CN117074050A - Trolley wheel rotation abnormality detection method based on visual detection - Google Patents
Trolley wheel rotation abnormality detection method based on visual detection Download PDFInfo
- Publication number
- CN117074050A CN117074050A CN202310982931.5A CN202310982931A CN117074050A CN 117074050 A CN117074050 A CN 117074050A CN 202310982931 A CN202310982931 A CN 202310982931A CN 117074050 A CN117074050 A CN 117074050A
- Authority
- CN
- China
- Prior art keywords
- wheel
- trolley
- bolt
- detection
- image
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 85
- 230000005856 abnormality Effects 0.000 title claims abstract description 27
- 230000000007 visual effect Effects 0.000 title claims abstract description 23
- 238000012544 monitoring process Methods 0.000 claims abstract description 46
- 230000002159 abnormal effect Effects 0.000 claims abstract description 31
- 238000004364 calculation method Methods 0.000 claims abstract description 12
- 238000013135 deep learning Methods 0.000 claims abstract description 8
- 238000012545 processing Methods 0.000 claims description 20
- 238000000034 method Methods 0.000 claims description 15
- 238000006073 displacement reaction Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 6
- 230000001960 triggered effect Effects 0.000 claims description 5
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 claims description 4
- 229910052782 aluminium Inorganic materials 0.000 claims description 4
- 230000002146 bilateral effect Effects 0.000 claims description 4
- 238000003708 edge detection Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 238000005299 abrasion Methods 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 3
- 238000009499 grossing Methods 0.000 claims description 3
- 238000005245 sintering Methods 0.000 abstract description 7
- 239000003550 marker Substances 0.000 abstract description 4
- 238000001931 thermography Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M17/00—Testing of vehicles
- G01M17/007—Wheeled or endless-tracked vehicles
- G01M17/013—Wheels
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
- G01J5/48—Thermography; Techniques using wholly visual means
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M1/00—Testing static or dynamic balance of machines or structures
- G01M1/12—Static balancing; Determining position of centre of gravity
- G01M1/122—Determining position of centre of gravity
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Aviation & Aerospace Engineering (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The application provides a trolley wheel rotation abnormality detection method based on visual detection, and belongs to the field of wheel rotation abnormality detection. According to the application, through the sintering trolley wheel rotation abnormality monitoring system, the position of the bolt on the hub surface of the sintering trolley is extracted and positioned by using a deep learning algorithm, and the rotation speed of the wheel is calculated by combining the values of the radius, the travelling speed and the like of the wheel. The infrared thermal imaging camera monitors the temperature condition of the contact position of the wheel and the rail in real time, and assists in judging the generation reason of the wheel locking phenomenon. According to the application, the bolt module of the sintering trolley wheel is detected by a deep learning algorithm to position the wheel rotation calculation target, so that the auxiliary pasting of a marker is not needed, and the workload and human error are reduced to a certain extent; through multiple judging conditions, the detailed classification of the rotation abnormality of the trolley wheels is realized. And repeating detection by using three groups of high-definition monitoring cameras, eliminating accidental errors, and judging the type of abnormal rotation by combining with the thermal infrared imager module. The scientificity and the reliability of the detection system are improved.
Description
Technical Field
The application relates to a trolley wheel rotation abnormality detection method based on visual detection, and belongs to the field of wheel rotation abnormality detection.
Background
The application CN201910023167.2 discloses a trolley wheel rotating speed monitoring system and a trolley wheel rotating speed monitoring method, wherein a rotating speed monitoring module realizes the marking of trolley wheels by using markers, and further performs wheel rotating speed monitoring by using a machine vision method. The technology has the defects that a plurality of marker modules are required to be stuck to each trolley wheel, and the workload is huge for a trolley system with a longer length; in addition, the pasting position and the extraction precision of the marker influence the speed calculation result, so that a large manual error exists.
Disclosure of Invention
The application aims to solve the problems in the prior art and further provides a trolley wheel rotation abnormality detection method based on visual detection.
The application aims at realizing the following technical scheme:
the trolley wheel rotation abnormality detection system based on visual detection comprises a photoelectric switch device, a thermal infrared imager module, a high-definition monitoring camera module and a data processing and control module, wherein trolley wheels are arranged on a trolley frame and walk along a trolley track; the photoelectric switch device, the thermal infrared imager module and the high-definition monitoring camera module are arranged on the aluminum profile bracket, the photoelectric switch device is positioned right above the trolley track, the emitter irradiates the tread of the rim of the trolley wheel, and the position is a; the thermal infrared imager module and the high-definition monitoring camera module are arranged on the central line of the wheel core of the trolley wheel in parallel and vertically irradiate the wheel; the high-definition monitoring camera module comprises 3 high-definition monitoring cameras, namely positions b, c and d, and the thermal infrared imager module is positioned at the position b; the data processing and controlling module is connected with the photoelectric switch device, the thermal infrared imager module and the high-definition monitoring camera module through a wired network.
The application discloses a detection method of a trolley wheel rotation abnormality detection system based on visual detection, wherein the distance among positions a, b, c and d is the trolley wheel distance.
The application discloses a detection method of a trolley wheel rotation abnormality detection system based on visual detection, which comprises the following specific steps of:
step one: after the trolley wheels enter the designated position a in the travelling process of the trolley, triggering the photoelectric switch device, transmitting signals of the photoelectric switch device to the data processing and control module, and recording the current trolley wheels as nth wheels;
step two: the data processing and control module outputs a control signal to control the high-definition camera module to acquire the trolley wheel image in the corresponding visual field; the high-definition monitoring cameras acquire images of trolley wheels, the corresponding trolley wheel numbers at the moment are respectively n-1, n-2 and n-3, the spacing distance of the 3 high-definition monitoring cameras is Lc and the distance between the trolley wheels is equal, and the height of the high-definition monitoring cameras is Hc and the height of the circle centers of the trolley wheels are equal;
step three, a step of performing; when the same wheel n runs to the detection positions b, c and d, two images are continuously acquired by using corresponding high-definition monitoring cameras at the detection positions, the acquisition interval time is t, the value of t is related to the number kappa of mounting bolts on the trolley wheels, the trolley speed v, the trolley angular speed omega and the radius r of the wheels, and the angular displacement l of a single bolt of the wheels which normally runs in the time t is as follows:
simultaneously, the angle of two adjacent mounting bolts on single platform truck wheel:
the rotation change condition of the bolt should satisfy l < l 1 The new position to which the bolt rotates after the time t is not overlapped with the position before the rotation of the adjacent bolt, and the new position can be obtained by the following formulas 1.1 and 1.2:
according to the formula 1.3, calculating the range of the acquisition interval time t, namely setting the continuous acquisition time of the monitoring camera;
step four: extracting bolts based on a deep learning target detection algorithm, firstly, acquiring 500-600 trolley wheel images in advance, manually marking the positions of the bolts in a square frame mode by using labelimg marking software, and then inputting yolo-v5 target detection networks to train a targeted prediction model;
step five, a step of performing a step of; in actual detection, an image P acquired by a high-definition monitoring camera is predicted through a yolo-v5 target detection model trained in the fourth step, and preliminary position information of all bolts is output; extracting a bolt region image P with the highest coordinate position A For a target object, combining an image processing algorithm, and further acquiring accurate position coordinates of the bolt:
firstly, smoothing a bolt area image P by adopting a bilateral filtering algorithm A The noise is removed, meanwhile, the edge information is reserved, and the degree of distinction between the bolt and the environment background is enhanced; taking the pixel mean value T of the neighborhood 5×5 range based on the image center point 1 Binarizing the image as a lower threshold to obtain a processed image P AS ，T 1 The calculation formula of (2) is as follows:
wherein x is the pixel value of the point, and the lower threshold value of the binarization algorithm can be calculated and obtained through a formula 1.4;
for image P AS Further extracting the edge of the mounting bolt, extracting by utilizing an edge detection function to obtain the outline of the edge of the bolt, and calculating a centroid calculation function to obtain the accurate centroid position coordinate z (x, y) of the bolt;
step six: in the same detection position, two images are acquired in the interval time t to calculate the barycenter coordinates, and the mounting bolt in the first image is the target of the time tThe positions of the target bolts are marked, the mounting bolts in the second image are the positions of the target bolts after t time, and the coordinates of the front and rear positions of the obtained target bolts are z respectively 1 (x 1 ,y 1 ) And z 2 (x 2 ,y 2 ) The distance r from the bolt to the center of the wheel is known z The angular displacement radian theta of the target bolt is calculated as follows:
radian of angular displacement and set threshold T 2 Comparing, if the threshold value is higher than the threshold value, the normal operation is performed, and if the threshold value is lower than the threshold value, the abnormal rotation phenomenon of the wheel occurs at the detection point; t (T) 2 The value of the arc value is set to be one third of the radian value of the wheel center of two adjacent bolts on the wheel;
step six: when the wheel rotates abnormally, the same wheel is subjected to judgment of the abnormal rotation phenomenon when the wheel runs to the set positions b, c and d, namely, the steps two to five are repeated three times, and if the abnormal rotation phenomenon of the wheel occurs, special conditions caused by uneven track can be eliminated;
step seven: the method comprises the steps of (a) judging the specific situation of abnormal rotation, namely locking or slipping of a wheel, wherein in the position b, an infrared thermal imager module is arranged, a temperature value of a contact surface of the wheel and a track is collected when a photoelectric switch device is triggered, when the temperature value exceeds a normal threshold range, locking caused by abnormal rotation of a wheel bearing is generated when the abnormal rotation phenomenon judged by a combined high-definition monitoring camera module occurs, and otherwise slipping caused by reduced wheel diameter after excessive wheel abrasion is generated.
The trolley wheel rotation abnormality detection method based on visual detection has the beneficial effects that:
1. the bolt module of the sintering trolley wheel is detected through the deep learning algorithm to position the wheel rotation calculation target, the auxiliary pasting of the marker is not needed, and the workload and human errors are reduced to a certain extent.
2. Through a machine vision image processing algorithm, the bolt outline is accurately extracted, the mass center is positioned, the calculation accuracy is improved, and the accuracy of the result is enhanced.
3. The application realizes the detailed classification of the wheel rotation abnormality of the trolley through multiple judging conditions. And repeating detection by using three groups of high-definition monitoring cameras, eliminating accidental errors, and judging the type of abnormal rotation by combining with the thermal infrared imager module. The scientificity and the reliability of the detection system are improved.
Drawings
Fig. 1 is an overall schematic diagram of a trolley wheel rotation abnormality detection system based on visual detection according to the present application.
Fig. 2 is a flowchart of a method for detecting wheel rotation abnormality of a dolly based on visual detection according to the present application.
Fig. 3 is a flowchart of extracting a bolt centroid by a visual algorithm in the method for detecting the abnormal rotation of the trolley wheel based on visual detection.
Fig. 4 is a schematic diagram of angular displacement measurement in the method for detecting abnormal rotation of a trolley wheel based on visual detection.
The reference numerals in the figures are: 1 is a photoelectric switch device; 2 is a trolley wheel; 3 is a trolley track; 4 is a thermal infrared imager module; 5 is a high definition monitoring camera module; and 6, a data processing and control module.
Detailed Description
The application will be described in further detail with reference to the accompanying drawings: the present embodiment is implemented on the premise of the technical solution of the present application, and a detailed implementation is given, but the scope of protection of the present application is not limited to the following embodiments.
Embodiment one: as shown in fig. 1-4, the abnormal rotation monitoring system of the wheel of the sintering trolley utilizes a deep learning algorithm to extract and position the bolt position of the hub surface of the sintering trolley, and calculates the rotation speed of the wheel by combining the values of the radius, the travelling speed and the like of the wheel. In addition, the system is designed with an infrared thermal imaging camera to monitor the temperature condition of the contact position of the wheels and the rail in real time, and assist in judging the reason of the locking phenomenon of the wheels.
The technical scheme of the application comprises the following detailed steps:
s1, a whole schematic diagram of a monitoring system is shown in FIG. 1: the system comprises a photoelectric switch device 1, a thermal infrared imager module 4, a high-definition monitoring camera module 5 and a data processing and control module 6. The trolley wheels 2 are arranged on the trolley frame and run from north to south along the trolley track 3; the photoelectric switch device 1, the thermal infrared imager module 4 and the high-definition monitoring camera module 5 are arranged on the aluminum profile bracket, the photoelectric switch device is arranged right above the trolley track 3 and irradiates the tread of the rim of the trolley wheel 2, and when the trolley wheel 2 walks to the position right below the photoelectric switch device 1, a wheel in-place signal is triggered; the thermal infrared imager module 4 and the high-definition monitoring camera module 5 are arranged parallel to the central line of the core of the trolley wheel 2 and vertically irradiate the wheel; the data processing module is placed in the machine room and connected with the field device through a wired network and is used for receiving, storing, analyzing and calculating various signals of the field device.
S2, the photoelectric switch module 1 is arranged right above the track and the wheel rim, and after the wheels enter the designated position a in the travelling process of the trolley, the photoelectric switch signal is triggered and transmitted to the data processing and control module 6. The current wheel is recorded as the nth wheel.
S3, the computer generates a control signal to control the high-definition camera module 5 to collect the wheel images in the corresponding vision field. The number of the high-definition monitoring cameras is 3, the high-definition monitoring cameras are arranged on the ground aluminum profile frame, are arranged parallel to the center line of the wheel core of the trolley wheel 2 and vertically irradiate towards the wheel core, the arrangement positions are shown in figures b, c and d, and the corresponding wheel numbers are n-1, n-2 and n-3 respectively. The spacing distance Lc is equal to the distance between the sintering trolley wheels, and the height Hc is equal to the height of the center of the wheel. Each camera is used for respectively and correspondingly acquiring an image of one wheel, and the set visual field is used for ensuring that the complete image of the wheel can be acquired.
S4, for the same wheel n, firstly, two images are continuously acquired by using a monitoring camera at a detection position, wherein the interval time is t, and the value of t is related to the number kappa of mounting bolts (black hexagons shown in fig. 4) on the trolley wheel 2, the trolley speed v, the trolley angular speed omega and the wheel radius r, and the angular displacement l of a single bolt of the wheel which normally runs in the time t is as follows:
simultaneously, the angle of two adjacent bolts on single platform truck wheel:
therefore, in order to judge the rotation change condition of the bolt, l is less than l 1 The new position to which the bolt rotates after the time t is not overlapped with the position before the rotation of the adjacent bolt, and the new position can be obtained by the following formulas 1.1 and 1.2:
s5, extracting bolts based on a deep learning target detection algorithm, wherein the visual algorithm flow is shown in fig. 3. Firstly, a certain number (about 500-600) of wheel images are collected in advance, bolt positions are manually marked in a square frame mode by using labelimg marking software, and then yolo-v5 target detection networks are input to train a targeted prediction model. The Yolo network has the advantages of high detection speed and good precision, and meets the requirement of continuous detection in a short time. And inputting the visible light image acquired by the camera into a trained model, and outputting the region position of each bolt.
S6, a certain background area interference exists in the bolt area extracted in the step S5, and the accurate barycenter coordinates of the bolt cannot be obtained. Therefore, the bolt with the highest coordinate position is taken as a target object, and the visual algorithm flow is shown in fig. 3 by combining an image processing algorithm, and the steps are as follows:
firstly, a bilateral filtering algorithm is adopted to smooth the extracted bolt area image and is used for removing noise, meanwhile, edge information is reserved, and the distinguishing degree of the bolt and the environment background is enhanced. Taking the pixel mean value T of the neighborhood 5×5 range based on the image center point 1 Image binarization as lower threshold, T 1 The calculation formula of (2) is as follows: where x is the pixel value of the point.
Further extracting the edge of the bolt, extracting a binarized image by utilizing an edge detection function (the threshold value is generated in real time as a binarization lower limit in the process of processing and calculating the binarization function, so as to realize the binarization of the self-adaptive threshold value and enhance the binarization processing effect), obtaining the contour of the edge of the bolt, and calculating the centroid calculation function of the contour of the edge of the bolt to obtain the accurate centroid position coordinate z (x, y) of the bolt.
S7, calculating mass center coordinates of two images acquired in interval time t at the same detection position, wherein as shown in fig. 4, a black solid hexagon is the position of the target bolt before t time, a dotted hollow hexagon is the position of the target bolt after t time, and the coordinates of the front position and the rear position are z respectively 1 (x 1 ,y 1 ) And z 2 (x 2 ,y 2 ) The distance r from the bolt to the center of the wheel is known z The angular displacement radian theta of the bolt is calculated as follows:
comparing the radian of the angular displacement with a set threshold value, wherein when the radian of the angular displacement is higher than the threshold value, the wheel runs normally, and when the radian of the angular displacement is lower than the threshold value, the wheel is locked at the detection point.
S8, the abnormal rotation phenomenon of the wheels is divided into self problems and external problems. The self problems include slipping phenomenon caused by the reduction of the wheel diameter after excessive wheel wear and wheel locking phenomenon caused by abnormal wheel bearing; external problems include occasional phenomena caused by uneven track. Therefore, multiple judgment needs to be carried out, and the abnormal rotation phenomenon judgment is carried out when the same wheel runs to the detection positions b, c and d, namely, the steps S4-S7 are repeated three times, and if the abnormal rotation phenomenon of the wheel occurs, the special situation caused by uneven track can be eliminated. (detection at position a may be due to the problem of the track at position a, and not necessarily the problem of the wheel itself, so that detection of an abnormality at positions b, c, and d is performed again on that wheel, and if abnormality is detected at several positions, it can be basically determined that abnormality of the wheel itself is not related to the track
S9, further judging the locking and slipping phenomenon of the wheel, and collecting the temperature value of the contact surface of the wheel and the track when the photoelectric switch is triggered at the detection position b by the thermal infrared imager. The thermal infrared imager can further distinguish whether the abnormal type is wheel locking or wheel slipping after detecting the abnormality of the wheel, wherein the wheel locking is contact and relative friction between the wheel and the rail, and the temperature can be abnormally increased; wheel slip is when the wheels and the rails are not in contact, and the temperature is normal. Therefore, when the temperature value exceeds the normal threshold range and the abnormal rotation phenomenon judged by the visible light camera is combined, the locking phenomenon caused by abnormal wheel bearing is generated. Otherwise, the slipping phenomenon caused by the small wheel diameter after the excessive wheel abrasion occurs.
Embodiment two: as shown in fig. 1, the method for detecting the abnormal rotation of the trolley wheel based on visual detection according to the present embodiment includes:
1. firstly, for the trolley wheel n which is travelling, the trolley wheel n rolls to a photoelectric switch trigger point a along the track to trigger a photoelectric in-place signal, and at the moment, the monitoring cameras at the computer control positions b, c and d collect trolley wheel images in the corresponding vision fields. At this time, the numbers of the wheels corresponding to the positions b, c and d are respectively n-1, n-2 and n-3. Each wheel is detected by the three detection positions b, c and d, and the detection process can be considered to be finished. Further, the detection matters of the single detection position are explained.
2. When the trolley wheels n run to the positions b, c and d, the monitoring camera is controlled to continuously acquire two images at t time intervals, t <0.82m/s is calculated according to the formula 1.3, wherein the radius of the wheels is 0.25m, the number of bolts is 6, and the running speed of the trolley is 0.32m/s. To clearly distinguish the variation of the bolts in successive images, t is one third of the maximum, i.e. t=0.27 s.
3. And (3) acquiring a certain number (500-600) of trolley wheel images in advance, wherein the images are consistent with the images acquired by the monitoring cameras at the positions b, c and d, marking the positions of bolts on the wheels by using labelimg image marking software, and collecting the marked images into a data set J1. And inputting the data set J1 into a network for model training by using an open source yolo-v5 deep learning target detection network as a basis.
4. In the actual detection, the detection of the position b is taken as an example. The monitoring camera collects an image P, inputs the yolo-v5 target detection model trained in the step 3 to conduct bolt position prediction, and the model automatically outputs the approximate position coordinate information of the bolts on the wheels. Selecting the bolt region coordinate P with highest image center position A [(745,214),(776,246)]And cut into images P A 。
5. And step four, a certain redundant boundary exists at the extracted bolt position, and accurate boundary information needs to be further acquired, so that an accurate center is calculated. Further processing by adopting an image processing algorithm: firstly, smoothing an image P by adopting a bilateral filtering algorithm A The noise is removed, meanwhile, the edge information is reserved, and the degree of distinction between the bolt and the environment background is enhanced; then taking the image center point as the basis, the opencv algorithm automatically takes the pixel mean value of the neighborhood 5 multiplied by 5 range as the lower threshold value to carry out image binarization, and the processed image P is obtained AS The method comprises the steps of carrying out a first treatment on the surface of the Finally, extracting P by edge detection function AS Performing centroid calculation function calculation on the edge contour of the bolt to obtain accurate centroid position coordinate z1 (761,231) of the bolt;
6. and (3) carrying out centroid extraction operation in the steps 4-5 on two images continuously shot at the same detection position, calculating mass center coordinates as z1 (761,231), calculating z2 (815,264), knowing that the pixel distance from a bolt to a wheel center is about 300 pixels according to the formula 1.5, and calculating the angular displacement radian theta= 0.4252 of the bolt. And comparing theta with a set threshold value T2 (the value of T2 is set to be one third of the radian value of the center of the wheel relative to two adjacent bolts on the wheel, namely, T2 (2 pi/6/3) =0.349), so that theta > T2 represents normal operation of the wheel. In this example, the normal rotation of the wheel is detected at the detection position b, and thus the subsequent step determination does not affect the detection result.
7. If the abnormal rotation of the wheel occurs in the step 6, the abnormal rotation judgment in the steps 4 to 6 is also performed when the wheel is to travel to the detection positions c and d. If the detection results of the positions b, c and d on the same wheel are abnormal, the wheel can be judged to have abnormal rotation.
8. According to the detection result of the step 7, if no abnormal rotation is detected, the next trolley wheel detection process is carried out; if a rotational abnormality is detected, it is necessary to further determine the specific cause of the rotational abnormality. The abnormal rotation is divided into two conditions of wheel locking and wheel slipping. When the wheel locks, the wheel and the rail can generate dry friction, so that the local temperature is abnormally increased; when the wheels slip, the wheels are not contacted with the rail, and temperature abnormality can not exist. And at the detection position b, detecting the temperature values of the wheels and the track surface by using a thermal infrared imager, wherein the wheel locking phenomenon is generated when the temperature is higher than a normal threshold range (the normal running temperature threshold of the trolley is (30-60 degrees)), otherwise, the slipping phenomenon is generated.
In the foregoing, the present application is merely preferred embodiments, which are based on different implementations of the overall concept of the application, and the protection scope of the application is not limited thereto, and any changes or substitutions easily come within the technical scope of the present application as those skilled in the art should not fall within the protection scope of the present application. Therefore, the protection scope of the application should be subject to the protection scope of the claims.
Claims (3)
1. The trolley wheel rotation abnormality detection system based on visual detection is characterized by comprising a photoelectric switch device (1), a thermal infrared imager module (4), a high-definition monitoring camera module (5) and a data processing and control module (6), wherein the trolley wheel (2) is arranged on a trolley frame and walks along a trolley track (3); the photoelectric switch device (1), the thermal infrared imager module (4) and the high-definition monitoring camera module (5) are arranged on the aluminum profile bracket, the photoelectric switch device (1) is positioned right above the trolley track (3) and the emitter irradiates the tread of the rim of the trolley wheel (2), and the position is a; the thermal infrared imager module (4) and the high-definition monitoring camera module (5) are parallelly arranged on the central line of the wheel core of the trolley wheel (2) and vertically irradiate to the wheel; the high-definition monitoring camera module (5) comprises 3 high-definition monitoring cameras, namely positions b, c and d, and the thermal infrared imager module (4) is positioned at the position b; the data processing and control module (6) is connected with the photoelectric switch device (1), the thermal infrared imager module (4) and the high-definition monitoring camera module (5) through a wired network.
2. The detection method of the trolley wheel rotation abnormality detection system based on visual detection according to claim 1, wherein the spacing between the positions a, b, c and d is a trolley wheel spacing.
3. The detection method of the trolley wheel rotation abnormality detection system based on visual detection according to claim 1 or 2, characterized in that the trolley wheel rotation abnormality detection method based on visual detection specifically comprises the following steps:
step one: after a trolley wheel (2) enters a designated position a in the travelling process of the trolley, triggering a photoelectric switch device (1), transmitting a signal of the photoelectric switch device (1) to a data processing and control module (6), and recording the current trolley wheel (2) as an nth wheel;
step two: the data processing and control module (6) outputs a control signal to control the high-definition camera module (5) to acquire images of trolley wheels (2) in the corresponding view; the high-definition monitoring cameras acquire images of trolley wheels (2), the numbers of the corresponding trolley wheels (2) are respectively n-1, n-2 and n-3, the spacing distance of the 3 high-definition monitoring cameras is Lc and equal to the distance between the trolley wheels, and the height of the high-definition monitoring cameras is Hc and equal to the height of the circle center of the trolley wheels (2);
step three, a step of performing; when the same wheel n runs to the detection positions b, c and d, two images are continuously acquired by using corresponding high-definition monitoring cameras at the detection positions, the acquisition interval time is t, the value of t is related to the number kappa of mounting bolts (7) on the trolley wheels (2), the trolley speed v, the trolley angular speed omega and the wheel radius r, and the angular displacement l of a single bolt of the normally running wheel in the t time is as follows:
simultaneously, the angle of two adjacent mounting bolts on single trolley wheel (2):
the rotation change condition of the bolt should satisfy l < l 1 The new position to which the bolt rotates after the time t is not overlapped with the position before the rotation of the adjacent bolt, and the new position can be obtained by the following formulas 1.1 and 1.2:
according to the formula 1.3, calculating the range of the acquisition interval time t, namely setting the continuous acquisition time of the monitoring camera;
step four: extracting bolts based on a deep learning target detection algorithm, firstly, acquiring 500-600 trolley wheel images in advance, manually marking the positions of the bolts in a square frame mode by using labelimg marking software, and then inputting yolo-v5 target detection networks to train a targeted prediction model;
step five, a step of performing a step of; in actual detection, an image P acquired by a high-definition monitoring camera is predicted through a yolo-v5 target detection model trained in the fourth step, and preliminary position information of all bolts is output; extracting a bolt region image P with the highest coordinate position A For a target object, combining an image processing algorithm, and further acquiring accurate position coordinates of the bolt:
firstly, smoothing a bolt area image P by adopting a bilateral filtering algorithm A The noise is removed, meanwhile, the edge information is reserved, and the degree of distinction between the bolt and the environment background is enhanced; taking the pixel mean value T of the neighborhood 5×5 range based on the image center point 1 Binarizing the image as a lower threshold to obtain a processed image P AS ，T 1 The calculation formula of (2) is as follows:
wherein x is the pixel value of the point, and the lower threshold value of the binarization algorithm can be calculated and obtained through a formula 1.4;
for image P AS Further extracting the edge of the mounting bolt, extracting by utilizing an edge detection function to obtain the outline of the edge of the bolt, and calculating a centroid calculation function to obtain the accurate centroid position coordinate z (x, y) of the bolt;
step six: in the same detection position, two images are acquired in the interval time t for calculating the barycenter coordinates, the mounting bolt in the first image is the target bolt position before the time t, the mounting bolt in the second image is the target bolt position after the time t, and the coordinates of the front position and the rear position of the obtained target bolt are z respectively 1 (x 1 ,y 1 ) And z 2 (x 2 ,y 2 ) The distance r from the bolt to the center of the wheel is known z The angular displacement radian theta of the target bolt is calculated as follows:
radian of angular displacement and set threshold T 2 Comparing, if the threshold value is higher than the threshold value, the normal operation is performed, and if the threshold value is lower than the threshold value, the abnormal rotation phenomenon of the wheel occurs at the detection point; t (T) 2 The value of the arc value is set to be one third of the radian value of the wheel center of two adjacent bolts on the wheel;
step six: when the wheel rotates abnormally, the same wheel is subjected to judgment of the abnormal rotation phenomenon when the wheel runs to the set positions b, c and d, namely, the steps two to five are repeated three times, and if the abnormal rotation phenomenon of the wheel occurs, special conditions caused by uneven track can be eliminated;
step seven: the method is characterized in that the specific situation of abnormal rotation is further judged, the phenomenon of wheel locking or slipping is further judged, an infrared thermal imager module (4) is arranged at the position b, the temperature value of the contact surface of the wheel and the rail is collected when the photoelectric switch device (1) is triggered, when the temperature value exceeds a normal threshold range, the phenomenon of abnormal rotation judged by the combined high-definition monitoring camera module (5) is generated, namely the phenomenon of locking caused by abnormal wheel bearing is generated, and otherwise, the phenomenon of slipping caused by reduced wheel diameter after excessive wheel abrasion is generated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310982931.5A CN117074050A (en) | 2023-08-07 | 2023-08-07 | Trolley wheel rotation abnormality detection method based on visual detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310982931.5A CN117074050A (en) | 2023-08-07 | 2023-08-07 | Trolley wheel rotation abnormality detection method based on visual detection |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117074050A true CN117074050A (en) | 2023-11-17 |
Family
ID=88707167
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310982931.5A Pending CN117074050A (en) | 2023-08-07 | 2023-08-07 | Trolley wheel rotation abnormality detection method based on visual detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117074050A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117634989A (en) * | 2024-01-25 | 2024-03-01 | 深圳市大力鸿震智能脚轮科技有限公司 | Caster quality assessment method, system and storage medium based on artificial intelligence |
-
2023
- 2023-08-07 CN CN202310982931.5A patent/CN117074050A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117634989A (en) * | 2024-01-25 | 2024-03-01 | 深圳市大力鸿震智能脚轮科技有限公司 | Caster quality assessment method, system and storage medium based on artificial intelligence |
| CN117634989B (en) * | 2024-01-25 | 2024-05-28 | 深圳市大力鸿震智能脚轮科技有限公司 | Caster quality assessment method, system and storage medium based on artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104608799B (en) | Based on information fusion technology Railway wheelset tread damage on-line checking and recognition methodss | |
| CN117074050A (en) | Trolley wheel rotation abnormality detection method based on visual detection | |
| CN110862033A (en) | Intelligent early warning detection method applied to coal mine inclined shaft winch | |
| CN110047070B (en) | Method and system for identifying rail wear degree | |
| CN106446807A (en) | Well lid theft detection method | |
| CN107392885A (en) | A kind of method for detecting infrared puniness target of view-based access control model contrast mechanism | |
| JPH11344334A (en) | Method for deciding traction of vehicle | |
| CN107576667A (en) | A kind of railway rail clip abnormality detection system based on linear array thermal camera | |
| JP4032727B2 (en) | Lane boundary detection device | |
| CN103051872A (en) | Method for detecting conveyor belt deviation based on image edge extraction | |
| CN112197715B (en) | Elevator brake wheel and brake shoe gap detection method based on image recognition | |
| CN106813569A (en) | A kind of automobile tire 3-D positioning method based on line-structured light | |
| CN111259718A (en) | Escalator retention detection method and system based on Gaussian mixture model | |
| CN107392216B (en) | Method for quickly identifying circumferential seams of shield tunnel segments based on gray data | |
| CN107067752A (en) | Automobile speedestimate system and method based on unmanned plane image | |
| CN115331000A (en) | ORB algorithm-based bow net running state detection method | |
| CN109060828A (en) | A kind of locomotive wheel thread defect image detecting system | |
| CN107977531A (en) | A kind of method that ground resistance hard measurement is carried out based on image procossing and field mathematical model | |
| CN113033443B (en) | Unmanned aerial vehicle-based automatic pedestrian crossing facility whole road network checking method | |
| CN107472298A (en) | The detection method and system of wheel diameters | |
| CN114572273B (en) | Railway vehicle wheel set tread 3D image detection method | |
| CN113222907B (en) | Detection robot based on curved rail | |
| CN112285111A (en) | Pantograph front carbon sliding plate defect detection method, device, system and medium | |
| JP3390515B2 (en) | Position matching method, vehicle speed calculation method, and vehicle speed calculation device | |
| CN111127442B (en) | Trolley wheel shaft defect detection method and device |
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 |