CN210386980U - Machine vision-based intelligent cooling bed control system - Google Patents
Machine vision-based intelligent cooling bed control system Download PDFInfo
- Publication number
- CN210386980U CN210386980U CN201921422712.7U CN201921422712U CN210386980U CN 210386980 U CN210386980 U CN 210386980U CN 201921422712 U CN201921422712 U CN 201921422712U CN 210386980 U CN210386980 U CN 210386980U
- Authority
- CN
- China
- Prior art keywords
- cooling bed
- control
- image
- acquisition device
- bed
- 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
Abstract
The utility model discloses a machine vision-based cooling bed intelligent control system, which comprises a cooling bed, an image acquisition device, a control host, a communication server and a monitoring display device, wherein the cooling bed is controlled and executed by a PLC (programmable logic controller), and the image acquisition device is matched with the device and arranged above the cooling bed and used for acquiring image information of a cooling bed region; the control host is respectively connected with the image acquisition device and the monitoring display equipment phase control, and is connected with the cooling bed phase control through the communication server; the communication server is also connected with the monitoring display equipment. The utility model discloses realize that cold bed control is by the manual intelligent conversion of automation of tradition, realize the high accuracy control and the quick logic processing of cold bed control technique, have the significance to promoting the intelligent development of iron and steel industry process, have apparent effect to improving iron and steel industry full flow production efficiency.
Description
Technical Field
The utility model relates to a hot rolled steel plate production field, the more specifically cold bed intelligence control system based on machine vision that says so.
Background
The cooling bed is an important part of the production process of the medium plate, on one hand, the cooling bed provides enough air cooling time for the steel plate, releases stress, ensures the performance and the plate shape, and on the other hand, the cooling bed plays a role in caching the production process and controls the whole production rhythm. At present, almost all the cooling beds of the hot rolling medium plate production line in China are manually controlled and mechanically automated methods, and the operation has obvious defects:
①, the current cooling bed operation is mainly characterized in that specific operators judge whether the steel plate moves under the cooling bed or not according to own experience, and the like, so that the requirement of standardized production cannot be met.
②, the efficiency is low, because the production line is in real-time operation state, the individual can not realize the real-time control to the whole cooling bed area, which causes the low efficiency, even becomes the factor restricting the production, and can not meet the requirement of the matching of the rolling process.
③, the control precision is poor, different teams and groups have different operation details, and different people have different operations, which causes the subjective influence on the control of the cooling bed and the control of the steel plate to be large, and the requirement of high-precision error-free production cannot be met.
Therefore, the conversion of the traditional manual control to the automatic intelligent control of the cooling bed is realized, the method has important significance for improving the cost advantage and the technical advantage of the steel enterprise, and in the industrial 4.0 era, the cooling bed control technology with high-precision control and rapid logic processing is developed, cooling bed control equipment matched with the cooling bed control technology is developed, the method has important significance for promoting the intelligent development of the steel industrial process, and the method has obvious effect for improving the whole process production efficiency of the steel industry.
SUMMERY OF THE UTILITY MODEL
The utility model discloses a cold bed intelligence control system based on machine vision, its main aim at overcome the above-mentioned not enough and shortcoming that prior art exists.
The utility model discloses the system is in the construction process, including the acquirement of cold bed target area image, cut apart and handle the image that acquires, GPU is technical processing with higher speed, the data that gather from the industry camera promptly transmit for the host computer through data acquisition's software and hardware system, trigger the image processing module and the logic processing module of host computer end, convert original image data into the data transmission that actually is available through a series of image processing procedure and give the logic end, the logic end carries out the logic control of whole cold bed through image processing's data, send control command and be used for cold bed action control for PLC, send the command simultaneously and be used for information updating control for the service end.
On the other hand, the system controls the cooling bed by means of machine vision, and mainly realizes information perception and dynamic perception of equipment to a cooling bed area, feeds back a perception result to a logic part to give a command and correction, and finally realizes deep learning of various control data generated by the cooling bed by constructing a deep learning machine neural network and an intelligent algorithm so as to realize self-state judgment and control of the cooling bed.
The technical scheme adopted by the scheme is as follows:
a machine vision-based intelligent cooling bed control system comprises a cooling bed, an image acquisition device, a control host, a communication server and a monitoring display device, wherein the cooling bed is controlled and executed by a PLC (programmable logic controller), and the image acquisition device is matched with the device and arranged above the cooling bed and used for acquiring image information of a cooling bed area; the control host is respectively connected with the image acquisition device and the monitoring display equipment phase control, and is connected with the cooling bed phase control through the communication server; the communication server is also connected with the monitoring display equipment.
Furthermore, the image acquisition device comprises six industrial cameras with high resolution, laser transmitters and trigger signal controllers, the six industrial cameras are respectively and symmetrically arranged above two sides of the cooling bed, the laser transmitters are used for assisting in positioning the position of the steel plate, the trigger signal controllers are respectively connected with the industrial cameras and the control host, and the trigger signal controllers are activated by image signals of the industrial cameras and then are connected with the control host.
Furthermore, the control host comprises an image processing module and a logic processing module, the image processing module is respectively connected with the image acquisition device and the logic processing module, and the logic processing module is connected with the cooling bed phase through the communication server in a control mode.
Furthermore, the logic processing module is also constructed with a deep learning convolution neural training network and an intelligent algorithm, and is used for deep learning and training various control logic instructions generated by the cooling bed, and intelligently identifying and judging the control state of the cooling bed.
A use method of a machine vision-based cooling bed intelligent control system comprises the following specific steps:
the method comprises the following steps: acquiring a cold bed image, namely acquiring an image of a cold bed area by an image acquisition device arranged above a cold bed, and transmitting collected image data information to a control host to finish data acquisition and transmission;
step two: image processing, namely, the control host performs target selection and perspective transformation processing on the image data information uploaded in the step one;
step three: filtering the image, namely filtering the image subjected to perspective transformation in the step two by adopting a median filtering denoising algorithm to remove noise points of the image;
step four: performing steel plate image segmentation, namely performing threshold segmentation on the target image obtained in the step three by adopting an OTSU threshold processing algorithm to completely extract the steel plate target image from the image, forming a whole or partial pixel set of the steel plate target image, and finally finding out a partial or whole outline of the steel plate target image;
step five: and image data is accelerated, a GPU image acceleration processor is adopted to convert and transmit original image data to a logic end of a control host, the logic end carries out logic control on the whole cooling bed through the obtained image processing data, sends a control instruction to a PLC for controlling the cooling bed behavior, and sends a command to a communication server for information updating control.
Further, the image processing in the second step includes the following specific steps:
(1) selecting a target area, namely adaptively selecting four coordinate points in an image, wherein a quadrangle formed by enclosing the four coordinate points comprises the target area to be processed;
(2) perspective transforming the image, mapping the image of the selected target area from a two-dimensional view plane to a three-dimensional view plane by using the following mapping formula,
wherein, in the step (A),
representing points before transformation, before transformation𝑤A value of 1, (U, V, 1) in three-dimensional representation, mapped to three-dimensional by matrix transformation
；
Then the image is mapped again from the three-dimensional view plane to the two-dimensional view plane by using the following mapping formula,
and completing perspective transformation processing of the image.
Further, the specific steps of adaptively selecting four coordinate points in the step (1) are as follows:
A. acquiring relevant parameters of the camera, wherein the relevant parameters of the camera comprise a field range and an angle α of camera shooting, and the arrangement height of the camerah0Distance l from camera to laser line0Distance l from camera to the nearest side of cooling bed1And the width l of the laser line obtained by cutting2；
B. Calculating to obtain the horizontal straight line distance l from the camera to the nearest side of the cooling bed by using the calculation formula of the side length of the right-angled triangle through the acquired camera parameters3And the horizontal linear distance l from the camera to the laser line4；
C. Width l of laser line obtained by cutting2Obtain the coordinates w of two end points1And w2；
D. By calculating l4And l3And the coordinate point w1And w2Four coordinate points p of the target area can be obtained1，p2，p3，p4。
Furthermore, the logic control in the fifth step further comprises the step of using a convolutional neural training network and an intelligent algorithm for constructing deep learning, and performing deep learning and training on various control logic instructions generated by the cooling bed, and intelligently identifying and judging the control state of the cooling bed.
Through the above description and illustration of the present invention, compared with the prior art, the present invention has the advantages that:
according to the scheme, the control of the cooling bed is realized by means of machine vision, the equipment can sense and dynamically sense the information of the cooling bed area, can feed back to a logic part according to a sensing result to give a command and correct, and finally, by constructing the learning of a deep learning machine neural network and an intelligent algorithm, the learning of various control data generated by the cooling bed is realized, and the self-state judgment and control of the cooling bed are realized. The scheme realizes high-precision control and rapid logic processing control of equipment, realizes intelligent operation of a cooling bed control technology, promotes intelligent development of the steel industry process, and has a remarkable effect of improving the whole process production efficiency of the steel industry.
Drawings
Fig. 1 is a system configuration diagram of the present invention.
Fig. 2 is a schematic diagram of the distribution structure of the image capturing device of the present invention.
Fig. 3 is a schematic diagram of the industrial camera of the present invention.
Detailed Description
The following further describes embodiments of the present invention with reference to the accompanying drawings.
As shown in fig. 1 to 3, an intelligent cooling bed control system based on machine vision includes a cooling bed 1 controlled and executed by a PLC, an image acquisition device 2, a control host 3, a communication server 4, and a monitoring display device 5, where the image acquisition device 2 is installed above the cooling bed 1 in a matching manner, and is used to acquire image information of a cooling bed area; the control host 3 is respectively connected with the image acquisition device 2 and the monitoring display device 5 in a control mode, and the control host 3 is connected with the cooling bed 1 in a control mode through the communication server 4; the communication server 4 is also connected with the monitoring display device 5.
Further, as shown in fig. 1 and 2, the image capturing device 2 includes six industrial cameras 21 with high resolution, six laser transmitters 22 symmetrically disposed above two sides of the cooling bed 1, and a trigger signal controller (not shown in the figures), wherein the laser transmitters 22 are used for assisting in positioning the position of the steel plate, the trigger signal controller is respectively connected to the industrial cameras 21 and the control host 3, and the trigger signal controller is activated by the image signal of the industrial cameras 21 and then connects the industrial cameras 21 to the control host 3.
Furthermore, the control host comprises an image processing module and a logic processing module, the image processing module is respectively connected with the image acquisition device and the logic processing module, and the logic processing module is connected with the cooling bed phase through the communication server in a control mode.
Furthermore, the logic processing module is also constructed with a deep learning convolution neural training network and an intelligent algorithm, and is used for deep learning and training various control logic instructions generated by the cooling bed, and intelligently identifying and judging the control state of the cooling bed.
A use method of a machine vision-based cooling bed intelligent control system comprises the following specific steps:
the method comprises the following steps: acquiring a cold bed image, namely acquiring an image of a cold bed area by an image acquisition device arranged above a cold bed, and transmitting collected image data information to a control host to finish data acquisition and transmission;
step two: image processing, namely, the control host performs target selection and perspective transformation processing on the image data information uploaded in the step one;
step three: filtering the image, namely filtering the image subjected to perspective transformation in the step two by adopting a median filtering denoising algorithm to remove noise points of the image;
step four: performing steel plate image segmentation, namely performing threshold segmentation on the target image obtained in the step three by adopting an OTSU threshold processing algorithm to completely extract the steel plate target image from the image, forming a whole or partial pixel set of the steel plate target image, and finally finding out a partial or whole outline of the steel plate target image;
the purpose of image segmentation is to segment a target or an interested part in an obtained image, and parts needing processing in a steel plate image have two aspects, namely laser line segmentation and laser line data extraction. The laser line segmentation aims at realizing the positioning of the steel plate at the position of the cooling bed, so as to depict the distribution condition of the steel plate on the cooling bed; the positioning of the steel plate on the cooling bed is related to whether a new steel plate can be put on the cooling bed or not, and the distribution of the steel plate on the cooling bed is related to the operation strategy of the cooling bed. The accurate position depiction is the first step of realizing intelligent control of the cooling bed, and the processing capacity directly reflects the intelligent processing capacity of the cooling bed.
The OTSU is a method for determining a threshold value by calculating the maximum variance of the gray distribution of an image based on a gray histogram of the image, and binarizes the image through the threshold value. The algorithm first assumes that there are two classes of pixels in the image, namely the target pixel and the background pixel, and in general, to classify the data, it can be calculated that its intra-class variance is minimized or its inter-class variance is maximized. The OTSU method is to find the threshold value when the inter-class variance is maximized, because larger inter-class variance means smaller probability of false score.
The mathematical expression process of the OTSU method is as follows:
for any image I (x, y), the image size is M × N, the segmentation threshold of the foreground background is assumed to be T, the segmentation threshold is 255 if the segmentation threshold is larger than the threshold, otherwise the segmentation threshold is 0, the proportion of target pixels belonging to the foreground in the whole image is w0, the gray average value of the target pixels is u0, the proportion of pixels belonging to the background in the whole image is w1, the gray average value of the target pixels is u1, the gray average value of the whole image is u, and the inter-class variance is g; the number of pixels in the image smaller than the threshold T is N0, and the number of pixels equal to or larger than the threshold T is N1.
For the OTSU method, T when the variance is maximized is the threshold value.
Step five: and image data is accelerated, a GPU image acceleration processor is adopted to convert and transmit original image data to a logic end of a control host, the logic end carries out logic control on the whole cooling bed through the obtained image processing data, sends a control instruction to a PLC for controlling the cooling bed behavior, and sends a command to a communication server for information updating control.
Further, the image processing in the second step includes the following specific steps:
(1) selecting a target area, namely adaptively selecting four coordinate points in an image, wherein a quadrangle formed by enclosing the four coordinate points comprises the target area to be processed;
(2) perspective transforming the image, mapping the image of the selected target area from a two-dimensional view plane to a three-dimensional view plane by using the following mapping formula,
wherein, in the step (A),
representing points before transformationBefore conversion𝑤A value of 1, (U, V, 1) in three-dimensional representation, mapped to three-dimensional by matrix transformation
；
Then the image is mapped again from the three-dimensional view plane to the two-dimensional view plane by using the following mapping formula,
and completing perspective transformation processing of the image.
Further, as shown in fig. 3, the specific step of adaptively selecting four coordinate points in step (1) is as follows:
A. acquiring relevant parameters of the camera, wherein the relevant parameters of the camera comprise a field range and an angle α of camera shooting, and the arrangement height h of the camera0Distance l from camera to laser line0Distance l from camera to the nearest side of cooling bed1And the width l of the laser line obtained by cutting2；
B. Calculating to obtain the horizontal straight line distance l from the camera to the nearest side of the cooling bed by using the calculation formula of the side length of the right-angled triangle through the acquired camera parameters3And the horizontal linear distance l from the camera to the laser line4；
C. Width l of laser line obtained by cutting2Obtain the coordinates w of two end points1And w2；
D. By calculating l4And l3And the coordinate point w1And w2Four coordinate points p of the target area can be obtained1，p2，p3，p4。
Furthermore, the logic control in the fifth step further comprises the step of using a convolutional neural training network and an intelligent algorithm for constructing deep learning, and performing deep learning and training on various control logic instructions generated by the cooling bed, and intelligently identifying and judging the control state of the cooling bed.
The basic principle of median filtering is: firstly, a proper square area is determined to be used as a neighborhood of a certain central pixel point, all the pixel points in the selected neighborhood are sorted according to the gray value, and the middle value of the group of sequences is taken as the gray value of the selected central pixel point to be output. Defining the intermediate value as follows, if the group of sequences contains odd number of pixel points, the gray value of the intermediate pixel point can be used as an output value; and if the group of sequences contains even number of pixel points, summing the gray values of the middle two pixel points and taking the average value as an output value. Because the window can slide, the image can be conveniently smoothed by sliding up and down, left and right.
The basic idea of median filtering can be expressed as:
wherein the content of the first and second substances,
the operation pixel point is represented and displayed,
meaning that all values in the window are arranged in a monotonic order (from large to small or small to large) and then intermediate values are taken. It assumes that the center pixel of the window is a noise point, and the pixels in the neighborhood of the center pixel are signals except the noise point, and after arranging the signal points according to the monotone sequence, the gray value of the middle point is given to the selected center pixel. However, the precondition is that a threshold needs to be set to distinguish the signal point from the noise point, and the value of the signal point is substituted for the value of the noise point, so that the median filtering operation is meaningful, and the selection of the threshold is particularly critical. Therefore, if the threshold is properly selected, the median filtering can accurately distinguish and remove noise points, and meanwhile, image detail information is protected from being lost.
Through the above description and illustration of the present invention, compared with the prior art, the present invention has the advantages that:
according to the scheme, the control of the cooling bed is realized by means of machine vision, the equipment can sense and dynamically sense the information of the cooling bed area, can feed back to a logic part according to a sensing result to give a command and correct, and finally, by constructing the learning of a deep learning machine neural network and an intelligent algorithm, the learning of various control data generated by the cooling bed is realized, and the self-state judgment and control of the cooling bed are realized. The scheme realizes high-precision control and rapid logic processing control of equipment, realizes intelligent operation of a cooling bed control technology, promotes intelligent development of the steel industry process, and has a remarkable effect of improving the whole process production efficiency of the steel industry.
The above is only the concrete implementation of the present invention, but the design concept of the present invention is not limited to this, and all the design concepts are to be utilized to improve the present invention insubstantially, and all the behaviors belonging to the infringement of the protection scope of the present invention should be considered.
Claims (4)
1. The utility model provides a cold bed intelligence control system based on machine vision which characterized in that: the system comprises a cooling bed, an image acquisition device, a control host, a communication server and monitoring display equipment, wherein the cooling bed is controlled and executed by a PLC (programmable logic controller), and the image acquisition device is arranged above the cooling bed in a matching manner and is used for acquiring image information of a cooling bed area; the control host is respectively connected with the image acquisition device and the monitoring display equipment phase control, and is connected with the cooling bed phase control through the communication server; the communication server is also connected with the monitoring display equipment.
2. The machine vision-based cooling bed intelligent control system of claim 1, characterized in that: the image acquisition device comprises industrial cameras with high resolution, laser transmitters and trigger signal controllers, wherein the industrial cameras are arranged above two sides of the cooling bed respectively and symmetrically, the laser transmitters are used for assisting in positioning the position of the steel plate, the trigger signal controllers are respectively connected with the industrial cameras and the control host, and the trigger signal controllers are activated by image signals of the industrial cameras and then are connected with the control host.
3. The machine vision-based cooling bed intelligent control system of claim 1, characterized in that: the control host comprises an image processing module and a logic processing module, the image processing module is respectively connected with the image acquisition device and the logic processing module, and the logic processing module is connected with the cooling bed phase through the communication server in a control mode.
4. The machine vision-based cooling bed intelligent control system of claim 3, characterized in that: the logic processing module is also constructed with a deep learning convolution nerve training network and an intelligent algorithm, and is used for deep learning and training various control logic instructions generated by the cooling bed and intelligently recognizing and judging the control state of the cooling bed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201921422712.7U CN210386980U (en) | 2019-08-29 | 2019-08-29 | Machine vision-based intelligent cooling bed control system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201921422712.7U CN210386980U (en) | 2019-08-29 | 2019-08-29 | Machine vision-based intelligent cooling bed control system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN210386980U true CN210386980U (en) | 2020-04-24 |
Family
ID=70340952
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201921422712.7U Active CN210386980U (en) | 2019-08-29 | 2019-08-29 | Machine vision-based intelligent cooling bed control system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN210386980U (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110404985A (en) * | 2019-08-29 | 2019-11-05 | 福建三钢闽光股份有限公司 | A kind of cold bed intelligence control system and its application method based on machine vision |
| CN113521683A (en) * | 2021-08-27 | 2021-10-22 | 吉林师范大学 | Intelligent physical ability comprehensive training control system |
| CN115301747A (en) * | 2022-09-28 | 2022-11-08 | 江苏精益智控科技有限公司 | Cooling bed area material tracking system and method based on multi-target identification and tracking |
-
2019
- 2019-08-29 CN CN201921422712.7U patent/CN210386980U/en active Active
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110404985A (en) * | 2019-08-29 | 2019-11-05 | 福建三钢闽光股份有限公司 | A kind of cold bed intelligence control system and its application method based on machine vision |
| CN113521683A (en) * | 2021-08-27 | 2021-10-22 | 吉林师范大学 | Intelligent physical ability comprehensive training control system |
| CN113521683B (en) * | 2021-08-27 | 2022-08-05 | 吉林师范大学 | Intelligent physical ability comprehensive training control system |
| CN115301747A (en) * | 2022-09-28 | 2022-11-08 | 江苏精益智控科技有限公司 | Cooling bed area material tracking system and method based on multi-target identification and tracking |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN210386980U (en) | Machine vision-based intelligent cooling bed control system | |
| CN106041937B (en) | A kind of control method of the manipulator crawl control system based on binocular stereo vision | |
| TWI394087B (en) | Method and apparatus for tracking target object | |
| CN104361314B (en) | Based on infrared and transformer localization method and device of visual image fusion | |
| CN106886216B (en) | Robot automatic tracking method and system based on RGBD face detection | |
| CN108335331B (en) | Binocular vision positioning method and equipment for steel coil | |
| WO2016055031A1 (en) | Straight line detection and image processing method and relevant device | |
| CN110509273B (en) | Robot manipulator detection and grabbing method based on visual deep learning features | |
| CN109159113B (en) | Robot operation method based on visual reasoning | |
| CN114241298A (en) | Tower crane environment target detection method and system based on laser radar and image fusion | |
| CN103646249A (en) | Greenhouse intelligent mobile robot vision navigation path identification method | |
| CN104616002A (en) | Facial recognition equipment used for judging age groups | |
| CN112906797A (en) | Plane grabbing detection method based on computer vision and deep learning | |
| TW201516969A (en) | Visual object tracking method | |
| CN113822810A (en) | Method for positioning workpiece in three-dimensional space based on machine vision | |
| CN114092478B (en) | Anomaly detection method | |
| CN110348344B (en) | Special facial expression recognition method based on two-dimensional and three-dimensional fusion | |
| CN115541030A (en) | Method and device for identifying temperature distribution of blast furnace top charge level and storage medium | |
| CN113034526B (en) | Grabbing method, grabbing device and robot | |
| CN104992448A (en) | Automatic positioning method for robot picking grapes in a damage-free way | |
| CN109313708B (en) | Image matching method and vision system | |
| WO2020037466A1 (en) | Image matching method and visual system | |
| CN111369497B (en) | Walking type tree fruit continuous counting method and device | |
| CN102303314B (en) | Carbon bowl center and guide groove positioning device in industrial production and positioning method thereof | |
| CN116740036A (en) | Method and system for detecting cutting point position of steel pipe end arc striking and extinguishing plate |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GR01 | Patent grant | ||
| GR01 | Patent grant |