CN116883623A

CN116883623A - Bridge floor roughness model determining method based on three-dimensional laser point cloud scanning technology

Info

Publication number: CN116883623A
Application number: CN202310900620.XA
Authority: CN
Inventors: 高康; 王诚; 杨震; 孙逊; 吴刚
Original assignee: Southeast University
Current assignee: Southeast University
Priority date: 2023-07-21
Filing date: 2023-07-21
Publication date: 2023-10-13

Abstract

The invention discloses a bridge floor roughness model determining method based on a three-dimensional laser point cloud scanning technology, which comprises the following steps of 1, determining a bridge floor scanning area; step 2, setting a target ball; step 3, scanning and obtaining bridge surface point cloud data; step 4, preprocessing bridge deck point cloud data; step 5, obtaining a normalized integral bridge surface point cloud matrix; and 6, establishing a bridge deck roughness model. According to the invention, the bridge floor roughness model is accurately and rapidly built according to actual conditions, the defect that the existing international standard ISO8608 bridge floor roughness model lacks authenticity is overcome, and the built model can be integrated with the bridge model into a whole bridge model with real bridge floor roughness. The invention can accurately establish the bridge floor roughness model especially aiming at the bridge floor with serious damage, can reconstruct the three-dimensional space form of the bridge floor completely and with high precision, is used for real bridge analysis and verification to timely discover the problems of apparent deformation, internal damage and the like of the in-service bridge, and provides a new thought for the daily operation and management of the bridge.

Description

Bridge floor roughness model determining method based on three-dimensional laser point cloud scanning technology

Technical Field

The invention relates to the field of bridge structure detection, in particular to a bridge deck roughness model determining method based on a three-dimensional laser point cloud scanning technology.

Background

The bridge is an important component of the traffic system in China, and is particularly important for safety monitoring. At present, the numerical simulation analysis by establishing a bridge model is an important method for evaluating the safety of a bridge, and how to establish the bridge model which accords with actual engineering is a precondition for accurately evaluating the health condition of a bridge. However, how to accurately simulate the bridge deck roughness is a big problem, especially in the field of detecting bridge parameters by using a car-bridge coupling method, the bridge deck roughness is an important influencing parameter.

At present, the traditional building mode of bridge deck roughness is simulated by using the international standard ISO 8608, and the method randomly generates road surface roughness of different grades by using power spectral density, and the generated road surface roughness is close to most application scenes, but has certain difference with reality, and particularly has larger difference on the road surface with serious damage. The damage class of bridge pavement is divided into three stages: slight, medium and severe, the standard simulation of deck roughness is only applicable to undamaged and slightly damaged deck conditions. For moderately damaged and severely damaged decks, if the roughness model is built using specifications, larger errors will be created, thereby affecting the final analysis results.

In recent decades, laser scanning technology has been fully developed, and three-dimensional laser point cloud scanning technology has become a fast way of acquiring spatial data. The three-dimensional laser scanning system not only can rapidly acquire three-dimensional point cloud data with abundant details, high density and high precision of a large-scale scene, but also can obtain more accurate point cloud data information by combining secondary development processing with other software for the output point cloud data, and is used for engineering practice.

Therefore, thanks to the development of laser scanning technology, a new method for establishing a bridge floor roughness model is possible.

Disclosure of Invention

Aiming at the defects of the prior art, the invention provides a bridge floor roughness model determining method based on a three-dimensional laser point cloud scanning technology, which extracts bridge surface point cloud data through a three-dimensional laser scanner, further preprocesses the point cloud data to extract point cloud elevation data information, and then uses finite element software to establish a bridge floor roughness model according to the elevation data.

In order to solve the technical problems, the invention adopts the following technical scheme:

A bridge floor roughness model determining method based on a three-dimensional laser point cloud scanning technology comprises the following steps.

Step 1, determining a bridge deck scanning area: and determining a bridge deck scanning area and setting an area boundary according to the bridge deck section to be measured.

Step 2, setting a target ball: segmenting the bridge deck scanning area determined in the step 1 along the longitudinal direction to form a plurality of bridge deck measurement segments; at least one measuring station and a plurality of target balls are distributed in each bridge deck measuring section, and at least 3 identical target balls can be scanned by two adjacent measuring stations.

And step 3, scanning and acquiring the bridge surface point cloud data, which specifically comprises the following steps.

And 3-1, judging the damage degree of each bridge deck measurement section.

Step 3-2, determining scanning parameters: determining scanning parameters of a three-dimensional scanner in a corresponding measuring station according to the bridge deck damage degree of the bridge deck measurement section obtained in the step 3-1, wherein the scanning parameters comprise scanning precision and scanning time;

step 3-3, scanning: starting from a first measuring station, the three-dimensional scanner scans 360 degrees according to the scanning parameters determined in the step 3-2; and so on until the three-dimensional scanning in all stations is completed.

Step 3-4, extracting point cloud data: the three-dimensional scanner takes the three-dimensional scanner as an origin, and further obtains the space relative coordinates (x, y, z) of each scanning point relative to the origin; wherein, the x direction is the transverse bridge direction, the y direction is the longitudinal bridge direction, and the z direction is the bridge deck elevation direction.

Step 4, preprocessing bridge surface point cloud data: adopting a point cloud simplifying algorithm combining a 3D-SIFT feature point extraction algorithm and an octree algorithm to reduce noise and render all the bridge surface point cloud data acquired in the step 3; and then, integrating all the bridge surface point cloud data after noise reduction and rendering into the same coordinate system.

Step 5, obtaining a normalized integral bridge surface point cloud matrix: importing the whole bridge surface point cloud data after the pretreatment in the step 4 into point cloud software for processing, normalizing the bridge floor elevation in the whole bridge surface point cloud data by taking a pavement plane as a reference to obtain a normalized relative elevation z ', thus obtaining normalized whole bridge surface point Yun Juzhen (x, y, z') in txt format, and recording the number of rows and columns of a normalized whole bridge surface point cloud matrix.

Step 6, building a bridge deck roughness model: and (5) importing the normalized integral bridge surface point cloud matrix obtained in the step (5) into finite element software, and establishing a bridge floor roughness model.

In step 1, the region boundaries include a transverse boundary and a longitudinal boundary, and the specific setting method is as follows.

Step 1-1, setting a transverse boundary: the transverse boundary includes a start line and a finish line, and the start line and the finish line positions of the bridge deck scanning area are identified by using lines at the bridge deck cross section center line positions.

Step 1-2, setting a longitudinal boundary: the longitudinal boundary comprises an upper boundary line and a lower boundary line, and the set distance positions of the two sides of the bridge deck scanning area to the inner side are respectively marked by lines and marked as the upper boundary line and the lower boundary line.

In step 2, the method for setting the target ball specifically includes the following steps.

Step 2-1, determining a segmentation line: the interface between two adjacent deck measurement segments is defined as the segment line.

Step 2-2, setting a measuring station: one station is provided at each center of the start line, each segment line, and the finish line.

Step 2-3, setting a target ball: a target ball is arranged at the joint point of the starting line, each segment line and the finishing line with the upper boundary line and the lower boundary line, each station measuring position and the centers of the upper boundary line and the lower boundary line on two sides of each bridge deck measuring segment.

In step 3-2, bridge deck damage levels for measuring segments of bridge deck include mild damage, moderate damage and severe damage; when the damage is slight damage or medium damage, the scanning precision is set to be +/-2-5 mm, and the scanning time is not more than 20 minutes; when the damage is serious, the precision of the scanner is set to +/-1-2 mm, and the scanning time is not more than 30 minutes.

In the step 4, a point cloud simplifying algorithm is adopted, and the method for carrying out noise reduction and rendering on bridge deck point cloud data comprises the following steps of.

Step 4-1, obtaining an initial point cloud picture: and (3) performing de-duplication processing on all the bridge surface point cloud data obtained in the step (3) to obtain an initial point cloud image.

Step 4-2, extracting a strong characteristic point cloud picture: extracting strong characteristic point cloud pictures in the initial point cloud pictures by adopting a 3D-SIFT characteristic point extraction algorithm; wherein the strong characteristic point cloud image is a point cloud with a characteristic curvature of not less than 0.1.

Step 4-3, secondary de-duplication: and (3) removing the strong characteristic point cloud image extracted in the step (4-1) from the initial point cloud image to obtain a weak characteristic point cloud image.

Step 4-4, octree filtering: and (3) filtering the weak feature point cloud image obtained in the step (4-3), and amplifying the weak feature point cloud information in the weak feature point cloud image to obtain a filtered and amplified weak feature point cloud image.

Step 4-5, constructing a preprocessing point cloud picture: and (3) combining the strong characteristic point cloud image in the step 4-2 and the filtering amplification weak characteristic point cloud image in the step 4-4 to obtain a preprocessed point cloud image.

In step 4-2, when the 3D-SIFT feature point extraction algorithm is adopted, and the feature point cloud intensity is larger than but close to the point cloud intensity set value, a strong feature point cloud image in the initial point cloud image is extracted by increasing the Gaussian convolution scale factor sigma value.

In step 4-2, when the 3D-SIFT feature point extraction algorithm is adopted to extract the key point clouds, if the Gaussian convolution scale factor sigma is determined, multiplying the Gaussian convolution scale factor sigma by a multiplication factor k with different scales, so as to determine the number of the extracted key point clouds; the greater the k value, the greater the number of key point clouds extracted will be.

In step 4-4, the side length of the subcubes is increased by introducing a proportion coefficient alpha and a proportion factor epsilon, so that weak characteristic point cloud information is amplified; wherein, the value range of alpha is 1.0-2.0, and the value range of epsilon is 1.0-1.3.

In the step 6, the method for establishing the bridge deck roughness model comprises the following steps of.

Step 6-1, adjusting an output format: outputting normalized integral bridge surface point cloud data (x, y, z') into txt format; the first, second and third columns of the normalized integral bridge surface point cloud data in txt format are respectively corresponding to x, y and z' coordinates of each acquisition point.

And 6-2, establishing bridge deck reference datum points of roughness point cloud data, which specifically comprise the following steps.

Step 6-2-1, establishing a three-dimensional bridge entity model: in finite element software, according to the size information of a bridge deck section to be measured in a bridge design drawing, a three-dimensional bridge entity model without surface roughness and with equal proportion is established; the length and the width of the three-dimensional bridge solid model are sequentially the scanning total length and the scanning total width of a bridge deck scanning area; the height of the three-dimensional bridge solid model is the sum of the height of the bridge deck and the thickness of the concrete layer; and then, selecting the upper surface of the three-dimensional bridge solid model as a bridge deck roughness datum reference surface, and setting the midpoint of the leftmost side line of the bridge deck roughness datum reference surface as the origin of coordinates of the bridge deck roughness datum reference surface.

Step 6-2-2, mesh division: performing grid division on the three-dimensional bridge entity model established in the step 6-2-1; the transverse number of the grids is the number of columns of the normalized integral bridge surface point cloud matrix minus one, and the longitudinal number of the grids is the number of rows of the normalized integral bridge surface point cloud matrix minus one; each grid is a bridge deck unit, and four corner points of each grid correspond to four bridge deck unit nodes respectively.

Step 6-2-3, establishing a bridge deck unit node set: and establishing all bridge deck unit nodes contained in the bridge deck roughness base reference surface after grid division as a bridge deck unit node set-1, establishing a Job Job-1, and selecting 'Write Input' to obtain numbers and coordinates inp files I of all bridge deck units and bridge deck unit nodes in the set-1.

And 6-3, creating an inp file II containing the roughness elevation coordinate, which specifically comprises the following steps.

Step 6-3-1, reading the initial row position of the inp file one: and reading the initial row position of the bridge deck unit and the initial row position of the bridge deck unit node in the inp file I.

Step 6-3-2, reading bridge deck node numbers: and (3) respectively reading the number of each bridge deck unit node aiming at all bridge deck unit nodes at the starting row positions read in the step (6-3-1).

And 6-3-3, extracting coordinates of all bridge deck unit joints.

Step 6-3-4, coordinate substitution: and (3) reading bridge deck node numbers according to the step 6-3-2 by using the normalized integral bridge deck points Yun Juzhen in txt format, and replacing corresponding coordinates in the step 6-3-3 with coordinates in a normalized integral bridge deck point cloud matrix so as to form an inp file II containing roughness elevation coordinates.

In step 5, the normalization method of the relative elevation z' comprises the following steps:

and 5-1, establishing a pavement plane equation, specifically.

0＝d ₀ +d ₁ x+d ₂ y-z

Wherein d ₀ 、d ₁ 、d ₂ And e is a plane equation coefficient respectively, and can be obtained by solving a matrix method.

Step 5-2, normalizing the elevation to form a relative elevation z ', wherein the calculation formula of the relative elevation z' is as follows:

the invention has the following beneficial effects:

1. the invention can realize the real restoration of the bridge floor roughness, can accurately acquire the bridge floor elevation information to establish a real bridge floor roughness model, and is used for various calculation and analysis of bridges.

2. The invention overcomes the defect that the bridge floor roughness model established by the ISO 8608 specification lacks authenticity in the past when the bridge model is established, and ignores the influence of the bridge floor roughness, thereby analyzing the bridge performance parameters more accurately.

3. The invention adopts a non-contact scanning mode, and the laser scans the bridge deck to extract the point cloud information, so that the road surface structure is not damaged potentially due to contact with the bridge deck.

4. The detection system of the invention has simple composition, and consists of a high-speed accurate laser range finder, a three-dimensional laser scanner consisting of a group of reflecting prisms capable of guiding laser and scanning at uniform angular velocity, and a handheld computer for controlling scanning setting and data recording.

5. The system development comprehensive cost of the invention is lower, while the price of the three-dimensional laser scanner is higher, the detection system can be recycled for a plurality of times, and the service life of the laser scanner is longer due to the adoption of a non-contact scanning mode, thereby reducing the equipment update and greatly reducing the use cost. On the other hand, the software used for the subsequent processing of the point cloud information is the open source software and can be used for free.

6. The three-dimensional laser scanner is convenient to arrange, simple and convenient to operate, and automatic acquisition of point cloud data is realized without excessive manual intervention in the acquisition process.

7. The detection equipment of the method has higher mobility, and can be matched with a detection vehicle to go to a detection place at any time, carry out detection scanning anywhere and transmit data in real time.

8. The detection precision of the method can be adjusted in real time, the adjustment method is simple, the scanning precision of the laser scanner can be adjusted according to the required point cloud precision, and when the precision exceeds the current required precision, the model of the laser scanner can be replaced to improve the precision, so that the method basically only relates to equipment replacement of the laser scanner, and the feasibility and convenience of the method in practical engineering application are greatly improved.

Drawings

Fig. 1 is a layout diagram of a three-dimensional laser scanning instrument.

Fig. 2 is a composition diagram of a laser scanning system apparatus.

Fig. 3 is a laser scanner field layout.

Fig. 4 is an effect diagram of initial scan bridge portion point cloud data.

Fig. 5 is a flow chart of point cloud data reduction and noise reduction, and rendering.

Fig. 6 is a partial point cloud data reduced noise reduction and rendering effect diagram.

Fig. 7 is a partial node number in the deck roughness reference point set-1.

Fig. 8 is an illustration of bridge section node numbers and coordinates.

Fig. 9 is a graph of the change in the coordinates of a portion of the nodes given roughness values to the deck reference points.

FIG. 10 is a graph of a partial bridge model with true deck roughness generated using an abaqus running script file.

FIG. 11 is a graph showing the modeling and amplifying effect of roughness on a bridge deck with severe damage at a certain place.

Fig. 12 is a flowchart of a bridge floor roughness model determination method based on a three-dimensional laser point cloud scanning technology.

The method comprises the following steps:

1-1, a first target ball; 1-2, a second target ball; 1-3, a third target ball; 1-4. Target ball four; 1-5, a target ball five; 1-6, a target ball six; 1-7, target sphere seven; 1-8, target ball eight; 1-9. Target sphere nine; 1-10. Target ball ten; 1-11. Target sphere eleven; 1-12. Target ball twelve; 1-13. Target ball thirteen; 1-14. Target sphere fourteen;

2-1, measuring station I; 2-2, measuring station II; 2-3, measuring station III; 2-4, measuring station IV;

3-1, a starting line; 3-2, segmenting the second line; 3-3, segmenting line three; 3-4, a finish line;

4-1, upper boundary line; 4-2. Lower boundary line.

1. A three-dimensional scanner; a PC terminal; 3. an initial coordinate system of the three-dimensional scanner; 4. road surface measuring points; 5. and (5) laser.

Detailed Description

The invention will be described in further detail with reference to the accompanying drawings and specific preferred embodiments.

In the description of the present invention, it should be understood that the terms "left", "right", "upper", "lower", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, are merely for convenience in describing the present invention and simplifying the description, and do not indicate or imply that the apparatus or element being referred to must have a specific orientation, be configured and operated in a specific orientation, and "first", "second", etc. do not indicate the importance of the components, and thus are not to be construed as limiting the present invention. The specific dimensions adopted in the present embodiment are only for illustrating the technical solution, and do not limit the protection scope of the present invention.

As shown in fig. 12, the bridge floor roughness model determining method based on the three-dimensional laser point cloud scanning technology comprises the following steps.

Step 1, determining a bridge deck scanning area: and determining a bridge deck scanning area and setting an area boundary according to the bridge deck section to be measured. As shown in fig. 1 and 3, the region boundaries include a lateral boundary and a longitudinal boundary, and the specific arrangement method is as follows.

Step 1-1, setting a transverse boundary: the lateral boundaries include a start line 3-1 and a finish line 3-4, and the start line and finish line locations of the bridge deck scan area are identified at the bridge deck cross-section centerline locations, preferably using red paint lines. The specific identification method is preferably as follows:

A. drawing a # -symbol at the center line position of the bridge deck cross section by using red paint, wherein the width of the drawn line is 0.5cm, the diameter of the circle is 5cm, and the cross line crossing points in the symbol are the starting point and the ending point of the measured bridge section and are the marking positions of 2-1 and 2-4 in figure 1 respectively.

B. And when the thin line is perpendicular to the road shoulder line and passes through the starting point position and the ending point position to stretch and cross at the central position of the width of the real line of the white outer boundary of the bridge deck edge, the marking position of the thin line represents the position of the starting point line and the ending point line of the bridge deck section, namely the transverse edge of the measured bridge section, namely the starting point line 3-1 and the ending point line 3-4 marked in figure 1 respectively.

Step 1-2, setting a longitudinal boundary: the longitudinal boundary comprises an upper boundary line 4-1 and a lower boundary line 4-2, and the set distance positions of the road shoulder edges at the two sides of the bridge deck scanning area to the inner side are respectively marked by lines and marked as an upper boundary line and a lower boundary line. The specific identification method comprises the following steps: the placement of thin lines along the solid line width centerline of the white outer boundary of the road edge identifies the longitudinal boundary of the deck, as shown in fig. 1.

Step 2, setting a target ball: according to the resolution ratio of the three-dimensional scanner and the required point cloud data precision, the bridge floor scanning area determined in the step 1 is segmented longitudinally to form a plurality of bridge floor measuring segments; at least one measuring station and a plurality of target balls are distributed in each bridge deck measuring section, and at least 3 identical target balls can be scanned by two adjacent measuring stations.

In the present application, a single three-dimensional scanner 1 may be provided to sequentially scan from the first measuring station in view of cost. Alternatively, a three-dimensional scanner may be deployed per station.

The scanning bridge floor area is preferably divided into three sections, the length of each section is 8m, and the road width is 4m.

The specific setting method of the target ball specifically comprises the following steps.

Step 2-3, setting a target ball: the method for measuring the bridge deck comprises the steps of respectively setting a target ball at the joint point of a starting line, each section line and a finishing line with an upper boundary line and a lower boundary line, each station measuring position and the centers of the upper boundary line and the lower boundary line on two sides of each bridge deck measuring section, wherein the specific setting method comprises the following steps:

A. a target ball is arranged at the midpoint position and two end positions of the thin line representing the starting line and the finishing line of the bridge deck section, as shown by 1-1, 1-7, 1-8 and 1-14 in figure 1, and a target ball is arranged at the station measuring position identified by 2-1 and 2-4.

B. A target ball is arranged at intervals of 10m along the boundary lines on two longitudinal sides of the bridge deck, as shown in 1-1, 1-2, 1-3, 1-4, 1-5, 1-6, 1-7, 1-8, 1-9, 1-10, 1-11, 1-12, 1-13 and 1-14 in the attached figure 1.

The target balls on the upper boundary line and the lower boundary line correspond to each other in longitudinal position, and the corresponding two target balls are called a group of target balls, such as 1-1 and 1-8, 1-2 and 1-9, 1-3 and 1-10, and the like, respectively, and are respectively target ball groups. Any two adjacent sets of target balls are arranged in a rectangular shape so that every two adjacent stations can scan at least 3 identical target balls.

And 3-1, judging the damage degree of each bridge deck measurement section.

The bridge deck damage degree of the bridge deck measurement section includes slight damage, medium damage and serious damage according to relevant regulations in the quality inspection and acceptance Specification of highway pavement engineering (JTG F80/1-2004).

Step 3-2, determining scanning parameters: and (3) determining scanning parameters of the three-dimensional scanner in the corresponding measuring station according to the bridge deck damage degree of the bridge deck measurement section obtained in the step (3-1), wherein the scanning parameters comprise scanning precision and scanning time.

When the damage is slight damage or medium damage, the scanning precision is set to be +/-2-5 mm, and the scanning time is not more than 20 minutes; when the damage is serious, the precision of the scanner is set to +/-1-2 mm, and the scanning time is not more than 30 minutes.

In the embodiment, a shielding object of the bridge deck is cleaned, and the degree of damage of the scanned bridge deck is judged to be in a second grade according to personal experience and in combination with relevant regulations in the highway pavement engineering quality acceptance Specification (JTG F80/1-2004); thus, the scanner resolution was set to 2mm and the scan time was 20 minutes.

In the segmented scanning process, in order to ensure the accuracy of extracting point cloud data, the method can be used for scanning for multiple times.

In this embodiment, the specific scanning method includes the following steps.

A. The scanning device used in this embodiment is preferably a Trimble x7 standing laser scanner equipped with a support bracket and a PC terminal 2 for hand-held control and data transmission, and the Trimble x7 standing laser scanner is composed of a high-speed accurate laser range finder and a set of reflecting prisms that can guide laser light and scan at a uniform angular velocity, as shown in fig. 2.

B. The scanning equipment is arranged, the arrangement position is positioned at the measuring station 2-1 in the figure 1, the arrangement mode is shown in the figure 3, the Trimble x7 three-dimensional laser scanner is stably arranged at a height of 1.6m from the ground by using the support bracket, and a personnel holds a control and data transmission PC terminal outside a measuring area.

C. The measurement principle, for example, is to measure point-P point cloud information of the road surface measurement point 4 in fig. 3: the emitting device of the three-dimensional scanner emits a laser pulse wave signal, as shown by laser 5 in fig. 3, the laser pulse wave signal is reflected diffusely by the surface of a solid point P, and is transmitted back to the receiver along the opposite direction of almost the same path, the distance S between the target point P and the three-dimensional scanner is calculated, meanwhile, the control encoder is matched to measure the observed value alpha of the transverse scanning angle and the observed value beta of the longitudinal scanning angle of each laser pulse, so as to obtain the space relative coordinates (X, Y, Z) of each scanning point and a measuring station, the three-dimensional scanning measurement is generally the self-defined coordinates of the instrument, the X axis is in the transverse scanning plane (transverse bridge direction), the Y axis is vertical to the X axis (longitudinal bridge direction) in the transverse scanning plane, the Z axis is vertical to the transverse scanning plane (bridge surface elevation direction), the coordinate system is shown as 3 in fig. 3, and the horizontal direction is rotated by the servo motor to complete horizontal 360-degree scanning, so as to obtain the three-dimensional point cloud data.

D. And (4) starting scanning, and recording point cloud data by the handheld PC end, wherein partial point cloud effect is shown in figure 4.

E. And (3) scanning the segmented pavement sequentially by using a three-dimensional laser scanner, and repeating the method in 3.2 for the measuring stations 2-2, 2-3 and 2-4 in the figure 1 sequentially until all measuring stations are scanned, so as to complete the acquisition of all point cloud data.

Step 3-4, extracting point cloud data: each three-dimensional scanner takes the scanner as an origin, and further obtains the space relative coordinates (x, y, z) of each scanning point relative to the origin.

Step 4, preprocessing bridge surface point cloud data

In the scanning process, the precision of laser scanning is affected due to the difference of road surface damage degrees and the difference of scanner working environments, so that the extracted point cloud characteristic information is strong and weak. The regular and slightly damaged bridge floor area has strong point cloud characteristic information extracted by a scanner, and the formed point cloud image is a strong characteristic point cloud image; and the bridge floor damages the serious area, because the very irregular of the damaged surface, the characteristic information of the point cloud extracted is not strong, appear the weak characteristic point cloud picture.

Adopting a point cloud simplifying algorithm combining a 3D-SIFT feature point extraction algorithm and an octree algorithm to reduce noise and render all the bridge surface point cloud data acquired in the step 3; and then, integrating all the bridge surface point cloud data after noise reduction and rendering into the same coordinate system.

A. Constructing a scale space, wherein the scale space is represented by a Gaussian pyramid, and the scale space of the 3D point cloud can be represented by convolving a changed Gaussian kernel function with the point cloud coordinates, namely:

wherein L (x, y, z, sigma) represents the scale space of the point cloud, G (x, y, z, sigma) represents the Gaussian convolution kernel function, I (x, y, z) represents the point cloud coordinates, sigma represents the Gaussian convolution scale factor (sigma usually takes on a value)The algorithm is used for adjusting the blurring degree or the smoothness degree of the filter, and the algorithm sets the strength of the point cloud characteristic according to sigma. When the characteristic point cloud intensity is larger than but close to the point cloud intensity set value, a strong characteristic point cloud image in the initial point cloud image is extracted by increasing the Gaussian convolution scale factor sigma value.

B. And detecting extreme points, and generating Gaussian pyramid of the point cloud by using multiplication factors k of different scales. The value range of k is preferably 1.0-1.6, and generally the larger the k value is, the more the number of point clouds participating in the detection of the extreme point is, the more the number of key point clouds is. When the extracted bridge surface point cloud information is not obvious, the k value can be properly increased, so that the enough number of the key point clouds can be reserved after filtering, and the integrity of the simplified point cloud information is improved.

The extremum formula for the Gaussian difference scale function detection stability is as follows:

G(x，y，z，k ⁱ σ)＝L(x，y，z，k ⁱ σ)-I(x，y，z，k ⁱ σ)

wherein i is 0, s+2, s is the number of layers in each pyramid group.

The key point is an extreme point, the current pixel point is compared with the adjacent point, and whether the value is an extreme value in the surrounding is determined (the key point cloud information is ensured not to be filtered by checking the extreme point, and the key point cloud information is ensured not to be distorted).

C. Determining key point principal direction

Firstly, calculating the direction and gradient characteristics of each neighborhood point, wherein the specific calculation formula is as follows:

wherein:

L _x ＝L(x+1，y，z)-L(x-1，y，z)

L _y ＝L(x，y+1，z)-L(x，y-1，z)

L _z ＝L(x，y，z+1)-L(x，y，z-1)

m (x, y, z) is the amplitude in the neighborhood window of the key point, θ (x, y, z) is the direction angle,is the pitch angle.

And then, using the gradient histogram to count the gradient and the direction of the pixels of the neighborhood points in the region, and determining the main direction of the key points.

D. Building feature descriptors

Dividing the neighborhood of the key point into n multiplied by n sub-areas, wherein each sub-area is a seed point, each seed has 8 directions (the value range of n is generally 2-6, the value is generally 4, the 8 directions of each seed are corresponding at the moment, the simplifying speed of the point cloud characteristic information is accelerated), the coordinate axis direction is rotated to the main direction of the key point, the points in the neighborhood are distributed into the corresponding sub-areas, the gradient values of the points in the sub-areas are calculated and distributed to the 8 directions, the weight of the gradient values are calculated, the gray gradient histogram of the 8 directions is counted, the characteristic vector is obtained, and the normalization processing is carried out, so that the characteristic descriptor is obtained.

According to the application, the gradient values are distributed to 8 directions, so that even if data in one direction is distorted, the ratio of the sum of the gradient values in other directions is far larger than that in one direction, and therefore, the error of the feature descriptors is reduced to the maximum extent, and the accuracy of the filtered point cloud is ensured.

The specific algorithm of the octree voxel filtering method is as follows: constructing a first cube according to a point cloud space with the maximum size point number M, wherein the volume of the first cube is V, the length, the width and the height of the first cube are A, B, C respectively, and the side length S of the subcubes is set as follows:

dividing the first cube into n subcubes, then n is:

where ceil () is a round-up function.

Alpha is a proportionality coefficient ranging from 1.0 to 2.0 and usually takes a value of 1.3.

Epsilon is a scale factor ranging from 1.0 to 1.3, usually taking a value of 1.1.

The feature of vulnerability cloud information is amplified by increasing the side length of the subcubes by appropriately increasing the values of α and ε.

Calculating the center of gravity (x) of the non-empty cube ₀ ，y ₀ ，z ₀ ) Establishing a center of gravity point set, wherein:

and finding out the nearest neighbor point of the gravity center point according to the octree, and constructing a new point set, wherein the points in the point set are the filtered points of the octree. The flow is shown in fig. 5, and the simplified noise reduction and rendering effect is shown in fig. 6.

Step 4-5, constructing a preprocessing point cloud picture: and (3) combining the strong characteristic point cloud image in the step 4-2 and the filtering and amplifying weak characteristic point cloud image in the step 4-4 to obtain a preprocessed point cloud image, which is also called integral bridge surface point cloud data. In the process of noise reduction and rendering, each group of data is registered and spliced by taking the point cloud information of the three-dimensional laser scanning target ball in the point cloud image as a reference point to form integral bridge surface point cloud data.

Step 5, obtaining a normalized integral bridge surface point cloud matrix: importing the whole bridge surface point cloud data after the pretreatment in the step 4 into point cloud software (preferably open source software Cloudcompare) for processing, normalizing the bridge floor elevation in the whole bridge surface point cloud data by taking the road surface plane as a reference to obtain normalized relative elevation z ', thus obtaining normalized whole bridge surface point Yun Juzhen (x, y, z') in txt format, and recording the number of rows and columns of the normalized whole bridge surface point cloud matrix.

The normalization method of the relative elevation z' comprises the following steps.

And 5-1, establishing a pavement plane equation, specifically.

0＝d ₀ +d ₁ x+d ₂ y-z

And 5-2, carrying out elevation normalization, wherein a specific calculation formula is as follows.

Where z' is the relative elevation.

Step 6, building a bridge deck roughness model: outputting the normalized integral bridge surface point cloud matrix obtained in the step 5 into txt format, and importing finite element software (preferably Python) to establish a bridge floor roughness model.

The method for establishing the bridge deck roughness model preferably comprises the following steps.

Step 6-2-1, establishing a three-dimensional bridge entity model: in finite element software Abaqus, according to the size information of a bridge deck section to be measured in a bridge design drawing, a three-dimensional bridge entity model without surface roughness and with equal proportion is established; the length and the width of the three-dimensional bridge solid model are sequentially the scanning total length and the scanning total width of a bridge deck scanning area; the height of the three-dimensional bridge solid model is the sum of the height of the bridge deck and the thickness of the concrete layer; and then, selecting the upper surface of the three-dimensional bridge solid model as a bridge deck roughness datum reference surface, and setting the midpoint of the leftmost side line of the bridge deck roughness datum reference surface as the origin of coordinates of the bridge deck roughness datum reference surface.

Step 6-3-1, reading the initial row position of the inp file one: and reading the initial row position of the bridge deck unit and the initial row position of the bridge deck unit node in the inp file I. The specific method is preferably as follows: in Python, a numpy library is imported and the inp file is typed using the open command. The start line position of the Node (Node) and the start line position of the Element (Element) are acquired by lines.

Step 6-3-2, reading bridge deck node numbers: respectively reading the serial number of each bridge deck unit node aiming at all bridge deck unit nodes at the starting line positions read in the step 6-3-1; the specific method is preferably as follows: the lines command is used to read the initial line position and the last line position of the node number contained in the Set with "Nset, nset=set-1" as the first line in the inp file one, and the two lines of the initial line and the last line and the number information contained between the two lines are all converted into a character string form. Each node number is separated by comma and contains space and carriage return symbol, as shown in fig. 7, each number needs to be distinguished by comma, and space and carriage return symbol are removed by segmentation using split command, so as to obtain the number of each bridge deck node.

Step 6-3-3, extracting coordinates of all bridge deck unit joints; the specific method is preferably as follows: all nodes of the bridge, i.e. the previous row from the start row position of the Node (Node) to the start row position of the Element (Element), are traversed by the for statement, and all nodes comprised between the two are traversed, as shown in fig. 8. And extracting the coordinates of nodes with the same number as bridge floor nodes in the traversed nodes, namely extracting the coordinates of all bridge floor roughness reference points.

Step 6-3-4, coordinate substitution: and (3) reading bridge deck node numbers according to the step 6-3-2 by using the normalized integral bridge deck points Yun Juzhen in txt format, and replacing corresponding coordinates in the step 6-3-3 with coordinates in a normalized integral bridge deck point cloud matrix so as to form an inp file II containing roughness elevation coordinates. The specific method is preferably as follows: and opening a txt format file of bridge deck point cloud elevation data through an open command, and sequentially replacing x, y and z coordinates of the extracted bridge deck roughness reference datum point with coordinate values of a first column, a second column and a third column in the txt file by using a for command according to the sequence of traversing node numbers, wherein the coordinate values are shown in figure 9. And storing the inp file after the coordinate values are replaced as a new inp file II.

And 6-4, running the compiled script file in finite element software Abaqus to generate a bridge deck roughness model with authenticity. Fig. 10 is a roughness modeling effect diagram of a certain area of the bridge deck, and fig. 11 shows a roughness modeling amplification effect diagram of the bridge deck with serious damage at a certain place.

According to the invention, the bridge floor roughness model is accurately and rapidly built according to actual conditions, the defect that the existing international standard ISO8608 bridge floor roughness model lacks authenticity is overcome, and the built model can be integrated with the bridge model into a whole bridge model with real bridge floor roughness. The invention can accurately establish the bridge floor roughness model especially aiming at the bridge floor with serious damage, can reconstruct the three-dimensional space form of the bridge floor completely and with high precision, is used for real bridge analysis and verification to timely discover the problems of apparent deformation, internal damage and the like of the in-service bridge, and provides a new thought for the daily operation and management of the bridge.

The preferred embodiments of the present invention have been described in detail above, but the present invention is not limited to the specific details of the above embodiments, and various equivalent changes can be made to the technical solution of the present invention within the scope of the technical concept of the present invention, and all the equivalent changes belong to the protection scope of the present invention.

Claims

1. A bridge floor roughness model determining method based on a three-dimensional laser point cloud scanning technology is characterized by comprising the following steps of: the method comprises the following steps:

step 1, determining a bridge deck scanning area: according to the bridge deck section to be measured, determining a bridge deck scanning area and setting an area boundary;

step 2, setting a target ball: segmenting the bridge deck scanning area determined in the step 1 along the longitudinal direction to form a plurality of bridge deck measurement segments; at least one measuring station and a plurality of target balls are distributed in each bridge deck measuring section, and at least 3 identical target balls can be scanned by two adjacent measuring stations;

step 3, scanning and acquiring bridge surface point cloud data, which specifically comprises the following steps:

step 3-1, judging the damage degree of each bridge deck measurement section;

step 3-2, determining scanning parameters: determining scanning parameters of the three-dimensional scanner in a corresponding measuring station according to the bridge deck damage degree of the bridge deck measurement section obtained in the step 3-1, wherein the scanning parameters comprise scanning precision and scanning time;

step 3-3, scanning: starting from a first measuring station, the three-dimensional scanner scans 360 degrees according to the scanning parameters determined in the step 3-2; and the like until the three-dimensional scanning in all measuring stations is completed;

Step 3-4, extracting point cloud data: the three-dimensional scanner takes the three-dimensional scanner as an origin, and further obtains the space relative coordinates (x, y, z) of each scanning point relative to the origin; wherein, the x direction is a transverse bridge direction, the y direction is a longitudinal bridge direction, and the z direction is a bridge deck elevation direction;

step 4, preprocessing bridge surface point cloud data: adopting a point cloud simplifying algorithm combining a 3D-SIFT feature point extraction algorithm and an octree algorithm to reduce noise and render all the bridge surface point cloud data acquired in the step 3; then, integrating all the bridge surface point cloud data after noise reduction and rendering into the same coordinate system;

step 5, obtaining a normalized integral bridge surface point cloud matrix: importing the whole bridge surface point cloud data after the pretreatment in the step 4 into point cloud software for processing, normalizing bridge floor elevation in the whole bridge surface point cloud data by taking a pavement plane as a reference to obtain normalized relative elevation z ', thus obtaining normalized whole bridge surface point Yun Juzhen (x, y, z') in txt format, and recording the number of rows and columns of a normalized whole bridge surface point cloud matrix;

2. The bridge floor roughness model determination method based on the three-dimensional laser point cloud scanning technology as claimed in claim 1, wherein the bridge floor roughness model determination method is characterized by comprising the following steps: in step 1, the region boundary includes a transverse boundary and a longitudinal boundary, and the specific setting method is as follows:

step 1-1, setting a transverse boundary: the transverse boundary comprises a starting line and a finishing line, and the starting line and the finishing line of the bridge deck scanning area are marked by lines at the position of the central line of the cross section of the bridge deck;

3. The bridge floor roughness model determination method based on the three-dimensional laser point cloud scanning technology as claimed in claim 2, wherein the bridge floor roughness model determination method is characterized by comprising the following steps: in step 2, the setting method of the target ball specifically comprises the following steps:

step 2-1, determining a segmentation line: defining the interface between two adjacent bridge deck measurement segments as a segment line; step 2-2, setting a measuring station: a measuring station is arranged at the centers of the starting line, each segment line and the finishing line respectively;

4. The bridge floor roughness model determination method based on the three-dimensional laser point cloud scanning technology as claimed in claim 1, wherein the bridge floor roughness model determination method is characterized by comprising the following steps: in step 3-2, bridge deck damage levels for measuring segments of bridge deck include mild damage, moderate damage and severe damage; when the damage is slight damage or medium damage, the scanning precision is set to be +/-2-5 mm, and the scanning time is not more than 20 minutes; when the damage is serious, the precision of the scanner is set to +/-1-2 mm, and the scanning time is not more than 30 minutes.

5. The bridge floor roughness model determination method based on the three-dimensional laser point cloud scanning technology as claimed in claim 1, wherein the bridge floor roughness model determination method is characterized by comprising the following steps: in step 4, a method for reducing noise and rendering bridge deck point cloud data by adopting a point cloud reduction algorithm comprises the following steps: step 4-1, obtaining an initial point cloud picture: performing de-duplication processing on all the bridge surface point cloud data obtained in the step 3 to obtain an initial point cloud image;

step 4-2, extracting a strong characteristic point cloud picture: extracting strong characteristic point cloud pictures in the initial point cloud pictures by adopting a 3D-SIFT characteristic point extraction algorithm; wherein the strong characteristic point cloud image is that the characteristic curvature of the pointing cloud is not less than 0.1;

step 4-3, secondary de-duplication: removing the strong characteristic point cloud picture extracted in the step 4-1 from the initial point cloud picture to obtain a weak characteristic point cloud picture; step 4-4, octree filtering: filtering the weak characteristic point cloud image obtained in the step 4-3, and amplifying the weak characteristic point cloud information in the weak characteristic point cloud image to obtain a filtered and amplified weak characteristic point cloud image;

6. The bridge floor roughness model determination method based on the three-dimensional laser point cloud scanning technology as claimed in claim 5, wherein the bridge floor roughness model determination method is characterized by comprising the following steps: in step 4-2, when the 3D-SIFT feature point extraction algorithm is adopted, and the feature point cloud intensity is larger than but close to the point cloud intensity set value, a strong feature point cloud image in the initial point cloud image is extracted by increasing the Gaussian convolution scale factor sigma value.

7. The bridge floor roughness model determination method based on the three-dimensional laser point cloud scanning technology as claimed in claim 6, wherein the bridge floor roughness model determination method is characterized by comprising the following steps: in step 4-2, when the 3D-SIFT feature point extraction algorithm is adopted to extract the key point clouds, if the Gaussian convolution scale factor sigma is determined, multiplying the Gaussian convolution scale factor sigma by a multiplication factor k with different scales, so as to determine the number of the extracted key point clouds; the greater the k value, the greater the number of key point clouds extracted will be.

8. The bridge floor roughness model determination method based on the three-dimensional laser point cloud scanning technology as claimed in claim 5, wherein the bridge floor roughness model determination method is characterized by comprising the following steps: in step 4-4, the side length of the subcubes is increased by introducing a proportion coefficient alpha and a proportion factor epsilon, so that weak characteristic point cloud information is amplified; wherein, the value range of alpha is 1.0-2.0, and the value range of epsilon is 1.0-1.3.

9. The bridge floor roughness model determination method based on the three-dimensional laser point cloud scanning technology as claimed in claim 1, wherein the bridge floor roughness model determination method is characterized by comprising the following steps: in the step 6, the method for establishing the bridge deck roughness model comprises the following steps:

step 6-1, adjusting an output format: outputting normalized integral bridge surface point cloud data (x, y, z') into txt format; the first, second and third columns of the normalized integral bridge surface point cloud data in txt format are respectively corresponding to the x, y and z' coordinates of each acquisition point; step 6-2, establishing bridge deck reference datum points of roughness point cloud data, which specifically comprises the following steps:

step 6-2-1, establishing a three-dimensional bridge entity model: in finite element software, according to the size information of a bridge deck section to be measured in a bridge design drawing, a three-dimensional bridge entity model without surface roughness and with equal proportion is established; the length and the width of the three-dimensional bridge solid model are sequentially the scanning total length and the scanning total width of a bridge deck scanning area; the height of the three-dimensional bridge solid model is the sum of the height of the bridge deck and the thickness of the concrete layer; then, selecting the upper surface of the three-dimensional bridge solid model as a bridge deck roughness datum reference surface, and setting the midpoint of the leftmost side line of the bridge deck roughness datum reference surface as the origin of coordinates of the bridge deck roughness datum reference surface; step 6-2-2, mesh division: performing grid division on the three-dimensional bridge entity model established in the step 6-2-1; the transverse number of the grids is the number of columns of the normalized integral bridge surface point cloud matrix minus one, and the longitudinal number of the grids is the number of rows of the normalized integral bridge surface point cloud matrix minus one; each grid is a bridge deck unit, and four corner points of each grid correspond to four bridge deck unit nodes respectively;

Step 6-2-3, establishing a bridge deck unit node set: establishing all bridge deck unit nodes contained in the bridge deck roughness standard reference surface after grid division as a bridge deck unit node set-1, establishing a Job Job-1, and selecting 'Write Input' to obtain numbers and coordinates inp files I of all bridge deck units and bridge deck unit nodes in the set-1;

step 6-3, creating an inp file II containing the roughness elevation coordinate, which specifically comprises the following steps:

step 6-3-1, reading the initial row position of the inp file one: reading the initial row position of the bridge deck unit and the initial row position of the bridge deck unit node in the inp file I;

step 6-3-2, reading bridge deck node numbers: respectively reading the serial number of each bridge deck unit node aiming at all bridge deck unit nodes at the starting line positions read in the step 6-3-1;

step 6-3-3, extracting coordinates of all bridge deck unit joints;

10. The bridge floor roughness model determination method based on the three-dimensional laser point cloud scanning technology as claimed in claim 1, wherein the bridge floor roughness model determination method is characterized by comprising the following steps: in step 5, the normalization method of the relative elevation z' comprises the following steps:

step 5-1, establishing a pavement plane equation, which specifically comprises the following steps:

0＝d ₀ +d ₁ x+d ₂ y-z

wherein d ₀ 、d ₁ 、d ₂ And e are plane equation coefficients respectively, and can be obtained by solving a matrix method;