A digital simulation pre-assembly method and system for steel structure

Abstract

The present invention relates to the technical field of image data processing, and proposes a method and system for digital simulation pre-assembly of steel structures, including: obtaining models of several components in the pre-assembly process of steel structures through point cloud scanning; obtaining the initial neighborhood of each point for several points in the model of the same component, and obtaining several neighborhood smoothing angles of each point; expanding the initial neighborhood of each point and obtaining the terminal neighborhood of each point; obtaining a complete evaluation of the segmentation of each point; segmenting the model to obtain several edges to be matched in the model; obtaining edge distribution arrays and matching edge arrays in the two models; obtaining an evaluation of the interface fit between the two models; constructing a splicing graph structure through threshold judgment; and obtaining the assembly order of components in the steel structure. The present invention aims to solve the problem that a plurality of successful matching positions are generated at the smooth position of a part of the surface of a structural part, thereby affecting the matching accuracy and thus affecting the assembly progress.

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing, and in particular to a steel structure digital simulation pre-assembly method and system. Background Art

[0002] Digital pre-assembly of steel structures belongs to reverse engineering. Different from forward engineering, reverse engineering converts existing product models into engineering design models in order to recreate existing product models. Using simulated pre-assembly before assembling steel structures can set the splicing order during the structural assembly process, and can automatically identify the matching of different components during the assembly process, which is conducive to reducing assembly errors caused by direct construction, thereby avoiding the problem of repeated disassembly and assembly affecting the accuracy of steel structures; in the process of assembling steel structures, the assembly process is judged by the model sequence recorded by manual dragging by technicians, and the recognition accuracy of manual assembly for massive parts is not high. The assembly process requires trying a variety of combinations, which affects the assembly progress.

[0003] In the process of assembly and matching between models, the existing technology judges the mutual fit of the model edges, aligns the edges to be matched through different edge sliding, and makes judgments and structural combinations through edge length thresholds; there are differences in the interface parts between different components in the steel structure, and the contact surface shapes of the components in the actual assembly process are different, resulting in multiple successful matching positions at the smooth surface positions of the component parts during the assembly process, affecting the matching accuracy. Summary of the invention

[0004] The present invention provides a digital simulation pre-assembly method and system for steel structures to solve the problem that the existing structural parts have multiple successful matching positions at the smooth surface, which affects the matching accuracy and further affects the assembly progress. The technical solutions adopted are as follows:

[0005] The present invention proposes a steel structure digital simulation pre-assembly method, which comprises the following steps:

[0006] Obtain models of several components during the pre-assembly process of the steel structure through point cloud scanning;

[0007] For a plurality of points in a model of the same component, an initial neighborhood of each point is obtained, wherein the initial neighborhood of each point includes a plurality of initial neighborhood points of each point, and a plurality of neighborhood smoothing angles of each point are obtained according to the spatial position distribution of each point and its initial neighborhood points; based on the neighborhood smoothing angle of each point and the spatial position distribution of the points in the model, the initial neighborhood of each point is expanded and a terminal neighborhood of each point is obtained;

[0008] According to the termination neighborhood of each point and several neighborhood smoothing angles, combined with the curvature of the point in the model, a complete segmentation evaluation of each point is obtained; combined with the termination neighborhood of each point, the model is segmented to obtain several edges of the model to be matched;

[0009] Calculate the matching distance between the edges to be matched of different models, and obtain the edge distribution array and the matching edge array in the two models according to the size relationship of the matching distance and the position distribution between the edges to be matched in the same model; obtain the interface fit evaluation between the two models according to the position distribution difference of each pair of edges in the edge distribution array and the matching edge array;

[0010] Based on the interface fit evaluation between different models, a splicing graph structure is constructed through threshold judgment; the graph structure is traversed to obtain the assembly order of components in the steel structure.

[0011] Optionally, the several neighborhood smoothing angles of each point are obtained by:

[0012] The model of any component is taken as the current model, and any point in the current model is taken as the target point. The normal of the plane where the target point is located and passes through the target point is obtained, and the angle between the line connecting the target point and any of its initial neighboring points and the normal is recorded as a neighborhood smoothing angle of the target point, and several neighborhood smoothing angles of the target point are obtained.

[0013] Optionally, the method of expanding the initial neighborhood of each point and obtaining the terminal neighborhood of each point includes the following specific methods:

[0014] According to the maximum and minimum values ​​of several neighborhood smoothing angles of the target point, the neighborhood smoothing angle value range of the target point is obtained, the window side length of the target point is expanded, and the newly added points in the window after each expansion are connected with the target point, and an angle is formed with the normal line, which is recorded as the expanded neighborhood angle;

[0015] If the expanded neighborhood angle belongs to the range of the neighborhood smoothing angle of the target point, the point corresponding to the expanded neighborhood angle is used as the neighborhood point of the target point; and so on, the window side length is continuously expanded until no new neighborhood points are generated after one expansion, and the expansion is stopped, and the area formed by the target point and its initial neighborhood points and the neighborhood points is recorded as the terminal neighborhood of the target point.

[0016] Optionally, the segmentation complete evaluation of each point is obtained by:

[0017] Represent the target point as a point , target point segmentation complete evaluation The calculation method is:

[0018]

[0019] in, represents the curvature of the target point, represents the number of points in the target point's terminating neighborhood, Indicates the number of points in the current model, and They respectively represent the minimum and maximum values ​​of several neighborhood smoothing angles of the target point.

[0020] Optionally, the model is segmented to obtain a plurality of edges to be matched of the model, including the following specific methods:

[0021] Arrange all points in the current model in descending order according to the size of the complete segmentation evaluation to obtain the segmentation point sequence of the current model; take the first point in the segmentation point sequence and its terminal neighborhood as a segmentation region, and remove several points corresponding to the segmentation region in the segmentation point sequence, and the remaining points in the segmentation point sequence constitute the first segmentation point sequence;

[0022] The first point in the first segmentation point sequence and the points in its termination neighborhood that exist in the first segmentation point sequence form a segmentation region, and a number of points corresponding to the segmentation region are removed from the first segmentation point sequence, and the remaining points in the first segmentation point sequence form a second segmentation point sequence;

[0023] This process is repeated in this way until all points in the segmentation point sequence of the current model are segmented to obtain a number of segmentation regions; the edge segments adjacent to each segmentation region are used as a number of edges to be matched of the current model.

[0024] Optionally, the edge distribution array and the matching edge array in the two models are specifically obtained by:

[0025] The edge to be matched in model a corresponding to the minimum value of all matching distances is recorded as the edge , the edge to be matched in model b corresponding to the minimum value is recorded as edge ;

[0026] 1) Get the edge The adjacent edges to be matched on model a are recorded as edges ;

[0027] 2) Place the edge The edge to be matched in model b corresponding to the minimum value among several matching distances is recorded as edge ;

[0028] 3) If the edge With edge intersect, continue to the next step; if the edge With edge If they do not intersect, return to step 2) and extract the edge to be matched in model b corresponding to the second smallest matching distance as the edge , and continue with step 3), and so on, until an edge that meets the conditions is found ;

[0029] 4) Follow step 1) to get the edge Other adjacent edges to be matched on model a are recorded as edges Repeat steps 2) and 3) for the edge Get edges that meet the criteria ; Similarly, for the edge of model a All adjacent edges to be matched are successfully matched with the edges to be matched on model b, and a number of edges to be matched are obtained;

[0030] The edge of model a and all its adjacent edges to be matched are arranged in the order traversed in the above matching process to obtain the edge distribution array of model a, and all the edges to be matched in model b that are successfully matched with the edges to be matched in model a are arranged in the order of the corresponding edges to be matched in the edge distribution array of model a to obtain the matching edge array of model b;

[0031] Get the marginal distribution array of model b and the matching marginal array of model a.

[0032] Optionally, the interface compatibility evaluation between the two models is obtained by:

[0033] Based on the position distribution relationship of different edges to be matched in the edge distribution array of model a and the matching edge array of model b, a position difference factor of each edge to be matched in the edge distribution array of model a and the matching edge array of model b is obtained;

[0034] The degree of interface compatibility between model a and model b The calculation method is:

[0035]

[0036] in, represents the number of edge pairs to be matched in the edge distribution array of model a and the matching edge array of model b, The edge distribution array of model a and the matching edge array of model b are represented by The position difference factor for the matched edges, represents the mean of the position difference factors of all edges to be matched in the edge distribution array of model a and the matching edge array of model b, represents the standard deviation of the position difference factors of all edges to be matched in the edge distribution array of model a and the matching edge array of model b;

[0037] Get the interface compatibility between model b and model a ,Will and The normalized mean value is used as the interface fit evaluation between model a and model b.

[0038] Optionally, the position difference factor of each edge to be matched in the edge distribution array of model a and the matching edge array of model b is specifically obtained by:

[0039] For the edge distribution array of model a and the matching edge array of model b, except for the first edge to be matched, i.e., edge With edge In addition, for any other pair of edges to be matched that belong to model a and edge Get the angle between the edge to be matched and the edge belonging to model b in the edge to be matched The angle is obtained, and the absolute value of the difference between the two angles is used as the position difference factor of the edge to be matched.

[0040] Optionally, the constructing of the mosaic graph structure by threshold judgment includes the following specific methods:

[0041] If the interface fit evaluation between any two models is greater than or equal to the interface fit threshold, the two models are matched successfully, and the edge distribution array corresponding to the maximum value of the interface fit degree of the two models as the benchmark and each edge to be matched in the matching edge array are used as several matching edge groups of the two models, and the two edges to be matched in the matching edge group are converted into edge forms corresponding to the matching distance selected by DDTW. According to the edge form, the corresponding order of pixels in the two edges to be matched are overlapped in the spatial domain to complete the splicing of the two models; if the interface fit evaluation between the two models is less than the interface fit threshold, the two models do not match;

[0042] The model of each component in the steel structure is regarded as a node in the graph structure. If the two models match successfully, the nodes corresponding to the two models are connected by undirected edges; if the two models do not match, no undirected edge connection is made between the corresponding nodes; all models are connected by undirected edges based on the matching results, and the resulting graph structure is recorded as a spliced ​​graph structure.

[0043] The present invention also proposes a digital simulation pre-assembly system for steel structures, which includes a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the above method when executing the computer program.

[0044] The beneficial effects of the present invention are as follows: the present invention splits the model of the component in the steel structure according to similarity growth, screens the smooth surface and the corresponding edge part, performs matching analysis and array construction according to the morphology between the edges to be matched, and thereby realizes the rotation positioning and matching between the models of different components; wherein in the process of analyzing the midpoint of the model, the neighborhood smooth angle is obtained by obtaining the angle between the connecting line and the normal between the target point and the neighborhood point, and similarity growth is performed on this, so that the target point obtains the termination neighborhood from the initial neighborhood, so that the termination neighborhood can reflect several neighborhood points that are in the same plane or close to the same plane as the target point, and avoids the point close to the edge position as the growth point to cause uneven growth area; after obtaining the termination area for the point, the point is quantified by combining the curvature of the point to segment the complete evaluation, and the point is reflected by the segmentation complete evaluation The possibility that the termination neighborhood of is a smooth interface surface is used, and this possibility is used for sorting, so as to complete the segmentation and edge extraction of the model; after obtaining the edges to be matched, matching is performed according to the morphological similarity and position distribution relationship between the edges to be matched in different models, and the edge distribution array and the matching edge array are obtained, and the position distribution difference between the edge pairs to be matched is further analyzed to obtain the interface fit evaluation, so as to quantify whether the morphology of the edges to be matched between different models is close, and then reflect whether the two models can be spliced; finally, the splicing graph structure is constructed through the matching relationship between the models, and by traversing the connection relationship between the nodes of the splicing graph structure, on the basis of the successful assembly and matching of the component models, the assembly order starting from the bottom structure is realized, so as to improve the matching accuracy and splicing efficiency of the component splicing in the steel structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0046] Figure 1 A schematic diagram of a process flow of a digital simulation pre-assembly method for a steel structure provided by an embodiment of the present invention;

[0047] Figure 2 It is a schematic diagram of steel structure components and calibration targets;

[0048] Figure 3 A schematic diagram of a steel structure component collected from one perspective;

[0049] Figure 4 Collect schematic diagrams for another perspective of steel structure components;

[0050] Figure 5 Schematic diagram of the synthesized steel structure component model. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] See also Figure 1 , which shows a flow chart of a steel structure digital simulation pre-assembly method provided by an embodiment of the present invention, the method comprising the following steps:

[0053] The purpose of the present invention is: to obtain a model only through screening according to a threshold value through the uneven points on the model boundary, and to split the edges to be matched and then splice them, the contact surfaces of the steel structure components are different in shape, so that the assembly process produces multiple assembling positions that affect the assembly accuracy; the present invention performs similarity growth through the surface smoothness characteristics of the model to be spliced, splits the edges of the current model for matching, and further rotates the structure according to the overlap of the point cloud after the three-dimensional steel structure is spliced, so as to ensure the accuracy of the simulated pre-assembly of the steel structure.

[0054] Step S001: Obtain models of several components in the steel structure pre-assembly process through point cloud scanning.

[0055] The purpose of this embodiment is to obtain a model by scanning the steel structure components using point clouds, assemble them according to the model and obtain the assembly order, so as to achieve pre-assembly of the steel structure components. Therefore, it is necessary to first scan the steel structure components at multiple points to obtain the digital models of each component.

[0056] Specifically, the steel structure components are placed and calibration targets are arranged to identify the structural contours of the components, such as Figure 2 As shown in the figure; due to the complex structure of the component itself, multi-faceted scanning can be used to fully identify the complex component, such as Figure 3 and Figure 4As shown, it is a schematic diagram of collecting components from different perspectives. During the collection process, adjacent collection perspectives are marked with serial numbers so that more point cloud areas sufficient for registration and fewer point cloud scanning blind spots appear in the subsequent point cloud matching process, and corresponding point cloud data is obtained for the components collected from each perspective.

[0057] Furthermore, the multi-view power data is fused using an iterative closest point (ICP) algorithm, wherein the ICP algorithm is a geometric optimization algorithm used to find the best correspondence between two point clouds. It is a prior art and will not be described in detail in this embodiment. In the fusion process, point cloud data from adjacent acquisition perspectives are fused, that is, point cloud data with adjacent serial numbers are fused. For example, there are five point clouds, 1, 2, 3, 4, and 5. ICP first merges 1 and 2 to obtain a new point cloud 6, which is placed at the positions of 1 and 2 to obtain 6, 3, 4, and 5. In the next step, 6 and 3 are merged. Finally, a model for synthesizing multi-view point cloud data is obtained, such as Figure 5 As shown; all components in the steel structure are collected and constructed according to the above method to obtain the model of each component.

[0058] It should be noted that the distribution of edge structural components required for the connection of components is first analyzed: the connection of steel structures depends on the nesting or bolts on the current components for fixed connection of points, so the uneven position on the outer surface of the current component is selected, that is, the edge part is extracted; and the point cloud corresponding to the model is smoothed before fitting, so it is necessary to screen out the existing uneven areas from the model for splicing. In the aligned point cloud, the intervals between points are the same, so similarity growth can be performed based on the idea of ​​regional growing and the difference in curvature between points.

[0059] Step S002: for several points in the model of the same component, the initial neighborhood of each point is obtained, and according to the spatial position distribution of each point and its initial neighborhood points, several neighborhood smoothing angles of each point are obtained; based on the neighborhood smoothing angle of each point and the spatial position distribution of the points in the model, the initial neighborhood of each point is expanded and the terminal neighborhood of each point is obtained.

[0060] It should be noted that during the growth process, it is necessary to avoid using points with smaller curvature near non-flat positions, that is, at the edge as growth points, which will lead to uneven growth areas and affect edge extraction and subsequent matching. In this case, the points in the model and the points in their neighborhood need to determine a plane reference (surface reference). The angle between the line between the point and the neighboring points and the vertical normal passing through the point is used to determine an angle range, and similarity growth is performed based on this, so as to avoid points with large plane changes but small distances from mistakenly forming the same growth area with neighboring points.

[0061] Preferably, in one embodiment of the present invention, the initial neighborhood of each point is obtained for a plurality of points in the model of the same component, and a plurality of neighborhood smoothing angles of each point are obtained according to the spatial position distribution of each point and its initial neighborhood points, including the specific method of:

[0062] Take any component model as the current model, any point in the current model as the target point, and construct the The size of the window, this embodiment uses Describe, record the window of the target point as the initial neighborhood of the target point, and the other points in the initial neighborhood except the target point as the initial neighborhood points of the target point; obtain the normal of the plane where the target point is located and passes through the target point, record the angle between the line connecting the target point and any one of its initial neighborhood points and the normal as a neighborhood smoothing angle of the target point, and obtain several neighborhood smoothing angles of the target point.

[0063] It should be noted that the more similar the neighborhood smoothing angles obtained by the initial neighborhood points of the target point are, the more likely the target point and its initial neighborhood points are to be in the same plane, that is, the surface position of the target point is smoother; and based on the extracted neighborhood smoothing angle as the judgment range, the initial neighborhood is expanded, and the points whose neighborhood smoothing angles are within the judgment range are screened to achieve similarity growth for the neighborhood of the target point, thereby obtaining the terminal neighborhood of the target point.

[0064] Preferably, in one embodiment of the present invention, based on the neighborhood smoothing angle of each point and the spatial position distribution of the points in the model, the initial neighborhood of each point is expanded and the terminal neighborhood of each point is obtained, including the specific method of:

[0065] According to the maximum and minimum values ​​of several neighborhood smooth angles of the target point, the neighborhood smooth angle value range of the target point is obtained, and the window side length of the target point is expanded with a step size of 1. The newly added point in the window after each expansion is connected with the target point, and an angle is formed with the normal line, which is recorded as the expanded neighborhood angle; if the expanded neighborhood angle belongs to the neighborhood smooth angle value range of the target point, the point corresponding to the expanded neighborhood angle is used as the neighborhood point of the target point, if not, it is not added to the neighborhood of the target point; and so on, the window side length is continuously expanded until no new neighborhood points are generated after one expansion, then the expansion is stopped, and the area formed by the target point, its initial neighborhood point and the neighborhood point is recorded as the terminal neighborhood of the target point.

[0066] So far, in the process of analyzing the midpoints of the model, the neighborhood smoothing angle is obtained by obtaining the angle between the line connecting the target point and the neighboring point and the normal line, and similarity growth is performed based on this, so that the terminal neighborhood of the target point is obtained from the initial neighborhood, so that the terminal neighborhood can reflect several neighboring points that are in the same plane or close to the same plane as the target point, while avoiding points close to the edge as growth points, which may lead to uneven growth areas.

[0067] Step S003, according to the termination neighborhood of each point and several neighborhood smoothing angles, combined with the curvature of the point in the model, obtain a complete segmentation evaluation of each point; combined with the termination neighborhood of each point, segment the model to obtain several edges to be matched of the model.

[0068] It should be noted that after the termination area of ​​the point is obtained, the model needs to be segmented according to the termination area of ​​the point, that is, to extract the smooth surfaces of the model, the termination area of ​​each point needs to be evaluated to ensure that the area of ​​the segmented smooth surface is large enough and the curvature is small enough. It is necessary to analyze the neighborhood smooth angle and combine the curvature of the point to perform calculations for complete segmentation evaluation.

[0069] Preferably, in one embodiment of the present invention, according to the termination neighborhood of each point and several neighborhood smoothing angles, combined with the curvature of the point in the model, a complete segmentation evaluation of each point is obtained, including the specific method of:

[0070] Get the curvature of each point in the current model (the curvature of the spatial surface, the calculation method is the existing technology, and this embodiment will not be repeated), and represent the target point as point , then the target point segmentation complete evaluation The calculation method is:

[0071]

[0072] in, represents the curvature of the target point, represents the number of points in the target point's terminating neighborhood, Indicates the number of points in the current model, and They respectively represent the minimum and maximum values ​​of several neighborhood smoothing angles of the target point.

[0073] It should be noted that, on the basis of the smaller curvature of the target point, the larger the number of points in its terminating neighborhood accounts for the number of points in the model, the more likely its terminating neighborhood is to be a completely grown interface smooth surface, otherwise it may be an area close to the edge causing uneven growth; and the minimum and maximum values ​​of several neighborhood smooth angles of the target point are compared. The closer the ratio is to 1 and the closer the two are, the smaller the range of the neighborhood smooth angle within the initial neighborhood of the target point, that is, the closer they are to the same plane, and the expanded terminating neighborhood is more likely to be a smooth surface.

[0074] Preferably, in one embodiment of the present invention, the model is segmented in combination with the termination neighborhood of each point to obtain a plurality of edges to be matched of the model, including the specific method of:

[0075] According to the above method, the termination neighborhood is obtained for each point in the current model, and the complete segmentation evaluation of each point is obtained. All points in the current model are arranged in descending order according to the size of the complete segmentation evaluation to obtain the segmentation point sequence of the current model; the first point in the segmentation point sequence and its termination neighborhood are used as a segmentation area, and several points corresponding to the segmentation area in the segmentation point sequence are removed, and the remaining points in the segmentation point sequence constitute the first segmentation point sequence (the arrangement order does not change); the first point in the first segmentation point sequence and the points in its termination neighborhood that exist in the first segmentation point sequence constitute a segmentation area, and similarly, several points corresponding to the segmentation area are removed in the first segmentation point sequence, and the remaining points in the first segmentation point sequence constitute the second segmentation point sequence; and so on, until all points in the segmentation point sequence of the current model are segmented, and several segmentation areas are obtained; the edge segments adjacent to each segmentation area are used as several edges to be matched of the current model.

[0076] At this point, after obtaining the termination area for the point, the point is quantified by combining the curvature of the point to segment the complete evaluation. The segmentation complete evaluation reflects the possibility that the termination neighborhood of the point is an interface smooth surface, and this possibility is used for sorting, thereby completing the segmentation and edge extraction of the model.

[0077] Step S004, calculating the matching distance between the edges to be matched of different models, and obtaining the edge distribution array and the matching edge array in the two models according to the size relationship of the matching distance and the position distribution between the edges to be matched in the same model; and obtaining the interface fit evaluation between the two models according to the position distribution difference of each pair of edges in the edge distribution array and the matching edge array.

[0078] It should be noted that since the connection between the two models requires a connection structure of the same shape, the model blocks that make up the connection structure have the same layout structure and can therefore fit together, so the same surface layout should have a similar edge structure; but since there may be pixel-level errors in the actual sampling and growth processes, there may not be exactly the same edges, and it is necessary to quantify the similarity of the edge structures based on DDTW matching between the edges to be matched, and then obtain the edge distribution array and matching edge array between different models.

[0079] Preferably, in one embodiment of the present invention, the matching distances between the edges to be matched of different models are calculated, and according to the size relationship of the matching distances and the position distribution between the edges to be matched in the same model, the edge distribution arrays and the matching edge arrays in the two models are obtained, including the specific method of:

[0080] Taking model a and model b as an example, for each to-be-matched edge of model a and each to-be-matched edge of model b, the matching distance is obtained by the DDTW matching algorithm, and the to-be-matched edge in model a corresponding to the minimum value of all matching distances is recorded as the edge , the edge to be matched in model b corresponding to the minimum value is recorded as edge ;

[0081] 1) Get the edge The adjacent edges to be matched on model a are recorded as edges (Since there are multiple adjacent edges to be matched, the matching distances corresponding to the edges to be matched are sorted and traversed from small to large);

[0082] 2) Place the edge The edge to be matched in model b corresponding to the minimum value among several matching distances is recorded as edge (Those that have been successfully matched will no longer participate in subsequent processing, that is, the edge is not included here );

[0083] 3) It should be noted that since the interfaces between different models should have similar surface structures, the corresponding edge positions will be restricted. With edge If they intersect, the corresponding edges Based on the edge The matched edges They should also have similar position distribution relationships, that is, the edge With edge intersect; if the two intersect, continue to the next step; if the two do not intersect, return to step 2) and extract the edge to be matched in model b corresponding to the second smallest matching distance as the edge , and continue with step 3), and so on, until an edge that meets the conditions is found ;

[0084] 4) Follow step 1) to get the edge Other adjacent edges to be matched on model a are recorded as edges Repeat steps 2) and 3) for the edge Get edges that meet the criteria ; Similarly, for the edge of model a All adjacent edges to be matched are successfully matched with the edges to be matched on model b, and then several edges to be matched are obtained;

[0085] The edge of model a and all its adjacent edges to be matched are arranged in the order traversed in the above matching process to obtain the edge distribution array of model a; all the edges to be matched in model b that are successfully matched with the edges to be matched in model a are arranged in the order of the corresponding edges to be matched in the edge distribution array of model a to obtain the matching edge array of model b.

[0086] Similarly, taking the edges to be matched in model b as a reference, the above method is used to search for edges to be matched successfully in model a, and several pairs of edges to be matched are also obtained, and further the edge distribution array of model b and the matching edge array of model a are obtained.

[0087] It should be noted that after analyzing the similarity of the morphological relationship and the position distribution of the edges to be matched, the edges to be matched in the two models are matched. After obtaining the edge distribution array and the matching edge array, it is necessary to analyze the position difference of each edge to be matched in the two arrays, and then select the edge to be matched according to its difference with the initial edge. and edge The angle difference is analyzed to obtain the interface fit evaluation.

[0088] Preferably, in one embodiment of the present invention, the interface fit evaluation between the two models is obtained according to the position distribution difference of each pair of edges in the edge distribution array and the matching edge array, and the specific method includes:

[0089] For the edge distribution array of model a and the matching edge array of model b, except for the first edge to be matched, i.e., edge With edge In addition, for any other pair of edges to be matched that belong to model a and edge Get the angle between the edge to be matched and the edge belonging to model b in the edge to be matched Get the angle, and take the absolute value of the difference between the two angles as the position difference factor of the edge to be matched; then the degree of interface fit between model a and model b is calculated based on the model a. The calculation method is:

[0090]

[0091] in, represents the number of edge pairs to be matched in the edge distribution array of model a and the matching edge array of model b, The edge distribution array of model a and the matching edge array of model b are represented by The position difference factor for the matched edges, represents the mean of the position difference factors of all edges to be matched in the edge distribution array of model a and the matching edge array of model b, Represents the standard deviation of the position difference factors of all edges to be matched in the edge distribution array of model a and the matching edge array of model b.

[0092] What needs to be explained is that by judging the distribution concentration of the position difference factors, that is, calculating the kurtosis between the position difference factors, the larger the kurtosis, the more concentrated the distribution, and the smaller the difference between the position difference factors, the closer the distribution similarity relationship between several pairs of matching edges is, and the accuracy of interface splicing between several pairs of matching edges is higher, and the interface fit is higher.

[0093] Further, for some edges to be matched in the edge distribution array of model b and the matching edge array of model a, the interface matching degree between model b and model a is obtained according to the above method. ,Will and The normalized result of the mean is used as the interface fit evaluation between model a and model b. In this embodiment, the sigmoid function is used for normalization.

[0094] At this point, after obtaining the edges to be matched, matching is performed based on the morphological similarity and position distribution relationship between the edges to be matched in different models, and the edge distribution array and the matching edge array are obtained. The position distribution differences between the edge pairs to be matched are further analyzed to obtain the interface fit evaluation, so as to quantify whether the morphology of the edges to be matched between different models is close, and then reflect whether the two models can be spliced, providing a basis for the splicing order of components in the subsequent steel structure.

[0095] Step S005: Based on the interface compatibility evaluation between different models, a splicing graph structure is constructed through threshold judgment; the graph structure is traversed to obtain the assembly order of the components in the steel structure.

[0096] Preferably, in one embodiment of the present invention, based on the interface fit evaluation between different models, the mosaic graph structure is constructed by threshold judgment, and the specific method includes:

[0097] An interface fit threshold is preset. In this embodiment, the interface fit threshold is described as 0.9. If the interface fit evaluation between any two models is greater than or equal to the interface fit threshold, the two models are successfully matched. The edge distribution array corresponding to the maximum value of the interface fit degree of the two models as the benchmark and each edge to be matched in the matching edge array are used as several matching edge groups of the two models. The two edges to be matched in the matching edge group are converted into edge forms corresponding to the matching distance selected by DDTW, that is, the edge forms formed by the matching arrangement of pixels in the DDTW matching process. According to the edge forms, the corresponding order pixels in the two edges to be matched are spatially overlapped to complete the splicing of the two models. If the interface fit evaluation between the two models is less than the interface fit threshold, the two models do not match.

[0098] Furthermore, the model of each component in the steel structure is regarded as a node in the graph structure. If the two models match successfully, the nodes corresponding to the two models are connected through undirected edges; if the two models do not match, no undirected edge connection is made between the corresponding nodes; all models are connected with undirected edges based on the matching results, and the final graph structure is recorded as a spliced ​​graph structure.

[0099] It should be noted that the interface fit evaluation is the edge matching degree between component models, which is used to judge the interface fit and splice, and then build a splicing graph structure based on the matching results; the steel structure is a layered structure, and building the upper structure will be more stable after building the underlying structure, so it is necessary to use the BFS (breadth-first search) graph traversal algorithm to start from the underlying structure, that is, the model with the most edges as the starting node, and finally obtain the assembly order.

[0100] Specifically, the number of edges to be matched in each model is counted, and the node corresponding to the model with the largest number of edges to be matched in the splicing graph structure is taken as the starting node. The BFS graph traversal algorithm is used to start from the starting node and visit adjacent nodes. The splicing order of the components corresponding to the model corresponding to the nodes is recorded in the order of visiting the nodes. Finally, the assembly order of the components in the steel structure is obtained by using the BFS graph traversal algorithm on the splicing graph structure.

[0101] So far, the models of components in the steel structure are split according to similarity growth, smooth surfaces and corresponding edge parts are screened, matching analysis and array construction are performed according to the morphology between the edges to be matched, and the rotation positioning and matching between the models of different components are realized. Finally, the splicing graph structure is constructed through the matching relationship between the models. By traversing the connection relationship between the nodes in the splicing graph structure, the assembly sequence starting from the bottom structure is realized based on the successful assembly and matching of the component models, thereby improving the matching accuracy and splicing efficiency of the component splicing in the steel structure.

[0102] Another embodiment of the present invention provides a digital simulation pre-assembly system for steel structures, which includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the above-mentioned method steps S001 to S005 are implemented.

[0103] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A digital simulation pre-assembly method for steel structure, characterized in that: The method comprises the following steps: Obtain models of several components during the pre-assembly process of the steel structure through point cloud scanning; For a plurality of points in a model of the same component, an initial neighborhood of each point is obtained, wherein the initial neighborhood of each point includes a plurality of initial neighborhood points of each point, and a plurality of neighborhood smoothing angles of each point are obtained according to the spatial position distribution of each point and its initial neighborhood points; based on the neighborhood smoothing angle of each point and the spatial position distribution of the points in the model, the initial neighborhood of each point is expanded and a terminal neighborhood of each point is obtained; According to the termination neighborhood of each point and several neighborhood smoothing angles, combined with the curvature of the point in the model, a complete segmentation evaluation of each point is obtained; combined with the termination neighborhood of each point, the model is segmented to obtain several edges of the model to be matched; Calculate the matching distance between the edges to be matched of different models, and obtain the edge distribution array and the matching edge array in the two models according to the size relationship of the matching distance and the position distribution between the edges to be matched in the same model; obtain the interface fit evaluation between the two models according to the position distribution difference of each pair of edges in the edge distribution array and the matching edge array; Based on the interface fit evaluation between different models, the splicing graph structure is constructed through threshold judgment; the graph structure is traversed to obtain the assembly order of components in the steel structure; The interface fit evaluation between the two models is obtained by: Based on the position distribution relationship of different edges to be matched in the edge distribution array of model a and the matching edge array of model b, a position difference factor of each edge to be matched in the edge distribution array of model a and the matching edge array of model b is obtained; The degree of interface compatibility between model a and model b The calculation method is: in, represents the number of edge pairs to be matched in the edge distribution array of model a and the matching edge array of model b, The edge distribution array of model a and the matching edge array of model b are represented by The position difference factor for the matched edges, represents the mean of the position difference factors of all edges to be matched in the edge distribution array of model a and the matching edge array of model b, represents the standard deviation of the position difference factors of all edges to be matched in the edge distribution array of model a and the matching edge array of model b; Get the interface compatibility between model b and model a ,Will and The normalized mean value is used as the interface fit evaluation between model a and model b.

2. A steel structure digital simulation pre-assembly method according to claim 1, characterized in that: The specific method for obtaining the several neighborhood smoothing angles of each point is as follows: The model of any component is taken as the current model, and any point in the current model is taken as the target point. The normal of the plane where the target point is located and passes through the target point is obtained, and the angle between the line connecting the target point and any of its initial neighboring points and the normal is recorded as a neighborhood smoothing angle of the target point, and several neighborhood smoothing angles of the target point are obtained.

3. A steel structure digital simulation pre-assembly method according to claim 2, characterized in that: The specific method of expanding the initial neighborhood of each point and obtaining the terminal neighborhood of each point includes: According to the maximum and minimum values ​​of several neighborhood smoothing angles of the target point, the neighborhood smoothing angle value range of the target point is obtained, the window side length of the target point is expanded, and the newly added points in the window after each expansion are connected with the target point, and an angle is formed with the normal line, which is recorded as the expanded neighborhood angle; If the expanded neighborhood angle belongs to the range of the neighborhood smoothing angle of the target point, the point corresponding to the expanded neighborhood angle is used as the neighborhood point of the target point; and so on, the window side length is continuously expanded until no new neighborhood points are generated after one expansion, and the expansion is stopped, and the area formed by the target point and its initial neighborhood points and the neighborhood points is recorded as the terminal neighborhood of the target point.

4. A steel structure digital simulation pre-assembly method according to claim 2, characterized in that: The segmentation and complete evaluation of each point is obtained by: Represent the target point as a point , target point segmentation complete evaluation The calculation method is: in, represents the curvature of the target point, represents the number of points in the target point's terminating neighborhood, Indicates the number of points in the current model, and They respectively represent the minimum and maximum values ​​of several neighborhood smoothing angles of the target point.

5. A steel structure digital simulation pre-assembly method according to claim 2, characterized in that: The method of dividing the model to obtain a plurality of edges to be matched of the model includes: Arrange all points in the current model in descending order according to the size of the complete segmentation evaluation to obtain the segmentation point sequence of the current model; take the first point in the segmentation point sequence and its terminal neighborhood as a segmentation region, and remove several points corresponding to the segmentation region in the segmentation point sequence, and the remaining points in the segmentation point sequence constitute the first segmentation point sequence; The first point in the first segmentation point sequence and the points in its termination neighborhood that exist in the first segmentation point sequence form a segmentation region, and a number of points corresponding to the segmentation region are removed from the first segmentation point sequence, and the remaining points in the first segmentation point sequence form a second segmentation point sequence; This process is repeated in this way until all points in the segmentation point sequence of the current model are segmented to obtain a number of segmentation regions; the edge segments adjacent to each segmentation region are used as a number of edges to be matched of the current model.

6. A steel structure digital simulation pre-assembly method according to claim 1, characterized in that: The edge distribution array and the matching edge array in the two models are specifically obtained by: The edge to be matched in model a corresponding to the minimum value of all matching distances is recorded as the edge , the edge to be matched in model b corresponding to the minimum value is recorded as edge ; The matching process includes: 1) Get the edge The adjacent edges to be matched on model a are recorded as edges ; 2) The edge The edge to be matched in model b corresponding to the minimum value among several matching distances is recorded as edge ; 3) If the edge With edge intersect, continue to the next step; if the edge With edge If they do not intersect, return to step 2) and extract the edge to be matched in model b corresponding to the second smallest matching distance as the edge , and continue with step 3), and so on, until an edge that meets the conditions is found ; 4) Follow step 1) to get the edge Other adjacent edges to be matched on model a are recorded as edges Repeat steps 2) and 3) for the edge Get edges that meet the criteria ; Similarly, for the edge of model a All adjacent edges to be matched are successfully matched with the edges to be matched on model b, and a number of edges to be matched are obtained; The edge of model a and all its adjacent edges to be matched are arranged in the order traversed in the above matching process to obtain the edge distribution array of model a, and all the edges to be matched in model b that are successfully matched with the edges to be matched in model a are arranged in the order of the corresponding edges to be matched in the edge distribution array of model a to obtain the matching edge array of model b; Get the marginal distribution array of model b and the matching marginal array of model a.

7. A steel structure digital simulation pre-assembly method according to claim 1, characterized in that: The specific method for obtaining the position difference factor of each edge to be matched in the edge distribution array of model a and the matching edge array of model b is as follows: For the edge distribution array of model a and the matching edge array of model b, except for the first edge to be matched, i.e., edge With edge In addition, for any other pair of edges to be matched that belong to model a and edge Get the angle between the edge to be matched and the edge belonging to model b in the edge to be matched The angle is obtained, and the absolute value of the difference between the two angles is used as the position difference factor of the edge to be matched.

8. A steel structure digital simulation pre-assembly method according to claim 1, characterized in that: The specific method of constructing the mosaic graph structure by threshold judgment is as follows: If the interface fit evaluation between any two models is greater than or equal to the interface fit threshold, the two models are successfully matched, and the edge distribution array corresponding to the maximum value of the interface fit degree of the two successfully matched models and each edge to be matched in the matching edge array are used as several matching edge groups of the two models, and the two edges to be matched in the matching edge group are converted into edge forms corresponding to the matching distance selected by DDTW. According to the edge form, the corresponding order of pixels in the two edges to be matched are overlapped in the spatial domain to complete the splicing of the two models; if the interface fit evaluation between the two models is less than the interface fit threshold, the two models do not match; The model of each component in the steel structure is regarded as a node in the graph structure. If two models are matched successfully, the nodes corresponding to the two models are connected through undirected edges. If two models do not match, no undirected edge connection is made between the corresponding nodes; all models are connected with undirected edges based on the matching results, and the resulting graph structure is recorded as a spliced ​​graph structure.

9. A digital simulation pre-assembly system for steel structures, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of a digital simulation pre-assembly method for steel structure as described in any one of claims 1-8 are implemented.

Images