A converter primary dust removal chimney smoke emission image recognition system and method

Abstract

This invention relates to an image recognition system and method for smoke emission from a converter primary dust removal chimney, belonging to the field of dust removal technology in the metallurgical industry. The technical solution is as follows: a camera (1) collects images of the dust removal chimney and transmits them to a recognition computer (2). The recognition computer (2) reads the converter blowing signal from the dust removal PLC (3) and determines whether it is in the blowing state. If it is in the blowing state, the recognition computer (2) reads the image from the camera (1), performs image analysis, calculates the smoke emission index, and transmits the alarm result to the dust removal PLC (3). An alarm is triggered on the primary monitoring screen, and the dust removal PLC (3) drives the audible and visual alarm device (4) to sound the alarm. The positive effects of this invention are: the invention has low investment costs. The manual judgment of chimney smoke emission operation by the on-site observation and monitoring screen is replaced by an image recognition system, which reduces the intensity of manual labor. At the same time, it realizes full-process online detection, eliminates the problem of omissions in manual observation, and improves the level of intelligent control.

Description

Technical Field

[0001] This invention relates to an image recognition system and method for smoke emission from a converter primary dust removal chimney, belonging to the field of dust removal technology in the metallurgical industry. Background Technology

[0002] In existing steel mill converters, a single dust removal system uses four electrostatic precipitators. Fields one and two handle 80% of the dust removal, while fields three and four handle the remaining 20%. When faults occur, such as electrode short circuits or transformer insulator damage, dust removal efficiency decreases, and smoke is easily emitted. Personnel must monitor the video feed from the dust removal chimney in real time; failure to detect smoke promptly can lead to environmental accidents. Summary of the Invention

[0003] The purpose of this invention is to provide an image recognition system and method for smoke emission from the primary dust removal chimney of a converter. This system is a post-event early warning device. Upon detecting smoke, it displays an alarm image on the dust removal operation screen and triggers an audible and visual alarm. Upon receiving the alarm, the dust removal operator immediately picks up the dust removal nozzle to prevent environmental accidents and immediately inspects the equipment and process operations. Once the problem is identified, it is addressed or rectified, resolving the aforementioned issues in the background technology.

[0004] The technical solution of this invention is:

[0005] A converter primary dust removal chimney smoke emission image recognition system includes a camera, a recognition computer, a dust removal PLC, and an audible and visual alarm device. The camera captures images of the dust removal chimney and transmits them to the recognition computer. The recognition computer reads the converter blowing signal from the dust removal PLC and determines whether it is in the blowing state. If it is in the blowing state, the recognition computer reads the image from the camera, performs image analysis, calculates the smoke emission index, and determines it as low, medium, or high smoke emission according to a set standard. The alarm result is transmitted to the dust removal PLC, which then alarms on the primary monitoring screen and activates the audible and visual alarm device.

[0006] Furthermore, the recognition computer performs image analysis on the dust removal chimney images captured by the camera, which is divided into two steps: chimney opening image recognition and smoke image recognition. Only after the chimney opening image recognition is successful will the smoke image recognition program begin.

[0007] Furthermore, the recognition program of the computer first locates the chimney in the captured image of the dust removal chimney. It sets the pixel coordinate range of the dust removal chimney in the captured image as (x1, h1), (x1, h2), (x2, h1), and (x2, h2). It scans line by line in the vertical H direction and calculates the average brightness of two adjacent windows in the horizontal X direction. Gradient judgment is performed: a decrease in brightness gradient indicates the left edge of the dust removal chimney, and an increase in brightness gradient indicates the right edge. The overall success rate of the dust removal chimney recognition in the H direction is calculated, yielding the center pixel position of the chimney opening and the chimney width Hd. The coordinates of the center pixel position of the chimney opening are (h0, x0). If the overall success rate is lower than the set standard value, the recognition program outputs information indicating that the recognition conditions are not met, and the program cannot proceed to the smoke image recognition process.

[0008] Further, after successful image recognition of the chimney opening, a smoke recognition program is performed. Using the coordinates (h0, x0) of the center pixel position of the chimney opening as the center, a scanning path is initiated with a radius equal to the chimney width Hd, incrementing by one pixel each time until the set maximum smoke detection position Hh is reached. The scanning path follows an inverted U-curve, calculating the average brightness of two adjacent windows and performing gradient judgment. A decreasing brightness gradient indicates the left edge of the smoke from the dust removal chimney, while an increasing brightness gradient indicates the right edge. For the identified smoke center position, errors are eliminated by gradually diffusing smoke from the chimney opening. The overall success rate of smoke recognition in the inverted U-direction is calculated and compared with a set standard value to determine if the chimney is emitting smoke. If smoke is emitted, the difference between the total average brightness of the pixels in the emitting area and the average brightness of the entire recognition area is calculated. Values ​​≥-30, ≥-80 and <-30, and <-80 are used to determine three levels of smoke intensity: large, medium, and small.

[0009] A method for image recognition of smoke emission from a converter primary dust removal chimney, comprising the following steps:

[0010] The camera captures images of the dust removal chimney and transmits them to the recognition computer. The recognition computer reads the converter blowing signal from the dust removal PLC and determines whether it is in the blowing state. If it is in the blowing state, the recognition computer reads the image from the camera, performs image analysis, calculates the smoke index, and judges it as low, medium, or high smoke level according to the set standard. The alarm result is transmitted to the dust removal PLC, and an alarm is triggered on the primary monitoring screen. The dust removal PLC then activates the audible and visual alarm device.

[0011] Furthermore, the more specific steps are as follows:

[0012] ① The camera captures images of the dust removal chimney and transmits them to the recognition computer. The recognition computer reads the converter blowing signal from the dust removal PLC and determines whether it is in the blowing state. If it is not in the blowing state, the recognition computer will not read the camera image. If it is in the blowing state, the recognition computer will read the camera image.

[0013] ② After the computer reads the image from the camera, it analyzes the image by comparing the image pixels. This is divided into two steps: chimney image recognition and smoke image recognition. If chimney image recognition fails, the smoke image recognition program will not be entered. If chimney image recognition is successful, the smoke image recognition program will be entered.

[0014] ③ The computer's recognition program first locates the chimney in the captured image of the dust removal chimney. It sets the pixel coordinate range of the chimney in the captured image (x1, h1), (x1, h2), (x2, h1), and (x2, h2). It scans line by line in the vertical H direction and calculates the average brightness of two adjacent windows in the horizontal X direction. Gradient judgment is performed: a decrease in brightness gradient indicates the left edge of the chimney, and an increase in brightness indicates the right edge. The overall success rate of chimney recognition in the H direction is calculated, yielding the center pixel position of the chimney opening and the chimney width Hd. The coordinates of the center pixel position of the chimney opening are (h0, x0). If the overall success rate is lower than the set standard value, the recognition program outputs a message indicating that the recognition conditions are not met, and the program cannot proceed to the smoke image recognition process. If the overall success rate is higher than the set standard value, the recognition program outputs a message indicating that the recognition conditions are met, and the program proceeds to the smoke image recognition process.

[0015] ④ After successful image recognition at the chimney opening, a smoke recognition program is performed. Using the coordinates (h0, x0) of the center pixel position at the chimney opening as the center, a scanning path is initiated with a radius equal to the chimney width Hd, incrementing by one pixel each time until the set maximum smoke detection position Hh is reached. The scanning path follows an inverted U-curve. The average brightness of two adjacent windows is calculated, and gradient judgment is performed. A decreasing brightness gradient indicates the left edge of the smoke from the dust removal chimney, while an increasing brightness gradient indicates the right edge. The smoke center position is identified by gradually diffusing smoke from the chimney opening, eliminating incorrect positions. The overall success rate of smoke recognition in the inverted U-direction is calculated and compared with a set standard value to determine if the chimney is emitting smoke. If smoke is emitted, the difference between the total average brightness of the pixels in the emitting area and the average brightness of the entire recognition area is calculated. Values ​​≥-30, ≥-80 and <-30, and <-80 are used to determine three levels of smoke intensity: large, medium, and small.

[0016] The improvement of this invention lies in that the camera signal is only read and analyzed when the converter is in the blowing state. The time spent in the converter blowing state accounts for only 1 / 3 of the total time, which improves the utilization rate of the recognition computer. At the same time, when the recognition computer detects smoke, it will make an alarm record on the dust removal operation screen and save the smoke photo by time in a designated folder, which facilitates the historical query of alarm data.

[0017] The positive effects of this invention are: low investment cost, replacing manual judgment of chimney smoke emission operation by observing and monitoring the screen at the post with an image recognition system, reducing the intensity of manual labor, while realizing full online detection, eliminating the problem of omissions in manual observation, and improving the level of intelligent control. Attached Figure Description

[0018] Figure 1 This is a system diagram of an embodiment of the present invention;

[0019] Figure 2 The chimney opening pixel coordinates are shown in this embodiment of the invention.

[0020] Figure 3 The coordinates for identifying the smoke pixels in this embodiment of the invention;

[0021] Figure 4 This is a flowchart of an embodiment of the present invention;

[0022] In the picture: 1. Camera; 2. Recognition computer; 3. Dust removal PLC; 4. Audible and visual alarm device. Detailed Implementation

[0023] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0024] Combined with appendix Figure 1 and 4 This embodiment provides an image recognition system for smoke emission from a converter primary dust removal chimney, including a camera 1, a recognition computer 2, a dust removal PLC 3, and an audible and visual alarm device 4. The camera 1 captures images of the dust removal chimney and transmits them to the recognition computer 2. The recognition computer 2 reads the converter blowing signal from the dust removal PLC 3 and determines whether it is in the blowing state. If it is in the blowing state, the recognition computer 2 reads the image from the camera 1 and performs image analysis to calculate the smoke emission index (0-100%). Based on the set standard, it is judged as a small, medium, or large smoke emission level. The alarm result is transmitted to the dust removal PLC 3, and an alarm is triggered on the primary monitoring screen. The dust removal PLC 3 drives the audible and visual alarm device 4 to sound an alarm, reminding the staff to check the dust removal facilities.

[0025] The recognition computer 2 performs image analysis on the dust removal chimney image captured by the camera 1, which is divided into two steps: chimney opening image recognition and smoke image recognition. Only after the chimney opening image recognition is successful will the smoke image recognition program be entered, thereby improving the image recognition accuracy.

[0026] Combined with appendix Figure 2, the recognition program of the recognition computer 2 first locates the chimney position in the collected dust removal chimney image, captures the pictures in the camera, and artificially sets the chimney pixel coordinate ranges (x1, h1), (x1, h2), (x2, h1), (x2, h2). It scans line by line from bottom to top in the vertical H direction, calculates the average brightness of two adjacent windows from left to right in the horizontal X direction, and makes a gradient judgment. A decreasing brightness gradient indicates the left edge of the chimney, and an increasing brightness gradient indicates the right edge of the chimney. The overall recognition success rate of the chimney in the H direction is comprehensively calculated to obtain the central pixel position of the chimney mouth and the width Hd of the chimney mouth. The central pixel position of the chimney mouth is the intersection point of the vertical center line and the horizontal center line, and the intersection point coordinates are (H0, X0). If the overall recognition success rate is lower than the standard value, the system outputs information indicating that the recognition conditions are not met.

[0027] Among them, (x1, h1) = (527, 600)

[0028] (x1, h2) = (527, 660)

[0029] (x2, h1) = (673, 600)

[0030] (x2, h2) = (673, 660)

[0031] Scan pixel by pixel in the vertical H direction and scan by gradient in the horizontal X direction. The gradient value is 5.

[0032] Boundary judgment formula: Take the average gray value YC_PJ of the pixels within the chimney pixel coordinate range. Scan line by line in the vertical H1 direction, and the average brightness d1 in the horizontal X1 direction. Scan line by line in the vertical H2 direction, and the average brightness d2 in the horizontal X2 direction. According to the picture positioning, identify the chimney brightness difference YC_LDC. When d1 > YC_PJ and (d1 - d2) > YC_LDC, determine the left boundary of the chimney. When d1 < YC_PJ and (d2 - d1) > YC_LDC, determine the right boundary of the chimney. Add the boundary coordinates to the list and perform list averaging to obtain the chimney coordinates: H0 = 647, X0 = 599, Hd = 75.

[0033] Combined with the appendix Figure 3After successful image recognition at the chimney opening, a smoke image recognition program is performed. Using the coordinates (h0, x0) of the center pixel position at the chimney opening as the center, a scanning path is initiated with a radius equal to the chimney width Hd, incrementing by one pixel each time until the set maximum smoke detection position Hh is reached. The scanning path follows an inverted "U" curve, calculating the average brightness of two adjacent windows and performing gradient judgment. A decreasing brightness gradient indicates the left edge of the smoke from the dust removal chimney, while an increasing brightness gradient indicates the right edge. For the identified smoke center position, errors are eliminated by gradually diffusing smoke from the chimney opening. The overall success rate of smoke recognition in the inverted "U" direction is calculated and compared with a set standard value to determine if the chimney is emitting smoke. If smoke is emitted, the difference between the total average brightness of the pixels in the emitting area and the average brightness of the entire recognition area is calculated. Values ​​≥-30, ≥-80 and <-30, and <-80 are used to determine three levels of smoke emission: large, medium, and small.

[0034] Where Hd=75, Hh=160;

[0035] The method for determining the flue gas boundary is the same as that for determining the chimney boundary, using an inverted "U" direction for scanning coordinates. If the flue gas exceeds 50% of the total recognition area, the recognition is successful.

[0036] The smoke detection accuracy of this invention is 98%.

[0037] A method for recognizing smoke emissions from a converter primary dust removal chimney, characterized by the following steps:

[0038] ① Camera 1 captures images of the dust removal chimney and transmits them to the recognition computer 2. The recognition computer 2 reads the converter blowing signal from the dust removal PLC 3 and determines whether it is in the blowing state. If it is not in the blowing state, the recognition computer 2 will not read the image from camera 1. If it is in the blowing state, the recognition computer 2 will read the image from camera 1.

[0039] ②After the computer 2 reads the image from the camera 1, it analyzes the image by comparing the image pixels. This is divided into two steps: chimney image recognition and smoke image recognition. If the chimney image recognition fails, the smoke image recognition program will not be entered. If the chimney image recognition is successful, the smoke image recognition program will be entered.

[0040] ③ The recognition program of computer 2 first locates the chimney in the captured image of the dust removal chimney. It sets the pixel coordinate range of the dust removal chimney in the captured image as (x1, h1), (x1, h2), (x2, h1), and (x2, h2). It scans line by line in the vertical H direction and calculates the average brightness of two adjacent windows in the horizontal X direction. Gradient judgment is performed: a decrease in brightness gradient indicates the left edge of the dust removal chimney, and an increase in brightness gradient indicates the right edge. The overall success rate of the dust removal chimney recognition in the H direction is calculated, yielding the center pixel position of the chimney opening and the chimney width Hd. The coordinates of the center pixel position of the chimney opening are (h0, x0). If the overall success rate is lower than the set standard value, the recognition program outputs a message indicating that the recognition conditions are not met, and the program cannot proceed to the smoke image recognition program. If the overall success rate is higher than the set standard value, the recognition program outputs a message indicating that the recognition conditions are met, and the program proceeds to the smoke image recognition program.

[0041] ④ After successful image recognition at the chimney opening, a smoke recognition program is performed. Using the coordinates (h0, x0) of the center pixel position at the chimney opening as the center, a scanning path is initiated with a radius equal to the chimney width Hd, incrementing by one pixel each time until the highest smoke detection position Hh is set. The scanning path follows an inverted U-curve. The average brightness of two adjacent windows is calculated, and gradient judgment is performed. A decreasing brightness gradient indicates the left edge of the smoke from the dust removal chimney, while an increasing brightness gradient indicates the right edge. The smoke center position is identified by gradually diffusing smoke from the chimney opening, eliminating incorrect positions. The overall success rate of smoke recognition in the inverted U-direction is calculated and compared with a set standard value to determine if the chimney is emitting smoke. If smoke is emitted, the difference between the total average brightness of the pixels in the emitting area and the average brightness of the entire recognition area is calculated. Values ​​≥-30, ≥-80 and <-30, and <-80 are used to determine three levels of smoke intensity: large, medium, and small.

[0042] In this embodiment:

[0043] Camera: Hikvision 3T66WDV3-L6MM;

[0044] Computer specifications: Windows 10 Professional, 16GB RAM, 11th Gen Intel® Core™ i7-11700 processor @ 2.50GHz;

[0045] Dust removal PLC: S7 416;

[0046] Audible and visual alarm device: SF-503 audible and visual alarm.

Claims

1. A converter primary dust removal chimney smoke emission image recognition system, characterized in that: Includes a camera (1), an identification computer (2), a dust removal PLC (3), and an audible and visual alarm device (4). The camera (1) collects images of the dust removal chimney and transmits them to the identification computer (2). The identification computer (2) reads the converter blowing signal from the dust removal PLC (3) and determines whether it is in the blowing state. If it is in the blowing state, the identification computer (2) reads the image from the camera (1) and performs image analysis to calculate the smoke index. According to the set standard, it is judged to be in three levels: small, medium, and large smoke. The alarm result is transmitted to the dust removal PLC (3) and an alarm is triggered on the first-level monitoring screen. The dust removal PLC (3) drives the audible and visual alarm device (4) to alarm. The recognition computer (2) performs image analysis on the dust removal chimney image captured by the camera (1), which is divided into two steps: chimney opening image recognition and smoke image recognition. Only after the chimney opening image recognition is successful will the smoke image recognition program be entered. The recognition program of the computer (2) first locates the chimney in the collected image of the dust removal chimney. It sets the pixel coordinate range of the dust removal chimney in the captured image of the dust removal chimney as (x1, h1), (x1, h2), (x2, h1), (x2, h2). It scans line by line in the vertical H direction and calculates the average brightness of two adjacent windows in the horizontal X direction. It performs gradient judgment. The brightness gradient decreases to the left edge of the dust removal chimney and increases to the right edge of the dust removal chimney. It calculates the total success rate of the dust removal chimney recognition in the H direction and obtains the center pixel position of the chimney opening and the chimney width Hd. The coordinates of the center pixel position of the chimney opening are (h0, x0). If the total success rate of recognition is lower than the set standard value, the recognition program outputs the information that the recognition conditions are not met and cannot enter the smoke image recognition program. If the total success rate of recognition is higher than the set standard value, the recognition program outputs the information that the recognition conditions are met and enters the smoke image recognition program. The smoke image recognition program uses the coordinates (h0, x0) of the center pixel position of the chimney opening as the center and scans a route starting from the chimney width Hd as the radius, incrementing by one pixel each time until the set highest smoke detection position Hh is reached. The scanning path follows an inverted U-curve, calculates the average brightness of two adjacent windows, and performs gradient judgment. A decrease in brightness gradient indicates the left edge of the smoke from the dust removal chimney, and an increase in brightness gradient indicates the right edge of the smoke from the dust removal chimney. The program identifies the center position of the smoke by gradually diffusing the smoke from the chimney opening, eliminating incorrect positions, and comprehensively calculates the total success rate of smoke recognition in the inverted U-direction. The total success rate is compared with a set standard value to determine whether the chimney is emitting smoke. If smoke is emitted, the program calculates the difference between the total average brightness of the pixels in the emitting area and the average brightness of the entire recognition area. The difference corresponds to ≥-30, ≥-80 and <-30, and <-80, respectively, and outputs three levels of smoke: large, medium, and small.

2. A method for recognizing smoke emissions from a converter primary dust removal chimney, characterized in that... The following steps are taken: the camera (1) collects the image of the dust removal chimney and transmits it to the recognition computer (2). The recognition computer (2) reads the converter blowing signal of the dust removal PLC (3) and determines whether it is in the blowing state. If it is in the blowing state, the recognition computer (2) reads the image of the camera (1) and performs image analysis to calculate the smoke index. According to the set standard, it is judged to be a small, medium and large smoke level. The alarm result is transmitted to the dust removal PLC (3) and alarms are set on the first-level monitoring screen. The dust removal PLC (3) drives the sound and light alarm device (4) to alarm. The specific steps are as follows: ① The camera (1) collects images of the dust removal chimney and transmits them to the recognition computer (2). The recognition computer (2) reads the converter blowing signal of the dust removal PLC (3) and determines whether it is in the blowing state. If it is not in the blowing state, the recognition computer (2) will not read the image of the camera (1). If it is in the blowing state, the recognition computer (2) reads the image of the camera (1). ②After the computer (2) reads the image from the camera (1), it analyzes the image by comparing the image pixels. The process is divided into two steps: chimney image recognition and smoke image recognition. If the chimney image recognition fails, the smoke image recognition program will not be entered. If the chimney image recognition is successful, the smoke image recognition program will be entered. ③ The recognition program of the computer (2) first finds the chimney position in the collected dust removal chimney image. In the captured dust removal chimney image, the pixel coordinate range of the dust removal chimney is set (x1, h1), (x1, h2), (x2, h1), (x2, h2). The vertical H direction is used for line-by-line scanning. The average brightness of two adjacent windows is calculated along the horizontal X direction. Gradient judgment is performed. The brightness gradient decreases to the left edge of the dust removal chimney, and the brightness gradient increases to the right edge of the dust removal chimney. The overall success rate of dust removal chimney recognition in the H direction is calculated to obtain the center pixel position of the chimney opening and the chimney width Hd. The coordinates of the center pixel position of the chimney opening are (h0, x0). If the overall recognition success rate is lower than the set standard value, the recognition program outputs the information that the recognition conditions are not met and cannot enter the smoke image recognition program. If the overall recognition success rate is higher than the set standard value, the recognition program outputs the information that the recognition conditions are met and enters the smoke image recognition program. ④ After successful image recognition at the chimney opening, a smoke recognition program is performed. Using the coordinates (h0, x0) of the center pixel position at the chimney opening as the center, a scanning path is initiated with a radius equal to the chimney width Hd, incrementing by one pixel each time until the set maximum smoke detection position Hh is reached. The scanning path follows an inverted U-curve. The average brightness of two adjacent windows is calculated, and gradient judgment is performed. A decreasing brightness gradient indicates the left edge of the smoke from the dust removal chimney, while an increasing brightness gradient indicates the right edge. The smoke center position is identified by gradually diffusing smoke from the chimney opening, eliminating incorrect positions. The overall success rate of smoke recognition in the inverted U-direction is calculated and compared with a set standard value to determine if the chimney is emitting smoke. If smoke is emitted, the difference between the total average brightness of the pixels in the emitting area and the average brightness of the entire recognition area is calculated. Values ​​≥-30, ≥-80 and <-30, and <-80 are used to determine three levels of smoke intensity: large, medium, and small.

Images