Steel structure pre-assembly modeling method and system based on Internet of Things
By screening key points from point cloud data, analyzing differences, calculating structural fineness, and adjusting model accuracy, the problem of insufficient pre-assembly accuracy of steel structures in the existing technology is solved, and the modeling accuracy and assembly quality are improved.
Patent Information
- Application Number
- CN202411319508.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-23
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-09-23
AI Technical Summary
The existing pre-assembly methods of steel structures cannot effectively distinguish the fineness of different parts, resulting in insufficient modeling accuracy, affecting the model accuracy and assembly quality.
By obtaining point cloud data of steel parts to be assembled, filtering key points, analyzing the differences between key points, calculating the structural fineness, adjusting the model accuracy, and pre-assembly.
It improves the accuracy of steel structure modeling and the effectiveness of pre-assembly, ensures the accuracy of the model after assembly, and facilitates adjustment.
Smart Images

Figure CN118839414B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of 3D modeling using computer graphics, and in particular to a steel structure pre-assembly modeling method and system based on the Internet of Things. Background Art
[0002] Large-span steel structure buildings are a common form of modern architecture. Their construction involves the installation of many steel components, and the installation requires high precision. If the installation error is large, it may damage the appearance of the building or even reduce the stress performance of the internal structure of the building, which may cause safety hazards to the corresponding building structure. Therefore, in the construction process of steel structure buildings, in order to ensure the construction quality, it is usually necessary to assemble pre-processed steel parts. However, since the parts usually used in steel structure buildings are heavy, it is difficult, time-consuming and labor-intensive to directly pre-assemble the physical parts, which will increase the manufacturing cost. Therefore, the existing pre-assembly method usually establishes a three-dimensional model of a steel part based on computer software, and performs virtual pre-assembly on the computer according to the obtained three-dimensional model, thereby improving the accuracy and speed of assembly.
[0003] At present, when pre-assembling steel structures, a fixed three-dimensional model is usually established for a single steel component based on the characteristics of multiple key points (such as corner points), and it is virtually pre-assembled with the three-dimensional model of the adjacent structure on a computer to provide a reference for the installation of the steel structure project. However, since the steel structures in actual projects usually have different shapes and are relatively precise and complex, the degree of precision of different parts is different, and the required modeling accuracy is also different. The existing methods cannot distinguish them, which may lead to too many key points of simple structures being selected and the modeling accuracy being set too high, which interferes with the accuracy of the obtained model; or too few key points of precise structures being selected and the modeling accuracy is insufficient, resulting in low model accuracy. Summary of the invention
[0004] In order to solve the technical problem of low model accuracy, the present invention aims to provide a steel structure pre-assembly modeling method and system based on the Internet of Things. The technical solutions adopted are as follows:
[0005] The present invention provides a steel structure pre-assembly modeling method based on the Internet of Things, the method comprising:
[0006] Acquire point cloud data of a plurality of steel parts to be assembled, and acquire key points from the point cloud data, wherein the key points are points used to characterize local features of the steel parts to be assembled;
[0007] Obtaining a structural refinement degree according to the differences between the key points, and obtaining a model accuracy according to the structural refinement degree;
[0008] The steel parts to be assembled are adjusted and pre-assembled using the model accuracy, and the pre-assembled model is displayed.
[0009] Further, obtaining key points from the point cloud data includes:
[0010] Performing clustering processing on the point cloud data to obtain multiple result clusters;
[0011] Analyze the distribution similarity of each point in each result cluster, and determine the degree of deviation of each point relative to the result cluster it belongs to based on the distribution similarity;
[0012] The contribution degree of each point to the local feature within a preset search range is determined according to the deviation degree, and the key point is obtained by screening according to the contribution degree.
[0013] Furthermore, the analysis of the distribution similarity of each point in each result cluster includes:
[0014] In the current result cluster, the distance from each point to the cluster center is obtained, and the distance mean is calculated based on the distance from each point to the cluster center;
[0015] Determine the degree of fluctuation according to the difference between the distance of each point to the cluster center and the mean of the distance;
[0016] The distribution similarity of the current result cluster is calculated using the fluctuation degree, the distance mean and the number of points in the current result cluster.
[0017] Furthermore, determining the degree of deviation of each point relative to the result cluster in which it is located according to the degree of distribution similarity includes:
[0018] In the current result cluster, the farthest distance is obtained from the distances from each point to the cluster center, and the deviation degree of the current point is calculated based on the farthest distance, the distribution similarity and the distance from the current point to the cluster center.
[0019] Further, determining the contribution degree of each point to the local feature within a preset search range according to the deviation degree, and selecting and obtaining the key point according to the contribution degree includes:
[0020] An information acquisition step, in the current result cluster, acquiring the curvature of the current point, the laser emission intensity, the mean curvature of the remaining points in a preset search range, and the mean laser emission intensity, wherein the preset search range is a range formed by the mean distance corresponding to the current result cluster with the current point as the center and the distance mean calculated based on the distance from each point in the current result cluster to the cluster center;
[0021] a contribution degree calculation step, calculating the contribution degree of the current point to the local feature according to the deviation degree, the difference between the curvature of the current point and the mean value of the curvature, and the difference between the laser emission intensity of the current point and the mean value of the laser emission intensity;
[0022] A first repetition step, repeatedly executing the information acquisition step to the contribution degree calculation step until the contribution degree of each point in all result clusters is calculated;
[0023] In the key point screening step, the contribution degree of each point in all result clusters is normalized to obtain the normalized contribution degree, and multiple target contribution degrees are screened from the normalized contribution degrees according to a preset contribution degree threshold, and the points corresponding to the target contribution degrees are used as key points.
[0024] Furthermore, obtaining the structural refinement according to the differences between the key points includes:
[0025] a similarity group determination step, analyzing the difference between the current key point and adjacent key points, and determining a plurality of similar key points of the current key point according to the difference, wherein the current key point and the plurality of similar key points form a similarity group;
[0026] A second repetition step, repeatedly executing the similarity group determination step until similarity groups corresponding to all key points are obtained;
[0027] The structural refinement calculation step is to calculate the structural refinement of the area where the current similarity group is located according to the minimum contribution degree of the key points in the current similarity group, the average value of the difference degree, and the difference between the number of key points in the current similarity group and the total number of key points in other similarity groups;
[0028] The third repetition step is to repeatedly execute the structure refinement calculation step until the structure refinement corresponding to each similarity group is obtained.
[0029] Further, the steel parts to be assembled include the current steel parts and the steel parts to be assembled with the current steel parts;
[0030] The obtaining of the model accuracy according to the structural refinement level comprises:
[0031] A regional accuracy acquisition step, calculating the regional accuracy according to the difference between the first structural fineness and the second structural fineness, and the difference between the first fine mean and the second fine mean, wherein the first structural fineness is the structural fineness of the region where the current similar group in the current steel component is located, the second structural fineness is the structural fineness of the region where the corresponding similar group in the steel component assembled with the current steel component is located, the first fine mean is the mean of all first structural finenesses in the current steel component, and the second fine mean is the mean of all second structural finenesses in the steel component assembled with the current steel component;
[0032] A model accuracy acquisition step, calculating and obtaining the model accuracy of the area where the current similarity group in the current steel component is located according to the first structural refinement and the area accuracy;
[0033] The fourth repetition step is to repeat the region accuracy acquisition step to the model accuracy acquisition step until the model accuracy corresponding to the region where all similar groups in the current steel component are located is acquired.
[0034] Furthermore, the use of model accuracy to adjust and pre-assemble the steel parts to be assembled includes:
[0035] The model accuracy is used to adjust the original accuracy of each area in the three-dimensional model corresponding to the current steel component to obtain the adjusted accuracy, and the adjusted model is obtained based on the adjusted accuracy. The adjusted model is assembled to obtain the assembled model.
[0036] Further, after acquiring key points from the point cloud data, the method further includes:
[0037] The assembly error is calculated and displayed based on the distance between the first key point and the second key point and the difference between the first direction vector and the second direction vector, wherein the first key point is the current key point in the current steel component, the second key point is the matching key point corresponding to the current key point in the steel component assembled with the current steel component, the first direction vector is the direction vector corresponding to the tangent of the first key point, and the second direction vector is the direction vector corresponding to the tangent of the second key point.
[0038] The present invention also provides a steel structure pre-assembly modeling system based on the Internet of Things, the system comprising:
[0039] A data acquisition module, used to acquire point cloud data of the steel parts to be assembled, and to acquire key points from the point cloud data, wherein the key points are points used to characterize local features of the steel parts to be assembled;
[0040] An accuracy calculation module, used for obtaining a structural refinement according to the differences between the key points, and obtaining a model accuracy according to the structural refinement;
[0041] The assembly and display module is used to adjust and pre-assemble the steel parts to be assembled using the model accuracy, and display the pre-assembled model.
[0042] The present invention has the following beneficial effects: by screening out key points that can represent local features from point cloud data, obtaining the structural refinement of different areas based on the differences between the key points, obtaining the model accuracy according to the structural refinement, and using the model accuracy to adjust the modeling accuracy of the three-dimensional model of the steel parts to be assembled, thereby obtaining an accurate three-dimensional model, and pre-assembling the steel structure based on multiple three-dimensional models, thereby improving the accuracy of steel structure modeling and the effectiveness of pre-assembly, and facilitating the staff to make timely adjustments based on the pre-assembly model. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are 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 work.
[0044] Figure 1 A schematic diagram of a flow chart of a steel structure pre-assembly modeling method based on the Internet of Things provided by one embodiment of the present invention;
[0045] Figure 2 A schematic structural diagram of a steel structure pre-assembly modeling system based on the Internet of Things provided by another embodiment of the present invention. DETAILED DESCRIPTION
[0046] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a steel structure pre-assembly modeling method and system based on the Internet of Things proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0047] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0048] The specific scheme of the steel structure pre-assembly modeling method and system based on the Internet of Things provided by the present invention is described in detail below with reference to the accompanying drawings.
[0049] See also Figure 1, which shows a method flow chart of a steel structure pre-assembly modeling method based on the Internet of Things provided by an embodiment of the present invention. The steel structure pre-assembly modeling method based on the Internet of Things includes the following steps:
[0050] S101, obtaining point cloud data of a plurality of steel parts to be assembled, and obtaining key points from the point cloud data, where the key points are points used to characterize local features of the steel parts to be assembled.
[0051] In this step, point cloud data of the surface of the physical steel parts to be assembled can be obtained through devices such as three-dimensional laser scanners, depth cameras, and binocular cameras, and the point cloud data can be preprocessed by filtering, denoising, etc. to obtain preprocessed point cloud data. The preprocessed point cloud data includes multiple points, and then the contribution of all points to local features is analyzed, and the points with greater contribution are used as key points.
[0052] The contribution degree is calculated based on the curvature, laser emission intensity, degree of deviation, etc. of each point. When the curvature and laser emission intensity of the current point are significantly different from those of the surrounding points, the contribution degree is greater, indicating that the current point can better reflect the local feature information of the steel component than the surrounding points. The current point can be determined as a key point, for example, a point at the edge, corner or thread of a steel component, and pre-assembly is performed based on these key points.
[0053] In addition, the greater the deviation of the current point, the farther the current point deviates from the cluster center of the result cluster where it is located, and the more likely it is an outlier. Outliers cannot provide more local feature information. Therefore, outliers need to be filtered out when screening key points.
[0054] S102, obtaining a structural refinement according to the difference between the key points, and obtaining a model accuracy according to the structural refinement.
[0055] In this step, the similar key points of each key point are first determined based on the differences between the key points, and then the structural refinement is obtained based on the similar key points, and then the model accuracy is obtained using the structural refinement.
[0056] If the number of similar key points is greater and the contribution of similar key points is greater, the similarity between the similar key points and the corresponding key points will be greater, and the structural refinement of the locations of the similar key points and the corresponding key points will be greater. The greater the structural refinement, the greater the need for modeling accuracy, and then the model accuracy will also be greater.
[0057] S103, adjusting and pre-assembling the steel parts to be assembled using the model accuracy, and displaying the pre-assembled model.
[0058] In this step, a three-dimensional model of the steel parts to be assembled is constructed using model accuracy, and multiple three-dimensional models are pre-assembled to obtain and display the pre-assembled model.
[0059] The Internet of Things-based steel structure pre-assembly modeling method provided by an embodiment of the present invention screens out key points that can represent local features from point cloud data, obtains the structural refinement of different areas based on the differences between the key points, obtains the model accuracy according to the structural refinement, and uses the model accuracy to adjust the modeling accuracy of the three-dimensional model of the steel parts to be assembled, thereby obtaining an accurate three-dimensional model, and pre-assembles the steel structure based on multiple three-dimensional models, thereby improving the accuracy of steel structure modeling and the effectiveness of pre-assembly, and facilitating the staff to make timely adjustments based on the pre-assembly model.
[0060] In some embodiments of the present invention, obtaining key points from point cloud data includes:
[0061] The point cloud data is clustered to obtain multiple result clusters. Specifically, the point cloud data can be clustered by clustering algorithms such as K-means clustering, DBSCAN (Density-Based Spatial Clustering of Applications with Noise), and Gaussian mixture clustering. Preferably, the point cloud data is clustered by DBSCAN. When clustering by using the DBSCAN algorithm, the parameter neighborhood radius can be set to the mean of the Euclidean distances between all adjacent points in the point cloud data, and the minimum number of samples can be set to 5.
[0062] Analyze the distribution similarity of each point in each result cluster, and determine the degree of deviation of each point relative to the result cluster it belongs to based on the distribution similarity. Specifically, in each result cluster, analyze the distribution similarity of each point in each result cluster based on the Euclidean distance from each point to the cluster center. The greater the distribution similarity, the higher the degree of aggregation of the points in the current result cluster. Conversely, the smaller the distribution similarity, the lower the degree of aggregation of the points in the current result cluster.
[0063] The contribution of each point to the local features within the preset search range is determined according to the degree of deviation, and the key points are obtained by screening according to the contribution degree.
[0064] It should be noted that the 3D laser scanner obtains point cloud data based on the difference in the ability of different positions on the surface of the physical steel parts to be assembled to reflect the laser beam. The point cloud data includes the 3D coordinates and morphological information of the surface of the steel parts to be assembled. Since the point cloud data obtained by the laser scanner is of high accuracy, there may be redundancy in the point cloud information, and the point cloud data is disordered, making it difficult to analyze the overall point cloud data of the surface of the steel parts to be assembled. Therefore, it is necessary to simplify the point cloud data and select key points from the simplified point cloud data. Specifically, the simplification can be performed by voxel downsampling.
[0065] In some embodiments of the present invention, since the distribution of point clouds is related to the laser reflectivity and roughness of different positions when the laser scanner scans the physical steel parts to be assembled, if the laser reflectivity of a certain area of the steel parts to be assembled is relatively similar, the corresponding point cloud quantity and distribution are also relatively similar. The smoothness of each point in the area is similar and the distance from the laser scanner is similar, so the laser reflectivity of each point is similar, and the point cloud distribution in the area is relatively similar. For example, there is no smooth area on the surface of the steel parts to be assembled that has its own texture such as threads. Based on this, the distribution similarity of each point is analyzed in the following way:
[0066] In the current result cluster, the distance from each point to the cluster center is obtained, and the mean distance is calculated based on the distance from each point to the cluster center. Specifically, the Euclidean distance from each point in the result cluster to the cluster center is calculated based on the result cluster, and the mean of all Euclidean distances is calculated to obtain the mean Euclidean distance.
[0067] The degree of fluctuation is determined based on the difference between the distance from each point to the cluster center and the mean distance. Specifically, the degree of fluctuation of the current result cluster is calculated based on the difference between the Euclidean distance from each point to the cluster center and the mean Euclidean distance.
[0068] The distribution similarity of the current result cluster is calculated using the fluctuation degree, distance mean, and the number of points in the current result cluster. It is specifically calculated using the following formula:
[0069] ;
[0070] In the formula, Indicates the distribution similarity of each point in the jth result cluster, Indicates the number of points in the jth result cluster where the i-th point is located. represents the mean of the Euclidean distances from each point in the jth result cluster to the cluster center, Represents the Euclidean distance from the i-th point in the j-th result cluster to the cluster center. Indicates the fluctuation degree of the jth result cluster. The smaller the fluctuation degree, the The larger it is, the more similar the distribution of points in the jth result cluster is.
[0071] The above distribution similarity calculation is performed on each result cluster, thereby obtaining the distribution similarity of all result clusters.
[0072] In some embodiments of the present invention, due to the existence of object skewness (for example, when the laser scanner emits a laser beam, the surface tangent plane normal vector of the physical steel structure to be assembled does not coincide with the direction of the laser beam, resulting in the laser beam not being fully received when it returns) and environmental interference (for example, during the laser scanner scanning process, the ambient temperature, air pressure, etc. interfere with the beam and electromagnetic wave transmission), the laser reflection ability of certain positions in the obtained point cloud data is different from the actual laser reflection ability of the position. Therefore, the point cloud at the corresponding position usually does not follow the geometric or physical structure of the steel component. Based on this, the key points are screened by analyzing the differences in the point cloud data at different positions. The differences in the point cloud data at different positions can be expressed by the degree of deviation, specifically:
[0073] In the current result cluster, the farthest distance is obtained from the distances from each point to the cluster center, and the deviation degree of the current point is obtained based on the farthest distance, the distribution similarity and the distance from the current point to the cluster center.
[0074] Indicatively, taking each result cluster as a unit, the Euclidean distance from each point in the current result cluster to the cluster center is calculated, and the maximum value of all Euclidean distances is taken as the farthest Euclidean distance. The deviation degree of the current point is calculated based on the farthest Euclidean distance, the Euclidean distance from the current point to the cluster center, and the distribution similarity degree corresponding to the current result cluster:
[0075] ;
[0076] In the formula, Indicates the degree of deviation of the i-th point relative to its result cluster. represents the Euclidean distance from the i-th point in the j-th result cluster to the cluster center, represents the farthest Euclidean distance from each point in the jth result cluster to the cluster center. Indicates the degree of distribution similarity of the jth result cluster.
[0077] The degree of deviation at the current point When it is larger, it means that the difference between the current point i and the other points in the current result cluster j is larger, and the greater the degree of its deviation from the center of the cluster, the more it can be said that the current point i is likely to be an abnormal point. The laser reflection ability of this abnormal point is different from that of the normal structure, and it cannot reflect the physical structure of the current steel part well. When selecting key points, the degree of deviation needs to be considered to avoid using abnormal points as key points. On the contrary, when the degree of deviation of the current point is The smaller the value is, the smaller the difference between the current point i and the other points in the current result cluster j is, and the smaller the degree of its deviation from the center of the cluster is, which indicates that the current point i is likely to be a normal point, which can well reflect the physical structure of the current steel component.
[0078] The deviation degree of each point in the current result cluster is calculated through the above deviation degree calculation process, and then the deviation degree of each point in all result clusters is calculated.
[0079] In some embodiments of the present invention, in addition to the degree of deviation of each point, it is also necessary to consider the contribution of each point to the local features of the current steel component during the key point screening process. For example, points at the edge, corner or thread of the steel component to be assembled can provide enough local feature information, which has a greater contribution to pre-assembly and thus has a greater contribution degree. Based on this, this embodiment starts from the fact that the points at the edge, corner or thread of the steel component to be assembled are often quite different from their adjacent points, and calculates the contribution degree of each point. Specifically, it includes the following steps:
[0080] The information acquisition step obtains, within the current result cluster, the curvature of the current point, the laser emission intensity, the curvature mean of the remaining points within the preset search range, and the laser emission intensity mean. The preset search range is a range formed by the radius of the distance mean corresponding to the current result cluster centered at the current point. The remaining points are the points remaining within the preset search range excluding the current point. The distance mean is calculated based on the distance from each point in the current result cluster to the cluster center. More specifically, the distance mean is the Euclidean distance mean.
[0081] In this step, the curvature and laser emission intensity of all points in the preset search range are obtained by taking the result cluster as the unit, and the curvature mean of the remaining points is calculated based on the curvature of all points except the current point. , the laser emission intensity of all points except the current point is calculated to obtain the average laser emission intensity of the remaining points , where the curvature of the current point is Represents the curvature at the i-th point, the laser emission intensity at the current point represents the laser emission intensity at the i-th point.
[0082] The contribution degree calculation step calculates the contribution degree of the current point to the local feature according to the deviation degree, the difference between the curvature of the current point and the curvature mean, and the difference between the laser emission intensity of the current point and the laser emission intensity mean.
[0083] In this step, the contribution of the current point is calculated by comprehensively considering the deviation degree of the current point, the difference between the curvature of the current point and the mean curvature, and the difference between the laser emission intensity of the current point and the mean laser emission intensity. The contribution degree of the current point is calculated by the following formula:
[0084] ;
[0085] In the formula, Indicates the contribution of the i-th point to its local feature. represents the curvature at the i-th point, represents the mean curvature of the remaining points in the search range corresponding to the i-th point, represents the laser reflection intensity at the i-th point, represents the average laser reflection intensity of the remaining points in the search range corresponding to the i-th point, Indicates the degree of deviation of the i-th point relative to its result cluster.
[0086] Degree of deviation The larger it is, the greater the possibility that the i-th point is an outlier. represents the difference between the curvature of the ith point and the mean curvature, represents the difference between the laser emission intensity at the ith point and the mean laser emission intensity, and The larger it is, the greater the contribution of the i-th point to the local feature.
[0087] The first repetition step is to repeatedly execute the information acquisition step to the contribution degree calculation step until the contribution degree of each point in all result clusters is calculated.
[0088] In this step, the information acquisition step is repeated until the contribution degree is obtained. The contribution degree of all points in the current result cluster is first calculated, and then the contribution degree of each point in all result clusters is calculated.
[0089] In the key point screening step, the contribution degree of each point in all result clusters is normalized to obtain the normalized contribution degree, and multiple target contribution degrees are screened from the normalized contribution degree according to the preset contribution degree threshold, and the points corresponding to the target contribution degrees are taken as key points.
[0090] In this step, the contribution of all points is normalized to (0, 1), the normalized contribution greater than the contribution threshold is taken as the target contribution, and the point corresponding to the target contribution is taken as the key point m. The contribution threshold is set according to experience and demand, and can be set to 0.4, 0.5, 0.6, etc. Preferably, it is set to 0.5 in this embodiment.
[0091] The Internet of Things-based steel structure pre-assembly modeling method provided by the embodiment of the present invention can better retain the local feature information of the steel parts to be assembled by screening the key points through the contribution degree. It can represent the local characteristics, is more conducive to the subsequent structural refinement analysis, and further improves the assembly accuracy.
[0092] In some embodiments of the present invention, the model accuracy is affected by the structural refinement at different positions. For example, for the area with a smooth surface of a steel component, the structure at the corresponding position itself contains less detailed information, so the modeling accuracy of the area can be reduced to retain its contour framework, thereby reducing the computational complexity; but for the threaded part or the area with fine texture of the steel component, the model accuracy of the corresponding position needs to be improved to retain the detailed information of the corresponding position as much as possible. Therefore, based on the key points that can express local features obtained above, the structural refinement at different positions is further analyzed, and the structural refinement is also the required degree of modeling accuracy. Specifically, the following steps are included:
[0093] The similarity group determination step analyzes the difference between the current key point and the adjacent key point, and determines multiple similar key points of the current key point according to the difference, and the current key point and the multiple similar key points form a similarity group. Among them, the adjacent key point is the key point adjacent to the current key point.
[0094] In this step, the difference between the current key point and the adjacent key points is calculated from the dimensions of position and contribution:
[0095] ;
[0096] In the formula, Indicates the degree of difference between the mth key point and its adjacent key points, represents the Euclidean distance between two points, represents the mean of the Euclidean distances between all key points, Indicates the contribution of the mth key point to its local features. Indicates the contribution of adjacent key points to their local features. represents the distance difference between the mth key point and the adjacent key points, represents the normalized result of the difference in contribution between the mth key point and its adjacent key points, and The larger the The larger it is, the lower the similarity between the mth key point and its adjacent key points.
[0097] After calculating the degree of difference as described above for all key points adjacent to the current key point, multiple degrees of difference are obtained. The degree of difference is normalized to (0, 1), and the normalized degree of difference that is less than the preset degree of difference threshold is used as the target degree of difference, and the adjacent key points corresponding to the target degree of difference are used as similar key points of the current key point, and the current key point and multiple similar key points form a similarity group. The degree of difference threshold is set according to experience or demand, and can be set to 0.4, 0.5, 0.6, etc. Preferably, in this embodiment, the degree of difference threshold is set to 0.5.
[0098] The second repetition step is to repeat the similarity group determination step until similarity groups corresponding to all key points are obtained.
[0099] In this step, the next key point is taken as the new current key point, and the similarity group determination step is repeated until similarity groups of all key points are obtained.
[0100] The step of calculating the structural refinement is to calculate the structural refinement of the area where the current similarity group is located according to the minimum contribution degree of the key points in the current similarity group, the mean value of the difference degree, and the difference between the number of key points in the current similarity group and the total number of key points in other similarity groups.
[0101] In this step, the structural refinement of the corresponding position is determined based on the difference analysis between multiple similarity groups. Specifically, the structural refinement of each similarity group is calculated by the following formula:
[0102] ;
[0103] In the formula, Indicates the structural refinement of the region where the kth similarity group is located. represents the number of all key points in the kth similarity group where the mth key point is located, represents the mean number of all key points in the remaining similar groups, Indicates the minimum contribution of each key point in the k-th similarity group (that is, the minimum contribution degree), represents the mean value of the difference between each key point in the kth similarity group, It represents the difference between the number of all key points in the kth similarity group where the mth key point is located and the remaining similarity groups.
[0104] The more key points there are in the kth similarity group, the greater the contribution of the kth similarity group to its local features. The smaller the difference between the key points in the kth similarity group, the finer the structure of the location of the kth similarity group, and the greater the need for modeling accuracy.
[0105] The third repetition step is to repeatedly perform the structural refinement calculation step until the structural refinement corresponding to each similarity group is obtained, and the structural refinement is calculated. Normalized to (0,1), the structural refinement after normalization is , based on the normalized structural refinement Perform subsequent model accuracy calculations.
[0106] In some embodiments of the present invention, in the actual assembly process, some steel parts to be assembled may have a relatively simple overall structure and fewer key points, which makes the overall modeling of the steel parts to be assembled not very precise, resulting in precision differences in the registration area when it is registered with other high-precision steel parts to be assembled, which in turn causes the registration area to not completely match, affecting the assembly accuracy. Therefore, by analyzing the difference between the area where the kth group in the current steel part is located and the corresponding area in the steel parts to be assembled, the model accuracy of the area where the kth group in the current steel part is located is adjusted, specifically including the following steps:
[0107] The regional accuracy acquisition step calculates the regional accuracy based on the difference between the first structural fineness and the second structural fineness, and the difference between the first fine mean and the second fine mean.
[0108] The steel parts to be assembled include the current steel parts and the steel parts assembled with the current steel parts. The first structural refinement level is the structural refinement level of the area where the current similar group k in the current steel parts is located, and the second structural refinement level is the structural refinement level of the area where the current similar group k in the steel parts assembled with the current steel parts is located. The structural refinement of the area, the first refined mean is the average of all first structural refinements in the current steel component, the second fineness average It is the average of the fineness of all the secondary structures in the steel parts assembled with the current steel part.
[0109] Schematically, the regional accuracy is calculated by the following formula:
[0110] ;
[0111] In the formula, Indicates the regional accuracy corresponding to the kth similarity group of the current steel component; represents the normalized structural refinement of the region where the k-th similarity group in the current steel component is located, that is, the first structural refinement; Indicates the normalized structural refinement of the region corresponding to the similar group in the steel component assembled with the current steel component, that is, the second structural refinement; The first fine mean represents the mean of the first structural refinement levels corresponding to all similar groups in the current steel component; It represents the average of the second structural refinement levels of all corresponding similar groups in the steel components assembled with the current steel component, that is, the second refinement mean. represents the difference between the first structural refinement and the second structural refinement, is the difference between the first and second fine means, is the normalization function, The value range of is [-1,1]. The corresponding similarity group is determined based on the matching key points in the two steel parts to be assembled.
[0112] It should be noted that the kth similarity group in the current steel component and the corresponding similarity group in the steel component assembled with the current steel component are determined through key point matching. Specifically, the key points in the current steel component and the steel component assembled with the current steel component are matched by a corner point matching algorithm. The similarity groups where the matched key points in the two steel components are located are the kth similarity group in the current steel component and the corresponding similarity group in the steel component assembled with the current steel component.
[0113] The model accuracy acquisition step calculates and obtains the model accuracy of the area where the current similarity group in the current steel component is located according to the first structural refinement and the area accuracy.
[0114] ;
[0115] In the formula, represents the model accuracy of the region where the kth similarity group in the current steel component is located, Indicates the first structural refinement level of the region where the kth similarity group is located in the current steel component, Indicates the regional accuracy corresponding to the kth similarity group in the current steel component.
[0116] The fourth repetitive step is to repeat the regional accuracy acquisition step to the model accuracy acquisition step until the model accuracy corresponding to the region where all similar groups in the current steel component are located is acquired.
[0117] After obtaining the model accuracy, the model accuracy is used to adjust the original accuracy of each area in the three-dimensional model corresponding to the current steel component to obtain the adjusted accuracy, and the corresponding areas are adjusted based on the adjusted accuracy to obtain the adjusted model. The adjusted model is assembled to obtain the assembled model and the assembled model is displayed. The staff can use the assembled model to predict the effect of the physical steel component to be assembled after assembly, and can also adjust the physical steel component to be assembled according to the assembled model.
[0118] The adjusted accuracy can specifically be the product of the original accuracy and the model accuracy. The adjusted accuracy makes the modeling of the corresponding area more refined, which can avoid the difference between the steel parts with lower original accuracy and the steel parts with higher original accuracy, so that the registration areas between the two steel parts can match, thereby improving the accuracy of splicing.
[0119] In the actual virtual splicing process, information such as model accuracy, key points, and similar groups can be directly input into the 3D modeling software to directly obtain the adjusted model, and virtual splicing between the adjusted models can be performed in the 3D modeling software.
[0120] In some embodiments of the present invention, after acquiring key points from the point cloud data, the method further includes:
[0121] The assembly error is calculated and displayed based on the distance between the first key point and the second key point, and the difference between the first direction vector and the second direction vector. The first key point is the current key point in the current steel component. The second key point is the matching key point in the steel part assembled with the current steel part corresponding to the current key point , the first direction vector is the direction vector corresponding to the tangent line of the first key point, and the second direction vector is the direction vector corresponding to the tangent line of the second key point.
[0122] Specifically, the second key point is matched with the first key point through the key point matching algorithm. Compared with other key points in the steel component assembled with the current steel component, the second key point is closest to the first key point. The Euclidean distance between the first key point and the second key point is calculated, and the first direction vector and the second direction vector corresponding to the first key point and the second key point are obtained, and then the assembly error is calculated:
[0123] ;
[0124] In the formula, Indicates assembly error, Indicates the first key point With the second key point The Euclidean distance between Represents the direction vector of the tangent line of the first key point, that is, the first direction vector, The direction vector representing the tangent line of the second key point, that is, the second direction vector. It represents the difference between the first direction vector and the second direction vector, that is, the vector angle between the first direction vector and the second direction vector. The larger the vector angle, the greater the assembly error. The larger the value is. The direction vector of the key point tangent line refers to the tangent line to the position of the key point, and the direction along the tangent line pointing to the positive half axis of the x-axis is recorded as the direction vector of the current tangent line. The key point matching algorithm can be a corner point detection algorithm.
[0125] Assembly error The larger the size, the greater the defect in the splicing. The staff can adjust the size of the splicing error according to the splicing error. Adjustments such as grinding are performed on the physical steel parts to be assembled to improve the size, shape, etc. of the physical steel parts to be assembled, so that the physical steel parts to be assembled have better assembly accuracy when they are assembled according to the assembled model.
[0126] In another embodiment of the present invention, the present invention also provides a steel structure pre-assembly modeling system based on the Internet of Things, such as Figure 2 As shown, the system includes:
[0127] The data acquisition module 201 is used to acquire point cloud data of the steel parts to be assembled, and to acquire key points from the point cloud data. The key points are points used to characterize local features of the steel parts to be assembled.
[0128] The accuracy calculation module 202 is used to obtain the structural refinement according to the differences between the key points, and obtain the model accuracy according to the structural refinement.
[0129] The assembly and display module 203 is used to adjust and pre-assemble the steel parts to be assembled using the model accuracy, and to display the pre-assembled model.
[0130] The Internet of Things-based steel structure pre-assembly modeling system provided in this embodiment corresponds to the above-mentioned Internet of Things-based steel structure pre-assembly modeling method, which will not be repeated here.
[0131] In summary, the Internet of Things-based steel structure pre-assembly modeling method and system provided by the embodiments of the present invention screens out key points that can represent local features from point cloud data, obtains the structural refinement of different areas based on the differences between the key points, obtains the model accuracy according to the structural refinement, and uses the model accuracy to adjust the modeling accuracy of the three-dimensional model of the steel parts to be assembled, thereby obtaining an accurate three-dimensional model, and pre-assembles the steel structure based on multiple three-dimensional models, and obtains the assembly error through the differences between the key points, thereby improving the accuracy of steel structure modeling and the effectiveness of pre-assembly, and facilitating the staff to adjust the physical steel structure to be assembled according to the assembly error in a timely manner.
[0132] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0133] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.
Claims
1. A steel structure pre-assembly modeling method based on the Internet of Things, characterized in that: The method comprises: Acquire point cloud data of a plurality of steel parts to be assembled, and acquire key points from the point cloud data, wherein the key points are points used to characterize local features of the steel parts to be assembled; Obtaining a structural refinement degree according to the differences between the key points, and obtaining a model accuracy according to the structural refinement degree; Using the model accuracy, the steel parts to be assembled are adjusted and pre-assembled, and the pre-assembled model is displayed; The obtaining of the structural refinement according to the differences between the key points comprises: a similarity group determination step, analyzing the difference between the current key point and adjacent key points, and determining a plurality of similar key points of the current key point according to the difference, wherein the current key point and the plurality of similar key points form a similarity group; A second repetition step, repeatedly executing the similarity group determination step until similarity groups corresponding to all key points are obtained; The structural refinement calculation step is to calculate the structural refinement of the area where the current similarity group is located according to the minimum contribution degree of the key points in the current similarity group, the average value of the difference degree, and the difference between the number of key points in the current similarity group and the total number of key points in other similarity groups; A third repetition step, repeatedly executing the structure refinement calculation step until the structure refinement corresponding to each similarity group is obtained; The steel parts to be assembled include the current steel parts and the steel parts to be assembled with the current steel parts; The obtaining of the model accuracy according to the structural refinement level comprises: A regional accuracy acquisition step, calculating the regional accuracy according to the difference between the first structural fineness and the second structural fineness, and the difference between the first fine mean and the second fine mean, wherein the first structural fineness is the structural fineness of the region where the current similar group in the current steel component is located, the second structural fineness is the structural fineness of the region where the corresponding similar group in the steel component assembled with the current steel component is located, the first fine mean is the mean of all first structural finenesses in the current steel component, and the second fine mean is the mean of all second structural finenesses in the steel component assembled with the current steel component; A model accuracy acquisition step, calculating and obtaining the model accuracy of the area where the current similarity group in the current steel component is located according to the first structural refinement and the area accuracy; A fourth repetition step, repeating the region accuracy acquisition step to the model accuracy acquisition step until the model accuracy corresponding to the region where all similar groups in the current steel component are located is acquired; The method of adjusting and pre-assembling the steel parts to be assembled by using the model accuracy includes: The model accuracy is used to adjust the original accuracy of each area in the three-dimensional model corresponding to the current steel component to obtain the adjusted accuracy, and the adjusted model is obtained based on the adjusted accuracy. The adjusted model is assembled to obtain the assembled model.
2. The steel structure pre-assembly modeling method based on the Internet of Things according to claim 1 is characterized in that: The acquiring key points from the point cloud data comprises: Performing clustering processing on the point cloud data to obtain multiple result clusters; Analyze the distribution similarity of each point in each result cluster, and determine the degree of deviation of each point relative to the result cluster it belongs to based on the distribution similarity; The contribution degree of each point to the local feature within a preset search range is determined according to the deviation degree, and the key point is obtained by screening according to the contribution degree.
3. The steel structure pre-assembly modeling method based on the Internet of Things according to claim 2 is characterized in that: The analysis of the distribution similarity of each point in each result cluster includes: In the current result cluster, the distance from each point to the cluster center is obtained, and the distance mean is calculated based on the distance from each point to the cluster center; Determine the degree of fluctuation according to the difference between the distance of each point to the cluster center and the mean of the distance; The distribution similarity of the current result cluster is calculated using the fluctuation degree, the distance mean and the number of points in the current result cluster.
4. The steel structure pre-assembly modeling method based on the Internet of Things according to claim 2 is characterized in that: Determining the degree of deviation of each point relative to the result cluster to which it belongs based on the degree of distribution similarity includes: In the current result cluster, the farthest distance is obtained from the distances from each point to the cluster center, and the deviation degree of the current point is calculated based on the farthest distance, the distribution similarity and the distance from the current point to the cluster center.
5. The method for pre-assembly modeling of steel structures based on the Internet of Things according to claim 2, characterized in that: Determining the contribution degree of each point to the local feature within a preset search range according to the deviation degree, and selecting and obtaining the key point according to the contribution degree, includes: An information acquisition step, in the current result cluster, acquiring the curvature of the current point, the laser emission intensity, the mean curvature of the remaining points in a preset search range, and the mean laser emission intensity, wherein the preset search range is a range formed by the mean distance corresponding to the current result cluster with the current point as the center and the distance mean calculated based on the distance from each point in the current result cluster to the cluster center; a contribution degree calculation step, calculating the contribution degree of the current point to the local feature according to the deviation degree, the difference between the curvature of the current point and the mean curvature, and the difference between the laser emission intensity of the current point and the mean laser emission intensity; A first repetition step, repeatedly executing the information acquisition step to the contribution degree calculation step until the contribution degree of each point in all result clusters is calculated; In the key point screening step, the contribution degree of each point in all result clusters is normalized to obtain the normalized contribution degree, and multiple target contribution degrees are screened from the normalized contribution degrees according to a preset contribution degree threshold, and the points corresponding to the target contribution degrees are used as key points.
6. The steel structure pre-assembly modeling method based on the Internet of Things according to claim 1 is characterized in that: After acquiring key points from the point cloud data, the method further includes: The assembly error is calculated and displayed based on the distance between the first key point and the second key point and the difference between the first direction vector and the second direction vector, wherein the first key point is the current key point in the current steel component, the second key point is the matching key point corresponding to the current key point in the steel component assembled with the current steel component, the first direction vector is the direction vector corresponding to the tangent of the first key point, and the second direction vector is the direction vector corresponding to the tangent of the second key point.
7. A steel structure pre-assembly modeling system based on the Internet of Things, characterized in that: The system is used to implement the steps of a steel structure pre-assembly modeling method based on the Internet of Things as described in any one of claims 1 to 6, and the system includes: A data acquisition module, used to acquire point cloud data of the steel parts to be assembled, and to acquire key points from the point cloud data, wherein the key points are points used to characterize local features of the steel parts to be assembled; An accuracy calculation module, used for obtaining a structural refinement according to the differences between the key points, and obtaining a model accuracy according to the structural refinement; The assembly and display module is used to adjust and pre-assemble the steel parts to be assembled using the model accuracy, and display the pre-assembled model.
Citation Information
Patent Citations
Steel structure factory building pre-assembling method based on BIM
CN107273607A
Steel bridge simulation pre-assembly method integrating ground type laser scanning and handheld type laser scanning
CN114332402A
Cited By
Three-dimensional model establishment method applied to plane art
CN121259246A
