CN110598239B - Application method based on track area point cloud big data - Google Patents
Application method based on track area point cloud big data Download PDFInfo
- Publication number
- CN110598239B CN110598239B CN201910652405.6A CN201910652405A CN110598239B CN 110598239 B CN110598239 B CN 110598239B CN 201910652405 A CN201910652405 A CN 201910652405A CN 110598239 B CN110598239 B CN 110598239B
- Authority
- CN
- China
- Prior art keywords
- point cloud
- section
- point
- points
- plane
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000013461 design Methods 0.000 claims abstract description 32
- 230000006870 function Effects 0.000 claims abstract description 17
- 238000011478 gradient descent method Methods 0.000 claims abstract description 12
- 238000013136 deep learning model Methods 0.000 claims abstract description 9
- 238000013507 mapping Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000000638 solvent extraction Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 abstract description 14
- 238000004364 calculation method Methods 0.000 abstract description 11
- 230000009545 invasion Effects 0.000 abstract description 10
- 238000000265 homogenisation Methods 0.000 abstract description 6
- 238000005259 measurement Methods 0.000 abstract description 6
- 230000000903 blocking effect Effects 0.000 abstract description 5
- 230000009467 reduction Effects 0.000 abstract description 5
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 239000013589 supplement Substances 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000002040 relaxant effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Length Measuring Devices With Unspecified Measuring Means (AREA)
Abstract
The invention discloses an application method based on point cloud big data of an orbit region, which belongs to the field of point cloud big data application, and can realize point cloud blocking, point cloud noise reduction and point cloud homogenization pretreatment on the point cloud data obtained by three-dimensional laser scanning, so that coordinate calculation of key points of which the ten measurement sections of the actual section are in intrusion limit degree is more accurate for any mileage; according to the design route constructed by the two-dimensional diagrams of the plane and the longitudinal section, ten key point coordinates of the theoretical section are determined, and the intrusion limit value of the section, the equipment dimensions of an evacuation platform, a contact net and the like are more accurately calculated; according to the calculated ten key point invasion limit values of the section, a deep learning model is established, the functional relation between the plane and vertical section parameters and the invasion limit values is learned, and then the extreme value of the function is obtained by adopting a gradient descent method, so that the aim of optimizing the design line is fulfilled.
Description
Technical Field
The invention relates to the field of point cloud big data application, in particular to an application method based on the point cloud big data of a track area.
Background
After the subway tunnel is completed, due to the influence of factors such as construction errors, measurement errors and structural deformation, deviation exists between the actual tunnel and the theoretical tunnel corresponding to the original design line, a plurality of sections are selected from the whole tunnel according to the same distance interval for measuring the deviation, the intrusion limit value of each section is calculated, the deviation of the tunnel can be judged according to the intrusion limit value of each section, and then whether the original design line scheme meets the driving requirement is judged.
When the limit value of the section does not meet the requirement, the original design line is required to be adjusted, the limit value of each section is calculated again, the train track can be paved until the adjusted design line meets the driving requirement, and the equipment size installed on the inner wall of the tunnel is required to be calculated after the track is paved. The specific operation is as follows: the measuring staff hand-push rail inspection trolley advances along the tunnel, acquires data such as the section mileage, the inclination angle and the track gauge of the tunnel, calculates the equipment size according to the measured data and the adjusted design line, gives the calculated equipment size data to equipment manufacturers, subscribes equipment with corresponding sizes, and finally installs the subscribed equipment on the inner wall of the tunnel. However, the following two drawbacks exist throughout the process using the above method:
1. the intrusion value for each section cannot be calculated in real time. The traditional section measurement method only measures the coordinate positions of the specific ten points of each section, and can calculate the intrusion limit value of the section by combining with the design circuit scheme, however, when the design circuit scheme is adjusted, the original measured coordinate positions of the ten points of each section can change, the intrusion limit value of each section can be calculated only by re-measuring the coordinate positions of the ten points of each section, a great amount of time is consumed for re-measuring, and the intrusion limit value of each section can be calculated after the measurement work is completed.
2. The equipment size error calculated from the measured value is large. The traditional method for calculating the equipment size is based on the track gauge measurement value of the track inspection trolley, the track gauge measurement value is easily influenced by human factors, measurement data are inaccurate, and then calculation errors of the equipment size are increased. When the calculated equipment size is smaller, the gap between the train and the equipment is increased, the equipment with strict distance requirements cannot meet the standard requirements, and the equipment needs to be re-ordered; when the calculated equipment size is bigger, the distance between the train and the equipment is too small, the train and the equipment can possibly rub and collide, the running safety of the train is threatened, the equipment is required to be reduced, the size is reduced, the running safety of the train is ensured, the equipment is required to be subjected to secondary treatment no matter the calculated equipment size is bigger or smaller, and the material and time are wasted.
At present, three-dimensional point cloud data of the inner wall of a tunnel such as various complicated, irregular, standard or nonstandard tunnels are directly collected into a computer through three-dimensional laser scanning, and a three-dimensional solid model of the tunnel is constructed through modeling, however, due to the huge amount of point cloud data, when all the point cloud data directly participate in operation, a great amount of time is required, the computer can not calculate results due to insufficient operation memory, and the problems of noise, uneven density and the like of the point cloud data in part of intervals are caused, and further preprocessing is required to be carried out on the point cloud data so as to improve the searching speed and accuracy of the point cloud data.
Disclosure of Invention
1. Technical problem to be solved
Aiming at the problems existing in the prior art, the invention aims to provide an application method based on the large data of the point cloud in the track area, which can realize the point cloud blocking, the point cloud noise reduction and the point cloud homogenization pretreatment on the point cloud data obtained by three-dimensional laser scanning, so that the coordinate calculation of the key points of which the ten measuring section intrusion limit degrees of any mileage corresponds to the actual section is more accurate; according to the design route constructed by the two-dimensional diagrams of the plane and the longitudinal section, ten key point coordinates of the theoretical section are determined, and the intrusion limit value of the section, the equipment dimensions of an evacuation platform, a contact net and the like are more accurately calculated; according to the calculated ten key point invasion limit values of the section, a deep learning model is established, the functional relation between the plane and vertical section parameters and the invasion limit values is learned, and then the extreme value of the function is obtained by adopting a gradient descent method, so that the aim of optimizing the design line is fulfilled.
2. Technical proposal
In order to solve the problems, the invention adopts the following technical scheme.
The application method based on the track area point cloud big data comprises the following steps:
s1: acquiring tunnel point cloud data through three-dimensional laser scanning;
s2: preprocessing point cloud data:
SA: and (3) partitioning the point cloud: dividing the point cloud data into a plurality of files according to mileage values of an original design line for storage, and adding indexes for the files;
SB: denoising the point cloud: denoising the point cloud data;
SC: homogenizing point cloud: homogenizing the density of the point cloud data;
s3: and (3) establishing a design line model: constructing a plane two-dimensional graph and a longitudinal section two-dimensional graph in a line through geometric relations among three parameters in the designed line, and constructing a theoretical tunnel model corresponding to the designed line according to the plane two-dimensional graph and the longitudinal section two-dimensional graph;
s4: calculating ten-point coordinates of a theoretical section: ten key points of a section corresponding to any mileage are extracted from the theoretical tunnel model, and ten key point coordinates of the theoretical section corresponding to the mileage are calculated;
s5: calculating actual cross section ten-point coordinates: finding out ten key point coordinates of an actual section corresponding to the same mileage as the S4 from the point cloud data after the pretreatment of the S2;
s6: calculating an intrusion value and a device size: according to the ten key point coordinates of the theoretical section obtained in the step S4 and the ten key point coordinates of the actual section obtained in the step S5, the intrusion limit value of the section and the size of equipment are calculated, a deep learning model is built according to the calculated intrusion limit value of the section, and then the extremum of a function is obtained to obtain an optimized design line, so that the point cloud data obtained by three-dimensional laser scanning can be subjected to point cloud blocking, point cloud noise reduction and point cloud homogenization pretreatment, and the coordinate calculation of the ten key points measuring the intrusion limit value degree of the section corresponding to the actual section in any mileage is more accurate; according to the design route constructed by the two-dimensional diagrams of the plane and the longitudinal section, ten key point coordinates of the theoretical section are determined, and the intrusion limit value of the section, the equipment dimensions of an evacuation platform, a contact net and the like are more accurately calculated; according to the calculated ten key point invasion limit values of the section, a deep learning model is established, the functional relation between the plane and vertical section parameters and the invasion limit values is learned, and then the extreme value of the function is obtained by adopting a gradient descent method, so that the aim of optimizing the design line is fulfilled.
Further, the specific steps of the SA are as follows:
SA1: let Pm0 equal Ps, pm1 equal Ps+0.01;
SA2: according to F (Ω) 1 ，Ω 2 ) And Pm0 and Pm1 to obtain normal plane equations SPm0 and SPm1 of the curve at the Ps;
SA3: finding points between SPm0 and SPm1 from the file and storing the points in a file 0;
SA4: let Pm0 equal to Pm1, pm1 equal to Pm1+0.01;
SA5: and repeating the steps SA1-4 until m0> e, wherein in SA, for the section of any mileage m on a new line, the distances from all points in the range from the filem-1 to the filem+1 to the section at the mileage m are required to be calculated, the point with the distance smaller than 1 cm from the section is projected on the section, so that the point cloud of the section can be obtained, after the point cloud data are processed by adopting the point cloud partitioning, when the point cloud data of any section are searched, the speed of searching the corresponding data from the file name can be increased, and the accuracy of ten key points in the section is obviously improved, thereby reducing the error in the calculation of the intrusion value and the equipment size, and realizing the efficient management and the utilization of the point cloud data.
Further, the specific steps of the SB are as follows:
SB1: let m=s+0.005;
SB2: according to F (Ω) 1 ，Ω 2 ) And Pm to obtain a normal plane equation SPm of a curve at the position of Pm;
SB3: projecting points in the file onto the SPm, and mapping the plane into a two-dimensional coordinate system of the plane SPm;
SB4: fitting an actual circle center O and a radius R in the two-dimensional coordinate system according to a least square method;
SB5: let m equal m+0.01;
SB6: according to F (Ω) 1 ，Ω 2 ) And Pm to obtain a normal plane equation SPm of a curve at the position of Pm;
SB7: projecting points in the file onto the SPm, mapping a plane into the two-dimensional coordinate system of the plane SPm;
SB8: deleting points with the distance between the points and O being more than R+0.01 or less than R-0.01 in the two-dimensional coordinate system, and fitting out the actual circle center O and the radius R according to a least square method;
SB9: and repeating the step SB5-8 until m0> e, and reducing the influence of noise points on the three-dimensional modeling, so that the processed point cloud data can more accurately reflect the boundary of the inner wall of the actual tunnel, and the simulation of the actual tunnel model is facilitated.
Further, the specific steps of the SC are as follows:
SC1: counting the number of point clouds in each segmented file;
SC2: when the statistical quantity is more than 10000, 8000 point clouds are randomly extracted, and the original point cloud data are saved and replaced;
SC3: when the number is less than 2000, fitting all points in the file into a standard circle, calculating coordinates of ten key points in the circle, and storing and replacing the original point cloud data, so that the supplement of the data deficiency part is effectively realized, and the coordinate calculation of the ten key points of the actual section is more accurate.
Further, the three parameters in S3 are the intersection point coordinate, the length of the relaxation curve, and the radius of the arc, and by determining the parameters of the intersection point coordinate, the length of the relaxation curve, and the radius of the arc in the design line, the two-dimensional plane map and the two-dimensional vertical plane map can be quickly constructed.
Further, the specific steps of S5 and S6 are as follows:
sa: according to F (Ω) 1 ，Ω 2 ) Obtaining a normal plane equation SPm of a curve at the position of the Pm by a line center point Pm corresponding to the mileage;
sb: finding out points with the distance smaller than 0.01 from the SPm in the point cloud data and projecting the points onto the SPm;
sc: mapping the plane into a two-dimensional coordinate system of the plane SPm;
sd: fitting out the actual circle center O and the radius R on the SPm according to a least square method;
se: finding ten key points on the SPm, and calculating the intrusion value of the ten key points and the size of the equipment;
sf: adding the intrusion values of ten identical position points in the section with the interval of 5M to obtain an intrusion value matrix M;
sg: altering F (Ω) within custom ranges 1 ，Ω 2 ) Obtaining a plurality of groups of different F (omega) 1 ，Ω 2 ) Repeating the steps Sa-Sf to obtain a plurality of groups of corresponding intrusion value matrixes M;
sh: constructing a deep neural network, and integrating multiple groups of line parameters F (omega 1 ，Ω 2 ) As input, corresponding to a plurality of groups of intrusion limit matrixes as output, and adjusting network weight parameters until the error of the intrusion limit is smaller than a set threshold;
si: obtaining line parameter F (Ω) 1 ，Ω 2 ) And obtaining the extremum of the function according to the functional relation of the intrusion value matrix M and the gradient descent method, and obtaining the optimized design route.
Further, the P is m Represents F (Ω) 1 ，Ω 2 ) The point of the upper mileage m, s represents F (Ω 1 ，Ω 2 ) Is expressed as F (Ω) 1 ，Ω 2 ) Is the end point mileage of said F (omega) 1 ，Ω 2 ) Representing the line modulation without modulation in rectangular coordinate systemLine centerline equation for a slope, where Ω 1 represents a profile parameter and Ω 2 represents a plane parameter.
Further, the extremum of the function in S6 is obtained by using a gradient descent method.
3. Advantageous effects
Compared with the prior art, the invention has the advantages that:
(1) According to the scheme, point cloud blocking, point cloud noise reduction and point cloud homogenization pretreatment are carried out on the point cloud data obtained by three-dimensional laser scanning, so that coordinate calculation of ten key points of which the actual section corresponds to the cross section intrusion limit degree is more accurate; according to the design route constructed by the two-dimensional diagrams of the plane and the longitudinal section, ten key point coordinates of the theoretical section are determined, and the intrusion limit value of the section, the equipment dimensions of an evacuation platform, a contact net and the like are more accurately calculated; according to the calculated ten key point invasion limit values of the section, a deep learning model is established, the functional relation between the plane and vertical section parameters and the invasion limit values is learned, and then the extreme value of the function is obtained by adopting a gradient descent method, so that the aim of optimizing the design line is fulfilled.
(2) In SA, for a section of any mileage m on a new line, the distances from all points in the file-1 to the section of the mileage m are required to be calculated, the points with the distance smaller than 1 cm from the section are projected on the section, so that the point cloud of the section can be obtained, after the point cloud data in SA are processed by adopting point cloud partitioning, when the point cloud data of any section are searched, the speed of searching the corresponding data from file names can be increased, the accuracy of ten key points in the section is obviously improved, thereby reducing the error in the calculation of the intrusion value and the equipment size, and realizing the efficient management and utilization of the point cloud data.
(3) After the point cloud denoising processing is adopted for the point cloud data in SB, the influence of the noise points on the three-dimensional modeling can be reduced, so that the processed point cloud data can more accurately reflect the boundary of the inner wall of the actual tunnel, and the simulation of the actual tunnel model is facilitated.
(4) After the point cloud homogenization treatment is adopted for the point cloud data in the SC, the supplement of the insufficient data is effectively realized, so that the coordinate calculation of ten key points of the actual section is more accurate.
(5) And S3, respectively determining the intersection point coordinate, the length of the moderation curve and the radius of the circular arc, and rapidly constructing a plane two-dimensional graph and a longitudinal two-dimensional graph by determining the parameters of the intersection point coordinate, the length of the moderation curve and the radius of the circular arc in a design line.
(6)P m Represents F (Ω) 1 ，Ω 2 ) The point where the upper mileage is m, s represents F (Ω 1 ，Ω 2 ) E represents F (Ω) 1 ，Ω 2 ) Is the end point mileage, F (Ω) 1 ，Ω 2 ) And (3) a line central line equation of a rectangular coordinate system without line adjustment and slope adjustment is shown, wherein omega 1 represents a longitudinal section parameter, and omega 2 represents a plane parameter.
(7) And S6, the extremum of the function is obtained by adopting a gradient descent method.
Drawings
FIG. 1 is a block diagram of the main structure of the present invention;
FIG. 2 is a schematic diagram of a point cloud partition structure according to the present invention;
FIG. 3 is a schematic diagram of the structure of the point cloud before denoising;
fig. 4 is a schematic diagram of a structure of the point cloud after denoising according to the present invention;
FIG. 5 is a partial circuit diagram of a longitudinal section of the present invention;
FIG. 6 is a plan view of a partial circuit diagram of the present invention;
FIG. 7 is a schematic view of the structure of the coordinate axis rotation translation transformation of the present invention;
FIG. 8 is a schematic cross-sectional view of ten key points of the present invention;
FIG. 9 is a schematic view of a profile parameter configuration of the present invention;
fig. 10 is a schematic structural diagram of the planar parameter of the present invention.
Detailed Description
The drawings in the embodiments of the present invention will be combined; the technical scheme in the embodiment of the invention is clearly and completely described; obviously; the described embodiments are only a few embodiments of the present invention; but not all embodiments, are based on embodiments in the present invention; all other embodiments obtained by those skilled in the art without undue burden; all falling within the scope of the present invention.
In the description of the present invention, it should be noted that the positional or positional relationship indicated by the terms such as "upper", "lower", "inner", "outer", "top/bottom", etc. are based on the positional or positional relationship shown in the drawings, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the apparatus or elements referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In the description of the present invention, it should be noted that, unless explicitly specified and limited otherwise, the terms "mounted," "configured to," "engaged with," "connected to," and the like are to be construed broadly, and may be either fixedly connected, detachably connected, or integrally connected, for example; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.
Example 1:
referring to fig. 1, an application method based on the big data of the point cloud of the track area includes the following steps:
s1: acquiring tunnel point cloud data through three-dimensional laser scanning;
s2: preprocessing point cloud data:
SA: and (3) partitioning the point cloud: dividing the point cloud data into a plurality of files according to mileage values of an original design line, storing the files, and adding indexes for the files;
SB: denoising the point cloud: denoising the point cloud data;
SC: homogenizing point cloud: homogenizing the density of the point cloud data;
s3: and (3) establishing a design line model: constructing a plane two-dimensional graph and a longitudinal section two-dimensional graph in a line through geometric relations among three parameters in the designed line, and constructing a theoretical tunnel model corresponding to the designed line according to the plane two-dimensional graph and the longitudinal section two-dimensional graph;
s4: calculating ten-point coordinates of a theoretical section: ten key points of a section corresponding to any mileage are extracted from the theoretical tunnel model, and ten key point coordinates of the theoretical section corresponding to the mileage are calculated;
s5: calculating actual cross section ten-point coordinates: finding out ten key point coordinates of an actual section corresponding to the same mileage as the S4 from the point cloud data after the pretreatment of the S2;
s6: calculating an intrusion value and a device size: and (3) according to the ten key point coordinates of the theoretical section obtained in the step (S4) and the ten key point coordinates of the actual section obtained in the step (S5), calculating the intrusion limit value of the section and the size of equipment, constructing a deep learning model according to the calculated intrusion limit value of the section, and then obtaining the extremum of the function to obtain the optimized design line.
Referring to fig. 2, the specific steps of sa are as follows:
SA1: let Pm0 equal Ps, pm1 equal Ps+0.01;
SA2: according to F (Ω) 1 ，Ω 2 ) And Pm0 and Pm1 to obtain normal plane equations SPm0 and SPm1 of the curve at the Ps;
SA3: finding points between SPm0 and SPm1 from the file and storing the points in the file 0;
SA4: let Pm0 equal to Pm1, pm1 equal to Pm1+0.01;
SA5: and repeating the steps SA1-4 until m0> e, calculating the distances from all points in the file-1 to the file+1 to the cross section at the mileage m in SA for the cross section of any mileage m on a new line, and projecting the points with the cross section distance smaller than 1 cm on the cross section to obtain the point cloud of the cross section.
Referring to FIGS. 3-4, which are schematic diagrams before denoising and after denoising, SB comprises the following steps:
SB1: let m=s+0.005;
SB2: according to F (Ω) 1 ，Ω 2 ) And Pm to obtain a normal plane equation SPm of a curve at the position of Pm;
SB3: projecting points in the file onto the SPm, and mapping the plane into a two-dimensional coordinate system of the plane SPm;
SB4: fitting an actual circle center O and a radius R in a two-dimensional coordinate system according to a least square method;
SB5: let m equal m+0.01;
SB6: according to F (Ω) 1 ，Ω 2 ) And Pm to obtain a normal plane equation SPm of a curve at the position of Pm;
SB7: projecting points in the file onto the SPm, and mapping the plane into a two-dimensional coordinate system of the plane SPm;
SB8: deleting points with the distance between the points and O being more than R+0.01 or less than R-0.01 in the two-dimensional coordinate system, and fitting the actual circle center O and the radius R according to a least square method;
SB9: and repeating the step SB5-8 until m0> e, and reducing the influence of noise points on the three-dimensional modeling, so that the processed point cloud data can more accurately reflect the boundary of the inner wall of the actual tunnel, and the simulation of the actual tunnel model is facilitated.
The specific steps of SC are as follows:
SC1: counting the number of point clouds in each segmented file;
SC2: when the statistical quantity is more than 10000, 8000 point clouds are randomly extracted, and original point cloud data are saved and replaced;
SC3: when the number is smaller than 2000, fitting all points in the file into standard circles, calculating coordinates of ten key points in the circles, storing and replacing original point cloud data, wherein the three-dimensional laser scanner is fixed at a specific position of a tunnel, the closer the three-dimensional laser scanner is to the tunnel section, the denser the scanned point cloud data are, the point cloud data acquired in the tunnel section with a larger distance are insufficient, the point cloud data are thinned by adopting a random value method in the section with a large point cloud density, the redundant data part is effectively processed, the deficient key point data are obtained by adopting a circle fitting method in the section with a small point cloud density, and the supplement of the insufficient data part is effectively realized, so that the coordinate calculation of ten key points of an actual section is more accurate.
Referring to fig. 5, three parameters in S3 are the intersection point coordinates, the length of the relaxation curve, and the radius of the arc, respectively, and by determining the parameters of the intersection point coordinates, the length of the relaxation curve, and the radius of the arc in the design line, the two-dimensional plane map and the two-dimensional longitudinal plane map can be quickly constructed.
Modeling of the vertical section in S3:
the straight line between points a to C and the solid line portion of the arc form a partial line drawing of the vertical section. Point D, E is the intersection of the straight line and the circular arc, and point F is the center of the circular arc. Is provided withIs the included angle of two straight lines, the angle +.>According to the cosine law, the following steps are carried out:
wherein a, b, c are the lengths of the line segments BC, AC, AB.
The included angles of the straight lines AB, BC and X coordinate axes are respectively
In practice, the straight line corresponding to BF is an angular bisector, i.e. the included angle between the straight line AB and the X coordinate axis isThereby obtaining the straight line corresponding to BFEquation:
y＝tanα(x-x b )+y b
length l of line segment BF bf Can be according to the triangular formulaObtaining the F point coordinate
Equation (x-x) for obtaining circles f ) 2 +(y-y f ) 2 ＝r b 2 Substituting the equation of the straight lines AB and BC into a circular equation, solving the equation set to obtain the coordinates (x d ，y d )，(x f ，y f )。
The coordinates of the key point D, E are obtained in the above steps, and the size of the x value is compared, so that whether a given arbitrary point falls on a straight line or an arc can be judged. At that time, the given point is on the AB straight line, x can be substituted into the equation of the AB straight line to obtain the y value; when x is d ＜x＜x e When the given point is on the DE arc, substituting x into the equation of the circle to obtain the y value; when x > x e When a given point is on the BC straight line, x can be substituted into the equation of the BC straight line to determine the y value.
Referring to fig. 6, modeling of the plane in S3:
the solid line portions of the straight line, the relaxing curve, and the circular arc between the points a to C constitute a planar partial line diagram. Point D, E is the intersection of the front relaxation curve with the straight line and the circular arc, point F, G is the intersection of the rear relaxation curve with the straight line and the circular arc, and point H is the center of the circular arc. Is provided withIs the included angle of two straight lines, the angle +.>According to the cosine law, the following steps are carried out:
wherein a, b, c are the lengths of the line segments BC, AC, AB.
The included angles of the straight lines AB, BC and X coordinate axes are respectively
The length of the front and rear relaxation curves corresponding to the point B and the radius of the arc are known: l (L) b1 ，l b2 ，
r b According to the corresponding moderation curve formulaCan calculate the rotation angle beta of the front and back relaxation curves b1 ，
β b2 Angle subtraction can calculate the angle of arcAccording to the formula->Can calculate the abscissa x of the front and rear relaxation curves in the relative coordinate system l1 ，x l2 According to the formula->The ordinate y of the front and rear relaxation curves in the relative coordinate system can be obtained l 1，y l2 。
Referring to fig. 7, according to the conversion relationship of rotation and translation in the figure, the following formula is obtained:
x＝x′cos(θ)+y′sin(θ)+x 0
y＝-y′sin(θ)+y′cos(θ)+y 0
let the starting point D coordinate of the relaxation curve be (x) d ，y d ) According to the coordinate axis conversion formula, the moderation curve can be obtainedLine end point D coordinate is (x) e ，y e ) The F, G point coordinates can be found by the same method to be (x f ，y f )，
(x g ，y g ) Substituting the coordinates of the G point into the equation of the straight line BC, the unknown number x can be obtained according to deduction d And y is d The coordinates (x) of the E, F, G point can be obtained accordingly e ，y e )，(x f ，y f )，(x g ，y g )。
The coordinates of the key point D, E, F, G are obtained in the steps, and the size of the x value is compared, so that whether a given arbitrary point falls in a straight line, an arc or a moderation curve section can be judged. When x is less than x d When a given point is on an AB straight line, x can be substituted into an equation of the AB straight line to obtain a y value;
when x is d ＜x＜x e When a given point is on the pre-DE relaxation curve, a coordinate system is established by taking the point D as an original point, the point AB as an x coordinate axis and the perpendicular line of the point AB as a y axis, and the formula is adopted
The relative coordinates of the length L of the relaxation curve (where X is L=l) s Coordinates of the front relaxation curve) according to a coordinate conversion formula, the coordinate value of any point on the front relaxation curve can be obtained;
when x is e ＜x＜x f When a given point is on the EF arc, x can be substituted into the equation of the BC straight line to obtain the y value. In the same way, the coordinate value of any point on the circular arc can be obtained by adopting a mode of obtaining a front relaxation curve.
When x > x f When the given point is on the FG post-relaxation curve, the pre-relaxation curve is calculated, and the G point is selected as the origin point relative to the coordinate system. The coordinate value of any point on the post-relaxation curve can be obtained.
Referring to fig. 8, the specific steps of S5 and S6 are as follows:
sa: according to F (Ω) 1 ，Ω 2 ) Obtaining a normal plane equation SP of a curve at a position of a line center point Pm corresponding to mileagem；
Sb: finding out a point with a distance smaller than 0.01 from SPm in the point cloud data and projecting the point onto the SPm;
sc: mapping the plane into a two-dimensional coordinate system of the plane SPm;
sd: fitting out the actual circle center O and the radius R on the SPm according to a least square method;
se: finding ten key points on the SPm, and calculating the intrusion value of the ten key points and the size of the equipment;
sf: adding the intrusion values of ten identical position points in the section with the interval of 5M to obtain an intrusion value matrix M;
sg: altering Ω within custom scope 1 ，Ω 2 To obtain a plurality of groups of different F' (omega) 1 ，Ω 2 ) Repeating the steps Sa-Sf to obtain a plurality of groups of corresponding intrusion value matrixes M;
sh: constructing a deep neural network, and integrating multiple groups of line parameters F (omega 1 ，Ω 2 ) As input, corresponding to a plurality of groups of intrusion limit matrixes as output, and adjusting network weight parameters until the error of the intrusion limit is smaller than a set threshold;
si: obtaining line parameter F (Ω) 1 ，Ω 2 ) And obtaining the extremum of the function according to the function relation of the intrusion value matrix M and the gradient descent method, and obtaining the optimized design route.
Referring to FIGS. 9-10, P m Represents F (Ω) 1 ，Ω 2 ) The point where the upper mileage is m, s represents F (Ω 1 ，Ω 2 ) E represents F (Ω) 1 ，Ω 2 ) Is the end point mileage, F (Ω) 1 ，Ω 2 ) And (3) a line central line equation which represents the slope of an unconditioned line in a rectangular coordinate system, wherein omega 1 represents a longitudinal section parameter, omega 2 represents a plane parameter, and the extremum of the function in S6 is obtained by adopting a gradient descent method.
According to the scheme, point cloud blocking, point cloud noise reduction and point cloud homogenization pretreatment are carried out on the point cloud data obtained by three-dimensional laser scanning, so that coordinate calculation of ten key points of which the actual section corresponds to the cross section intrusion limit degree is more accurate; according to the design route constructed by the two-dimensional diagrams of the plane and the longitudinal section, ten key point coordinates of the theoretical section are determined, and the intrusion limit value of the section, the equipment dimensions of an evacuation platform, a contact net and the like are more accurately calculated; according to the calculated ten key point invasion limit values of the section, a deep learning model is established, the functional relation between the plane and vertical section parameters and the invasion limit values is learned, and then the extreme value of the function is obtained by adopting a gradient descent method, so that the aim of optimizing the design line is fulfilled.
Related symbol meaning: xi represents the coordinate value of the i-th point x;
yi represents the coordinate value of the ith point y;
l i2 the length of the ith post-mitigation curve is shown.
The above; is only a preferred embodiment of the present invention; the scope of the invention is not limited in this respect; any person skilled in the art is within the technical scope of the present disclosure; equivalent substitutions or changes are made according to the technical proposal of the invention and the improved conception thereof; are intended to be encompassed within the scope of the present invention.
Claims (7)
1. The application method based on the track area point cloud big data is characterized by comprising the following steps of:
s1: acquiring tunnel point cloud data through three-dimensional laser scanning;
s2: preprocessing point cloud data:
SA: and (3) partitioning the point cloud: dividing the point cloud data into a plurality of files according to mileage values of an original design line for storage, and adding indexes for the files;
SB: denoising the point cloud: denoising the point cloud data;
SC: homogenizing point cloud: homogenizing the density of the point cloud data;
s3: and (3) establishing a design line model: constructing a plane two-dimensional graph and a longitudinal section two-dimensional graph in a line through geometric relations among three parameters in the designed line, and constructing a theoretical tunnel model corresponding to the designed line according to the plane two-dimensional graph and the longitudinal section two-dimensional graph; the three parameters are respectively an intersection point coordinate, a relaxation curve length and an arc radius parameter;
s4: calculating ten-point coordinates of a theoretical section: ten key points of a section corresponding to any mileage are extracted from the theoretical tunnel model, and ten key point coordinates of the theoretical section corresponding to the mileage are calculated;
s5: calculating actual cross section ten-point coordinates: finding out ten key point coordinates of an actual section corresponding to the same mileage as the S4 from the point cloud data after the pretreatment of the S2;
s6: calculating an intrusion value and a device size: and (3) calculating the intrusion limit value of the cross section and the size of equipment according to the ten key point coordinates of the theoretical cross section obtained in the step (S4) and the ten key point coordinates of the actual cross section obtained in the step (S5), constructing a deep learning model according to the calculated intrusion limit value of the cross section, and then solving the extremum of the function to obtain the optimized design line.
2. The application method based on the track area point cloud big data according to claim 1, wherein the application method is characterized in that: the SA comprises the following specific steps:
SA1: let Pm0 equal Ps, pm1 equal Ps+0.01;
SA2: obtaining normal plane equations SPm0 and SPm1 of the curve at the Ps according to F (omega 1, omega 2) and Pm0 and Pm1;
SA3: finding points between SPm0 and SPm1 from the file and storing the points in a file 0;
SA4: let Pm0 equal to Pm1, pm1 equal to Pm1+0.01;
SA5: repeating steps SA 1-SA 4 until m0> e;
the Pm representsF（ Ω1 ， Ω2 ) The upper mileage is the point of m, s representsF（ Ω1 ， Ω2 ) Is the starting mileage of (2), where e representsF（ Ω1 ， Ω2 ) End mileage of (1), saidF（ Ω1 ， Ω2 ) Line centreline equation representing the slope of the line without line adjustment in rectangular coordinate system, saidΩ1Representing a longitudinal section parameter, saidΩ2Representing the plane parameters.
3. The application method based on the track area point cloud big data according to claim 2, wherein the application method is characterized in that: in the SA, for the section of any mileage m on a new line, the distances from all points in the index file from the filem-1 to the filem+1 to the section of the mileage m are required to be calculated, and the point with the distance smaller than 1 cm from the section is projected on the section, so that the point cloud of the section can be obtained.
4. The application method based on the track area point cloud big data according to claim 3, wherein the application method is characterized in that: the SB comprises the following specific steps:
SB1: let m=s+0.005;
SB2: according toF（ Ω1 ， Ω2 ) And Pm to obtain a normal plane equation SPm of a curve at the position of Pm;
SB3: projecting points in the file onto the SPm, and mapping the plane into a two-dimensional coordinate system of the plane SPm;
SB4: fitting an actual circle center O and a radius R in the two-dimensional coordinate system according to a least square method;
SB5: let m equal m+0.01;
SB6: according toF（ Ω1 ， Ω2 ) And Pm to obtain a normal plane equation SPm of a curve at the position of Pm;
SB7: projecting points in the file onto the SPm, mapping a plane into the two-dimensional coordinate system of the plane SPm;
SB8: deleting points with the distance between the points and O being more than R+0.01 or less than R-0.01 in the two-dimensional coordinate system, and fitting out the actual circle center O and the radius R according to a least square method;
SB9: steps SB 5-SB 8 are repeated until m0> e.
5. The application method based on the track area point cloud big data according to claim 1, wherein the application method is characterized in that: the specific steps of the SC are as follows:
SC1: counting the number of point clouds in each segmented file;
SC2: when the statistical quantity is more than 10000, 8000 point clouds are randomly extracted, and the original point cloud data are saved and replaced;
SC3: and when the number is smaller than 2000, fitting all points in the file into a standard circle, calculating coordinates of ten key points in the circle, and storing and replacing the original point cloud data.
6. The application method based on the track area point cloud big data according to claim 1, wherein the application method is characterized in that: the specific steps of S5 and S6 are as follows:
sa: according toF（ Ω1 ， Ω2 ) Obtaining a normal plane equation SPm of a curve at the position of the Pm by a line center point Pm corresponding to the mileage;
sb: finding out points with the distance smaller than 0.01 from the SPm in the point cloud data and projecting the points onto the SPm;
sc: mapping the plane into a two-dimensional coordinate system of the plane SPm;
sd: fitting out the actual circle center O and the radius R on the SPm according to a least square method;
se: finding ten key points on the SPm, and calculating the intrusion value of the ten key points and the size of the equipment;
sf: adding the intrusion values of ten identical position points in the section with the interval of 5M to obtain an intrusion value matrix M;
sg: altering within custom scopeF（ Ω1 ， Ω2 ) Obtaining a plurality of groups of differentF（ Ω1 ， Ω2 ) Repeating the steps Sa-Sf to obtain a plurality of groups of corresponding intrusion value matrixes M;
sh: constructing a deep learning model, and combining multiple groups of line parametersF（ Ω1 ， Ω2 ) As input, corresponding to a plurality of groups of intrusion limit matrixes as output, and adjusting network weight parameters until the error of the intrusion limit is smaller than a set threshold;
si: obtaining line parametersF（ Ω1 ， Ω2 ) With the intrusion value matrix MAnd obtaining the extremum of the function according to the function relation and the gradient descent method, and obtaining the optimized design route.
7. The application method based on the track area point cloud big data according to claim 1, wherein the application method is characterized in that: and S6, the extremum of the function is obtained by adopting a gradient descent method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910652405.6A CN110598239B (en) | 2019-07-19 | 2019-07-19 | Application method based on track area point cloud big data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910652405.6A CN110598239B (en) | 2019-07-19 | 2019-07-19 | Application method based on track area point cloud big data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110598239A CN110598239A (en) | 2019-12-20 |
| CN110598239B true CN110598239B (en) | 2023-11-17 |
Family
ID=68853040
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910652405.6A Active CN110598239B (en) | 2019-07-19 | 2019-07-19 | Application method based on track area point cloud big data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110598239B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111813775B (en) * | 2020-06-03 | 2023-03-24 | 中铁第四勘察设计院集团有限公司 | Tunnel point cloud data processing method and device and storage medium |
| CN112199802A (en) * | 2020-08-15 | 2021-01-08 | 中建安装集团有限公司 | Pipeline prefabricating and installing method based on track region point cloud big data |
| CN111968233A (en) * | 2020-08-15 | 2020-11-20 | 中建安装集团有限公司 | Method for installing BIM (building information modeling) and point cloud positioning and track interval communication equipment |
| CN112012635A (en) * | 2020-08-15 | 2020-12-01 | 中建安装集团有限公司 | Civil air defense door construction method based on point cloud big data |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105756711A (en) * | 2016-03-02 | 2016-07-13 | 中交第二航务工程局有限公司 | Tunnel construction primary support limit invasion monitoring analysis early-warning method based on three-dimensional laser scanning |
| WO2017101150A1 (en) * | 2015-12-14 | 2017-06-22 | 深圳先进技术研究院 | Method and device for calibrating structured-light three-dimensional scanning system |
| CN106930784A (en) * | 2017-03-08 | 2017-07-07 | 中交第二航务工程局有限公司 | Tunnel monitoring method based on 3 D laser scanning |
| CN108171796A (en) * | 2017-12-25 | 2018-06-15 | 燕山大学 | A kind of inspection machine human visual system and control method based on three-dimensional point cloud |
| CN108596169A (en) * | 2018-03-12 | 2018-09-28 | 北京建筑大学 | Piecemeal signal conversion based on video streaming image and object detection method and device |
| CN108917712A (en) * | 2018-07-10 | 2018-11-30 | 湖南城市学院 | A kind of Tunnel automation monitoring system and method based on three-dimensional laser scanning technique |
-
2019
- 2019-07-19 CN CN201910652405.6A patent/CN110598239B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017101150A1 (en) * | 2015-12-14 | 2017-06-22 | 深圳先进技术研究院 | Method and device for calibrating structured-light three-dimensional scanning system |
| CN105756711A (en) * | 2016-03-02 | 2016-07-13 | 中交第二航务工程局有限公司 | Tunnel construction primary support limit invasion monitoring analysis early-warning method based on three-dimensional laser scanning |
| CN106930784A (en) * | 2017-03-08 | 2017-07-07 | 中交第二航务工程局有限公司 | Tunnel monitoring method based on 3 D laser scanning |
| CN108171796A (en) * | 2017-12-25 | 2018-06-15 | 燕山大学 | A kind of inspection machine human visual system and control method based on three-dimensional point cloud |
| CN108596169A (en) * | 2018-03-12 | 2018-09-28 | 北京建筑大学 | Piecemeal signal conversion based on video streaming image and object detection method and device |
| CN108917712A (en) * | 2018-07-10 | 2018-11-30 | 湖南城市学院 | A kind of Tunnel automation monitoring system and method based on three-dimensional laser scanning technique |
Non-Patent Citations (2)
| Title |
|---|
| A Deep Learning Approach to Mid-air Gesture Interaction for Mobile Devices from Time-of-Flight Data;Thomas Kopinski等;《MOBIQUITOUS 2016》;20161128;全文 * |
| 基于激光雷达的地铁隧道侵限检测方法;刘璐宾等;《机械工程与自动化》;20180607(第03期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110598239A (en) | 2019-12-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110598239B (en) | Application method based on track area point cloud big data | |
| CN107402001B (en) | Ultrahigh-rise building construction deviation digital inspection system and method based on 3D scanning | |
| CN106871880B (en) | The method that rib position, half-breadth and height carry out ship closure is calculated using three-dimensional coordinate | |
| CN110095060A (en) | Steel construction rapid quality detection method based on 3-D scanning technology | |
| CN113587835A (en) | Method for checking and accepting bridge engineering quality by using three-dimensional laser scanning technology | |
| CN102721371B (en) | Method for calculating sag of power transmission line | |
| US20230112991A1 (en) | Method of high-precision 3d reconstruction of existing railway track lines based on uav multi-view images | |
| CN110763194B (en) | Method for monitoring surface subsidence of mining area by ground three-dimensional laser scanning and without target | |
| CN110888143B (en) | Bridge through measurement method based on unmanned aerial vehicle airborne laser radar | |
| CN110348125B (en) | Visual discrimination method for stability of block | |
| CN103884291B (en) | Building surface plastic deformation monitoring method based on NURBS parametric surface | |
| CN109255837B (en) | Construction method of efficient B-spline surface for laser radar point cloud data processing | |
| CN102609940A (en) | Method for processing errors generated by point cloud registration in process of surface reconstruction of measuring object by using ground laser scanning technique | |
| CN111968233A (en) | Method for installing BIM (building information modeling) and point cloud positioning and track interval communication equipment | |
| CN106092137B (en) | The outdoor calibrator (-ter) unit and method of a kind of vehicle-mounted three-dimensional laser pavement detection system | |
| CN116993928B (en) | Urban engineering mapping method and system based on unmanned aerial vehicle remote sensing technology | |
| CN109682303A (en) | Curve PC track girder precision three-dimensional detection method based on BIM technology | |
| CN106482700A (en) | A kind of digitized House Property Area Surveying measuring method of sketch direct mapping | |
| CN108801221B (en) | Quick and fine dereferencing method for surface mine slope rock mass joint scale based on digital photogrammetry | |
| CN103344297B (en) | The reservoir storage survey method of underground water seal cave depot | |
| CN113591182A (en) | Tunnel wind speed single-point measurement sensor arrangement and wind speed correction method | |
| CN110111421B (en) | Method and device for mobile mapping point cloud | |
| CN114332402A (en) | Ground-based and handheld laser scanning-fused steel bridge simulation pre-assembly method | |
| CN113865570A (en) | Steel structure circular stand column verticality measuring method | |
| CN113689394A (en) | Tunnel primary support flatness detection method based on three-dimensional point cloud |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20230914 Address after: No.9, Sanlihe Road, Haidian District, Beijing 100089 Applicant after: CHINA CONSTRUCTION BASE BACILITIES Co.,Ltd. Applicant after: CHINA CONSTRUCTION SOUTH INVESTMENT CO.,LTD. Applicant after: CHINA CONSTRUCTION INDUSTRIAL & ENERGY ENGINEERING GROUP Co.,Ltd. Applicant after: Shenzhen Qianhai Xirui big data Culture Co.,Ltd. Applicant after: GUANGZHOU METRO DESIGN & RESEARCH INSTITUTE Co.,Ltd. Applicant after: Zhongyitian Construction Engineering Technology (Shenzhen) Co.,Ltd. Address before: No. 6 Wenlan Road, Qixia District, Nanjing City, Jiangsu Province, 210000 Applicant before: CHINA CONSTRUCTION INDUSTRIAL & ENERGY ENGINEERING GROUP Co.,Ltd. Applicant before: Shenzhen Qianhai Xirui big data Culture Co.,Ltd. Applicant before: GUANGZHOU METRO DESIGN & RESEARCH INSTITUTE Co.,Ltd. Applicant before: Zhongyitian Construction Engineering Technology (Shenzhen) Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |