KR19990050927A - How to check the center of coke oven extruder - Google Patents

Abstract

The present invention relates to a method for identifying the center of gravity of an extruder, a system for accurately confirming that the center of the coke oven and the extrusion ram of the extruder are located at the center of the coke oven when stopped in front of the coke oven to extrude the dried coke into the extruder. It is a method of checking the center of a coke extruder for the purpose of preventing accidents, such as furnace damage, by moving the extruder driving motor to a desired position even in a noisy environment by using an image recognition device.

Description

How to check the center of coke oven extruder

The present invention is to determine whether the center of the coke oven and the center of the extruder ram exactly coincide with each other when trying to extrude the coal charged in the coke oven mobile equipment after 16 hours of solidifying the extruder ram if Control and calibrate as much distance as you want by moving the extruder drive motor to prevent large accidents caused by furnace damage or mismatch of the center of the coke extruder. It is about.

Coke furnace operation is a process of manufacturing coke, an gray-white solid compound, by charging coking coal into a sealed coke oven, blocking the air, and then drying it at a high temperature. The main process is charging by using a charging vehicle and an extruder. It consists of extrusion work, guide work using guide cars, and fire fighting work using tanks. The coke oven, which is the main equipment of the coke manufacturing process, has an alternating structure of a carbonization chamber for coal drying and a combustion chamber for supplying heat sources for the drying. The mobile equipment includes a charging car for charging coal in a carbonization chamber, an extruder for performing an extrusion operation, a guide car for safely inducing coke to a fire truck when extruding, and a fire hydrant for extinguishing extruded high temperature coke. Consists of.

As a method for controlling the center of gravity of a conventional extruder, as shown in FIG. 1, a distance measuring encoder 66 installed in the extruder driving motor 2 is used to calculate a traveling distance and stop the oven in a corresponding oven. . Since the extruder has a heavy weight of 400 tons or more, the encoder 66 has a backslash of the extruder itself generated during the stop of the extruder driven on the rail, and the slip between the rail and the extruder wheel is generated, so that the correct position Stop is difficult. As a device for correcting this, a center mark is controlled by using an infrared oven number recognition device that recognizes the position of the oven by installing a mark for each oven in the furnace and reading the mark by the mark lead device of the extruder. However, this method does not cope with the occurrence of errors due to the deformation of the furnace body such as shrinkage and expansion in the oven because all the devices are installed outside of the oven has a lot of difficulties in controlling the center error.

In the patent application of "Central error control method of coke oven carbonization chamber" (patent application 96-72565) filed by Pohang Iron & Steel Co., Ltd., the center error control was performed by using an image recognition device. However, when there are many changes in the surrounding environment, such as midnight or when the lighting changes are severe, unstable measurement occurs and there are many difficulties in detecting the boundary line, resulting in poor control or abnormal data.

Accordingly, the present inventors have conducted research and experiments to improve the disadvantages of the conventional method described above, and proposed the present invention based on the results. The present invention recognizes an image using an image recognition apparatus and uses the image to identify the boundary line of the present invention. The purpose of the present invention is to provide a method for identifying the center of the extruder with coke oven that eliminates the center error by detecting the boundary line using the detection algorithm and then moving the position of the extruder according to the center error value for driving the extruder driving motor with the information. There is this.

1 is a block diagram of a conventional extruder center center check device,

2 is a block diagram of an extruder positive center check device according to the present invention,

3 is a data flow diagram of the present invention;

4 is a flowchart of program execution of the present invention;

5 is a raw image and image processing result (weekly) using the present invention,

6 is a profile (weekly) for gray values in the present invention,

7 is a histogram (weekly) for gray values in the present invention,

8 is a result of the original image and image processing using the present invention (night),

9 is a profile (night) for gray values in the present invention,

10 is a histogram (night) for gray values in the present invention.

<Description of the symbols for the main parts of the drawings>

1: Infrared oven number recognition device 2: Extruder driving motor

3: program logic controller 4: computer

5: monitoring screen 6: oven door

7: CCD camera 8: image processing board

9: Image Processing System 10: Communication Interface

50: start signal 51: start signal determination unit

52: image recognition unit 53: boundary detection unit

54: center error calculation unit 55: external storage device

56: program execution file 57: initialization

58: left camera 59: right camera

60: camera adjustment and recognition unit 61: image processing unit

62: right image calculator 63: left image calculator

64: file storage section 65: control start section

66: Encode

In order to achieve the above object, the present invention performs image processing only on an image of a rectangular shape among images received from a CCD camera, and detects one boundary line in each area after two image processing processes. After receiving the image with two channels in 640 × 480 pixel size, it is temporarily stored in the buffer to apply the proposed image processing algorithm, then processed and stored in the form of a file on the hard disk. Expression Performing average filtering using an average filtering mask such as A line Sobel operation mask is used to detect only vertical edges using a line Sobel mask such as the following. After storing images captured by the left and right cameras, the edge detection algorithm is proposed. It detects the edge of the image of the camera, and the result is displayed on the screen so that the user can easily check the result on the screen, and the error in the program logic controller to control the extruder-driven motor It is characterized in that it comprises the steps of giving information and using the control, and the step of performing control by identifying the center of gravity of the extruder with coke oven.

The system adopting the method of the present invention is a system for correcting the center of gravity by acquiring the left and right images through two cameras 7, calculating the center of gravity error using an edge detection algorithm, and then calculating the positional deviation. Same as 2.

The configuration of the present invention is as follows. The system is divided into three parts: first, an image processing board 8 for acquiring an image captured by the CCD camera 7; Secondly, an image processing system 9 and a computer 4 for obtaining an edge by using the proposed algorithm and expressing the result in an image configuration suitable for a user; Finally, it consists of a communication interface 10 showing the operator the edge detection results obtained here.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings. The image acquisition unit that receives the images from the two CCD cameras 7 installed in the front of the extruder, reduces the amount of computation in the image and takes the rectangular image out of the images received from the CCD cameras 7 in order to consider only the important area for edge detection. Only image processing was performed. In order to shorten the computation time, two areas having a size of 200 × 100 are selected for the rectangular area, and one boundary line is detected for each area after the image processing process is completed. The image grabber frame grabber is a DT3851-4 board, which accepts images in two channels by 640 × 480 pixels through two CCD cameras 7 and temporarily stores them in a buffer to apply the proposed image processing algorithm. After saving the image, the image is processed and saved in the form of a file on the hard disk.

Over time, the measurement environment gets worse, so the CCD image contains a lot of noise. Especially in the case of night measurement, the effect of noise is more serious than daytime due to the effect of lighting. In order to reduce the influence of noise, average filtering using an average filtering mask such as Equation 1 was performed on a rectangular image.

[Equation 1]

In addition, the following line Sobel mask was used to improve the accuracy of the input image. The boundary between the extruder ram and the box stay is a nearly vertical component edge. Therefore, in the present invention, a line Sobel operation mask is used to detect only edges of vertical parts. A mask as in Equation 2 calculates the amount of change on the left and right about the reference position. In the case of the boundary part, the amount of change between the two areas between the extruder ram and the box stay is calculated so that the value is larger than the surrounding area.

[Equation 2]

After the above process, the edge of the image is calculated by the proposed algorithm. Edge is the most basic of the measurement of the image. It provides useful information about the boundary of the object and reduces the amount of data to be processed. It is widely used when interpreting or recognizing images. Indeed, in the automation of classification and recognition of image patterns, the performance is greatly influenced by the validity of this edge detection result.

Therefore, in the present invention, an edge detection method using a first order differential operator having a relatively stable characteristic against high frequency image noise is used. Edge detection method using first-order differential operator is simple to use and relatively strong in high frequency noise and practically used, but it requires thresholding on the gradation edge image to clarify the edge part. . In the present invention, since the system is exposed to the outside, it is greatly affected by the external environment change. Therefore, many problems occurred in the selection of the threshold value and the thresholding process was not performed. However, the average filtering is performed on random noises to remove noises first, the edges are clearly generated after the Sobel operation, and the line with the maximum slope is selected as the edges through the first derivative. Is almost eliminated.

Using the above method, first, the images captured by the left and right cameras are stored, and the edge of each image is detected using the proposed edge detection algorithm.

The result is a program that allows the user to see the result on the screen. Also, to control the extruder driving motor, error information was given to the program logic controller to be used for the control.

Therefore, the software for positive center confirmation control of the present invention will be described in detail with reference to FIGS. 3 and 4. Implemented a Windows program based on Windows 95. The program consists of two parts. One is program initialization part that inputs extruder ram, oven box stay and image to left and right cameras, and the other is boundary detection which finds boundary by receiving images continuously from camera in real time under initialized environment. Part. First, the initialization part will be described. (1) The reference screen will be determined. When the initial program is run, the image is input from the left and right cameras. The camera image input detects the point of time when the extruder stops at the front of the oven through PLC communication and receives the left and right images through the CCD. The focus of the image is configured to be automatically adjusted.

Next, after the image recognition is completed in the initialization operation, a program that calculates the center-of-center correction error is operated. This process does not appear on the screen, but the program proceeds behind the MMI screen.

Referring to the result of installing the center of gravity check system developed in the present invention installed in the extruder No. 2 in the Gwangyang 2 coke factory, the method is operated in the cab by measuring the center error when the extruder is stopped at the target oven location It is a method of guiding the operator to the central state through the human-machine interface (MMI) screen and storing the central error data so that it can be managed for each oven. The purpose of providing the center error to the operator is to enable the operator to verify the difference between the center error measured in the positive centering system and the actual center error that can be measured during extrusion. The purpose of managing the center error data for each oven is to remove the center error that is continuously generated by changing the position of the infrared sensor installed in the lower part of the oven with respect to the oven where the center error is continuously constant.

The figures shown below show day / night results as a result of image processing for center error measurement from the original image obtained from the camera.

FIG. 5 is a data measured during the day and shows a result of detecting edges using edge detection algorithm after removing noise by applying average filtering to the original image and the original image obtained from the first camera around 1 pm. . 6 shows a profile when the average filtering is performed on the original image, and shows that a lot of noise components are removed as a result of the average filtering. The gray level value changes most around the 300th pixel and 450th pixel, which are boundary detection areas.

FIG. 7 shows a histogram that shows the distribution of gray values, and the shape of the average filtering resulted in a sharp change in shape.

8, 9 and 10 show the measurement results at 9 pm as data measured at night. In the case of nighttime, the original image shows a lot of noise, but the average filtering shows that the noise is almost eliminated. In the case of edge detection, the detection is as accurate as daytime.

From the above results, it was confirmed from the measured data that the image acquisition and the boundary line detection are accurately performed during the day and night. Therefore, the center error measured based on the detected boundary line is also measured accurately.

As described above, according to the present invention, by measuring the center error by using the image recognition, it is possible to measure the center error caused by the deformation of the furnace body, it is possible to remove the center error by moving the position of the extruder according to the measurement result There is an advantage that the cork extrusion can be performed in the state where the center of the ram and the carbonization chamber of the extruder coincide.

Claims (1)

Image processing is performed only on the rectangular image among the images received from the CCD camera. After the image processing process is completed, one boundary line is detected in each area, and the image is displayed in two channels through two CCD cameras. In order to apply the proposed image processing algorithm after accepting the size, it is temporarily stored in the buffer, processed and stored in the form of a file on the hard disk. Performing average filtering using an average filtering mask such as A line Sobel operation mask is used to detect only vertical edges using a line Sobel mask such as the following. After storing images captured by the left and right cameras, the edge detection algorithm is proposed. It detects the edge of the image of the camera, and the result is displayed on the screen so that the user can easily check the result on the screen, and the error in the program logic controller to control the extruder-driven motor A method of confirming the center of a coke oven extruder, comprising the steps of using the control to give information, and confirming the center of gravity of the coke oven extruder.