements of subsequent morphological recognition.

[0052] The design parameter set of the steel truss girder components and the measured point cloud data at the nodes are imported into the three-dimensional coordinate space. The reference coordinate system of the design model is constructed based on the component installation posture vector, and the point cloud data is mapped to the same coordinate system through rotation and translation matrices.

[0053] To accurately identify the geometric differences between the point cloud and the design model, a constraint-driven layer-by-layer matching algorithm is used in the alignment process. This layer-by-layer matching algorithm is implemented through the following steps:

[0054] Set the reference axis constraint of the component, and perform initial linear fitting on the point set in the point cloud that is consistent with the axis direction of the component to obtain the initial alignment frame.

[0055] Based on the theoretical boundary of the node plate shape, surface constraints are established, and the point cloud surface points are gradually moved closer to the theoretical surface through least squares distance optimization.

[0056] Based on the theoretical fillet radius of the weld toe, curve constraints are established, and a weighted iterative nearest-point algorithm is used to gradually converge the point cloud of the weld toe region to the theoretical curve.

[0057] After the above matching is completed, the three-dimensional offset vector from each point in the point cloud to the corresponding position in the design model is calculated, and all offset vectors are sorted according to the component surface coordinates to generate a basic geometric error field. The basic geometric error field is stored in the form of a vector set and is used to characterize the actual deviation of the component's shape at each position.

[0058] To extract geometric features at different size levels from the basic geometric error field, a multi-scale decomposition operator is constructed for the error field. The multi-scale decomposition operator is constructed using a Gaussian kernel function, with its scale σ set in increments of 0.5 mm, 2 mm, and 10 mm, to separate micro-scale weld toe features, meso-scale indentation features, and large-scale component posture changes.

[0059] The basic geometric error field is convolved with Gaussian kernels of different scales σ to obtain morphological response fields at different scales. Difference between the morphological response fields of adjacent scales yields the feature enhancement results for the corresponding scales; these differences serve as the geometric feature information output by the morphological filtering model.

[0060] Extract them in ascending order of scale:

[0061] Microscale: Local curvature abrupt change in the transition of weld toe fillet;

[0062] Medium-scale: medium-sized deformations such as localized indentations and depressions;

[0063] Large scale: The overall posture of the component shows signs of skewness, twisting, and bending.

[0064] The feature information at different scales above is aggregated to form a multi-scale morphological deviation set.

[0065] To analyze the coupling relationship between morphological deviations at different scales, a geometric deviation correlation weight matrix is ​​constructed using a multi-scale morphological deviation set as input. The steps for constructing the weight matrix are as follows:

[0066] Calculate the gradient direction for each deviation vector in the multi-scale morphological deviation set;

[0067] Calculate the coupling coefficient between scales based on the directional similarity of the deviation vector;

[0068] A three-dimensional matrix is ​​formed based on the coupling coefficient, and each element is used to characterize the degree of correlation between different scale deviations.

[0069] By applying the geometric deviation correlation weight matrix to the point cloud interpolation processing of the three-dimensional discrete surface of the component, the spatial position of each node on the surface is weighted and corrected, so that the multi-scale features can reflect the mutual influence relationship in the geometric space.

[0070] The interpolation correction adopts a node interpolation method based on radial basis functions. The coordinates of each node in the point cloud are updated by weighted shape-driven offsets, and finally a refined feature point cloud describing the coupling relationship between the local and global shapes of the component is obtained.

[0071] The refined feature point cloud is input into the morphological correction model. The morphological correction model is based on a multi-scale morphological deviation set and a geometric deviation correlation weight matrix, and calculates the deviation representation of the component at each scale through the following steps:

[0072] Calculate the cumulative intra-scale deviation of each node in the point cloud, which is defined as the cumulative magnitude of the deviation vector of that node at the same scale.

[0073] The consistency index of deviation direction at the calculation node is defined as the average cosine of the angle between the deviation vectors of multiple scales, and is used to describe the directional coordination of multi-scale deformation.

[0074] The cumulative deviation and directional consistency are reconstructed into a matrix form based on the node coordinates to make it spatially indexable for subsequent modeling steps.

[0075] Finally, the cumulative deviation matrix and the directional consistency matrix are combined to form a multi-scale geometric feature matrix. This matrix serves as the basic data for fine modeling of the steel truss, providing accurate input for subsequent node topology construction and stress path simulation.

[0076] Based on the multi-scale geometric feature matrix, a node spatial coordinate system is constructed, and a node construction topology network is established. The scattered nodes in the point cloud are reconstructed into component constraint boundaries, forming a three-dimensional constraint set of rods, plates, and welds at the nodes.

[0077] After obtaining the multi-scale geometric feature matrix, in order to extract the true force orientation and local geometric change trend of the components at the nodes, the principal direction components of the consistency index of deviation directions at each scale are calculated. The principal direction components are determined using the eigenvector decomposition method, and the specific steps are as follows:

[0078] The consistency index of each scale deviation direction is used to construct a direction coordination matrix according to the node coordinate position. Each element of the matrix is ​​used to characterize the cosine of the direction angle between two scale deviation vectors.

[0079] The eigenvector decomposition of the directional coordination matrix is ​​performed, and the eigenvector corresponding to the largest eigenvalue is selected as the principal axis direction of the node to represent the actual force path of the component;

[0080] The eigenvectors corresponding to the second and third largest eigenvalues ​​are selected as the local geometric perturbation direction and the plate normal direction of the node, which are used to describe the geometric changes of the node details.

[0081] The node spatial coordinate system is constructed by using the above eigenvector decomposition method, so that the three coordinate axes correspond to the actual force path direction of the component, the local geometric disturbance direction, and the plate normal direction, respectively, providing a unified reference direction for subsequent geometric reconstruction.

[0082] To accurately identify weld boundaries, plate boundaries, and member end boundaries, the nodal spatial coordinate system is mapped to the nodal scatter area of ​​the point cloud data. The relative position of each measurement point in the point cloud is determined in the nodal spatial coordinate system based on its coordinate projection.

[0083] To perform boundary category classification, a point-by-point clustering algorithm based on distance constraints is adopted. This algorithm is implemented through the following steps:

[0084] Three distance thresholds are set based on weld width, plate edge thickness, and rod end cross-sectional dimensions. These thresholds are used to distinguish the spatial distribution characteristics of different geometric boundaries in the point cloud. The three-dimensional Euclidean distance between any two points is calculated and compared with the distance threshold of the corresponding category. Under the premise of satisfying the distance threshold constraints, a point cluster is formed by a point-by-point expansion method. Each point cluster corresponds to one of the weld boundary, plate boundary, or rod end boundary. The primary boundary set structurally describes the three types of boundaries at the nodes, providing basic classification data for subsequent boundary surface reconstruction.

[0085] To construct boundary surfaces that satisfy the geometric continuity of component connections, the boundaries are finely reconstructed based on the axial correspondence between the primary boundary set and the node spatial coordinate system. The specific implementation is as follows:

[0086] Weld area: The arc fitting method is used to fit the set of points on the weld toe edge, and the fitting function adopts the equation of a circular arc curve. Where x and y are the lateral and longitudinal coordinates of a measurement point in the measured point cloud within the fitted plane coordinate system, used to describe the position of the weld toe boundary point. a and b are the lateral and longitudinal coordinates of the arc fitting center. Obtained by least squares optimization, it represents the center position of the theoretical weld toe fillet in the actual measurement space. r is the fitted weld toe fillet radius. It reflects the actual fillet size of the weld transition region and is an important geometric parameter for local weld toe fatigue sensitivity analysis.

[0087] Plate boundary region: Geometric reconstruction is performed using planar constraints, and planar equations are established based on the plate's normal direction. ; where the coefficient , , The parameter d is obtained by optimizing the minimum distance between the surface points, taking the normal direction of the plate in the nodal space coordinate system.

[0088] End boundary region of the member: The arc-straight line joint fitting method of the cross section profile is used to make the end profile meet the design cross section size requirements of the member.

[0089] The above fitting results are input into the joint solution step, and the continuity function is constrained. ; The coordinates of the theoretical connection point generated during the boundary splicing process of the i-th region (such as the weld region). The coordinates of the actual connection points are provided by adjacent regions (such as plate regions or rod end regions). Continuity correction is performed on the contact points between multiple regions to obtain the constrained boundary surface that satisfies the geometric continuity of the component connection.

[0090] After completing the geometric reconstruction of each local boundary surface, the boundary surfaces are topologically combined based on the physical connection relationship between the rods and plates and the attachment structure of the welds in three-dimensional space. The combination process includes the following steps:

[0091] Using the node spatial coordinate system as a reference, the weld boundary surface is projected onto the contact area of ​​the end boundary surface of the member;

[0092] The intersection line between the plate boundary surface and the weld boundary surface is extracted to ensure that the intersection line remains continuous in three-dimensional space;

[0093] Multiple local geometric segments are spliced ​​together according to the connection sequence of rod-plate-weld to ensure that the three-dimensional surfaces are consistent in both the normal and tangential directions.

[0094] Once assembled, a three-dimensional constraint set of components is formed, which includes the spatial orientation of nodes, local structural scale, and stress correlation. This provides basic boundary conditions for subsequent component behavior simulation, node stress path derivation, and refined digital modeling.

[0095] The three-dimensional constraint set is input into the multi-source data fusion model to simulate the transmission path of the design and construction logic, and outputs the correlation constraint probability matrix between nodes.

[0096] After obtaining the three-dimensional constraint set of the component, in order to further extract key parameters describing the geometric continuity and stress matching at the nodes, feature quantization processing is performed on the spatial boundary parameters representing the weld surface, plate surface, and end surface of the rod in the constraint set. The feature quantization processing is carried out according to the following steps:

[0097] The directional angle between the normal directions of adjacent surfaces is calculated and represented by a directional deviation value. A smaller directional deviation value indicates a more continuous surface connection. The directional deviation value is quantified by measuring the rate of change of the normal direction.

[0098] Curvature sequences are extracted along the principal curvature directions on each surface, and the magnitude of curvature change per unit length is taken as the rate of curvature change. The rate of curvature change is used to describe the degree of surface bending and its local geometric features.

[0099] The angle between the normal direction of the curved surface and the direction representing the actual force path of the component in the nodal spatial coordinate system is compared, and the size of the angle is converted into the force direction matching degree. The smaller the angle, the higher the force direction matching degree.

[0100] The three features mentioned above are represented as geometric continuity index, rate of curvature change and force direction matching degree, respectively. The three quantitative features of each surface are combined in a fixed order to form a multi-source feature vector set, providing a unified input data format for subsequent node construction association graph construction.

[0101] To simulate the propagation path of the construction logic at the nodes, a node construction association graph is constructed based on a multi-source feature vector set. The construction steps are as follows:

[0102] Using the contact area between surfaces as the geometric adjacency condition, when two surfaces are actually connected, the relationship is recorded, and the area ratio of the contact area is calculated as the geometric adjacency weight.

[0103] Based on the deviation between the directional distribution of each surface in the node space coordinate system and the direction of the force path, the deviation value is converted into a force direction weight, and the smaller the deviation, the greater the weight.

[0104] If there is a weld connection between curved surfaces, the weld size, weld toe radius, and weld inclination angle are used as input parameters, and the contact weight is calculated using a fixed weighting formula, where the weight increases as the weld size increases.

[0105] The geometric adjacency weights, force orientation weights, and surface contact weights are combined according to a preset ratio to obtain the connection weights between surfaces. These surfaces are used as nodes, and the connection weights are used as edge weights to form a weighted topology, which is then used as input for the subsequent constraint propagation process.

[0106] To quantify the geometric dependencies and stress correlations between nodes, a weighted topology is input into the multi-source data fusion model. The multi-source data fusion model is implemented in the following manner:

[0107] By comparing the degree of difference between multi-source feature vectors, the similarity value of each pair of nodes is calculated; the smaller the difference, the greater the similarity. The similarity value serves as the initial value for node feature similarity.

[0108] An attenuation coefficient is set for the paths in the topology based on their length. As the path length increases, the attenuation coefficient decreases by a fixed proportion, thereby reducing the impact of long-distance node associations.

[0109] Based on the aforementioned two results, the node feature similarity values ​​and the attenuated path weights are superimposed layer by layer. Through multiple rounds of iterative calculation, the constraint propagation strength gradually converges, ultimately yielding the initial association score matrix. This matrix is ​​used to describe the geometric dependence and force association between surfaces.

[0110] To ensure that the node construction relationship conforms to the connection logic of the actual structure, the initial association scoring matrix is ​​normalized so that all scoring values ​​in the matrix fall within the range of 0 to 1, which facilitates the comparison between different scoring items.

[0111] After normalization, a threshold for logical consistency in the construction is set. This threshold is determined through statistical analysis of a large number of node construction instances and is used to distinguish between reasonable and unreasonable connection relationships. Node pairs with scores higher than the threshold are considered to have valid associations.

[0112] Node pairs with scores not lower than the construction logic consistency threshold are recorded as valid associations, and their probability values ​​in the normalized scoring system are calculated to form a node association constraint probability matrix. This matrix reflects the geometric continuity and force connection relationship between surfaces at the nodes, providing constraints for subsequent design logic derivation and component modeling.

[0113] Based on the associated constraint probability matrix, constraint folding operation is performed on the multi-scale geometric feature matrix to establish the mapping relationship between the actual force path direction of the component and the actual geometric deviation of the node, and to generate a force sensitivity prediction curve.

[0114] After obtaining the node association constraint probability matrix, the node constraint strength in the matrix is ​​paired with the cumulative deviation at each scale in the multi-scale geometric feature matrix according to the correspondence between nodes and scales.

[0115] To ensure that the strength of nodal constraints affects the deviation amount, a weighted folding factor is applied to each cumulative deviation. The weighted folding factor is defined as follows:

[0116] The weighted folding coefficient is directly derived from the probability values ​​in the node association constraint probability matrix;

[0117] When the node constraint probability value is large, the weighted folding coefficient of the corresponding cumulative deviation increases, which amplifies the deviation and thus reflects the strong force path constraint effect.

[0118] When the node constraint probability value is small, the corresponding cumulative deviation is suppressed.

[0119] The cumulative deviations after applying weighted folding coefficients are arranged in scale order to form a multi-scale deviation reconstruction sequence, which is used to describe the changing trend of the deviations of nodes at different scales under the action of force constraints.

[0120] To analyze the behavior of nodal geometric deviations in different directions, the multi-scale deviation reconstruction sequence is projected onto the principal axis components representing the force path direction in the nodal spatial coordinate system.

[0121] The node space coordinate system contains three orthogonal directions: the force path direction, the local disturbance direction, and the plate normal direction.

[0122] Directional unfolding is achieved through the following steps: the deviation at each scale in the reconstructed sequence is projected along the force path direction to form the force path direction deviation component; similarly, the deviation is projected along the local disturbance direction to form the local disturbance deviation component; and then the deviation is projected along the normal direction of the plate to form the normal deviation component of the plate.

[0123] In this way, the multi-scale bias reconstruction sequence is split into three independent directional bias sets, so that the three types of bias sets can be used to analyze the response behavior of nodes under different structural orientations.

[0124] To obtain a trend reflecting the variation of nodal deviations with scale, scale fusion processing is performed based on the directional deviation set. The core principle of scale fusion processing is to use the trend of deviation changes in the force path direction as the dominant variable. The implementation method is as follows:

[0125] First, the set of force path direction deviations is sorted according to scale order, from the smallest scale to the largest scale.

[0126] Analyze the direction of change of the deviation as the scale increases. If the deviation increases with the scale, the trend is defined as an increasing sequence; if the deviation decreases with the scale, the trend is defined as a decreasing sequence.

[0127] Using the above trend as the dominant quantity, the local disturbance deviation component and the plate normal deviation component are adjusted synchronously according to the force path deviation trend, so that the three types of directional deviations maintain a consistent direction of change in the scale dimension.

[0128] Finally, a trend function is constructed to characterize the sensitivity of nodal geometric deviations to the force path response. The trend function is used to describe the degree and variation of the influence of nodal deviations on the force path at different scales.

[0129] To transform the trend function into a continuous expression, interpolation fitting is performed on the trend function within the scale domain. The interpolation fitting uses a fixed continuous interpolation method, and the steps are as follows:

[0130] The deviation trend values ​​corresponding to each scale in the trend function are selected as interpolation nodes;

[0131] Apply continuity constraints to the interpolation nodes to ensure that the fitted curve does not exhibit abrupt changes in the scale domain;

[0132] The curvature of the interpolation interval is determined by the variation range between adjacent scales of the node, so that the curve can accurately reflect the variation pattern of multi-scale deviation.

[0133] After interpolation and fitting, a stress sensitivity prediction curve is generated. This prediction curve describes the sensitivity of nodal geometric deviations to the stress path response, and can provide a basis for subsequent component stress analysis, fatigue-sensitive part identification, and design correction.

[0134] Based on the stress sensitivity prediction curve, the geometric deviation influence factor of each component position is calculated, and the influence factor is adaptively replaced with the component model to generate an optimized fine digital model of the steel truss beam.

[0135] After obtaining the stress sensitivity prediction curve, the curve values ​​at different scale locations are read and used as the deviation sensitivity values ​​of the component at the corresponding scale. The deviation sensitivity value reflects the degree of influence of geometric deviation on the stress path response.

[0136] The deviation sensitivity value is paired item by item with the directional deviation set at the component location. The directional deviation set includes the deviation along the force path direction, the deviation along the local disturbance direction, and the deviation along the plate normal direction.

[0137] To ensure that the deviation sensitivity affects the directional deviation set, a geometric deviation influence factor is calculated. The geometric deviation influence factor is defined as follows: at the same scale, the deviation amounts in the directional deviation set are weighted according to the deviation sensitivity value, so that the deviation amounts are amplified at scales with larger deviation sensitivity values ​​and suppressed at scales with smaller deviation sensitivity values.

[0138] The result obtained through the weighted calculation described above serves as the geometric deviation influence factor, used to characterize the degree to which geometric deviation contributes to the force path. The larger the influence factor, the more easily the geometric deviation at that location affects the direction of the force path.

[0139] To establish the correspondence between the geometric deviation influence factor and the actual spatial shape of the component, the geometric deviation influence factor is mapped to the three-dimensional geometric domain of the component. The mapping steps are as follows:

[0140] Each influencing factor is projected onto the three-dimensional surface of the component according to its corresponding spatial coordinate position;

[0141] An influence factor distribution field is formed on the surface of the component, so that each surface location corresponds to a unique influence factor value.

[0142] After mapping, the local morphological parameters of the component surface are adaptively adjusted according to the magnitude of the influence factor. These morphological parameters include weld fillet radius, plate edge thickness, rod end shape transition length, and local curvature value. The rules for adaptive adjustment are defined as follows:

[0143] In the region corresponding to the direction of the force path, when the influence factor increases, the adjustment weight of the morphological parameters is increased to make the component shape more consistent with the force path.

[0144] In the local disturbance direction and the plate normal direction, the weights are adjusted by assigning medium or low parameters according to the relative magnitude of the influence factor values.

[0145] The adjusted morphological parameters must maintain continuity with the original geometric boundaries of the component.

[0146] Through the above adaptive adjustment, the surface morphology of the component can be optimized according to the stress sensitivity, thereby enhancing the rationality of the component's stress response.

[0147] To ensure the digital model reflects the adaptively adjusted morphological changes, a component-by-component feature replacement operation is performed on the original digital model of the component. The feature replacement steps are as follows:

[0148] The adjusted morphological parameters are compared with the parameters at the corresponding positions in the original model;

[0149] For locations where there are discrepancies, replace the original model parameters with the adjusted parameters;

[0150] Ensure that the replaced parameters remain continuous in the length direction, cross-sectional direction, and node connection areas of the component, without any abrupt changes.

[0151] During feature replacement, all parameters must ensure the consistency of the orientation of the three coordinate axes in the node space coordinate system and satisfy the geometric continuity requirements defined by the component's three-dimensional constraint set.

[0152] After the replacement is completed, a detailed local model of the component with deviation compensation capability is generated. This model can reflect the geometric correction requirements that the component may have under the actual force path and has higher force expression accuracy.

[0153] To create a globally consistent digital model of the steel truss girder, the detailed local models of each component were reassembled according to the connection relationships between the members. The assembly method is as follows:

[0154] Based on the orientation relationship of the node spatial coordinate system, the local fine model is located at the node;

[0155] Three-dimensional geometric assembly is performed based on the connection relationships between rods, plates, and welds;

[0156] Local geometric smoothing is performed at the splicing points to maintain the continuity of the shape between the components.

[0157] After assembly, a refined digital model of the optimized steel truss is formed. This model can simultaneously reflect stress sensitivity characteristics, geometric deviation distribution characteristics, and structural continuity at the nodes, making it suitable for structural analysis and design optimization.

[0158] Example 2, please refer to Figure 2 As shown in this embodiment, the refined modeling system based on the digital design of steel trusses includes:

[0159] Component data acquisition module: acquires the design parameter set of steel truss components and the measured point cloud data at the nodes, and calculates a multi-scale geometric feature matrix containing the actual geometric deviation of the components through the morphological correction model;

[0160] The constraint generation module constructs a node spatial coordinate system based on the multi-scale geometric feature matrix and establishes a node construction topology network. It reconstructs the scattered nodes in the point cloud into component constraint boundaries, forming a three-dimensional constraint set of rods, plates and welds at the nodes.

[0161] Construction logic association calculation module: Input the three-dimensional constraint set into the multi-source data fusion model, simulate the transmission path of the construction logic, and output the association constraint probability matrix between nodes;

[0162] Sensitivity calculation module: Based on the associated constraint probability matrix, performs constraint folding operation on the multi-scale geometric feature matrix to establish the mapping relationship between the actual force path direction of the component and the actual geometric deviation of the node, and generates a force sensitivity prediction curve;

[0163] Model Adaptive Update Module: Based on the stress sensitivity prediction curve, calculate the geometric deviation influence factor of each component position, and adaptively replace the influence factor with the component model to generate an optimized fine digital model of the steel truss beam.

[0164] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A refined modeling method based on digital design of steel trusses, characterized by: include: The design parameter set of steel truss components and the measured point cloud data at the nodes are obtained, and a multi-scale geometric feature matrix containing the actual geometric deviation of the components is calculated through the morphological correction model. Based on the multi-scale geometric feature matrix, a node spatial coordinate system is constructed, and a node construction topology network is established. The scattered nodes in the point cloud are reconstructed into component constraint boundaries, forming a three-dimensional constraint set of rods, plates and welds at the node. The three-dimensional constraint set is input into the multi-source data fusion model to simulate the transmission path of the design and construction logic, and outputs the correlation constraint probability matrix between nodes. Based on the associated constraint probability matrix, constraint folding operation is performed on the multi-scale geometric feature matrix to establish the mapping relationship between the actual force path direction of the component and the actual geometric deviation of the node, and to generate a force sensitivity prediction curve. Based on the stress sensitivity prediction curve, the geometric deviation influence factor of each component position is calculated, and the influence factor is adaptively replaced with the component model to generate an optimized fine digital model of the steel truss beam.

2. The refined modeling method based on digital design of steel trusses according to claim 1, characterized in that: The step of calculating the multi-scale geometric feature matrix containing the true geometric deviations of the component through the morphological correction model includes: Spatial reference alignment is performed between the design parameter set of the steel truss girder and the measured point cloud data at the nodes. A constraint-driven layer-by-layer matching algorithm is used to extract the initial geometric deviation distribution of the girder shape and form a basic geometric error field. Based on the aforementioned basic geometric error field, a multi-scale decomposition operator is constructed. Through a scale-progressive morphological filtering model, geometric feature information of weld toe fillet, local indentation, and overall component posture change is separated to generate a multi-scale morphological deviation set. The geometric deviation correlation weight matrix is ​​constructed using the multi-scale morphological deviation set, and node-level interpolation correction is performed on the three-dimensional discrete surface of the component to obtain a refined feature point cloud describing the coupling relationship between the local and overall morphology of the component. The refined feature point cloud is input into the morphological correction model to calculate the cumulative deviation and orientation consistency index of the component at each scale, and the multi-scale geometric feature matrix is ​​generated accordingly.

3. The refined modeling method based on digital design of steel trusses according to claim 1, characterized in that: The steps for forming the three-dimensional constraint set of the member-plate-weld at the node include: Based on the principal direction component of the consistency index of deviation direction of each scale in the multi-scale geometric feature matrix, the eigenvector decomposition method is used to determine the principal axis direction of the node, and a node spatial coordinate system is constructed so that the node coordinate axes correspond to the actual force path of the component, the direction of local geometric disturbance and the normal direction of the plate, respectively. The node spatial coordinate system is mapped to the node scatter area in the point cloud data. The node scatter is classified by a point-by-point clustering algorithm based on distance constraints, generating a primary boundary set representing weld boundaries, plate boundaries and rod end boundaries. Based on the axial relationship between the primary boundary set and the node spatial coordinate system, the local boundary at the node is reconstructed by a joint solution method of arc fitting and plane constraint, so as to obtain the constraint boundary surface that satisfies the geometric continuity of the component connection. The constrained boundary surfaces are combined according to the connection topology of rod-plate-weld to form a three-dimensional constraint set of components with node spatial orientation, local structural scale and stress correlation.

4. The refined modeling method based on digital design of steel trusses according to claim 1, characterized in that: The step of defining the correlation constraint probability matrix between output nodes includes: The spatial boundary parameters of the weld surface, plate surface and rod end surface, which are characterized by the three-dimensional constraints of the components, are respectively subjected to feature quantization to form a multi-source feature vector set that includes geometric continuity index, curvature change rate and force direction matching degree. Based on the multi-source feature vector set, a node construction association graph is constructed. By assigning different connection weights to geometric adjacency relationships, force orientation relationships, and surface contact conditions, a weighted topology structure for simulating the design construction logic transmission path is generated. The weighted topology is input into the multi-source data fusion model. By calculating the similarity of node features and the path weight decay strategy, the constraint propagation strength between each node is iteratively obtained to obtain the initial association score matrix describing the geometric dependence and force correlation of the nodes. The initial association scoring matrix is ​​normalized, and valid association items are screened based on the construction logic consistency threshold. Finally, a node association constraint probability matrix that satisfies the requirements of node geometric continuity and force connection is output.

5. The refined modeling method based on digital design of steel trusses according to claim 1, characterized in that: The step of generating the force sensitivity prediction curve includes: The node constraint strength in the correlation constraint probability matrix is ​​paired with the cumulative deviation at each scale of the multi-scale geometric feature matrix. A weighted folding coefficient is applied to the deviation based on the probability value to obtain a multi-scale deviation reconstruction sequence that reflects the degree of force path constraint. Based on the multi-scale deviation reconstruction sequence, the direction is expanded according to the principal axis component of the force path direction in the node spatial coordinate system, so that the deviation components form analyzable directional deviation sets in the force path direction, the local disturbance direction and the plate normal direction, respectively. The directional deviation set is subjected to scale fusion processing. By taking the trend of deviation change of the force path direction as the dominant quantity, an increasing or decreasing sequence of deviation changes with scale is constructed to obtain a trend function characterizing the sensitivity of node geometric deviation to force path response. The trend function is continuously interpolated and fitted within the scale domain to form a force sensitivity prediction curve.

6. The refined modeling method based on digital design of steel trusses according to claim 1, characterized in that: The step of adaptively replacing the influencing factors with the component model to generate an optimized, refined digital model of the steel truss includes: Based on the stress sensitivity prediction curve, the deviation sensitivity values ​​of the components at each scale are extracted. The sensitivity values ​​are compared with the directional deviation set at each component location item by item, and the geometric deviation influence factor, which characterizes the contribution of geometric deviation to the stress path, is calculated. The geometric deviation influence factor is mapped to the three-dimensional geometric domain of the component. The local morphological parameters of the component surface are adaptively adjusted according to the magnitude of the influence factor, so that the adjustment weight of the morphological parameters of the region corresponding to the force path direction is higher than the adjustment weight of the local disturbance direction and the plate normal direction. Based on the adjusted morphological parameters, feature replacement operations are performed on the original digital model of the component, so that the replacement results simultaneously meet the consistency requirements of the node spatial coordinate system and the geometric continuity requirements of the three-dimensional constraint set, thereby forming a local fine model of the component with deviation compensation capability. The local fine model is reassembled according to the component connection relationship to generate an optimized fine digital model of the steel truss beam that is globally consistent and reflects the stress-sensitive characteristics.

7. A fine modeling system based on digital design of steel trusses, used to implement the fine modeling method based on digital design of steel trusses as described in any one of claims 1-6, characterized in that: include: Component data acquisition module: acquires the design parameter set of steel truss components and the measured point cloud data at the nodes, and calculates a multi-scale geometric feature matrix containing the actual geometric deviation of the components through the morphological correction model; The constraint generation module constructs a node spatial coordinate system based on the multi-scale geometric feature matrix and establishes a node construction topology network. It reconstructs the scattered nodes in the point cloud into component constraint boundaries, forming a three-dimensional constraint set of rods, plates and welds at the nodes. Construction logic association calculation module: Input the three-dimensional constraint set into the multi-source data fusion model, simulate the transmission path of the construction logic, and output the association constraint probability matrix between nodes; Sensitivity calculation module: Based on the associated constraint probability matrix, performs constraint folding operation on the multi-scale geometric feature matrix to establish the mapping relationship between the actual force path direction of the component and the actual geometric deviation of the node, and generates a force sensitivity prediction curve; Model Adaptive Update Module: Based on the stress sensitivity prediction curve, calculate the geometric deviation influence factor of each component position, and adaptively replace the influence factor with the component model to generate an optimized fine digital model of the steel truss beam.

Images