CN113808133B - Subway brake shoe fault detection method based on three-dimensional point cloud - Google Patents
Subway brake shoe fault detection method based on three-dimensional point cloud Download PDFInfo
- Publication number
- CN113808133B CN113808133B CN202111372550.2A CN202111372550A CN113808133B CN 113808133 B CN113808133 B CN 113808133B CN 202111372550 A CN202111372550 A CN 202111372550A CN 113808133 B CN113808133 B CN 113808133B
- Authority
- CN
- China
- Prior art keywords
- brake shoe
- point cloud
- point
- dimensional
- subway
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/08—Indexing scheme for image data processing or generation, in general involving all processing steps from image acquisition to 3D model generation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Quality & Reliability (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a subway brake shoe fault detection method based on three-dimensional point cloud, which specifically comprises the following steps: collecting three-dimensional point cloud data of the brake shoe component by using a three-dimensional industrial camera; down-sampling to process the point clouds and control the number of the point clouds; after down-sampling, calculating a mass center, traversing each point, calculating the distance between each point and the mass center, and filtering out point clouds which do not belong to main components; dividing the brake shoe by a pointCNN point cloud division network based on deep learning, and judging whether the brake shoe is missing or not; and point cloud processing and calculating a thickness brake shoe to judge whether abrasion occurs. The invention improves the accuracy of subway brake shoe fault detection, scientifically measures and refines the abrasion degree of the brake shoe, and clearly and visually expresses the abrasion degree; and more comprehensive guarantee is provided for the operation safety of the subway.
Description
Technical Field
The invention belongs to the field of subway brake shoe fault detection, and particularly relates to a subway brake shoe fault detection method based on three-dimensional point cloud.
Background
Urban rail transit has extremely important significance in China development, and with the rapid development of Chinese economy, the technical level and the construction scale of Chinese railway tracks make great progress. At present, the scale of operation lines, line under construction and passenger flow in China all occupy the first world, and China becomes a genuine large urban rail country. More and more operation lines are bound to be a great challenge to safety, so that the establishment of a train safety guarantee system becomes an essential link in the construction of a railway system. As an important component of a smart city, the construction of a smart subway can promote the production mode innovation, organization form innovation, management idea innovation and business mode innovation of the rail transit industry. The traditional manual detection is still the most main mode of train maintenance, and the manual detection is not in line with the trend under the background, so that the efficiency is low, and some detail hidden dangers cannot be easily detected. With the increasing maturity of technologies such as robots, deep learning, computer vision, intelligent dollies and the like, artificial intelligence appears in more fields to meet the requirements of human beings. The intelligent detection mode that combines together urban rail transit safety and artificial intelligence can replace artifical the detection to a certain extent, can promote the development of "wisdom subway".
Computer vision is applied to a large number of fields at present, machine vision is more and more mature in recent years, and the application is gradually mature from bar code detection of daily supermarkets, attendance of fingerprints on and off duty, face recognition attendance, expressway license plate recognition, aerial remote sensing measurement and control of landform and landform, movie special effect manufacturing, industrial production automation detection, medical image detection, to the field of aviation astronomy and the like. The application of subway fault detection is gradually emphasized, various subway appearance detection systems are in the research and development or application stage, but a large number of detection algorithms are researched and developed based on two-dimensional images. The two-dimensional image has many advantages, such as convenient acquisition, mature algorithm, easy transmission, etc. However, for some fault types, algorithms based on two-dimensional pictures cannot be used, and measurements such as dimensions and gaps of components can only be obtained through three-dimensional information. With the gradual development of subversive technologies such as machine vision, automatic driving and the like, the 3D camera is adopted to perform object recognition, behavior recognition and scene modeling, so that the related applications are more and more, and the 3D camera is the eyes of a terminal and a robot. The 3D camera solutions available on the market at present are the 3: structured light, binocular vision, light time of flight method. Structured light is certainly the best solution at present from the most widespread use point of view. The point cloud data can complete more detection tasks, fill the blank of the existing subway appearance detection mode, and establish a more perfect intelligent maintenance system.
The subway appearance detection system who has put into use now has a lot of limitations, and this kind of system has fixed the position of gathering the camera, takes place relative displacement with the camera when the subway passes through when the access station, utilizes line scanning imaging principle to accomplish the data of gathering. The acquisition system can only shoot the parts on the top surface when used at the bottom of the vehicle, and the important parts of some side surfaces and shielded parts cannot be detected and can not finish a plurality of fine measurements. This method cannot completely replace manual detection, and only reduces the workload of some workers to a certain extent. The brake part for directly rubbing the wheels to stop the train during subway braking is a brake shoe. The tile-shaped brake block made of cast iron or other materials holds the wheel tread tightly during braking, and the wheel stops rotating through friction. Its loss and disappearance can cause huge harm to subway driving safety.
From the above background, it is clear that four key points that must be solved for brake shoe failure detection using three-dimensional computer vision are: (1) the algorithm model can effectively inhibit the interference of environmental factors such as illumination, stains and the like, has strong robustness and overcomes the limitation of two-dimensional pictures. (2) The algorithm model needs to exert the characteristics of the point cloud to complete fine detection. (3) The algorithm model has to have the characteristics of high precision, high stability and generalization, and brake shoes of different vehicles and carriages can be identified and divided, so that the manual overhaul mode can be replaced, and the train running safety is ensured. (4) The method has the advantages that the efficiency is high, the train maintenance time can only be the operation idle period, the whole train needs to be checked and maintained in the idle period, the detection efficiency is high, and the algorithm model is required to accurately complete the detection of the item point in a short time.
Disclosure of Invention
Aiming at the problems, the invention provides a subway brake shoe fault detection method based on three-dimensional point cloud.
The invention discloses a subway brake shoe fault detection method based on three-dimensional point cloud, which comprises the following steps:
step 1: the intelligent trolley is automatically positioned to a designated position, and brake shoe component data of the subway car bottom to be overhauled are acquired by means of the three-dimensional industrial camera on the mechanical arm.
Step 2: point cloud pretreatment: down-sampling to process the point clouds and control the number of the point clouds; after down-sampling, calculating the mass center, traversing each point and calculating the distance between each point and the mass center, and filtering out point clouds which do not belong to the principal component.
And step 3: brake shoe is divided by a pointCNN point cloud division network based on deep learning, and whether the brake shoe is missing or not is judged:
s31: and loading a trained point cloud segmentation model, and performing segmentation processing on the point cloud processed in the last step as input.
S32: the divided point cloud labels are divided into two types, one is a brake shoe target, and the other is point cloud belonging to other parts; counting the percentage k of the total number N occupied by the point cloud number N belonging to the brake shoe label, namely:
if k is too low and approaches to 0, the brake shoe missing fault is judged to occur, namely:
brake shoe absence requires alerting the operator.
And 4, step 4: point cloud processing, and evaluating the wear state of the brake shoe:
s41: and fitting the optimal plane by using a random sampling consistency algorithm.
S42: points belonging to the optimal plane are input for upsampling.
S43: and (6) resampling.
S44: the greedy triangulation projection algorithm triangulates the directed point cloud.
S45: calculating the area of each triangular surface in the three-dimensional world coordinate system, and counting the surface areas of all the triangular surfaces:
setting a triangular surfaceS ΔThe three-dimensional coordinates of the vertexes of the three point clouds forming the triangular surface are
，
，
And calculating the lengths of three sides ab, ac and bc of the triangle by using a space distance formula:
the half perimeter p is:
the area of the triangle is obtained by the Helen formulaS Δ：
The whole reconstructed curved surface of brake shoemThe triangular surface is formed by counting the area of all the trianglesS i Sum of (a):
the total area S of the reconstructed curved surface of the brake shoe is obtained.
S46: the known brake shoe is a uniform arc-shaped object, and the mathematical relation between the thickness and the arc length and the surface area is established through calculus; arc length is known asLThe above equation finds the total area as S, and the relation for the thickness D is:
s47: comparing the calculated thickness of the brake shoe with threshold data to obtain a difference value, comparing a fault threshold value, and judging the wear state of the brake shoe;
and judging the interval in which the data of the detection result D is positioned, uploading the detection data to a database if the detection thickness of the brake shoe is normal, and alarming to remind a maintainer to carry out recheck if the detection thickness of the brake shoe is abnormal.
The invention has the beneficial technical effects.
1. With the assistance of the intelligent inspection trolley and the mechanical arm, the high-precision positioning and flexible mechanical arm can help the 3D camera to acquire image data of complex parts, the problem that some important parts in the visual field blind area cannot be detected is solved, and the comprehensiveness and reliability of detection are improved.
2. The invention designs a subway brake shoe size measurement mode for directly inputting and processing three-dimensional point cloud, and solves the problem that the prior two-dimensional image cannot detect and solve the pain point. The point cloud data has many advantages compared with the fault detection of the two-dimensional image, and effectively eliminates the defects caused by the self limitation of the two-dimensional image and the external interference such as: light, rain, dirt, and the like. The method designs and reconstructs the relationship between the surface area and the thickness of the curved surface, and scientifically measures and refines the wear degree of the brake shoe. The intermediate process can be expressed clearly and intuitively by using a mathematical expression.
3. The invention combines the point cloud segmentation network based on deep learning with the traditional point cloud processing method, reduces the influence of different types of brake shoe structures on detection, and ensures the robustness of segmentation and the measurement accuracy. This mode can be generalized to more part size measurements on subways. The whole algorithm can realize automatic brake shoe missing detection and abrasion detection, and provides more comprehensive guarantee for the operation safety of the subway.
Drawings
FIG. 1 is a flow chart of a subway brake shoe fault detection method based on three-dimensional point cloud.
FIG. 2 is a schematic view of triangulation.
Fig. 3 shows the curved surface reconstruction result.
Detailed Description
The invention is described in further detail below with reference to the figures and the detailed description.
The data acquisition mode of the invention is that the intelligent patrol car is accurately positioned to a designated acquisition place, and three-dimensional point cloud data (XYZ) and color information (RGB) of corresponding parts at the bottom of the subway car are acquired by a three-dimensional industrial camera carried by a mechanical arm of the car. The brake shoe fault detection can be completed by several steps, and the whole algorithm can be completed by the steps of training a segmentation model, preprocessing, point cloud segmentation, point cloud processing, parameter calculation and the like. The method for calculating the thickness of the brake shoe can accurately judge whether the brake shoe is worn or not, is more accurate and reliable than the observation by naked eyes, and can also complete the task of detecting the brake shoe loss.
The invention discloses a subway brake shoe fault detection method based on three-dimensional point cloud, which is shown in figure 1 and comprises the following steps:
step 1: three-dimensional point cloud data of the brake shoe component is acquired by using a three-dimensional industrial camera.
The collection platform is an intelligent inspection robot, is positioned to a designated area at the bottom of the subway train through automatic identification, and then transmits data of the collected brake shoe component back to the background for real-time processing through a three-dimensional camera on a carried mechanical arm. The imaging principle of the camera is structured light, which can generate high-precision point cloud data (XYZ) and high-quality color information (RGB).
Step 2: and (4) point cloud preprocessing.
The number of the original data point clouds collected by the camera is up to 100 ten thousand, and the algorithm is slow due to the huge number of the point clouds. Therefore, under the condition of ensuring that the point cloud information is not lost, the processing speed is improved by reducing the number of the point clouds. The down-sampling is used for filtering a large amount of noise points, reducing the number of points and simultaneously keeping the shape characteristics of the point cloud, and is used for preprocessing the point cloud operation, and a VoxelGrid (voxel) filter is used for down-sampling the point cloud. The method adopted by the VoxelGrid down-sampling is a point cloud grid deleting method, namely a voxelized grid method. Downsampling is performed by setting the size of a small cube, calculating its center of gravity (instead of the center) instead of all of its surrounding cube sizes, leaving only one center of gravity point. And finally, sampling the point cloud points to about 1 ten thousand points. Since the object is more concentrated, non-target objects appearing during the acquisition process, such as detected people or background, are filtered before segmentation. And calculating the mass center of the point cloud after down sampling, traversing the distance between each point and the mass center, and filtering out too far parts.
And step 3: brake shoe is segmented by a pointCNN (point cloud convolutional neural network) point cloud segmentation network based on deep learning, and whether brake shoe loss exists or not is judged:
the application of convolutional neural networks to two-dimensional images is mature, but the application of CNNs (convolutional neural networks) to three-dimensional space, especially to such disordered sets of point clouds, is now studied particularly rarely. The point cloud data is often irregular and unordered, and if the traditional convolution operation is directly applied, the shape information is lost, and the influence of the change of the order of the point set is also generated. In order to solve the above-mentioned problem, the PointCNN network proposes X-transformation (X transformation), and combines with the traditional convolution operation to form an X-conv module. All points need to be labeled in the point cloud segmentation, all point clouds are labeled into two types in the data, the parts belonging to the brake tile are labeled into one type, and all the other points are labeled into non-target types, so that the efficiency of the training model and the detection speed precision can be improved. When point cloud data is input, a trained segmentation model is loaded, all point clouds belonging to a target object are segmented and proposed, and a new point cloud is formed to be used as input of the next step. Counting the percentage k of the total number N occupied by the point cloud number N belonging to the brake shoe label, namely:
if k is too low and approaches to 0, the brake shoe missing fault is judged to occur, namely:
brake shoe absence requires alerting the operator.
And 4, step 4: point cloud processing, and evaluating the wear state of the brake shoe:
noise still exists in the number of generated target point clouds, and the side point clouds belonging to the gate tiles are possibly segmented into the target point clouds. At the moment, a point cloud plane is fitted by using a random sampling consistency algorithm, an optimal plane is fitted in the global point cloud, and all internal points only belonging to the external surface plane model are extracted. The number of interior points after fitting is too small to constitute a triangulated number. Therefore, the number of point clouds is increased by means of up-sampling. Upsampling is a process of surface reconstruction that helps you recover the original surface when there is less than expected point cloud data, and is a complicated guess hypothesis by interpolating the current point cloud data. The problem of structural gaps after upsampling is not solved, errors are generated when a small object is measured, a curved surface is not smooth or has holes due to direct reconstruction, the surface needs to be subjected to smoothing treatment and hole repairing in order to establish a complete model.
A point cloud is a point with three-dimensional coordinates, but independent and discrete from each other. Computing area and volume requires converting discrete points into three-dimensional data in a grid format. The greedy triangulation projection algorithm triangulates the directed point cloud (as shown in fig. 2). The method is suitable for the condition that the sampled point cloud comes from a continuous and smooth curved surface and the density change of the point cloud is uniform. The point cloud is projected into a certain two-dimensional coordinate plane through a normal line, and then the point cloud obtained through projection is triangulated in the plane, so that the topological connection relation of each point is obtained. In the process of the plane triangulation, a spatial region growing algorithm based on Delaunay (triangulation) is used. Triangulation is an extremely important preprocessing technique. And finally, determining topological connection among the original three-dimensional points according to the topological connection relation of the projection points in the plane, wherein the obtained triangular mesh is the reconstructed curved surface model (as shown in figure 3).
And calculating the area of each triangular surface in the three-dimensional world coordinate system, and counting the surface areas of all the triangular surfaces. Setting a triangular surfaceS ΔThe three-dimensional coordinates of the vertexes of the three point clouds forming the triangular surface are
，
，
And calculating the lengths of three sides ab, ac and bc of the triangle by using a space distance formula:
the half perimeter p is:
the area of the triangle is obtained by the Helen formulaS Δ：
The whole reconstructed curved surface of brake shoemThe triangular surface is formed by counting the area of all the trianglesS i Sum of (a):
the total area S of the reconstructed curved surface of the brake shoe is obtained.
The known brake shoe is a uniform arc-shaped object, and the mathematical relation between the thickness and the arc length and the surface area is established through calculus; arc length is known asLThe above equation finds the total area as S, and the relation for the thickness D is:
comparing the calculated thickness of the brake shoe with threshold data to obtain a difference value, comparing a fault threshold value, and judging the wear state of the brake shoe;
and judging the interval in which the data of the detection result D is positioned, uploading the detection data to a database if the detection thickness of the brake shoe is normal, and alarming to remind a maintainer to carry out recheck if the detection thickness of the brake shoe is abnormal.
The whole scheme of the invention has a perfect detection process, the improvement of the data acquisition mode can more comprehensively cover the detection part, and the defect of the target part caused by shielding or angle is avoided, so that a plurality of important part detection tasks cannot be completed. The new acquisition mode can shoot data of each angle of the brake shoe, and is more beneficial to the processing and judgment of a subsequent algorithm. The 3D camera is used for replacing the 2D camera, so that color collection can be achieved, point cloud data of higher dimensional information can be obtained, the problem of fault detection of more brake shoe size measurement can be solved through the point cloud, and the problem of common false alarm rate in the field of fault detection is reduced.
The deep learning model trained by a large number of samples has the advantages of strong generalization, strong adaptability, good transportability and the like. In the method, a pointCNN algorithm is used for segmentation and clustering, and input point clouds are divided into two types, namely target point clouds and non-target point clouds. The segmentation algorithm can be suitable for brake shoe components in different subways and different positions, and can stably complete segmentation work. By combining with a subsequent traditional point cloud processing method, the target point cloud can be segmented in a complex and dynamic environment, and the accuracy and the rapidness of subsequent processing can be ensured. The time consumed by one input point cloud segmentation process is about 0.8 s.
The brake shoe thickness wear measuring method designed by the invention utilizes the triangulation surface area calculation method to restore the brake shoe surface to the maximum extent to ensure the authenticity of the calculation result, and the final result is deduced through a very intuitive mathematical formula, so that the detection result is more accurate. Firstly, rely on the collection camera of high accuracy, secondly through the compensation of algorithm, under a large amount of data tests, the precision of calculated result and difference of true value are within 0.5mm, can satisfy the precision of measuring wearing and tearing completely. The method has excellent performance in real-time algorithm detection. The thickness detection of brake shoe has been thinned and powerful guarantee can be provided for subway driving safety, the health status of brake shoe is foreseen according to historical data trend of change probably, guarantees the reliability of brake.
Claims (1)
1. A subway brake shoe fault detection method based on three-dimensional point cloud is characterized by comprising the following steps:
step 1: the intelligent trolley is automatically positioned to a designated position, and brake shoe component data of the subway car bottom to be overhauled are acquired by means of a three-dimensional industrial camera on the mechanical arm;
step 2: point cloud pretreatment: down-sampling to process the point clouds and control the number of the point clouds; after down-sampling, calculating a mass center, traversing each point, calculating the distance between each point and the mass center, and filtering out point clouds which do not belong to main components;
and step 3: brake shoe is divided by a pointCNN point cloud division network based on deep learning, and whether the brake shoe is missing or not is judged:
s31: loading a trained point cloud segmentation model, and performing segmentation processing on the point cloud processed in the last step as input;
s32: the divided point cloud labels are divided into two types, one is a brake shoe target, and the other is point cloud belonging to other parts; counting the percentage k of the total number N of the point clouds belonging to the brake shoe labels, namely:
if k is too low and approaches to 0, the brake shoe missing fault is judged to occur, namely:
the brake shoe is missing, and an alarm needs to be given to a worker;
and 4, step 4: point cloud processing, and evaluating the wear state of the brake shoe:
s41: fitting the optimal plane by using a random sampling consistency algorithm;
s42: inputting points belonging to the optimal plane for up-sampling;
s43: resampling;
s44: carrying out triangulation on the directed point cloud by a greedy triangulation projection algorithm;
s45: calculating the area of each triangular surface in the three-dimensional world coordinate system, and counting the areas of all the triangular surfaces:
setting a triangular surface, wherein three point cloud vertexes forming the triangular surface have three-dimensional coordinates of
，
，
And calculating the lengths of three sides ab, ac and bc of the triangle by using a space distance formula:
the half perimeter p is:
determining the area of the triangular surface by the Helen formulaS Δ：
The whole reconstructed curved surface of brake shoemThe triangular surfaces are formed by counting the areas of all the triangular surfacesS i Sum of (a):
the total area S of the reconstructed curved surface of the brake shoe is obtained;
s46: the known brake shoe is a uniform arc-shaped object, and the mathematical relation between the thickness and the arc length and the surface area is established through calculus; arc length is known asLThe above equation finds the total area as S, and the relation for the thickness D is:
s47: comparing the calculated thickness of the brake shoe with threshold data, and judging the wear state of the brake shoe;
and judging the interval in which the data of the detection result D is positioned, uploading the detection data to a database if the detection thickness of the brake shoe is normal, and alarming to remind a maintainer to carry out recheck if the detection thickness of the brake shoe is abnormal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111372550.2A CN113808133B (en) | 2021-11-19 | 2021-11-19 | Subway brake shoe fault detection method based on three-dimensional point cloud |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111372550.2A CN113808133B (en) | 2021-11-19 | 2021-11-19 | Subway brake shoe fault detection method based on three-dimensional point cloud |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113808133A CN113808133A (en) | 2021-12-17 |
| CN113808133B true CN113808133B (en) | 2022-01-25 |
Family
ID=78938450
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111372550.2A Active CN113808133B (en) | 2021-11-19 | 2021-11-19 | Subway brake shoe fault detection method based on three-dimensional point cloud |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113808133B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114037703B (en) * | 2022-01-10 | 2022-04-15 | 西南交通大学 | Subway valve state detection method based on two-dimensional positioning and three-dimensional attitude calculation |
| CN114862957B (en) * | 2022-07-08 | 2022-09-27 | 西南交通大学 | Subway car bottom positioning method based on 3D laser radar |
| CN115063579B (en) * | 2022-08-19 | 2022-11-04 | 西南交通大学 | Train positioning pin looseness detection method based on two-dimensional image and three-dimensional point cloud projection |
| CN115308247B (en) * | 2022-10-11 | 2022-12-16 | 江苏昭华精密铸造科技有限公司 | Method for detecting deslagging quality of aluminum oxide powder |
| CN117710436A (en) * | 2024-02-06 | 2024-03-15 | 唐山百川智能机器股份有限公司 | Brake shoe thickness detection method based on unmanned aerial vehicle monocular camera |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8294712B2 (en) * | 2003-09-19 | 2012-10-23 | The Boeing Company | Scalable method for rapidly detecting potential ground vehicle under cover using visualization of total occlusion footprint in point cloud population |
| CA2864700A1 (en) * | 2014-09-25 | 2016-03-25 | Flatfab Inc. | System and method for generating planar section 3d shape representations |
| CN104455121A (en) * | 2014-12-18 | 2015-03-25 | 成都铁安科技有限责任公司 | Method and system for detecting residual thickness of brake shoe |
| CN107123161A (en) * | 2017-06-14 | 2017-09-01 | 西南交通大学 | A kind of the whole network three-dimensional rebuilding method of contact net zero based on NARF and FPFH |
| US10282834B1 (en) * | 2018-06-22 | 2019-05-07 | Caterpillar Inc. | Measurement platform that automatically determines wear of machine components based on images |
| CN108974044A (en) * | 2018-08-09 | 2018-12-11 | 铁路愿景欧洲有限公司 | Railroad track assets survey system |
| EP3671531A1 (en) * | 2018-12-17 | 2020-06-24 | Promaton Holding B.V. | Semantic segmentation of non-euclidean 3d data sets using deep learning |
| EP3767521A1 (en) * | 2019-07-15 | 2021-01-20 | Promaton Holding B.V. | Object detection and instance segmentation of 3d point clouds based on deep learning |
| CN111080609B (en) * | 2019-12-12 | 2020-12-15 | 哈尔滨市科佳通用机电股份有限公司 | Brake shoe bolt loss detection method based on deep learning |
| CN111091555B (en) * | 2019-12-12 | 2020-10-16 | 哈尔滨市科佳通用机电股份有限公司 | Brake shoe breaking target detection method |
| CN113139572B (en) * | 2021-03-09 | 2022-08-02 | 西南交通大学 | Image-based train air spring fault detection method |
| CN112991347B (en) * | 2021-05-20 | 2021-08-03 | 西南交通大学 | Three-dimensional-based train bolt looseness detection method |
-
2021
- 2021-11-19 CN CN202111372550.2A patent/CN113808133B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN113808133A (en) | 2021-12-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113808133B (en) | Subway brake shoe fault detection method based on three-dimensional point cloud | |
| Liu et al. | A review of applications of visual inspection technology based on image processing in the railway industry | |
| Yu et al. | 3D reconstruction of road surfaces using an integrated multi-sensory approach | |
| CN108960060A (en) | A kind of automatic driving vehicle pavement texture identifying system and method | |
| CN104392212B (en) | The road information detection and front vehicles recognition methods of a kind of view-based access control model | |
| CN106918299B (en) | A kind of line-structured light machine vision tire wear measurement method | |
| Rau et al. | Bridge crack detection using multi-rotary UAV and object-base image analysis | |
| CN109489584B (en) | Tunnel clearance detection system and tunnel clearance identification method based on 3D technology | |
| WO2023045299A1 (en) | Road surface technical condition detection method and device based on three-dimensional contour | |
| CN114241298A (en) | Tower crane environment target detection method and system based on laser radar and image fusion | |
| CN115482195B (en) | Train part deformation detection method based on three-dimensional point cloud | |
| CN113409240B (en) | Agricultural machinery behavior analysis and operation area statistical method based on Beidou positioning data | |
| CN110766785A (en) | Real-time positioning and three-dimensional reconstruction device and method for underground pipeline | |
| CN114037703B (en) | Subway valve state detection method based on two-dimensional positioning and three-dimensional attitude calculation | |
| Han et al. | Cantilever structure segmentation and parameters detection based on concavity and convexity of 3-D point clouds | |
| CN111023965A (en) | Steel coil identification and positioning method based on laser technology | |
| CN114488350B (en) | Mountain city short-time rainfall forecasting method based on radar map information processing | |
| CN111855667A (en) | Novel intelligent train inspection system and detection method suitable for metro vehicle | |
| CN113240682A (en) | Overturn-preventing construction driving map generation method and system for crawler crane | |
| CN116642420A (en) | Online glue type detection device and method for gluing automobile windshield glass | |
| CN115180476B (en) | Elevator traction sheave race form detection method based on three-dimensional reconstruction | |
| CN110728747A (en) | Vehicle three-dimensional outline drawing method based on laser point cloud data and OpenGL | |
| Sun et al. | A new single-view 3D pantograph reconstruction aided by prior CAD model | |
| Bulatov et al. | Context-based urban terrain reconstruction from images and videos | |
| Liu et al. | Method for real-time reconstruction of a transmission line based on the LiDAR point cloud data of a partial line segment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |