A coke mechanical strength detection method based on image recognition

Abstract

The invention relates to a coke mechanical strength detection method based on image recognition. The method comprises the following steps: obtaining an image of a coke sample, calculating and counting the total pixel diameter and the total number of the coke sample before and after a test, respectively counting each coke particle in a range of 60 to 80 mm and greater than 80 mm in the sample before the test, calculating the average particle size to obtain the average pixel particle size and the number of coke particles in a range of 60 to 80 mm and greater than 80 mm, calculating the average pixel particle size and the corresponding number of coke particles in a range of 10 to 20 mm, 20 to 40 mm, 40 to 60 mm, 60 to 80 mm and greater than 80 mm after the test, calculating the area ratio of the coke pixels in each particle size range, establishing wear resistance indexes Ar1 to Ar3, establishing anti-crushing indexes Rc1 to Rc2, and establishing a comprehensive mechanical strength index. The method achieves a quantitative description of the mechanical properties of the coke, replaces manual screening and weighing methods, improves the detection speed, and solves the problems of weak representativeness of the coke mechanical strength, poor sensitivity, and influence of the sampled coke particle size on the detection result.

Description

Technical Field

[0001] The present invention relates to the technical field of ironmaking in the steel industry, and in particular to a method for detecting the mechanical strength of coke based on image recognition. Background Art

[0002] The mechanical strength of metallurgical coke determines its crushing and powder generation properties within the blast furnace, crucially impacting the permeability of the entire blast furnace column, as well as the path and resistance of rising gas flows. The current method for testing the mechanical strength of metallurgical coke is based on the 1970s and 1980s methods of manual screening and weighing using rotary drums. Mechanical properties are expressed in terms of wear resistance (M40) and crushing strength (M10). M40 is the weight percentage of coke with a diameter greater than 40 mm after rotary drum operation, and M10 is the weight percentage of coke with a diameter less than 10 mm after rotary drum operation. This method can only roughly assess the degradation of coke after mechanical operation and has significant drawbacks. Its main issues include: 1. It cannot accurately measure the amount of coke degradation caused by mechanical operation. For example, it is difficult to measure the actual particle size of coke with a particle size less than 10 mm. 2. What is the actual diameter of coke with a particle size greater than 40 mm after rotary drum operation, and how much has the particle size decreased? 1. The method requires that coke with a particle size greater than 60 mm be tested, and then the coke with a particle size greater than 40 mm be weighed after the drum test. When coke with a particle size of 80 mm and coke with a particle size of 60 mm are taken, it is obvious that M40 will have a significant impact on the result. 2. The measurement results lack information on particle size distribution, and have poor guiding significance for blast furnaces and correspondence with the blast furnace smelting process. The blast furnace cannot adjust the smelting parameters and conduct detailed analysis through changes in M40 and M10. Summary of the Invention

[0003] The present invention provides a method for detecting the mechanical strength of coke based on image recognition, which realizes the quantitative description of the mechanical properties of coke, replaces the previous method of manual screening and weighing, improves the detection speed, and solves many problems such as the lack of representativeness of the mechanical strength of coke, poor sensitivity, and the influence of the sampled coke particle size on the detection results.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] A method for detecting the mechanical strength of coke based on image recognition comprises the following steps:

[0006] S1. Obtain an image of a coke sample, measure the particle size of the coke before and after the drum test, and fill all pixels in a square with a side length equal to the pixel diameter of the coke according to the number of pixels in the coke image;

[0007] S2. Calculate and count the total pixel diameter and total number of the coke sample before and after the test;

[0008] S3. Count each coke particle in the range of 60-80 mm and larger than 80 mm in the sample before the test, and calculate the average particle size to obtain the average pixel particle size and number of coke particles in the range of 60-80 mm and larger than 80 mm;

[0009] S4. Calculate the average pixel particle size and corresponding coke quantity of the coke in the ranges of 10-20 mm, 20-40 mm, 40-60 mm, 60-80 mm, and greater than 80 mm after testing the sample;

[0010] S5. Calculate the area ratio of coke pixels in each particle size range;

[0011] S6. Establish wear resistance indices Ar1 to Ar3, where Ar1 is the weight percentage of particles with a diameter less than 10 mm, Ar2 is the pixel area percentage of particles with a diameter of 10 to 20 mm after the test, and Ar3 is the percentage increase in the total pixel area of ​​the coke before and after the test;

[0012] S7, establishing anti-crush indexes Rc1 to Rc2;

[0013] Rc1 is the area ratio of pixels with particle size > 40 mm, which is the proportion of large particle size coke after mechanical movement of coke;

[0014] Rc2 is the percentage of coke fragmentation with a particle size greater than 80 mm, which is the proportion of coke fragmentation with large particle size after mechanical movement;

[0015] S8. Establish a comprehensive mechanical strength index;

[0016] M1=(D-D') / D×100% (1)

[0017] Wherein, D is the average pixel size of coke before the sample test; D' is the average pixel size of coke after the sample test; M1 represents the reduction ratio of the average pixel size of coke after mechanical movement.

[0018] Furthermore, in step S1 , an image of the coke sample is obtained by using a camera, and the camera is set at a distance of 0.2 to 5 meters above the sample.

[0019] Furthermore, in step S2, the coke test is image-based and the particle size calculation formula for each particle of coke is:

[0020]

[0021] Where d is the pixel size of each coke particle in the coke test; S is the actual area of ​​the real region corresponding to the entire coke image; k is the number of pixels contained in the area of ​​each coke instance; and HW is the total number of image pixels.

[0022] Furthermore, the calculation formula for the coke pixel area ratio of each particle size interval in step S5 is as follows:

[0023]

[0024] Among them, S i is the percentage of pixel area of ​​each size coke after the test; N' i is the amount of coke of each particle size after the test; i is the range of each particle size, which is 10, 20, 40, 60, 80, D' i D' is the average pixel diameter of each particle size range of coke after the test, and the average pixel particle size of coke in the ranges of 10-20 mm, 20-40 mm, 40-60 mm, 60-80 mm and >80 mm is expressed as 10 、D' 20 、D' 40 、D' 60 、D' 80 .

[0025] Furthermore, in step S6,

[0026] Ar2=S' 10-20 =D' 10 2 ×N' 10 / ∑(D' i 2 ×N' i )×100% (4)

[0027] Among them, S' 10-20 is the pixel area ratio of the coke particle size in the range of 10 to 20 mm after the test;

[0028]

[0029] Among them, D i is the average pixel diameter of each particle size range of coke before the test, N i is the amount of coke of each particle size before the test.

[0030] Furthermore, in step S7,

[0031] Rc1=S' 40-60 +S' 60-80 +S' >80 (6)

[0032] Among them, S' 40-60S' is the pixel area ratio of coke particles with a diameter of 40 to 60 mm after the test; 60-80 S' is the percentage of pixel area with particle size of 60-80 mm after the test; >80 The percentage of pixel area with particle size greater than 80 mm after the test;

[0033] Rc2=(N 80 -N' 80 ) / N 80 ×100% (7)

[0034] Among them, N 80 N' is the number of coke particles with a diameter greater than 80 mm; 80 It is the average pixel particle size of coke with a particle size greater than 80 mm.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1) Use digital images to accurately measure the particle size of coke before and after the test. By using the concept of pixel area, the particle size of each coke is quantitatively measured.

[0037] 2) Obtain detailed information on particle size distribution, average particle size, coke quantity, and coke pixel area change before and after mechanical action. By establishing several image-based measurement-based wear resistance indexes (Ar) and crushing resistance indexes (Rc), a quantitative description of the mechanical properties of coke can be achieved. This replaces the previous method of manual screening and weighing, speeding up the test while addressing many previous issues, such as the lack of representativeness of coke mechanical strength, poor sensitivity, and the influence of sampled coke particle size on test results.

[0038] 3) It can be used to guide the advancement of coking technology and the establishment of new standards for blast furnace coke, significantly improving the quality of metallurgical coke and optimizing the blast furnace smelting process. DETAILED DESCRIPTION

[0039] The specific embodiments of the present invention are further described below:

[0040] The present invention provides a method for detecting the mechanical strength of coke based on image recognition, comprising the following steps:

[0041] S1. Obtain an image of the coke sample. The camera is required to be 0.2 to 5 meters above the coke sample. The coke sample must ensure that the edges are clear and there is no overlapping between cokes. Use coke with a sample particle size greater than 60 mm and a weight of 50 ± 0.2 kg. Use image recognition processing software to measure the particle size of the coke before and after the drum test. Based on the number of pixels in the coke image, fill all pixels into a square with a side length equal to the pixel diameter of the coke.

[0042] S2. Use image processing software to perform image particle size calculation on the coke test to obtain the pixel particle size d of each coke in the coke test: use the DeblurGANv2 algorithm to deblur the obtained coke image; use the labelme tool to label the clear coke image, and perform data enhancement on the labeled coke image to obtain the coke image dataset required for training; by combining the coco dataset and the coke image dataset, train the improved MaskRCNN model to obtain the instance segmentation model for coke target detection; after deblurring the coke image collected in real time, use the trained instance segmentation model to perform detection, and obtain the number of coke N in the coke image and the corresponding actual area of ​​coke based on the detection results; use the coke area calculation formula to calculate the corresponding coke pixel particle size from the actual area of ​​coke obtained by formula 1;

[0043] The calculation formula for the pixel size of each coke particle is:

[0044]

[0045] Where d is the pixel size of each coke particle in the coke test; S is the actual area of ​​the real area corresponding to the entire coke image; k is the number of pixels contained in the range of each coke instance area; HW is the total number of image pixels; the actual physical meaning of the pixel size of coke is: fill the projected area of ​​coke on the plane with the side length of a solid square area, the pixel size of coke ≤ the diameter of the sieve hole that coke can pass through, and average the pixel size of each coke particle in the sample to obtain Record it as the average pixel particle size D of the coke in this batch of samples.

[0046] S3. Count each particle of coke with a particle size of 60 to 80 mm and larger than 80 mm in the sample and calculate the average particle size to obtain D 60 、D 80 , respectively, represent the average pixel particle size of coke between 60 and 80 mm and larger than 80 mm, in mm, N 60 、N 80 Respectively represent the number of coke in the range of 60-80mm and larger than 80mm in the sample, in pieces.

[0047] S4. Take out the sample after performing drum test. Drum test includes ASTM drum test, JIS drum test, Mikum drum test, ISO drum IRSID drum test, GB / T2006-94 Chinese national standard drum test, GB / T2006-80 Chinese national standard drum test.

[0048] S5. The sample after step 4 is sieved using a sieve with a mesh size of 10 mm, and the image particle size of the sieve material is detected according to steps S1 to S3, and the total number of cokes in the sample N', the average pixel diameter D' of coke, and the average pixel particle size D' of coke in the range of 10-20 mm, 20-40 mm, 40-60 mm, 60-80 mm and greater than 80 mm are obtained respectively. 10 、D' 20 、D' 40 、D' 60 、D' 80 , and the corresponding coke quantity N' 10 、N' 20 、N' 40 、N' 60 、N' 80 .

[0049] S6. Calculate the coke pixel area ratio S in each particle size range according to the following formula i , use S i Indicates the particle size distribution of coke after mechanical action.

[0050]

[0051] Among them, S i is the percentage of pixel area of ​​each size coke after the test; N' i is the number of cokes of each particle size after the test, in pieces; i is the range of each particle size, with values ​​of 10, 20, 40, 60, and 80, D' i D' is the average pixel diameter of each particle size range of coke after the test, and the average pixel particle size of coke in the ranges of 10-20 mm, 20-40 mm, 40-60 mm, 60-80 mm and >80 mm is expressed as 10 、D' 20 、D' 40 、D' 60 、D' 80 , unit: mm.

[0052] S7, establish wear resistance index Ar1 to Ar3 respectively,

[0053] Ar1 is the weight percentage of particles with a diameter less than 10 mm, indicating the weight of the powder produced by coke;

[0054] Ar2 is the percentage of pixel area where the particle size of coke after the test is in the range of 10–20 mm, indicating the extent to which coke produces small particle size coke;

[0055] Ar2=S' 10-20 =D' 10 2 ×N' 10 / ∑(D' i2 ×N' i )×100% (10)

[0056] Among them, S' 10-20 is the pixel area ratio of the coke particle size in the range of 10 to 20 mm after the test;

[0057] Ar3 is the percentage increase of the total pixel area of ​​coke before and after the test, indicating the percentage increase of the total pixel area of ​​coke after mechanical movement, reflecting the degree of coke breakage after mechanical movement;

[0058]

[0059] Among them, D i is the average pixel diameter of each particle size range of coke before the test, N i is the amount of coke of each particle size before the test.

[0060] S8. Establishing anti-crush indexes Rc1 to Rc2;

[0061] Rc1 is the area ratio of pixels with particle size > 40 mm, which is the proportion of large particle size coke after mechanical movement of coke;

[0062] Rc1=S' 40-60 +S' 60-80 +S' >80 (11)

[0063] Among them, S' 40-60 is the pixel area ratio of the coke particle size in the range of 40 to 60 mm after the test; S' 60-80 is the pixel area ratio of the coke particle size in the range of 60 to 80 mm after the test; S' >80 The percentage of pixel area with particle size greater than 80 mm after the test;

[0064] Rc2 is the percentage of coke fragmentation with a particle size greater than 80 mm, which is the proportion of coke fragmentation with large particle size after mechanical movement;

[0065] Rc2=(N 80 -N' 80 ) / N 80 ×100% (12)

[0066] Among them, N 80 N' is the number of coke particles with a diameter greater than 80 mm; 80 The average pixel particle size of coke in the range greater than 80 mm.

[0067] S9. Establish a comprehensive mechanical strength index:

[0068] M1=(D-D') / D×100% (13) M1 represents the percentage of reduction in the average pixel particle size of coke after mechanical movement. It can comprehensively reflect the deterioration properties of coke in the blast furnace and is a comprehensive reflection of wear resistance and anti-breakage performance. Its value is equal to the percentage of reduction in the average particle size of coke after mechanical action, that is, the degree of reduction in the coke particle size. The average particle size of coke after wear and beating can be directly calculated through this value. The smaller the value, the smaller the degree of reduction in the average particle size of coke after wear and beating, and the higher the strength of the coke.

[0069] The following examples are implemented under the premise of the technical solution of the present invention, and provide detailed implementation methods and specific operating processes, but the scope of protection of the present invention is not limited to the following examples. The methods used in the following examples are conventional methods unless otherwise specified.

[0070] [Example]

[0071] S1. Select coke with a particle size greater than 60 mm and a weight of 50 ± 0.2 kg, and obtain an image of the coke sample to ensure that the tiled edges are clear and there is no overlapping between cokes.

[0072] S2. Use image processing software to calculate the particle size of the coke test image to obtain the pixel size d of each particle of coke in the coke test.

[0073] The calculation formula for the pixel size of each coke particle is:

[0074]

[0075] Where d is the pixel size of each coke particle in the coke test; S is the actual area of ​​the real-world region corresponding to the entire coke image; k is the number of pixels within each coke instance; HW is the total number of image pixels. The actual physical meaning of the pixel size of coke is: fill the projected area of ​​coke on the plane with the side length of a solid square area, and the pixel size of coke is ≤ the diameter of the sieve hole that coke can pass through. The pixel size of each coke particle in the sample is averaged to obtain d, which is recorded as the average pixel size D of coke for the batch of samples.

[0076] S3. Count each particle of coke with a particle size of 60 to 80 mm and larger than 80 mm in the sample and calculate the average particle size to obtain D 60 、D 80 , respectively, represent the average pixel particle size of coke between 60 and 80 mm and larger than 80 mm, in mm, N 60 、N 80 Respectively represent the number of cokes in the range of 60 to 80 mm and larger than 80 mm in the sample, in pieces; the specific measurement data are shown in Table 1.

[0077] Table 1

[0078] <![CDATA[D 60(mm) ]]> 68.4 <![CDATA[N 60(个) ]]> 82 <![CDATA[D 80(mm) ]]> 86.6 <![CDATA[N 80(个) ]]> 36 <![CDATA[D (mm) ]]> 74.0 <![CDATA[N (个) ]]> 118

[0079] S4. Perform the GB / T2006-80 Chinese national standard drum test on the sample and then take it out.

[0080] S5. Perform image particle size detection on the oversize material. Obtain the total number of coke samples N', the average pixel diameter D' of coke, and the average pixel particle size D' of coke in the ranges of 10-20mm, 20-40mm, 40-60mm, 60-80mm and greater than 80mm. 10 、D' 20 、D' 40 、D' 60 、D' 80 , and the corresponding coke quantity N' 10 、N' 20 、N' 40 、N' 60 、N' 80 .

[0081] S6. Calculate the coke pixel area ratio S in each particle size range according to the following formula i , use S i Indicates the particle size distribution of coke after mechanical action.

[0082]

[0083] The specific measurement and calculation data are shown in Table 2.

[0084] Table 2

[0085]

[0086]

[0087] S7. Expression of wear resistance index: establish wear resistance index Ar1 to Ar3 respectively.

[0088] Ar1 is the weight percentage of particles with a diameter less than 10 mm, indicating the weight of the powder produced by coke;

[0089] Ar2 is the percentage of pixel area where the particle size of coke after the test is in the range of 10–20 mm, indicating the extent to which coke produces small particle size coke;

[0090] Ar2=S' 10-20 =D' 10 2 ×N' 10 / ∑(D' i 2 ×N' i )×100=14.4 2×201 / (867800)*100=4.8% (16)

[0091] Among them, S' 10-20 is the pixel area ratio of the coke particle size in the range of 10 to 20 mm after the test;

[0092] Ar3 is the percentage increase of the total pixel area of ​​coke before and after the test, indicating the percentage increase of the total pixel area of ​​coke after mechanical movement, reflecting the degree of coke breakage after mechanical movement;

[0093]

[0094] S8. Expression of anti-crush index, establishing anti-crush indexes Rc1 to Rc2 respectively;

[0095] Rc1 is the pixel area ratio of particle size > 40mm,

[0096] Rc1=S' 40-60 +S' 40-60 +S' >80 =(313124+341718+82867) / 867800×100%=85.0% (18)

[0097] It indicates the proportion of large particle size coke after mechanical movement of coke;

[0098] Rc2 is the percentage of coke fragments with a particle size greater than 80 mm.

[0099] Rc2=(N 80 -N' 80 ) / N 80 ×100=(36-12) / 36×100%=66.67%; (19)

[0100] It indicates the proportion of large-particle coke that breaks after mechanical movement of coke.

[0101] S9. Establish a comprehensive mechanical strength index

[0102] M1=(D-D') / D×100=(74.0-35.2) / 74.0×100%=52.43% (20)

[0103] It indicates the percentage of reduction in the average pixel particle size of coke after mechanical movement. It can comprehensively reflect the deterioration properties of coke in the blast furnace and is a comprehensive manifestation of wear resistance and anti-breakage performance. Its value is equal to the percentage of reduction in the average particle size of coke after mechanical action, that is, the degree of reduction in coke particle size. The average particle size of coke after wear and beating can be directly calculated through this value. The smaller the value, the smaller the degree of reduction in the average particle size of coke after wear and beating, and the higher the strength of the coke.

Claims

1. A method for detecting the mechanical strength of coke based on image recognition, characterized in that: The steps include: S1. Obtain an image of a coke sample, measure the particle size of the coke before and after the drum test, and fill all pixels in a square with a side length equal to the pixel diameter of the coke according to the number of pixels in the coke image; S2. Calculate and count the total pixel diameter and total number of the coke sample before and after the test; S3. Count each coke particle in the range of 60 to 80 mm and larger than 80 mm in the sample before the test, and calculate the average particle size to obtain the average pixel particle size and number of coke particles in the range of 60 to 80 mm and larger than 80 mm; S4. Calculate the average pixel particle size and corresponding coke quantity of the coke in the ranges of 10-20 mm, 20-40 mm, 40-60 mm, 60-80 mm, and greater than 80 mm after testing the sample; S5. Calculate the area ratio of coke pixels in each particle size range; S6. Establish wear resistance indices Ar1 to Ar3, where Ar1 is the weight percentage of particles with a diameter less than 10 mm, Ar2 is the pixel area percentage of particles with a diameter of 10 to 20 mm after the test, and Ar3 is the percentage increase in the total pixel area of ​​the coke before and after the test; Among them, S' 10-20 is the pixel area ratio of the coke particle size in the range of 10 to 20 mm after the test; Among them, D i is the average pixel diameter of each particle size range of coke before the test, N i is the amount of coke of each particle size before the test; S7, establishing anti-crush indexes Rc1 to Rc2; Rc1=S' 40-60 +S' 60-80 +S' >80 (3) Among them, S' 40-60 is the pixel area ratio of the coke particle size in the range of 40 to 60 mm after the test; S' 60-80 is the pixel area ratio of the coke particle size in the range of 60 to 80 mm after the test; S' >80 The percentage of pixel area with particle size greater than 80 mm after the test; Rc2=(N 80 -N' 80 ) / N 80 ×100% (4) Among them, N 80 N' is the number of coke particles with a diameter greater than 80 mm; 80 is the average pixel particle size of coke with a particle size greater than 80 mm; Rc1 is the area ratio of pixels with particle size > 40 mm, which is the proportion of large particle size coke after mechanical movement of coke; Rc2 is the percentage of coke fragmentation with a particle size greater than 80 mm, which is the proportion of coke fragmentation with large particle size after mechanical movement; S8. Establish a comprehensive mechanical strength index; M1=(D-D') / D×100% (5) Wherein, D is the average pixel size of coke before the sample test; D' is the average pixel size of coke after the sample test; M1 represents the reduction ratio of the average pixel size of coke after mechanical movement.

2. The method for detecting the mechanical strength of coke based on image recognition according to claim 1, characterized in that: In step S1 , an image of the coke sample is obtained by using a camera, which is placed 0.2 to 5 meters above the sample.

3. The method for detecting the mechanical strength of coke based on image recognition according to claim 1, characterized in that: In step S2, the coke test is imaged and the particle size is calculated using the following formula: Where d is the pixel size of each coke particle in the coke test; S is the actual area of ​​the real region corresponding to the entire coke image; k is the number of pixels contained in the area of ​​each coke instance; and HW is the total number of image pixels.

4. The method for detecting the mechanical strength of coke based on image recognition according to claim 1, characterized in that: The calculation formula for the coke pixel area ratio of each particle size interval in step S5 is as follows: Among them, S i is the percentage of pixel area of ​​each size coke after the test; N' i is the amount of coke of each particle size after the test; i is the range of each particle size, which is 10, 20, 40, 60, 80, D' i D' is the average pixel diameter of each particle size range of coke after the test, and the average pixel particle size of coke in the ranges of 10-20 mm, 20-40 mm, 40-60 mm, 60-80 mm and >80 mm is expressed as 10 、D' 20 、D' 40 、D' 60 、D' 80 .

Images