CN112319543A - Inspection device capable of identifying in-orbit and inspection method thereof - Google Patents
Inspection device capable of identifying in-orbit and inspection method thereof Download PDFInfo
- Publication number
- CN112319543A CN112319543A CN202011223678.8A CN202011223678A CN112319543A CN 112319543 A CN112319543 A CN 112319543A CN 202011223678 A CN202011223678 A CN 202011223678A CN 112319543 A CN112319543 A CN 112319543A
- Authority
- CN
- China
- Prior art keywords
- rail
- inspection
- control motor
- pixel point
- picture
- 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
- 238000007689 inspection Methods 0.000 title claims abstract description 90
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000006073 displacement reaction Methods 0.000 claims abstract description 14
- 230000003595 spectral effect Effects 0.000 claims description 18
- 230000036541 health Effects 0.000 claims description 13
- 238000001514 detection method Methods 0.000 claims description 12
- 230000011218 segmentation Effects 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 9
- 230000007547 defect Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000003708 edge detection Methods 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 238000002372 labelling Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims description 2
- 238000010276 construction Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 206010044565 Tremor Diseases 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000004575 stone Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61K—AUXILIARY EQUIPMENT SPECIALLY ADAPTED FOR RAILWAYS, NOT OTHERWISE PROVIDED FOR
- B61K9/00—Railway vehicle profile gauges; Detecting or indicating overheating of components; Apparatus on locomotives or cars to indicate bad track sections; General design of track recording vehicles
- B61K9/08—Measuring installations for surveying permanent way
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61D—BODY DETAILS OR KINDS OF RAILWAY VEHICLES
- B61D15/00—Other railway vehicles, e.g. scaffold cars; Adaptations of vehicles for use on railways
- B61D15/08—Railway inspection trolleys
- B61D15/12—Railway inspection trolleys power propelled
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61L—GUIDING RAILWAY TRAFFIC; ENSURING THE SAFETY OF RAILWAY TRAFFIC
- B61L23/00—Control, warning, or like safety means along the route or between vehicles or vehicle trains
- B61L23/04—Control, warning, or like safety means along the route or between vehicles or vehicle trains for monitoring the mechanical state of the route
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D21/00—Measuring or testing not otherwise provided for
- G01D21/02—Measuring two or more variables by means not covered by a single other subclass
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration by the use of local operators
-
- G06T5/70—
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- 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]
Abstract
The invention discloses a routing inspection device capable of identifying entering rails and a routing inspection method thereof, wherein the routing inspection device comprises a routing inspection vehicle and support shafts positioned on two sides of a rail; the bottom of the supporting shaft is provided with a jack fixing foot, the top of the supporting shaft is movably connected with the cantilever through a pulley, and the cantilever is fixedly provided with a rack; a limiting block is arranged on the rack; a first control motor and a second control motor are arranged between the top of the inspection vehicle and the shell, and output shafts of the first control motor and the second control motor are respectively provided with a gear meshed with the rack; a displacement sensor and an inspection assembly are arranged at the side end of the inspection vehicle; the inspection assembly comprises a supporting rod arranged on the inspection vehicle; a rotating motor is arranged on the supporting rod, an output shaft of the rotating motor is connected with the gyroscope, a four-shaft motor is arranged on the lower portion of the gyroscope, and an output shaft of the four-shaft motor is connected with the camera; a piezoelectric sensor is arranged on the rail; and a scanner is arranged at the bottom of the inspection vehicle.
Description
Technical Field
The invention belongs to the technical field of inspection vehicles, and particularly relates to an inspection device capable of identifying in-orbit and an inspection method thereof.
Background
With the development of economy and the progress of society, the transportation mode of people is more and more convenient, wherein the railway transportation is the travel mode that people select more at present. The health condition of the railway is an important index for whether the railway train can run safely, the health condition of the railway can be checked by a railway inspection vehicle, the railway inspection vehicle is generally large in equipment, and how to check the upper track and the lower track is an important subject. At present, inspection of a patrol vehicle is realized manually, so that the inspection is relatively complicated, and the construction burden of a detector is increased seriously; and at the in-process that the inspection vehicle patrolled and examined, need gather orbital relevant information, for example the rail has the crack etc. whether, need shoot to the rail and check this time, but at the in-process of examining and shooing, avoided having can not have the appearance of the situation such as many other jolts, the camera that leads to the camera focuses on inaccurate problem appears, and the camera focuses on inaccurate, influences very greatly, seriously influences the later stage to the judgement whether the rail is healthy.
Disclosure of Invention
The invention aims to provide a routing inspection device capable of identifying the entering of a rail and a routing inspection method thereof aiming at overcoming the defects in the prior art, so as to solve the problems of complicated entering of a routing inspection vehicle and poor routing inspection effect on the health of the rail.
In order to achieve the purpose, the invention adopts the technical scheme that:
a routing inspection device capable of identifying entering rails and a routing inspection method thereof comprise a routing inspection vehicle and support shafts positioned on two sides of rails; the bottom of the supporting shaft is provided with a jack fixing foot, the top of the supporting shaft is movably connected with the cantilever through a pulley, and the cantilever is fixedly provided with a rack; a limiting block is arranged on the rack; a first control motor and a second control motor are arranged between the top of the inspection vehicle and the shell, and output shafts of the first control motor and the second control motor are respectively provided with a gear meshed with the rack; a displacement sensor and an inspection assembly are arranged at the side end of the inspection vehicle; the inspection assembly comprises a supporting rod arranged on the inspection vehicle; a rotating motor is arranged on the supporting rod, an output shaft of the rotating motor is connected with the gyroscope, a four-shaft motor is arranged on the lower portion of the gyroscope, and an output shaft of the four-shaft motor is connected with the camera; a piezoelectric sensor is arranged on the rail; the bottom of the inspection vehicle is provided with a scanner; the first control motor, the second control motor, the scanner, the displacement sensor, the rotating motor, the gyroscope, the four-axis motor, the camera and the piezoelectric sensor are connected with an external calculator.
A method for inspecting an inspection device capable of recognizing a track, comprising:
s1, pushing the inspection vehicle to a position close to the rail, and installing a support shaft on one side of the rail;
s2, enabling the cantilever connected with the supporting shaft to penetrate through the space between the shell and the top of the inspection vehicle, and enabling the rack on the cantilever to be meshed and connected with the gears on the first control motor and the second control motor;
s3, mounting the other support shaft on the opposite side of the rail, starting a jack fixing foot, and lifting the cantilever and the inspection trolley to a preset height;
s4, controlling to start the first control motor and the second control motor, and driving the inspection vehicle to move towards the rail direction;
s5, when the inspection vehicle moves to the position of the limiting block, and the displacement sensor detects that the distance is not changed any more, controlling to close the first control motor and the second control motor;
s6, starting a jack fixing foot, placing the inspection vehicle on a rail, and drawing away the cantilever;
s7, enabling the inspection vehicle to move along a target rail, and enabling an external computer to receive picture information and piezoelectric signals acquired by a camera and a piezoelectric sensor in real time;
s8, the external computer carries out the fuzzy detection of the picture by Laplacian transformation, and sets a variance threshold, if the variance of the picture obtained by calculation is smaller than the variance threshold, the step S9 is carried out; if the variance of the picture is greater than or equal to the variance threshold, the process proceeds to step S10;
s9, controlling the rotating motor, the gyroscope and the four-axis motor to work, adjusting the focal distance between the camera and the rail, uploading the shot rail picture to the computer, and entering the step S8;
s10, labeling the picture target data set by adopting a semantic segmentation model;
s11, carrying out gray scale processing on the divided model;
s12, calculating a threshold value of a pixel point in the gray image by adopting local self-adaptive binarization, wherein if the gray value of the pixel point in the gray image is greater than the threshold value, the binarization result is 1 and represents a target rail; if the gray value of the pixel point is smaller than the threshold value, the binarization result is 0 and represents the rail background;
s13, carrying out edge detection on the image by using a CARRY operator, and calculating the current gradient strength of each pixel point;
s14, calculating the current gradient intensity G of the pixel pointMeasuringGradient intensity G of pixel points corresponding to normal railOften timesMaking a comparison if | GMeasuring-GOften timesIf the absolute value of | is greater than the threshold value, judging that the rail corresponding to the pixel point has a defect, and if | GMeasuring-GOften timesIf the absolute value of | is less than or equal to the threshold value, the process proceeds to step S9;
s15, converting the piezoelectric signals collected by the same section of the railway into digital signals, converting the time domain waveforms of the digital signals into power spectral density distribution patterns of frequency domains, comparing the power spectral density distribution patterns with the power spectral density distribution patterns of normal railway, and judging the health condition of the corresponding railway.
Preferably, in S8, performing blur detection on the picture by using Laplacian transform includes:
and measuring a second derivative of the picture by using a Laplacian operator for boundary detection, performing convolution calculation by using 1 channel of the picture and combining a kernel of 3x3, and calculating a variance value of the output picture.
Preferably, in S10, the construction of the semantic segmentation model is adopted, including:
s10.1, randomly dividing a target data set into a training sample set and a testing sample set;
and S10.2, training by adopting a training sample set, constructing a semantic segmentation model, and testing the precision of the model by adopting a test sample set.
Preferably, in S12, local adaptive binarization is used to calculate a threshold of a pixel point in the grayscale image, and if the grayscale value of the pixel point in the grayscale image is greater than the threshold, the binarization result is 1, which represents the target rail; if the gray value of the pixel point is less than or equal to the threshold value, the binarization result is 0, which represents the rail background, and the method comprises the following steps:
s12.1, setting the gray value of the image at the pixel point (x, y) as f (x, y);
s12.2, calculating a threshold value w (x, y) of each pixel point (x, y) in the image:
W(x,y)＝0.5*(max f(x+m,y+n)+min f(x+m,y+n))
s12.3, if f (x, y) > w (x, y), the binarization result is 1 and represents a target point of the character area; otherwise, the binarization result is 0, which represents the target point of the background area.
Preferably, in S13, performing edge detection on the image by using the CARRY operator, and calculating the current gradient strength of each pixel point, including:
s13.1, smoothing the image by adopting a Gaussian filter, and filtering noise;
s13.2, calculating the gradient strength G of each pixel point in the image:
wherein G isxAnd GyGradient values in the x and y directions, respectively;
s13.3 Sobel operator S in the x and y directionsxAnd SyRespectively as follows:
s13.4, if a window of 3x3 in the image is A and a gradient-calculating pixel point is e, performing convolution calculation by combining a Sobel operator to obtain gradient values of the pixel point e in the x direction and the y direction, wherein the gradient values are respectively as follows:
wherein, is a convolution symbol, sum represents the sum of all elements in the matrix, and a, b, c, d, e, f, g, h and i are parameters in the convolution matrix.
The routing inspection device capable of identifying the in-orbit and the routing inspection method thereof provided by the invention have the following beneficial effects:
the device has simple structure and ingenious conception, realizes the automatic track entry of the inspection vehicle through the cooperation of the cantilever, the support shaft and the jack fixing foot, effectively releases manpower and obviously improves the construction efficiency. In addition, the camera is adjusted according to the acquired image definition until a clear image picture is obtained, so that the accuracy of judging the rail defects in the later period is improved; meanwhile, calculating the gradient intensity of the rail image pixels acquired in real time, and judging the health condition of the rail according to the difference value of the acquired gradient intensity and the gradient intensity of the normal rail; and then, judging the health condition of the section of the railway by using the power spectral density distribution map of the rest railway sections which cannot be accurately judged. The invention adopts two methods to continuously judge the health condition of the rail on the same section of railway so as to reduce the influence of the environment on the pixel intensity gradient and improve the accuracy of pixel detection.
Drawings
Fig. 1 is a schematic diagram of a patrol vehicle approaching a rail of a patrol device capable of recognizing the approach of the rail.
Fig. 2 is a connection diagram of the inspection vehicle and the cantilever.
Fig. 3 is a supporting shaft fixing view.
Fig. 4 is a diagram of the movement of the inspection vehicle on the cantilever.
Fig. 5 is a track entering diagram of the inspection vehicle.
Fig. 6 is a patrol diagram of the patrol vehicle.
Wherein, 1, a first control motor; 2. a housing; 3. a scanner; 4. a second control motor; 5. a roller; 6. a rail; 7. a pulley; 8. a support shaft; 9. a jack fixing foot; 10. a displacement sensor; 11. a limiting block; 12. a routing inspection assembly; 13. a rack; 14. a cantilever; 15. a gear; 121. a support bar; 122. a rotating electric machine; 123. a gyroscope; 124. a four-axis motor; 125. a camera; 126. and (5) inspecting the vehicle.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
According to one embodiment of the application, referring to fig. 1-5, the routing inspection device capable of identifying the approach comprises a routing inspection vehicle 126 and support shafts 8 positioned on two sides of a rail 6; the bottom of the supporting shaft 8 is a jack fixing foot, and the supporting shaft 8 can ascend and descend through the jack.
The top of the supporting shaft 8 is movably connected with a cantilever 14 through a pulley 7, a rack 13 is fixedly installed on the cantilever 14, a limiting block 11 is installed on the rack 13, and the limiting block 11 is used for limiting the moving displacement of the inspection vehicle 126.
A first control motor 1 and a second control motor 4 are arranged between the top of the inspection vehicle 126 and the shell 2, and gears 15 meshed with the racks 13 are arranged on output shafts of the first control motor 1 and the second control motor 4.
During operation, the first control motor 1 and the second control motor 4 operate to drive the gear 15 thereon to rotate, and since the rack 13 cannot move, when the gear 15 rotates, the inspection vehicle 126 is driven to move.
The side end of the patrol vehicle 126 is provided with a displacement sensor 10 and a patrol assembly 12, wherein the displacement sensor 10 is used for detecting the movement displacement of the patrol vehicle 126, and when no change of displacement data is detected, the first control motor 1 and the second control motor 4 stop working.
The inspection assembly 12 comprises an inspection vehicle 126 arranged on the rail 6 and a support rod 121 arranged on the inspection vehicle 126, and the support rod 121 is used for supporting.
The support rod 121 is provided with a rotating motor 122, and the position of the rotating motor 122 can be selected according to the position between the actual rail 6 and the support rod 121.
An output shaft of the rotating motor 122 is connected with a gyroscope 123, a four-shaft motor 124 is mounted on the lower portion of the gyroscope 123, an output shaft of the four-shaft motor 124 is connected with a camera 125, and the gyroscope 123 is used for preventing and reducing shaking of the camera 125.
The four-axis motor 124 is used to adjust the position of the camera 125 in the vertical direction, and the rotating motor 122 is used to adjust the position of the camera 125 in the horizontal direction.
Therefore, under the matching use of the four-axis motor 124 and the rotating motor 122, the position of the camera 125 on the three-dimensional surface can be adjusted, so as to ensure the clarity of the shot picture.
A piezoelectric sensor is arranged on the rail 6; the rotary motor 122, gyroscope 123, four-axis motor 124, camera 125 and piezoelectric sensor are connected to an external calculator.
The external computer can receive the sensor signal and control the operation of the rotating motor 122 and the four-axis motor 124, so as to achieve the best shooting effect on the rail 6.
The working principle of the embodiment is as follows:
the position of the support rod 121 is adjusted to fix the camera 125 at a substantially fixed position without deviation, thereby ensuring that the camera 125 is within the position of the rail 6 being photographed.
After the main body is fixed, the four-axis motor 124 is controlled to operate, the position is automatically adjusted according to the function setting, the position of the camera 125 can be adjusted up, down, left and right, and the camera 125 is guaranteed to be always aligned with the central area of the detection object.
Simultaneously at the anti-shake gyroscope 123 on the base of four-axis motor 124, can carry out active anti-shake according to gyroscope 123 feedback, reduce and tremble, be favorable to taking the picture in the positive center and can not tremble.
The camera 125 for automatically shooting and storing the camera with high precision is used, so that the visual field shot by the camera is widened, and the acquisition work of the track related information is facilitated.
The device can ensure that the current motor position can be obtained in real time for closed-loop control, and the camera position is determined by matching with the external structure and the internal core controller of the inspection device, so that the camera position is more accurate.
According to the second embodiment of the application, the automatic rail entering inspection method for the rail inspection vehicle comprises the following steps:
s1, pushing the inspection vehicle 126 to a position close to the rail 6, and fixedly mounting a support shaft 8 on one side of the rail 6;
s2, enabling the cantilever 14 connected with the support shaft 8 to penetrate through the shell 2 and the top of the inspection vehicle 126, enabling the shell 2 to be in direct contact with the cantilever 14, and enabling the shell 2 to be used for bearing; and the rack 13 on the cantilever 14 is meshed with the gears 15 on the first control motor 1 and the second control motor 4;
s3, mounting the other support shaft 8 on the opposite side of the rail 6, starting a jack fixing foot, and lifting the cantilever 14 and the inspection trolley to a preset height;
s4, controlling to start the first control motor 1 and the second control motor 4, and driving the inspection vehicle 126 to move towards the direction of the rail 6;
s5, when the inspection vehicle 126 moves to the position of the limiting block 11 and the distance detected by the displacement sensor 10 is not changed any more, controlling to close the first control motor 1 and the second control motor 4;
s6, starting the jack fixing feet, placing the inspection vehicle 126 on the rail 6, and drawing away the cantilever 14;
s7, receiving the image information and the piezoelectric signal acquired by the camera 125 and the piezoelectric sensor in real time, that is, acquiring the piezoelectric information of the current rail 6 in real time by the piezoelectric sensor, and transmitting the piezoelectric information to an external computer; and transmitting pictures or photos captured by the camera 125 in real time to an external computer.
S8, carrying out fuzzy detection on the picture by adopting Laplacian transformation,
the Laplacian transformation is used for fuzzy detection of the picture, a Laplacian operator is used for measuring a second derivative of the picture for boundary detection, 1 channel of the picture is used for carrying out convolution calculation by combining a kernel of 3x3, and a variance value of the output picture is calculated.
Setting a variance threshold, and if the variance of the picture obtained by calculation is smaller than the variance threshold, entering the step S9; if the variance of the picture is greater than or equal to the variance threshold, the process proceeds to step S10.
And S9, if the variance of the picture is smaller than the variance threshold value, the external computer controls the rotating motor 122, the gyroscope 123 and the four-axis motor 124 to work, the focal distance between the camera 125 and the rail 6 is adjusted, the shot rail 6 picture is uploaded to the computer, and the step S8 is carried out until the definition meets the requirement, namely the variance of the picture is larger than or equal to the variance threshold value.
S10, labeling the picture target data set by adopting a semantic segmentation model, wherein the construction of the semantic segmentation model comprises the following steps:
s10.1, randomly dividing a target data set into a training sample set and a testing sample set;
and S10.2, training by adopting a training sample set, constructing a semantic segmentation model, and testing the precision of the model by adopting a test sample set.
And inputting the image with the definition meeting the requirement into a semantic segmentation model, and carrying out image target data set annotation, wherein annotation objects comprise rails 6, stones, water pits and backgrounds.
S11, carrying out gray scale processing on the divided model;
s12, calculating a threshold value of a pixel point in the gray image by adopting local self-adaptive binarization, wherein if the gray value of the pixel point in the gray image is greater than the threshold value, the binarization result is 1 and represents the target rail 6; if the gray value of the pixel point is smaller than the threshold value, the binarization result is 0, which represents the background of the rail 6, and the method specifically comprises the following steps:
s12.1, setting the gray value of the image at the pixel point (x, y) as f (x, y);
s12.2, calculating a threshold value w (x, y) of each pixel point (x, y) in the image:
W(x,y)＝0.5*(max f(x+m,y+n)+min f(x+m,y+n))
s12.3, if f (x, y) > w (x, y), the binarization result is 1 and represents a target point of the character area; otherwise, the binarization result is 0, which represents the target point of the background area.
According to the invention, the semantic segmentation model in the step S10 is used for labeling the picture target data set, the labeled objects are the rail 6, the stone, the puddle and the background, so that the data of the post-binarization processing are reduced, and then the binarization processing in the step S6 is used for rapidly calculating to obtain the target point rail 6.
Compared with the traditional method of directly carrying out binarization processing to obtain a target point, the method obtains the target data set by preliminarily screening the semantic segmentation model, and can greatly improve the calculation rate by carrying out binarization processing on the basis.
S13, carrying out edge detection on the image by using the CARRY operator, and calculating the current gradient strength of each pixel point, wherein the method specifically comprises the following steps:
s13.1, smoothing the image by adopting a Gaussian filter, and filtering noise;
s13.2, calculating the gradient strength G of each pixel point in the image:
wherein G isxAnd GyGradient values in the x and y directions, respectively;
s13.3 Sobel operator S in the x and y directionsxAnd SyRespectively as follows:
s13.4, if a window of 3x3 in the image is A and a gradient-calculating pixel point is e, performing convolution calculation by combining a Sobel operator to obtain gradient values of the pixel point e in the x direction and the y direction, wherein the gradient values are respectively as follows:
wherein, is a convolution symbol, sum represents the sum of all elements in the matrix, and a, b, c, d, e, f, g, h and i are parameters in the convolution matrix.
S14, calculating the current gradient intensity G of the pixel pointMeasuringGradient intensity G of pixel points corresponding to the normal rail 6Often timesMaking a comparison if | GMeasuring-GOften timesIf the absolute value of | is greater than the threshold, the defect of the rail 6 corresponding to the pixel point can be directly judged.
If | GMeasuring-GOften timesThe absolute value of | is less than or equal to the threshold value, and may be that the defect of the rail 6 is not obvious due to more uncertain factors of the rail 6 environment, such as a small crack and is not noticeable in the image, or the pit is too small to be represented in the form of an image. Therefore, in order to avoid the erroneous determination caused by this hidden trouble, the present invention further adopts step S15 to perform the determination.
S15, converting the piezoelectric signals collected by the same section of the railway 6 into digital signals, converting the time domain waveforms of the digital signals into power spectral density distribution patterns of frequency domains, comparing the power spectral density distribution patterns with the power spectral density distribution patterns of the normal railway 6, and judging the health condition of the corresponding railway 6.
When the inspection vehicle 126 is driven toward and past the length of railway rail 6, it causes low frequency vibrations of the rail 6, which are monitored by piezoelectric sensors mounted on the rail foot in the form of signal waves and transmitted to an external computer.
And converting the time domain waveform of the received digital signal into a power spectral density distribution map of a frequency domain, comparing the power spectral density distribution map with the power spectral density distribution map of the normal rail 6, and judging the health condition of the corresponding rail 6.
Comparing the detected power spectral density distribution of the rail 6 with the power spectral density distribution of the rail 6 of the normal rail 6, and if the difference value of the two is greater than a preset value, judging that the rail 6 is cracked or otherwise damaged; otherwise, it is judged that no crack or other damage has occurred to the rail 6.
If the detected power spectral density distribution of the rail 6 is mainly concentrated on 70kHz, and the power spectral density distribution of the rail 6 in the same time period (referring to the same season, the same month and the same time point in the historical data) of the same section of the rail 6 is mainly concentrated on 78kHz, and the difference between the two is 8kHz and is greater than the preset difference of 2kHz, the section of the rail 6 is judged to be cracked or otherwise damaged.
The device has a simple structure and ingenious conception, realizes automatic track entry of the inspection vehicle 126 through the cooperation of the cantilever 14, the support shaft 8 and the jack fixing foot 9, effectively releases manpower, and remarkably improves the construction efficiency. In addition, the invention adjusts the camera 125 according to the acquired image definition until a clear image picture is obtained, so as to improve the accuracy of the later rail 6 defect judgment; meanwhile, calculating the gradient intensity of the image pixels of the rail 6 acquired in real time, and judging the health condition of the rail 6 according to the difference value of the acquired gradient intensity and the gradient intensity of the normal rail 6; subsequently, the remaining portions of the rail 6 that cannot be accurately determined are used to determine the health of the section of rail 6 using the power spectral density profile. The invention adopts two methods to continuously judge the health condition of the rail 6 on the same section of railway so as to reduce the influence of the environment on the pixel intensity gradient and improve the accuracy of pixel detection.
While the embodiments of the invention have been described in detail in connection with the accompanying drawings, it is not intended to limit the scope of the invention. Various modifications and changes may be made by those skilled in the art without inventive step within the scope of the appended claims.
Claims (6)
1. The utility model provides a recognizable inspection device that puts into orbit which characterized in that: the system comprises a polling car and support shafts positioned on two sides of a rail; the bottom of the supporting shaft is provided with a jack fixing foot, the top of the supporting shaft is movably connected with the cantilever through a pulley, and the cantilever is fixedly provided with a rack; a limiting block is arranged on the rack; a first control motor and a second control motor are arranged between the top of the inspection vehicle and the shell, and output shafts of the first control motor and the second control motor are respectively provided with a gear meshed with the rack; a displacement sensor and an inspection assembly are installed at the side end of the inspection vehicle; the inspection assembly comprises a supporting rod arranged on the inspection vehicle; a rotating motor is arranged on the supporting rod, an output shaft of the rotating motor is connected with the gyroscope, a four-shaft motor is arranged on the lower portion of the gyroscope, and an output shaft of the four-shaft motor is connected with the camera; a piezoelectric sensor is arranged on the rail; a scanner is arranged at the bottom of the inspection vehicle; the first control motor, the second control motor, the scanner, the displacement sensor, the rotating motor, the gyroscope, the four-axis motor, the camera and the piezoelectric sensor are connected with an external calculator.
2. The inspection method of the inspection apparatus for the identifiable in-track inspection according to claim 1, comprising:
s1, pushing the inspection vehicle to a position close to the rail, and installing a support shaft on one side of the rail;
s2, enabling the cantilever connected with the supporting shaft to penetrate through the space between the shell and the top of the inspection vehicle, and enabling the rack on the cantilever to be meshed and connected with the gears on the first control motor and the second control motor;
s3, mounting the other support shaft on the opposite side of the rail, starting a jack fixing foot, and lifting the cantilever and the inspection trolley to a preset height;
s4, controlling to start the first control motor and the second control motor, and driving the inspection vehicle to move towards the rail direction;
s5, when the inspection vehicle moves to the position of the limiting block, and the displacement sensor detects that the distance is not changed any more, controlling to close the first control motor and the second control motor;
s6, starting a jack fixing foot, placing the inspection vehicle on a rail, and drawing away the cantilever;
s7, enabling the inspection vehicle to move along a target rail, and enabling an external computer to receive picture information and piezoelectric signals acquired by a camera and a piezoelectric sensor in real time;
s8, the external computer carries out the fuzzy detection of the picture by Laplacian transformation, and sets a variance threshold, if the variance of the picture obtained by calculation is smaller than the variance threshold, the step S9 is carried out; if the variance of the picture is greater than or equal to the variance threshold, the process proceeds to step S10;
s9, controlling the rotating motor, the gyroscope and the four-axis motor to work, adjusting the focal distance between the camera and the rail, uploading the shot rail picture to the computer, and entering the step S8;
s10, labeling the picture target data set by adopting a semantic segmentation model;
s11, carrying out gray scale processing on the divided model;
s12, calculating a threshold value of a pixel point in the gray image by adopting local self-adaptive binarization, wherein if the gray value of the pixel point in the gray image is greater than the threshold value, the binarization result is 1 and represents a target rail; if the gray value of the pixel point is smaller than the threshold value, the binarization result is 0 and represents the rail background;
s13, carrying out edge detection on the image by using a CARRY operator, and calculating the current gradient strength of each pixel point;
s14, calculating the current gradient intensity G of the pixel pointMeasuringGradient intensity G of pixel points corresponding to normal railOften timesMaking a comparison if | GMeasuring-GOften timesIf the absolute value of | is greater than the threshold value, judging that the rail corresponding to the pixel point has a defect, and if | GMeasuring-GOften timesIf the absolute value of | is less than or equal to the threshold value, the process proceeds to step S9;
s15, converting the piezoelectric signals collected by the same section of the railway into digital signals, converting the time domain waveforms of the digital signals into power spectral density distribution patterns of frequency domains, comparing the power spectral density distribution patterns with the power spectral density distribution patterns of normal railway, and judging the health condition of the corresponding railway.
3. The inspection method according to claim 2, wherein the step of performing fuzzy detection on the picture by using Laplacian transformation in the step S8 includes:
and measuring a second derivative of the picture by using a Laplacian operator for boundary detection, performing convolution calculation by using 1 channel of the picture and combining a kernel of 3x3, and calculating a variance value of the output picture.
4. The inspection method according to claim 2, wherein the S10 is implemented by constructing a semantic segmentation model including:
s10.1, randomly dividing a target data set into a training sample set and a testing sample set;
and S10.2, training by adopting a training sample set, constructing a semantic segmentation model, and testing the precision of the model by adopting a test sample set.
5. The inspection method according to claim 2, wherein in the step S12, the threshold value of the pixel points in the gray image is calculated by local adaptive binarization, and if the gray value of the pixel points in the gray image is greater than the threshold value, the binarization result is 1, which represents the target rail; if the gray value of the pixel point is less than or equal to the threshold value, the binarization result is 0, which represents the rail background, and the method comprises the following steps:
s12.1, setting the gray value of the image at the pixel point (x, y) as f (x, y);
s12.2, calculating a threshold value w (x, y) of each pixel point (x, y) in the image:
W(x,y)＝0.5*(max f(x+m,y+n)+min f(x+m,y+n))
s12.3, if f (x, y) > w (x, y), the binarization result is 1 and represents a target point of the character area; otherwise, the binarization result is 0, which represents the target point of the background area.
6. The inspection method according to claim 2, wherein in the step S13, edge detection is performed on the image by using a CARRY operator, and the step of calculating the current gradient strength of each pixel point includes:
s13.1, smoothing the image by adopting a Gaussian filter, and filtering noise;
s13.2, calculating the gradient strength G of each pixel point in the image:
wherein G isxAnd GyGradient values in the x and y directions, respectively;
s13.3 Sobel operator S in the x and y directionsxAnd SyRespectively as follows:
s13.4, if a window of 3x3 in the image is A and a gradient-calculating pixel point is e, performing convolution calculation by combining a Sobel operator to obtain gradient values of the pixel point e in the x direction and the y direction, wherein the gradient values are respectively as follows:
wherein, is a convolution symbol, sum represents the sum of all elements in the matrix, and a, b, c, d, e, f, g, h and i are parameters in the convolution matrix.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011223678.8A CN112319543A (en) | 2020-11-05 | 2020-11-05 | Inspection device capable of identifying in-orbit and inspection method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011223678.8A CN112319543A (en) | 2020-11-05 | 2020-11-05 | Inspection device capable of identifying in-orbit and inspection method thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112319543A true CN112319543A (en) | 2021-02-05 |
Family
ID=74316010
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011223678.8A Pending CN112319543A (en) | 2020-11-05 | 2020-11-05 | Inspection device capable of identifying in-orbit and inspection method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112319543A (en) |
-
2020
- 2020-11-05 CN CN202011223678.8A patent/CN112319543A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102297660B (en) | Measuring method of shield tunnel lining segment seam open width and apparatus thereof | |
| CN106226157B (en) | Concrete structure member crevices automatic detection device and method | |
| CN106248686A (en) | Glass surface defects based on machine vision detection device and method | |
| CN106066333A (en) | Curved surface part surface profile breakdown diagnosis devices and methods therefor | |
| JP5175528B2 (en) | Tunnel lining crack inspection system | |
| CN109709574B (en) | Seabed microtopography laser scanning imaging system and three-dimensional terrain reconstruction method | |
| CN113259579B (en) | Image acquisition device, method and equipment | |
| KR101094069B1 (en) | Calibration Device and Method for revision of Crack Detection System by Using Image Processing Technology | |
| CN209640237U (en) | Optical detection apparatus | |
| CN112308847A (en) | Rail inspection vehicle inspection device and inspection method thereof | |
| CN104239904A (en) | Non-contact detection method for external outline of railway vehicle | |
| CN110186934B (en) | Axle box rubber pad crack detection method and detection device | |
| CN113610054A (en) | Underwater structure disease depth detection method, system and device and storage medium | |
| CN112319543A (en) | Inspection device capable of identifying in-orbit and inspection method thereof | |
| CN116309447B (en) | Dam slope crack detection method based on deep learning | |
| CN115575402B (en) | Intelligent identification, calculation and judgment method for defects of inner wall of closed cylindrical part | |
| CN113222907B (en) | Detection robot based on curved rail | |
| CN114913131A (en) | Vertical absolute and relative vertical deformation detection method for ballastless track structure | |
| CN115661453A (en) | Tower crane hanging object detection and segmentation method and system based on downward viewing angle camera | |
| CN109254011B (en) | Bridge defect detection device and detection method based on monocular area-array camera | |
| CN113465549A (en) | Mark point-free bearing retainer contact angle measuring system and method based on vision | |
| TWI419762B (en) | Online measuring method and apparatus for work pieces of machine tools | |
| CN111957592A (en) | Railway wagon bogie sleeper spring sorting system and sorting method thereof | |
| CN108225175B (en) | Device and method for acquiring tooth surface contact mark data of bevel gear of tractor transmission system | |
| CN214138538U (en) | Track range unit and system based on vision measurement |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20210205 |