JPH09182953A - Method for detecting slag - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a slag detecting method which can prevent the mixture of slag by quickly and surely detecting the slag flowing in the discharge stream of molten metal. SOLUTION: In the detecting method, luminance of the discharge stream 2 of the molten metal is observed with a CCD camera 3 and a picture processing is applied to the obtd. luminance distribution picture, and a threshold value for dividing the molten metal and the slag is calculated. After executing a binarization processing by the setting threshold value, a laveling processing is executed to a connecting position having the luminance of the threshold value or higher and a specific number of picture elements or higher. When the calculated value of the area (the number of picture elements) of the slag part becomes the discriminated value or higher, it is discriminated as the starting time of the flow-out of the slag.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a slag detecting method for promptly and accurately detecting a discharge flow of molten metal, for example, slag flowing into molten steel discharged from a ladle during steel production.

[0002]

2. Description of the Related Art Steel is manufactured by melting raw metal such as scrap at a high temperature in an electric furnace or the like, but during melting, impurities in the raw metal and added processing agents become slag and float on the molten steel. Come on. If this slag mixes in the molten steel, the quality of the steel deteriorates, so it is extremely important to detect the slag that flows into the molten steel quickly and accurately to prevent the slag from mixing in order to maintain the quality of the steel. .

Conventionally, the detection of mixed slag has been performed by a method of visual observation by a human. However, in this method, it is difficult to perform objective management because of a large factor depending on human skill and intuition. For this reason, a method for detecting slag using a difference in physical properties between molten steel and slag, for example, a difference in viscosity or electrical conductivity has been developed, but a sufficient result in terms of accuracy has not been obtained.

A method for optically detecting slag by utilizing the difference in brightness between molten steel and slag has also been developed. For example, Japanese Patent Publication No. 59-263383 discloses a method for injecting molten metal into a container for molten metal. The molten metal injection flow or the surface of the molten metal in the container is imaged by an optical camera, and the amount of slag mixed in the molten metal in the container is detected based on the brightness difference between the molten metal and the slag in the image pickup signal of the camera. A method for detecting the inclusion of slag has been proposed. This method divides the imaging screen into a slag screen and a molten steel screen according to the boundary brightness between the slag and the molten steel, and calculates the slag area ratio [= (the number of slag screens) / (the number of slag screens + the number of molten steel screens)] in the molten steel. Then
The slag mixing is determined when this value becomes larger than a preset determination value.

However, in the calculation of the slag area ratio, all the screens having a larger luminance than the boundary brightness are regarded as slag screens, and the number is counted. There is a problem in that it is calculated as a screen, and slag is erroneously recognized and slag mixing is erroneously determined. Furthermore, although the boundary brightness between the slag and the molten steel, which is divided into the slag screen and the molten steel screen, that is, the setting of the threshold value is important for calculating the slag area ratio, the method of setting the boundary brightness is unknown, and it is detected. There is a defect that the accuracy is poor.

Further, as a method for optically detecting slag by utilizing the difference in brightness between molten steel and slag, Japanese Patent Laid-Open No. 2-2 is known.
In JP-A-51362, the widthwise radiant energy distribution of molten metal is measured, and the continuous maximum width portion of the measurement result is detected as the diameter of the molten metal flow, and the width of the diameter of the molten metal flow and its integrated value are measured over time. A slag outflow detection method has been proposed in which the slag outflow is determined when the increase in the two values is detected and the slag is outflowed. However, since the measurement area to be measured is only one line in the lateral direction of the width direction of the molten metal flow, if the molten metal outlet fluctuates up and down, it is moved out of the monitoring area and the position is manually set each time. However, since the threshold value must be set for each target steel grade because the brightness changes depending on the steel grade because binarization processing is performed with a constant threshold, detection accuracy is required. There is also a drawback of lacking stability.

Further, in Japanese Unexamined Patent Publication No. 7-260696, a process in which the surface of a tapped steel stream flowing from a converter into a ladle is photographed by a television camera or a CCD camera and the image is output to an image processing analysis device at a constant cycle. Then, in the image processing analysis device, every time there is an input image in the set monitoring area in the image, a gray level corresponding to the brightness is obtained for each pixel in the area, and then an average gray level in the monitoring area is calculated. A slag outflow detection method consisting of a process and a process of performing a comparison calculation with the moving average value up to the previous time each time the average gray level is obtained, and determining "slag outflow" when the difference exceeds the set value It is disclosed. However, this method calculates the gray level according to the brightness for each pixel of the input image in a fixed cycle, calculates the average gray level in the monitoring area, and calculates the average gray level every time when the average gray level is calculated. Since it calculates the difference, it means that changes in the average gray level, that is, the average brightness level, are compared.For example, when a small amount of slag is mixed, the change in the average brightness level is small and difficult to detect, and the timing for detecting slag discharge Is delayed.

[0008]

DISCLOSURE OF THE INVENTION In order to solve the above problems, the present inventors have earnestly studied a method for accurately and quickly detecting slag flowing into an exhaust flow of molten metal such as molten steel, and as a result, Utilizing the fact that the brightness of molten metal and that of slag are significantly different, a histogram of the brightness signal of molten metal is created, and the threshold value for separating molten metal and slag is automatically set, and then binarized to obtain a threshold value. Only the connected components with a certain number of pixels or more are labeled from the luminance image of the slag that is more than the value, and the time when the calculated value of the area (the number of pixels) of the slag portion sharply increases is the start time of the slag inflow. It was found that the slag can be accurately detected by making a determination.

The present invention has been completed based on this finding, and its object is a slag detecting method capable of accurately and quickly detecting the slag flowing into the discharge stream of molten metal and preventing the inclusion of the slag. To provide.

[0010]

The slag detection method according to the present invention for achieving the above object is to measure the brightness of a discharge stream of molten metal with a CCD camera and perform image processing on the brightness distribution chart obtained to melt the brightness. After calculating the threshold value for dividing into metal and slag and performing binarization processing with the set threshold value, labeling processing is performed on the connected component having the brightness equal to or higher than the threshold value and the specific pixel number or more, The structural feature is that the time when the calculated value of the area (pixel number) of the slag portion is equal to or larger than the determination value is determined as the slag inflow start time.

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, the slag detecting method of the present invention will be described in detail by taking molten steel as an example of molten metal. First, the discharge flow of molten steel is observed with a CCD camera, and the brightness signal is input to an arithmetic unit equipped with a histogram counter, and the brightness distribution of brightness levels 0 to 255 is obtained by taking the brightness on the horizontal axis and the frequency on the vertical axis. Create a diagram and use the discriminant analysis method to set brightness levels 0-2
A threshold value for separating molten steel and slag is set within a range of 50. Note that the brightness level is set in the range of 0 to 250 because
This is because an error occurs due to an abnormally high temperature portion of the molten steel, for example, an abnormally high temperature portion due to oxygen blowing, that is, an abnormally high luminance portion when the high luminance portion of the luminance levels 251 to 255 is added to the calculation range.

This threshold is set by the following process. From the luminance distribution diagram of the molten steel discharge flow shown in FIG. 2 and the luminance distribution diagram when only the slag flows out shown in FIG. 3, the threshold value α is first divided into the background luminance distribution and the molten steel luminance distribution using the discriminant analysis method. Ask for. The brightness level i as the range of 0 to 250, the frequency of the n _i luminance level i, the total power is N, the random variable P _i of the luminance level is represented by the following formula (1), P _i = n _i / N (1) The total average level μ _T of the image is expressed by equation (2).

The 0th-order average value ω (i) and the 1st-order average value μ (i) in the luminance distribution up to the luminance level i can be expressed by the following equations (3) and (4), respectively: ω (i) = ω (i−1) + P _i ... (3) μ (i) = μ (i−1) + i × P _i ... (4) When the image is divided into two classes according to the brightness level,
The variance between each class is expressed by the following equation (5). σ _B ² = [μ _T × ω (i) −μ (i)] ² / [ω (i) × [1−ω (i)]]… (5)

Therefore, the optimum threshold value, that is, the threshold value α for dividing the brightness distribution of the background and the molten steel can be obtained by the following equation (6). i ^* = σ _B ² (i ^* ) = max σ _B ² (i) (6)

Using this threshold value α, a threshold value β for separating molten steel and slag is set by the following procedure. 1. The frequency at which the brightness level is the threshold value α is the threshold value α to 2
Subtract from each frequency in the range of 50. 2. The luminance level is the threshold value α to threshold value α + (temporary average value−threshold value α) × 2, with the luminance level having the highest frequency in the threshold value α to 250 range as the temporary average value. Calculate the average value of molten steel in the range. 3. Considering the brightness distribution of molten steel as a normal distribution, the brightness level is threshold α to threshold α + (temporary average value-threshold α)
The standard deviation of the brightness distribution of the molten steel is calculated within the range of x2. 4. To obtain the brightness distribution range of the molten steel, the average value + standard deviation × 2 (2σ) is calculated, and the threshold β for separating molten steel and slag is calculated. The average value + standard deviation x 3
The threshold value β may be calculated by calculating (3σ).

According to the present invention, the brightness distribution chart is divided into molten steel and slag by using the threshold value β calculated in this way and automatically set, and binarized. The area (the number of pixels) of the slag portion is calculated by performing the labeling process as the slag portion. Then, by monitoring the change in the area (the number of pixels), it is determined that the time when the value becomes larger than a preset value is the time when the inflow of slag starts. In this case, disturbances such as splash having a small number of pixels can be removed from the calculation target by excluding those having a fixed number of pixels or less, for example, 50 pixels or less, from the calculation target and performing labeling processing. The accuracy can be improved.

After the tapping start signal is input in this operation,
Repeatedly and continuously performed until the slag is detected, and when the slag is detected, the alarm device is immediately activated to prevent the slag from being mixed. The processing speed of one cycle is 1
It is about 50 milliseconds. As described above, according to the present invention, the brightness of the discharge flow of molten steel is constantly observed by the CCD camera, and the change in the area (number of pixels) of the slag portion is monitored. Since it is detected as the inflow start time, it becomes possible to detect the slag flowing into the molten steel quickly and accurately.

[0018]

EXAMPLES The present invention will be described in detail below with reference to molten steel discharged from a ladle when producing steel as molten metal, but the slag detection method of the present invention is not limited to molten steel as molten metal. .

FIG. 4 is a block diagram illustrating a device applied to the slag detection method of the present invention, in which the CCD camera 3 constantly monitors the luminance signal of the molten steel flow 2 discharged from the ladle 1. The output signal from the CCD camera 3 is input to the image processing board 6 of the CPU 5 via the camera controller 4,
The luminance signal observed by the CCD camera 3 is displayed as an image on the video monitor 7. An image of the luminance signal input to the image processing board 6 is captured in a frame memory, and then arithmetic processing is performed to set a threshold value for dividing molten steel and slag and perform binarization processing. Next, from the binarized image, for example, only the connected component having 50 or more pixels is labeled to calculate the area (pixel number) of the slag portion, and if this value is equal to or greater than the determination value, slag inflow is detected. When it is determined that the signal is output, a signal is output from the output board 8 to the alarm device 9 and the PC monitor 10.

Next, the slag detection method of the present invention will be described based on the flow chart shown in FIG. First, after setting the measurement area of the CCD camera for observing the brightness of the discharge flow of molten steel, input the observed brightness signal to the image processing board of the CPU and capture the image in the frame memory,
The brightness distribution charts shown in FIGS. 2 and 3 are created by performing a calculation process using the histogram counter. Based on this luminance distribution chart, the threshold value β for dividing molten steel and slag is calculated and set by the method described above.

Image processing is performed using black as the molten steel and white as the slag according to the set threshold value β, and then binarization processing is performed, and the area of the slag is obtained from the portion displayed as white on the screen above the threshold value β. . Next, the area (pixel number) of this slag is subjected to labeling processing to calculate the area (pixel number) of the slag portion. In this case, disturbances such as splash can be eliminated by excluding those having a small number of pixels, for example, those having 50 pixels or less from the calculation target.

The area of the slag portion calculated in this way is compared with a judgment value which is set in advance as a criterion for judging slag detection, and when it becomes equal to or larger than the judgment value, it is judged as the slag inflow start time. The above operation is continuously repeated until the slag inflow is detected while the tapping start signal is input, and when the slag inflow is detected, an alarm signal is output to the alarm device to prevent slag mixture. Measures are taken. In addition, a series of these operations is processed at a speed of about 150 milliseconds per cycle.

[0023]

As described above, according to the slag detection method of the present invention, the brightness of the discharged flow of molten metal is observed by the CCD camera, and the created brightness distribution map is image-processed to separate it into molten metal and slag. The area (number of pixels) of the slag portion is calculated by calculating and setting a threshold value, binarizing the threshold value, and labeling only connected components having a brightness equal to or higher than the threshold value and having a specific number of pixels or more. Since it is determined that the slag inflow start time is calculated and this value becomes equal to or greater than the predetermined determination value, the slag inflow start time can be detected quickly and accurately. Further, the labeling process can eliminate a disturbance such as a splash having a small number of pixels, can improve the accuracy of slag detection, and effectively prevent deterioration of quality due to mixing of slag.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a processing procedure of a slag detection method of the present invention.

FIG. 2 is a luminance distribution diagram of a molten steel discharge flow without slag inflow.

FIG. 3 is a luminance distribution diagram when only slag flows out.

FIG. 4 is a block diagram illustrating an apparatus applied to the slag detection method of the present invention.

[Explanation of symbols]

 1 Ladle 2 Molten Steel Flow 3 CCD Camera 4 Camera Controller 5 CPU 6 Image Processing Board 7 Video Monitor 8 Output Board 9 Alarm Device 10 PC Monitor

Claims (1)

[Claims] 1. The brightness of the discharge flow of the molten metal is observed by a CCD camera, the brightness distribution chart obtained is image-processed to calculate a threshold value for dividing the molten metal into slag, and the binary value is set according to the set threshold value. After the conversion processing, the connected component that has a brightness equal to or higher than the threshold value and has a specific pixel number or more is labeled, and the value of the calculated slag area (pixel number) becomes the determination value or more. A slag detection method characterized in that time is determined to be a slag inflow start time.