CN115937098A - Power fitting crimping quality visual detection method - Google Patents
Power fitting crimping quality visual detection method Download PDFInfo
- Publication number
- CN115937098A CN115937098A CN202211398336.9A CN202211398336A CN115937098A CN 115937098 A CN115937098 A CN 115937098A CN 202211398336 A CN202211398336 A CN 202211398336A CN 115937098 A CN115937098 A CN 115937098A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- point
- model
- electric power
- cloud model
- 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.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Length Measuring Devices With Unspecified Measuring Means (AREA)
Abstract
The invention relates to a visual detection method for crimping quality of an electric power fitting, which comprises the following steps: s1, point cloud data acquisition: acquiring space point cloud data of the electric power fitting to be detected, and converting the space point cloud data into a universal txt file form as original point cloud data; s2, point cloud data denoising: denoising a point cloud model of the electric power fitting to be detected, and removing outliers in the original point cloud data; s3, point cloud model registration: unifying a reference coordinate system of the target point cloud model and a reference coordinate system of the reference point cloud model; s4, extracting point cloud model parameters: dividing the point cloud model of the electric power fitting to be detected, and extracting parameters of the point cloud model of the electric power fitting to be detected; and S5, point cloud difference value visualization, including point cloud difference value quantitative calculation and point cloud difference value visualization. Compared with the prior art, the method has the advantages of high precision, high efficiency, high accuracy and the like.
Description
Technical Field
The invention relates to the field of detection of crimping quality of power fittings, in particular to a visual detection method of crimping quality of power fittings.
Background
The strain clamp equal-pressure-connection type electric power fittings are used in a large number in connection of overhead transmission lines of China on a lead and a ground wire, bear passing charge loads, bear tension of the lead or the ground wire and are important stress and conduction equipment on the power line. However, the construction of the strain clamp belongs to hidden engineering, and the strain clamp is difficult to disassemble once being installed. Therefore, the installation quality of the tension clamp equal-pressure connection type electric power fitting is guaranteed, and the method has very important significance for guaranteeing reliable operation of a line and safe power supply.
At present, the quality detection method of the crimping type electric power fitting mainly comprises the following steps: manual appearance dimension measurement, grip strength test and X-ray nondestructive testing.
1) The manual external dimension measurement is to use a vernier caliper to measure and record the external dimension of the strain clamp before and after crimping, and mainly depends on manual measurement and visual identification. On one hand, because the measurement by using the vernier caliper is influenced by factors such as artificial operation errors, subjective judgment and the like, multiple measurement confirmation is often required, but even the result of multiple measurements by the same person has deviation, and the accuracy of the measurement result is difficult to ensure. On the other hand, repeated measurement and confirmation is time-consuming and labor-consuming, measurement data needs to be copied manually, and the problems that the records are complicated, the formats are difficult to unify, the results are difficult to store digitally and the like exist.
2) The grip test is a destructive sampling method performed in a laboratory, and the method can intuitively detect the external and internal crimping states of the electric power fitting. However, the grip test is a spot check test, and the crimping state of the hardware obtained by the test cannot represent the crimping state of the hardware used in the line; meanwhile, the spot check test causes the loss of hardware and a lead, and is not suitable for field detection.
3) The X-ray nondestructive testing utilizes the principle that rays absorbed by different metal structures are different, signals received by an imager are transmitted to a computer, and images of the internal crimping state of the electric power fitting can be conveniently, quickly and accurately captured under the condition that a detected object is not damaged. However, the method is mainly used for detecting the internal crimping quality of the electric power fitting, is the second step of the crimping quality detection, and cannot accurately measure the size of the external crimping.
The measurement of the size of the power fitting after crimping is particularly important, and the measurement is the first step for checking whether the crimping quality is qualified or not and is the most critical step. If the size does not reach the standard, the crimping is required to be carried out again, and the detection of the internal crimping quality is required only after the size reaches the standard. Therefore, how to achieve efficient, high-precision and digital detection of the sizes of the electric power fittings before and after crimping and realize visualization of detection results becomes a technical problem to be solved.
Disclosure of Invention
The invention aims to provide a visual detection method for the crimping quality of an electric power fitting to overcome the defects in the prior art.
The purpose of the invention can be realized by the following technical scheme:
a visual detection method for crimping quality of an electric power fitting comprises the following steps:
s1, point cloud data acquisition: acquiring space point cloud data of the electric power fitting to be detected, converting the space point cloud data into a universal txt file form, using the txt file form as original point cloud data, and establishing a point cloud model of the electric power fitting to be detected;
s2, point cloud data denoising: denoising a point cloud model of the electric power fitting to be detected by adopting a radius filtering algorithm, and removing outliers in original point cloud data;
s3, point cloud model registration: taking a point cloud model of a standard crimping power fitting as a reference point cloud model, taking a point cloud model of a power fitting to be detected as a target point cloud model, and unifying a reference coordinate system of the target point cloud model and a reference coordinate system of the reference point cloud model;
s4, point cloud model parameter extraction: dividing the point cloud model of the electric power fitting to be detected, and extracting parameters of the point cloud model of the electric power fitting to be detected through a model fitting algorithm;
s5, point cloud difference value visualization: and carrying out quantitative calculation on difference values of the registered reference point cloud model and the registered target point cloud model, converting the measured point cloud difference values into corresponding color parameters, endowing a corresponding color map to the three-dimensional model, and carrying out visual display on a detection result.
Further, the point cloud data in the step S1 is acquired by using a high-precision handheld laser radar scanner.
Further, the step S2 specifically includes the following steps:
s201, any point p in point cloud model of electric power fitting to be measured i ＝(x i ,y i ,z i ) T Determining a three-dimensional spherical space region with the spherical center as the center and the radius as r, and calculating the total number C of point clouds in the neighborhood of r r (p i )；
S202, counting total number C of point clouds in r neighborhood of point cloud model of to-be-measured electric power fitting r (p i ) Performing statistical analysis to determine a point cloud number judgment threshold C for removing noise point clouds T ：
Wherein the content of the first and second substances,
representing the average total number of point clouds in the r neighborhood of the point cloud model, N representing the number of points in the point cloud model, σ r Representing the standard deviation of the number of point clouds in the r neighborhood of the point cloud model, and k represents a confidence coefficient interval;
s203, judging a threshold value C according to the number of point clouds T And removing outliers in the original point cloud data.
Further, step S203 specifically includes: traversing point cloud data, e.g. point p i Point cloud total C r (p i ) Less than point cloud number judgment threshold C T Then the point p is considered i Removing noise point cloud data; otherwise, the point cloud data is reserved.
Further, the point cloud model registration in step S3 specifically includes the following steps:
s301, point cloud feature extraction: extracting a point cloud Global Feature Descriptor (GFD) by adopting a principal component analysis method and a coordinate transformation idea, extracting a point cloud local feature descriptor (FPFH) by adopting a rapid point feature histogram algorithm, and obtaining a point cloud Fusion Feature Descriptor (FFD) by integrating the GFD and the FPFH;
s302, point cloud rough registration: completing feature point matching by measuring the similarity of the point cloud fusion feature descriptor FFD, selecting m groups of matching point pairs with local invariant feature similarity from high to low rank according to the one-to-one correspondence of the matching point pairs, and calculating a corresponding rigid body transformation matrix to realize the preliminary coincidence of the point cloud model;
s303, point cloud fine registration: and on the basis of the rough point cloud registration, performing refined registration by adopting an ICP (inductively coupled plasma) algorithm to obtain an optimized point cloud model registration result.
Further, the parameters in step S4 include a crimping length and a crimping-to-moment.
Further, the point cloud model parameter extraction in step S4 specifically includes:
s401, point cloud model segmentation: selecting a proper initial clustering point in a specific area according to the point cloud model registration result, and realizing clustering segmentation of the point cloud model by adopting an improved k-means algorithm to obtain a final clustering segmentation result;
s402, axial layering of the point cloud model: the method comprises the following steps of intercepting a plurality of point clouds with certain thicknesses in a layering mode along the axial direction of the electric power fitting, and projecting each layer of intercepted point clouds on a corresponding plane perpendicular to a central axis to obtain layered point cloud slices;
s403, point cloud model parameter extraction: and extracting six linear boundaries of the point cloud slice, finding data points on the six linear boundaries, performing model fitting on the found data points to obtain boundary straight lines of the point cloud slice, and calculating parameters of the layered point cloud according to a linear equation.
Further, the clustering segmentation in step S401 divides the point cloud model into a compression region model and a non-compression region model.
Further, in step S403, an RANSAC algorithm is used to extract six straight line boundaries of the point cloud slice, and an overall least square method is used to perform model fitting on the found data points.
Further, the point cloud difference value visualization in step S5 specifically includes the following steps:
s501, point cloud difference value quantization calculation: determining the point correspondence between the target point cloud model and the reference point cloud model by adopting a k-d tree algorithm and a closest point correspondence idea, and calculating corresponding point cloud difference values by utilizing the point correspondence;
s502, point cloud difference value visualization method: and establishing a color map matrix, determining corresponding color parameters according to the linear proportional relation between the point cloud difference value and the maximum allowable error value, endowing the corresponding color map to a three-dimensional model, and performing visual display of the detection result.
Compared with the prior art, the invention has the following beneficial effects:
1. according to the invention, on the basis of analyzing the crimping model of the electric power fitting, the parameter detection problem of the electric power fitting is converted into the problem of model fitting of the outer contour form of the axial tangent plane and automatic extraction of the three-dimensional coordinate of the central point, a computer worker is realized by introducing an artificial intelligence algorithm, the automatic, digital and high-precision measurement of the external dimension parameters of the electric power fitting is effectively realized, the precision, the efficiency and the digital level of the quality detection of the electric power fitting are obviously improved, and the engineering application challenge is solved.
2. The invention adopts the point cloud fusion feature descriptor FFD which integrates the advantages of GFD and FPFH, solves the problem of limited description performance of a single point cloud feature descriptor, improves the registration accuracy of a point cloud model in a limited way, and improves the accuracy of point cloud parameter extraction and deviation calculation.
3. The invention adopts RANSAC algorithm and integral least square method to realize boundary line fitting of point cloud slices, and has higher noise point robustness and fitting accuracy.
4. The method and the device realize three-dimensional visualization of the structural size deviation of the electric power fitting, can facilitate operators to intuitively position the region and degree of structural damage of the electric power fitting in the first time, can realize quick judgment of the crimping quality of the electric power fitting, and effectively improve the detection efficiency and accuracy of the crimping quality of the electric power fitting.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a schematic diagram of a point cloud r neighborhood of the present invention;
FIG. 3 is a schematic diagram of the calculation of a point cloud r neighborhood parameter of the present invention;
FIG. 4 is a flow chart of the coarse registration of the point cloud of the present invention;
FIG. 5 is a flow chart of fine registration of the point cloud of the present invention;
FIG. 6 is a flow chart of an electric power fitting point cloud model segmentation algorithm of the present invention;
FIG. 7 is a schematic diagram illustrating a point cloud difference value quantization calculation according to the present invention;
FIG. 8 is a point cloud model initial position diagram of strain clamp #1 and strain clamp #2 of the present invention;
FIG. 9 is a result diagram of the rough registration of the point cloud models of strain clamp #1 and strain clamp #2 of the present invention;
FIG. 10 is a diagram showing the fine registration results of the point cloud models of the strain clamp #1 and the strain clamp #2 according to the present invention;
FIG. 11 is a point cloud model initial position diagram of strain clamp #1 and strain clamp #3 of the present invention;
FIG. 12 is a result diagram of the rough registration of the point cloud models of strain clamp #1 and strain clamp #3 of the present invention;
FIG. 13 is a diagram showing the fine registration results of the point cloud models of strain clamp #1 and strain clamp #3 of the present invention;
FIG. 14 is a graph of strain clamp #1 versus moment distribution of the present invention;
FIG. 15 is a graph of strain clamp #2 versus moment distribution of the present invention;
FIG. 16 is a graph of the distribution of the strain clamp #3 to the moment according to the present invention;
FIG. 17 is a graph showing the results of visualization of structural deviations for strain clamp #2 of the present invention;
fig. 18 is a graph showing the result of visualization of structural deviation of the strain clamp #3 of the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and the specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Example 1
As shown in fig. 1, a method for visually detecting crimping quality of an electric power fitting includes the following steps:
s1, point cloud data acquisition: and acquiring space point cloud data of the electric power fitting to be detected by adopting a high-precision handheld laser radar scanner, converting the space point cloud data into txt file form, and establishing a point cloud model of the electric power fitting to be detected as original point cloud data.
S2, point cloud data noise reduction: and denoising the point cloud model of the electric power fitting to be detected by adopting a radius filtering algorithm, and removing outliers in the original point cloud data.
The point cloud data acquired by the high-precision handheld laser radar scanner has ideal data quality, the noise point cloud contained in the point cloud data often appears in an outlier form, and the total data amount of the noise point cloud is relatively small. The method adopts a radius filtering algorithm to carry out filtering and denoising, and comprises the following specific steps:
s201, any point p in point cloud model of electric power fitting to be measured i ＝(x i ,y i ,z i ) T Determining a three-dimensional spherical space region with the radius of r and the spherical center of the three-dimensional spherical space region, and calculating the total number C of point clouds in the neighborhood of r r (p i )；
S202, counting the total number C of point clouds in r neighborhood of the point cloud model of the electric power fitting to be detected r (p i ) Performing statistical analysis to determine point cloud number judgment threshold C for removing noise point cloud T ：
Wherein the content of the first and second substances,
representing the average total number of point clouds in the r neighborhood of the point cloud model, N representing the number of points in the point cloud model, σ r Representing the standard deviation of the number of point clouds in the r neighborhood of the point cloud model, and k represents a confidence coefficient interval coefficient (k = 3);
s203, judging a threshold value C according to the number of point clouds T Removing the original point cloud numberAccording to the outliers, the outliers are specifically: traversing point cloud data, e.g. point p i Total number of point clouds C r (p i ) Less than point cloud number judgment threshold C T Then the point p is considered i Removing noise point cloud data; otherwise, the point cloud data is reserved.
S3, point cloud model registration: and taking the point cloud model of the standard crimping power fitting as a reference point cloud model, taking the point cloud model of the power fitting to be detected as a target point cloud model, and unifying a reference coordinate system of the target point cloud model and a reference coordinate system of the reference point cloud model.
S301, point cloud feature extraction: and extracting a point cloud Global Feature Descriptor (GFD) by adopting a principal component analysis method and a coordinate transformation idea, extracting a point cloud local feature descriptor (FPFH) by adopting a rapid point feature histogram algorithm, and obtaining a point cloud Fusion Feature Descriptor (FFD) by integrating the GFD and the FPFH.
The global feature descriptor GFD constructs a reference coordinate system corresponding to the electric power fitting through a Principal Component Analysis (PCA), and converts three-dimensional coordinate information (x, y, z) of the point cloud into distance and angle parameters under a spherical coordinate system by using a coordinate conversion idea so as to represent the spatial position relation of the point cloud relative to the electric power fitting. The method comprises the following specific steps:
1) Solving the centroid coordinates of the electric power fitting according to the point cloud distribution, and setting the centroid coordinates as the origin of a reference coordinate system;
2) Centralizing the point cloud data, calculating the covariance matrix C, performing characteristic decomposition on the covariance matrix C, and solving the characteristic value lambda 0 ≥λ 1 ≥λ 2 And its feature vector v 0 、v 1 、v 2 ；
3) Constructing a reference coordinate system corresponding to the electric power fitting: o = p c ，v o-x ＝v 2 ，v o-y ＝v 1 ，v o-z ＝v 0 ；
4) For point cloudsAny point p in the data i Computing its global feature descriptor in the above-mentioned reference coordinate system
Wherein:
wherein the parameter C =2 ε (f) xoy (p i ) -1, ∈ (x) stands for a step function and ∈ (0) =1; f. of xoy (p) then the plane equation representing the plane of the reference coordinate system xoy, f xoy (p)＝[(x,y,z) T -p c ]·v o-z 。
The point cloud local feature descriptor FPFH parameterizes the difference degree between adjacent points by calculating the deviation angle between the surface normal and the normal of the adjacent points of a certain feature point, thereby obtaining the complete description of the point cloud geometric attributes. The method comprises the following specific steps:
1) As shown in FIG. 2, a query point D is determined q K neighborhood of (c); calculating any two points D s And D t With an inter-euclidean distance d and a corresponding normal n s And n t Defining a local coordinate system UVW at one point thereof, as shown in fig. 3; obtaining a parameterized feature representation:
2) Computing the distance between all pairs of points in the k neighborhood
Four sets of values, which are used for representing the position relation between any two points; the simplified point feature histogram SPFH is calculated and weighted and summed to obtain the FPFH value.
S302, point cloud rough registration: as shown in fig. 4, feature point matching is completed by measuring the similarity of the point cloud fusion feature descriptor FFD, and by using the one-to-one correspondence of the matching point pairs, m groups of matching point pairs with local invariant feature similarity from high to low rank are selected to calculate a corresponding rigid body transformation matrix, so as to realize the preliminary registration of the point cloud model, and the specific steps are as follows:
1) Using FFD as similarity criterion to establish two corresponding point sets P i And Q i (ii) a Randomly selecting 3 characteristic point pairs (p) from all the characteristic point pairs i ,q i ) I =1,2,3; calculating the corresponding initial rotation matrix R by using a singular value decomposition method 0 And translation matrix T 0 ；
2) Traversing all the characteristic point pairs, and calculating any one characteristic point pair (p) j ,q j ) Measurement error e generated under action of initial rigid body transformation matrix j ：
e j ＝||p j -R 0 q j -T 0 ||
Will measure the error e j And an error determination threshold e t And (3) comparison: if e j ＜e t Then, consider the characteristic point pair (p) j ,q j ) Is a correct pair of characteristic points; otherwise, the characteristic point pair (p) is considered j ,q j ) Is an erroneous pair of characteristic points;
3) Counting the correct number n of pairs of feature points k (ii) a Selecting corresponding n k Obtaining the initial rotation matrix R by the maximum rigid body transformation result 0 And translation matrix T 0 。
S303, point cloud fine registration: and on the basis of point cloud rough registration, further reducing registration errors between the reference point cloud model and the target point cloud model so as to obtain a high-precision point cloud registration result. As shown in FIG. 5, the method of the invention adopts an ICP algorithm to perform point cloud model fine registration, and comprises the following specific steps: .
1) For any one reference point cloud p i Traversing the data points in the target point cloud set, and selecting the corresponding matching point q according to the principle of the closest point k To obtain two corresponding point sets P i And Q i ；
2) According to the corresponding point set P i And Q i Can construct the objective function f (R) k ,T k ) And solving the optimal rotation matrix R k And translation matrix T k ；
3) According to a rotation matrix R k And translation matrix T k To obtain a new target point cloud Q * (ii) a Calculating position optimized target point cloud Q * Average distance error d from reference point cloud P:
4) Judging whether the calculated average distance error d meets the convergence condition: if the average distance error d is smaller than the judgment threshold tau set by the algorithm, the algorithm is ended; otherwise, the algorithm continues iterative computation until the iterative times of the algorithm are greater than the preset maximum iterative times, and then the iterative computation is stopped. Obtaining a final rotation matrix R k And translation matrix T k 。
S4, extracting point cloud model parameters: and segmenting the point cloud model of the electric power fitting to be detected, and extracting parameters of the point cloud model of the electric power fitting to be detected through a model fitting algorithm.
S401, point cloud model segmentation: as shown in fig. 6, selecting an appropriate initial clustering point in a specific area according to the point cloud model registration result, and implementing clustering segmentation of the point cloud model by using an improved k-means algorithm, and dividing the point cloud model into a compression joint area model and a non-compression joint area model, specifically comprising the following steps:
1) For any one target point cloud p i Determining a point cloud data set P in the r neighborhood thereof i Calculating to obtain a curved surface equation of the r neighborhood by utilizing a quadratic surface fitting method; according to the first basic form of the surface equation, calculating to obtain the firstBasic quantities E, F, G; calculating to obtain second basic quantities L, M and N according to a second basic form of the curved surface equation; by using the first basic quantity and the second basic quantity, the target point cloud p can be calculated i Gaussian curvature K, mean curvature H and principal curvature K 1 And k is 2 ：
2) For arbitrary point cloud data p i And calculating to obtain its point cloud characteristic vector F (p) i )；
F(p i )＝[x i ,y i ,z i ,K i ,H i ,k 1i ,k 2i ]
3) For the registered target point cloud model, the spatial range of the registered target point cloud model is roughly determined; according to the priori knowledge, randomly selecting a clustering center from different point cloud model areas as an initial clustering center;
4) For arbitrary point cloud data p i Calculating it to each cluster center c i Point cloud feature distance D (p) i ,c j )：
D(p i ,c j )＝||F(p i )-F(c j )||
5) For arbitrary point cloud data p i According to it to each cluster center c i Determining the clustering region to which the point cloud characteristic distance belongs:
S(p i )＝argmin j D(p i ,c j )
wherein S is i Denoted as the ith clustering region, S i Cluster center of (a) is equal to c i (ii) a Thus obtaining all current clustering results S 1 ,S 2 ,…,S N }；
6) According to the current clustering result S 1 ,S 2 ,…,S N And updating the clustering centers of the clustering areas again to obtain a brand-new clustering center result { c } 1 ,c 2 ,…,c N }，
7) Repeating the steps 4) to 6) until the updated cluster center { c) } 1 ,c 2 ,…,c N And stopping clustering until no change occurs or the maximum iteration times are reached, and obtaining a final clustering segmentation result.
S402, axial layering of the point cloud model: and intercepting a plurality of point clouds with certain thickness in a layering mode along the axial direction of the electric power fitting, and respectively projecting each layer of intercepted point clouds on a corresponding plane perpendicular to the central axis to obtain a layering point cloud slice.
The laser radar point cloud data is a discrete data form, and if the outline of the electric power fitting point cloud model section is obtained by simply selecting a way of intersecting the point cloud and a plane, the number of outline point clouds is possibly small, and the actual electric power fitting model condition is not sufficiently restored. Therefore, the method introduces the idea of slicing, and replaces a pure plane with a point cloud slice with a certain thickness. Setting a proper axial layering number n, and cutting the point cloud model into n point cloud layering along the direction of the central axis of the electric power fitting (S) 1 ,S 2 ,…,S n }。
S403, point cloud model parameter extraction: extracting six straight line boundaries of the point cloud slice by using an RANSAC algorithm, finding data points on the six straight line boundaries, performing model fitting on the found data points by using an integral least square method to obtain a boundary straight line of the point cloud slice, and calculating parameters of the layered point cloud according to a linear equation.
For an electric power fitting, a main object of quality detection is performed in a crimping area of the electric power fitting; the electric power fitting is constructed by a regular cylinder before crimping, and is changed into a regular hexagonal prism after crimping, so that main parameters reflecting the crimping condition of the electric power fitting comprise crimping length and moment. The method adopts the idea of model fitting to fit the layered section model of the electric power fitting and extracts parameters such as the moment. The method comprises the following specific steps:
1) For renHierarchical data set S of meaning sub-model i ∈{S 1 ,S 2 ,…,S n Determining projection data in the axial direction, and performing linear detection and extraction on the projection data by using a RANSAC algorithm, so as to obtain six linear boundaries of layered data:
a. from the current hierarchical data set S i Randomly selecting two data points, and determining a corresponding linear equation y = kx + b according to the two data points;
b. traversing a hierarchical data set S i At any point p in i Calculating a point p i The distance to the fitted straight line is,
will d i And a judgment threshold value d t And (3) comparison: if d is i ＜d t Then, consider point p i Is the "local interior point" of the fitted straight line; otherwise, consider point p i Is the "outlier" of the fitted line;
c. number N of local points of statistical fitting straight line k (ii) a Selecting N k A maximum straight line model;
2) Dividing the 6 linear equations extracted in the step 1) into 3 groups of parallel lines, and calculating the distance d between the parallel lines of each group 1 、d 2 、d 3 (ii) a The corresponding pair of moments can be determined by solving the maximum of the distances of the three sets of parallel lines:
S L ＝max(d 1 ,d 2 ,d 3 )
3) According to the 6 linear equations extracted in the step 1), the corresponding 6 linear intersection points (x) can be determined j ,y j ,z j ) J =1,2, \8230;, 6, the center point of the layered cross-section can be determined from the coordinate values of the intersection points:
4) Determining hierarchical centers for all submodelsSet of points { c 1 ,c 2 ,…,c r H, will c 1 As a start and stop point, c r As the termination point, determining a fitting equation l of the central axis of the electric power fitting c (ii) a Respectively calculating the distance between the residual layered center and the fitted linear equation { d 2 ,d 3 …,d k-1 }; therefore, the maximum chord height h of the electric power fitting can be determined;
h＝max(d 2 ,d 3 ,…,d k-1 )
5) Counting and recording all the appearance size parameters of the electric power fitting: a crimp length L; crimping opposite moment S L (ii) a The maximum chord height h at the bend.
S5, point cloud difference value visualization: and carrying out quantitative calculation on difference values of the registered reference point cloud model and the registered target point cloud model, converting the measured point cloud difference values into corresponding color parameters, endowing a corresponding color map to the three-dimensional model, and carrying out visual display on a detection result.
S501, point cloud difference value quantization calculation: as shown in fig. 7, the point cloud difference value quantization calculation is to calculate the difference between the target point cloud model and the reference point cloud model by comparing the registered reference point cloud model and the target point cloud model; if the power fitting generates a certain deformation, the curved surface in fig. 7 has a corresponding deformation deviation. The specific calculation process of the point cloud difference value is as follows:
1) Aiming at any point q in electric power fitting target point cloud i The nearest data point p in the reference point cloud is found by the k-d tree nearest neighbor search algorithm i Calculating the point q i And point p i A distance d between M ；
In the formula (x) pi ,y pi ,z pi ) Represents a point p i (x) three-dimensional coordinate value of (c) qi ,y qi ,z qi ) Represents a point q i Three-dimensional coordinate values of (a).
2) Set point q i Is a point
If the electric power fitting does not have structural deviation, the curved surface does not have deformation deviation, and the point q is i And a point->
Completely coincide, then point p i And a point->
A distance d between R Equivalent to the point cloud registration error, can be represented by the root mean square error of the distance between the matching point pairs;
4) According to point p i Point q i And point
The physical corresponding point q can be calculated i And &>
A distance d between T This distance is also equivalent to the result of the quantitative calculation of the point cloud difference value.
S502, point cloud difference value visualization method: and determining a color map matrix, determining corresponding color parameters according to the linear proportional relation between the point cloud difference value and the maximum allowable error value, and endowing a three-dimensional model with a corresponding color map, so that the structural deviation is displayed in a three-dimensional visual form on the electric power fitting point cloud model. Setting the maximum allowable error value E max And the Color map matrix Color, if anyp i The calculation result of the point cloud difference is d T Then AND point p i The corresponding color matrix index may be determined as ind i ，
In formula (ind) max Representing the maximum index value of the color matrix.
In the embodiment, one strain clamp qualified in crimping and two strain clamps unqualified in crimping are selected to visually detect the crimping quality of the electric power fitting; wherein, the qualified strain clamp serial number of crimping is #1, and two unqualified strain clamps of crimping serial numbers are #2 and #3 respectively.
Collecting point cloud data of three strain clamps by using a handheld laser radar scanner, and performing noise elimination processing on the point cloud data of the three strain clamps; as shown in fig. 8 to 13, the point cloud model of the strain clamp #1 is used as a reference point cloud model, and the point cloud models of the strain clamp #2 and the strain clamp #3 are used as target point cloud models, and point cloud model registration operations are respectively performed.
After the registration is completed, parameters of the strain clamp #1, the strain clamp #2 and the strain clamp #3 are extracted respectively, the extracted parameter results are shown in table 1, and the moment distribution conditions are shown in fig. 14-16.
TABLE 1 parameter extraction results
| Measuring parameters | Strain clamp #1 | Strain clamp #2 | Strain clamp #3 |
| Length L of non-pressing area of aluminum tube 0 /mm | 230.45204 | 220.55126 | 217.89625 |
| Compression joint length L of steel anchor side aluminum pipe 1 /mm | 118.38518 | 121.16169 | 126.90010 |
| Crimping length L of aluminum pipe at side of lead 2 /mm | 278.87448 | 295.14273 | 291.11778 |
| Chord height h/mm of maximum bending position | 2.16661 | 4.66855 | 13.18316 |
| Total length L/mm of strain clamp aluminum pipe | 627.71171 | 636.85568 | 544.91798 |
As shown in fig. 17 and 18, after the strain clamp parameters are extracted, the strain clamp #2 and the strain clamp #3 are subjected to visual processing of structural deviation respectively, so as to obtain a final detection visual result.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations can be devised by those skilled in the art in light of the above teachings. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.
Claims (10)
1. The visual detection method for the crimping quality of the electric power fitting is characterized by comprising the following steps:
s1, point cloud data acquisition: acquiring space point cloud data of the electric power fitting to be detected, converting the space point cloud data into txt file form, using the txt file form as original point cloud data, and establishing a point cloud model of the electric power fitting to be detected;
s2, point cloud data denoising: denoising a point cloud model of the electric power fitting to be detected by adopting a radius filtering algorithm, and removing outliers in original point cloud data;
s3, point cloud model registration: taking a point cloud model of a standard crimped power fitting as a reference point cloud model, taking a point cloud model of a power fitting to be detected as a target point cloud model, and unifying a reference coordinate system of the target point cloud model and a reference coordinate system of the reference point cloud model;
s4, extracting point cloud model parameters: dividing the point cloud model of the electric power fitting to be detected, and extracting parameters of the point cloud model of the electric power fitting to be detected through a model fitting algorithm;
s5, point cloud difference value visualization: and carrying out quantitative calculation on difference values of the registered reference point cloud model and the registered target point cloud model, converting the measured point cloud difference values into corresponding color parameters, endowing a corresponding color map to the three-dimensional model, and carrying out visual display on a detection result.
2. The visual detection method for the crimping quality of the electric power fittings according to claim 1, wherein the point cloud data in the step S1 is acquired by a high-precision handheld laser radar scanner.
3. The visual detection method for the crimping quality of the electric power fittings according to claim 1, wherein the step S2 specifically comprises the following steps:
s201, any point p in point cloud model of electric power fitting to be measured i ＝(x i ,y i ,z i ) T Determining a three-dimensional spherical space region with the spherical center as the center and the radius as r, and calculating the total number C of point clouds in the neighborhood of r r (p i )；
S202, counting total number C of point clouds in r neighborhood of point cloud model of to-be-measured electric power fitting r (p i ) Performing statistical analysis to determine a point cloud number judgment threshold C for removing noise point clouds T ：
Wherein, the first and the second end of the pipe are connected with each other,
representing the average total number of point clouds in the r neighborhood of the point cloud model, N representing the number of points in the point cloud model, σ r Representing the standard deviation of the number of point clouds in the r neighborhood of the point cloud model, and k represents a confidence coefficient interval;
s203, judging a threshold value C according to the number of point clouds T And removing outliers in the original point cloud data.
4. The visual detection method for crimping quality of electric power fittings according to claim 3, wherein the step S203 specifically comprises: traversing point cloud data, e.g. point p i Total number of point clouds C r (p i ) Less than point cloud number judgment threshold C T Then the point p is considered i Removing noise point cloud data; otherwise, the point cloud data is reserved.
5. The visual detection method for crimping quality of electric power fittings according to claim 1, wherein the point cloud model registration in step S3 specifically comprises the following steps:
s301, point cloud feature extraction: extracting a point cloud global feature descriptor GFD by adopting a principal component analysis method and a coordinate transformation idea, extracting a point cloud local feature descriptor FPFH (fast point feature histogram) by adopting a fast point feature histogram algorithm, and obtaining a point cloud fusion feature descriptor FFD by integrating the GFD and the FPFH;
s302, point cloud rough registration: completing feature point matching by measuring the similarity of the point cloud fusion feature descriptor FFD, selecting m groups of matching point pairs with local invariant feature similarity from high to low rank according to the one-to-one correspondence of the matching point pairs, and calculating a corresponding rigid body transformation matrix to realize the preliminary coincidence of the point cloud model;
s303, point cloud fine registration: and on the basis of the rough point cloud registration, performing refined registration by adopting an ICP (inductively coupled plasma) algorithm to obtain an optimized point cloud model registration result.
6. The visual detection method for crimping quality of electric power fittings according to claim 1, wherein the parameters in the step S4 include crimping length and crimping-to-moment.
7. The visual detection method for crimping quality of electric power fittings according to claim 1, wherein the point cloud model parameter extraction in the step S4 specifically comprises:
s401, point cloud model segmentation: selecting a proper initial clustering point in a specific area according to the point cloud model registration result, and realizing clustering segmentation of the point cloud model by adopting an improved k-means algorithm to obtain a final clustering segmentation result;
s402, axial layering of the point cloud model: the method comprises the following steps of intercepting a plurality of point clouds with certain thicknesses in a layering mode along the axial direction of the electric power fitting, and projecting each layer of intercepted point clouds on a corresponding plane perpendicular to a central axis to obtain layered point cloud slices;
s403, point cloud model parameter extraction: and extracting six straight line boundaries of the point cloud slice, finding data points on the six straight line boundaries, performing model fitting on the found data points to obtain boundary straight lines of the point cloud slice, and calculating parameters of the layered point cloud according to a straight line equation.
8. The visual detection method for crimping quality of electric power fittings according to claim 7, wherein the clustering segmentation in the step S401 divides the point cloud model into a crimping area model and a non-crimping area model.
9. The visual detection method for crimping quality of electric power fittings according to claim 7, wherein in the step S403, a RANSAC algorithm is adopted for extracting six straight line boundaries of a point cloud slice, and an integral least square method is adopted for performing model fitting on the found data points.
10. The visual detection method for crimping quality of electric power fittings according to claim 1, wherein the point cloud difference visualization in the step S5 specifically comprises the following steps:
s501, point cloud difference value quantization calculation: determining the point correspondence between the target point cloud model and the reference point cloud model by adopting a k-d tree algorithm and a closest point correspondence idea, and calculating a corresponding point cloud difference value by utilizing the point correspondence;
s502, point cloud difference value visualization method: and establishing a color map matrix, determining corresponding color parameters according to the linear proportional relation between the point cloud difference value and the maximum allowable error value, endowing the corresponding color map to a three-dimensional model, and performing visual display of the detection result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211398336.9A CN115937098A (en) | 2022-11-09 | 2022-11-09 | Power fitting crimping quality visual detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211398336.9A CN115937098A (en) | 2022-11-09 | 2022-11-09 | Power fitting crimping quality visual detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115937098A true CN115937098A (en) | 2023-04-07 |
Family
ID=86549821
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211398336.9A Pending CN115937098A (en) | 2022-11-09 | 2022-11-09 | Power fitting crimping quality visual detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115937098A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117171695A (en) * | 2023-11-02 | 2023-12-05 | 北京建工环境修复股份有限公司 | Method and system for evaluating ecological restoration effect of antibiotic contaminated soil |
-
2022
- 2022-11-09 CN CN202211398336.9A patent/CN115937098A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117171695A (en) * | 2023-11-02 | 2023-12-05 | 北京建工环境修复股份有限公司 | Method and system for evaluating ecological restoration effect of antibiotic contaminated soil |
| CN117171695B (en) * | 2023-11-02 | 2024-01-02 | 北京建工环境修复股份有限公司 | Method and system for evaluating ecological restoration effect of antibiotic contaminated soil |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112906694B (en) | Reading correction system and method for transformer substation inclined pointer instrument image | |
| CN111336990A (en) | Tunnel section convergence rapid analysis method and device | |
| Bokhabrine et al. | 3D characterization of hot metallic shells during industrial forging | |
| CN111612850A (en) | Pig body size parameter measuring method based on point cloud | |
| CN111489432B (en) | Bent pipe reconstruction and allowance calculation method based on point cloud data | |
| CN113689415A (en) | Steel pipe wall thickness online detection method based on machine vision | |
| CN109508709B (en) | Single pointer instrument reading method based on machine vision | |
| CN113487722B (en) | Automatic concrete member detection method based on three-dimensional laser scanning method | |
| CN111814740B (en) | Pointer instrument reading identification method, device, computer equipment and storage medium | |
| CN115937098A (en) | Power fitting crimping quality visual detection method | |
| CN115993096A (en) | High-rise building deformation measuring method | |
| CN115690184A (en) | Tunnel face displacement measurement method based on three-dimensional laser scanning | |
| CN116518864A (en) | Engineering structure full-field deformation detection method based on three-dimensional point cloud comparison analysis | |
| CN114663672B (en) | Method and system for detecting corrosion of steel member of power transmission line tower | |
| CN108205645B (en) | Reference image quality evaluation method of heterogeneous image matching system | |
| CN113870326B (en) | Structural damage mapping, quantifying and visualizing method based on image and three-dimensional point cloud registration | |
| CN114998395A (en) | Effective embankment three-dimensional data change detection method and system | |
| CN115937116A (en) | Detection method for strain clamp of power transmission line | |
| CN112785711B (en) | Insulator creepage distance detection method and detection system based on three-dimensional reconstruction | |
| CN112884057B (en) | Point cloud data-based three-dimensional curved surface quality classification method and system and storage medium | |
| Liu et al. | Computer Vision‐Based Structural Displacement Monitoring and Modal Identification with Subpixel Localization Refinement | |
| Grélard et al. | Precise cross-section estimation on tubular organs | |
| CN114838695B (en) | Aviation conduit bending section measuring method based on multi-camera measuring system | |
| Guldur et al. | Laser-based automatic cross-sectional change detection for steel frames | |
| CN113592877B (en) | Method and device for identifying red line exceeding of pumped storage power station |
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 |