CN-119942344-B - 3D box Liang Dianyun data topological structure identification method based on continuous coherence

CN119942344BCN 119942344 BCN119942344 BCN 119942344BCN-119942344-B

Abstract

The invention discloses a 3D box Liang Dianyun data topological structure identification method based on continuous coherence, which comprises the steps of S1, obtaining point clouds of box girders, preprocessing the point clouds to generate a three-dimensional data set, S2, carrying out continuous coherence analysis according to the three-dimensional data set to obtain topological features of the box girders, S3, analyzing the topological features of the box girders to mark out areas with structural problems, and S4, carrying out physical interpretation on the topological features according to the marked areas to generate a health evaluation report. By adopting a continuous coherent technology and combining topological analysis, noise robustness, geometric mechanics interpretation and multi-scale feature capture of point cloud data, the invention aims to realize multi-scale, robust and highly-interpretative topological structure analysis of box Liang Dianyun data, can comprehensively capture complex topological features of box girders under different scales, overcomes the defects of the traditional geometric analysis method, and has higher robustness and reliability especially when the problems of noise, uneven point cloud data and the like are processed.

Inventors

WANG HUAN
JI JINGJING
ZHANG KAIKAI
FENG RENJIE
Yong Yifeng
ZHANG JIANHUI
CHEN ZHENGBIN
ZHANG TENGYUN
WANG FAN
LU HONGCHUN
QIU HAO
FENG XU
LI ZHILIN
YU JIE

Assignees

中国水利水电第七工程局有限公司

Dates

Publication Date: 20260512
Application Date: 20250115

Claims (5)

1. The 3D box Liang Dianyun data topological structure identification method based on continuous coherence is characterized by comprising the following steps of: S1, acquiring point clouds of a box girder, preprocessing the point clouds to generate a three-dimensional data set, wherein the method for preprocessing the point cloud data comprises the following steps of: S11, calculating the neighborhood density of each point in the point cloud, and eliminating abnormal points with the neighborhood density lower than a neighborhood density threshold value in the point cloud; s12, registering the point cloud with abnormal points removed; S13, reducing the volume of the point cloud after registration by adopting a coarsening data granularity method, dividing the whole data set by a plurality of point sets, wherein the centroid of each point set represents the point sets, and generating a three-dimensional data set, wherein the S13 specifically comprises the following steps: s131, taking the point cloud as a point set to be divided; S132, acquiring two points with the farthest distance in the point set, establishing a first subset and a second subset according to the acquired points, and distributing the rest points in the point set to the subsets nearby; s133, calculating the compactness degree of the point set, the first subset and the second subset, and the compactness degree The expression of (2) is specifically: Wherein m is the number of points in the point set D, p j is the points in the point set D, and C is the centroid of the point set D; judging whether a splitting condition is satisfied, wherein the splitting condition is specifically as follows: And is also provided with In the formula, To the extent of compactness of the first subset, To the extent of compactness of the second subset, For the degree of compactness of the point set, The total number of points of the point set D is n, and the total number of points of the point cloud is n; If yes, the first subset and the second subset are used as point sets to be divided, the S132 is returned, if not, the current point set is used as the simplified point cloud, and the S134 is entered; s134, establishing a three-dimensional data set according to all the simplified point clouds; s2, performing continuous coherent analysis according to the three-dimensional data set to obtain topological characteristics of the box girder; the dimensions of the topological feature include 0, 1, and 2 dimensions; Wherein, 0-dimensional topological feature represents a connected branch in the point cloud, 1-dimensional topological feature represents a closed loop or a hole, and 2-dimensional topological feature represents a cavity or a closed three-dimensional space; s3, analyzing topological characteristics of the box girder and marking out a region with structural problems, wherein the method for analyzing the topological characteristics of the box girder comprises the following steps: stress analysis is carried out on the box girder through finite element analysis, stress distribution of different areas under the action of external force is calculated, and the superposition degree of the topological feature and the stress concentration area is identified through comparing the stress distribution with the topological feature; Aiming at a geometric structure area with topological features, fatigue life evaluation is carried out by combining a mechanical model, the stress state and load history of the geometric structure area are analyzed, and the fatigue failure possibility of the geometric structure area under the action of long-time cyclic load is evaluated S4, performing physical interpretation on the topological features according to the marked areas to generate a health evaluation report, wherein the method for performing physical interpretation on the topological features comprises crack and breakage detection and cavity and material defect identification; the method for detecting the cracks and the breakage comprises the following steps: Detecting connectivity interruption or a new communication branch, which indicates that a local area of the box girder is broken or damaged, and analyzing the influence of the communication branch on the integral performance of the box girder through a geometric structure and mechanics so as to guide maintenance work; The method for identifying the material defects comprises the following steps: And extracting a 2-dimensional topological characteristic hollow cavity structure, and determining the potential threat of the cavity structure to structural safety by continuously and synchronously analyzing the specific position and shape of the cavity structure and combining mechanical analysis.
2. The method for identifying a 3D box Liang Dianyun data topology based on continuous synchronization according to claim 1, wherein in S11, the neighborhood density of the i-th point is The expression of (2) is specifically: In the formula, For the euclidean distance between the i-th point and the j-th point, For the selected cutoff distance.
3. The method for identifying a 3D box Liang Dianyun data topology based on continuous coherence according to claim 1, wherein S2 includes the following sub-steps: s21, calculating Euclidean distances between points based on each point of the three-dimensional data concentrated point cloud, and generating a distance matrix; S22, setting a neighborhood radius, creating a durable bar graph, analyzing the durability of the topological feature generated by each point cloud, and obtaining the topological feature of the box girder.
4. The method for identifying a 3D box Liang Dianyun data topology based on continuous synchronization according to claim 3, wherein the step S22 is specifically: Setting a neighborhood radius, gradually increasing the neighborhood radius under a continuous coherent framework, calculating topological features under the neighborhood radius of different scales, creating a durable bar graph according to the life cycle of the topological features, and reserving the topological features with the life cycle exceeding 10% of the whole time length as the topological features of the box girder; The neighborhood radius is used for determining whether the points are regarded as connected or not, and in response to the Euclidean distance between the two points being smaller than the neighborhood radius, the two points are connected in the topological space; the lifecycle is in particular the time at which the topological feature appears and disappears.
5. The method for identifying a 3D box Liang Dianyun data topology based on continuous coherence according to claim 1, wherein in S4, the health assessment report includes the whole connectivity of the box girder, local damage and areas requiring major monitoring or maintenance.

Description

3D box Liang Dianyun data topological structure identification method based on continuous coherence Technical Field The invention belongs to the field of data structure analysis, and particularly relates to a 3D box Liang Dianyun data topological structure identification method based on continuous coherence. Background In bridge engineering, box girders are used as key load-bearing members in bridges, buildings and infrastructures, and the structural stability and deformation conditions of the box girders are directly related to the safety of the whole engineering. The geometry of the box girder has a decisive influence on its load-carrying capacity and the mechanical response under load. Therefore, the comprehensive understanding and the accurate analysis of the structural state, the deformation condition and the potential damage of the box girder are important preconditions for guaranteeing the engineering safety. Along with the continuous development of 3D scanning, remote sensing technology and computer vision technology, 3D point cloud data are widely applied to multiple fields such as bridge detection, building modeling, medical imaging and the like. The 3D point cloud data can accurately capture the geometry of the object, providing rich spatial information, which enables engineers to develop detailed structural analysis based on the point cloud data. The analysis can help evaluate the structural health condition of the box girder, and can also identify the problems of deformation, damage, abrasion and the like of the box girder, thereby providing basis for subsequent quality evaluation, maintenance detection and design optimization. At present, the existing point cloud data analysis technology is mainly divided into analysis based on geometric features and a method based on machine learning. Geometric feature-based methods, such as KD-trees, quadtrees, etc., typically analyze the spatial distribution of point cloud data through local features and distance metrics. However, such methods rely mainly on local geometric information, and under complex stress environments, it is difficult to capture changes in the overall structure of the box girder. Especially when multi-scale topological features are involved, existing geometric analysis techniques tend to be frustrating. This means that in the face of complex structural deformations or potential damages, there may be limitations on the results of the geometric feature analysis, which do not fully reflect the actual situation of the box girder under various loads. On the other hand, machine learning methods, particularly deep learning techniques such as PointNet and PointCNN, are excellent in feature extraction and classification of point cloud data in recent years. The method can automatically extract useful features in the point cloud data through the end-to-end learning framework, and reduces the burden of manual feature engineering. However, machine learning approaches also face some challenges. First, deep learning models typically require a large amount of high quality training data, and in bridge engineering, it can be relatively difficult to obtain a comprehensive and precisely labeled point cloud data set. Second, while these models perform well in feature extraction, they lack the ability to directly interpret key topological features in the box girder structure, making it difficult to reveal topological relationships in point cloud data, such as connectivity, holes, etc. In summary, the prior art has a common limitation in the analysis of the point cloud data, that is, they rely mainly on geometric metrics, and it is difficult to effectively capture topology information in the box Liang Dianyun data. In particular, in the context of multi-scale analysis and complex structures, it is difficult for both geometric analysis methods and machine learning models to distinguish between true structural features and noise. Furthermore, they lack the ability to identify important topologies in the box girder, which is particularly important in assessing the overall structural stability of the box girder. Therefore, there is a need to introduce more comprehensive analysis means, such as a topology data analysis method based on continuous coherence, to overcome the limitations of these prior art techniques, especially in terms of multi-scale topology feature detection of complex structures. Disclosure of Invention Aiming at the defects in the prior art, the method for identifying the 3D box Liang Dianyun data topological structure based on continuous coherence solves the problems that multi-scale topology is difficult to identify, noise is sensitive and explanatory is poor in the existing 3D box Liang Dianyun data analysis. In order to achieve the aim of the invention, the technical scheme adopted by the invention is that the 3D box Liang Dianyun data topological structure identification method based on continuous coherence comprises the following steps: