CN113034480B - Blast furnace damage analysis method based on artificial intelligence and image processing - Google Patents
Blast furnace damage analysis method based on artificial intelligence and image processing Download PDFInfo
- Publication number
- CN113034480B CN113034480B CN202110356247.7A CN202110356247A CN113034480B CN 113034480 B CN113034480 B CN 113034480B CN 202110356247 A CN202110356247 A CN 202110356247A CN 113034480 B CN113034480 B CN 113034480B
- Authority
- CN
- China
- Prior art keywords
- furnace
- blast furnace
- crack
- degree
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 41
- 238000012545 processing Methods 0.000 title claims abstract description 23
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 17
- 230000007797 corrosion Effects 0.000 claims abstract description 33
- 238000005260 corrosion Methods 0.000 claims abstract description 33
- 230000008447 perception Effects 0.000 claims abstract description 17
- 238000005516 engineering process Methods 0.000 claims abstract description 6
- 238000000034 method Methods 0.000 claims description 26
- 230000006870 function Effects 0.000 claims description 13
- 210000001015 abdomen Anatomy 0.000 claims description 11
- 230000011218 segmentation Effects 0.000 claims description 10
- 230000000694 effects Effects 0.000 claims description 8
- 230000003628 erosive effect Effects 0.000 claims description 8
- 238000010586 diagram Methods 0.000 claims description 6
- 238000012423 maintenance Methods 0.000 claims description 6
- 238000013210 evaluation model Methods 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000000926 separation method Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 2
- 230000008030 elimination Effects 0.000 claims description 2
- 238000003379 elimination reaction Methods 0.000 claims description 2
- 230000007547 defect Effects 0.000 abstract description 9
- 238000013178 mathematical model Methods 0.000 abstract description 3
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 4
- 238000003723 Smelting Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 229910000831 Steel Inorganic materials 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 229910052742 iron Inorganic materials 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 229910052751 metal Inorganic materials 0.000 description 2
- 230000035515 penetration Effects 0.000 description 2
- 239000010959 steel Substances 0.000 description 2
- 229910001018 Cast iron Inorganic materials 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 239000000428 dust Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000013021 overheating Methods 0.000 description 1
- 238000009991 scouring Methods 0.000 description 1
- 239000002893 slag Substances 0.000 description 1
- 239000000779 smoke Substances 0.000 description 1
- 238000009628 steelmaking Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- C—CHEMISTRY; METALLURGY
- C21—METALLURGY OF IRON
- C21B—MANUFACTURE OF IRON OR STEEL
- C21B7/00—Blast furnaces
- C21B7/24—Test rods or other checking devices
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention discloses a blast furnace damage analysis method based on artificial intelligence and image processing, which solves the problem that the blast furnace damage degree still needs to be intelligently identified in the prior art. The invention comprises the following steps: a plurality of high temperature resistant CCD cameras are arranged at the furnace mouth and beside the furnace to collect images of the furnace surface; sending the infrared image acquired by the infrared camera into a trained semantic perception network to acquire a gas flow area at the edge of the furnace; extracting temperature distribution characteristic data of the edge gas flow to analyze the corrosion and damage condition of the furnace lining; extracting the surface characteristics of the furnace shell, compensating edge gas flow characteristic data, and analyzing the corrosion loss degree of the furnace lining by combining various characteristics; extracting a crack characteristic vector on the surface of the furnace body by an image processing technology; and constructing a blast furnace damage degree analysis model by combining the characteristics of the corrosion degree of the blast furnace lining and the crack degree of the blast furnace surface. The technology aims at the blast furnace damage degree, combines the furnace lining, the furnace shell and the surface defect characteristics of the furnace body, and builds a mathematical model to analyze the blast furnace damage degree.
Description
Technical Field
The invention relates to the field of artificial intelligence computer vision processing, in particular to a blast furnace damage analysis method based on artificial intelligence and image processing.
Background
The damage degree of the blast furnace is mainly reflected in the damage conditions of furnace lining, furnace body, furnace shell, furnace hearth and the like, and if the blast furnace with larger damage degree is not maintained or rebuilt in time, serious safety accident problem can be caused, and the quality of blast furnace smelting products is reduced. At present, blast furnace damage degree analysis and service life evaluation are performed, and blast furnace service conditions are analyzed mainly by operators observing furnace conditions and monitoring material distribution mineral reaction conditions in a blast furnace hearth, so that the blast furnace damage degree is further evaluated. The method has low accuracy, no real-time performance, manpower waste and low working efficiency. The blast furnace ironmaking is the basic guarantee of the processes of steelmaking, cast iron and the like of iron and steel enterprises, the quality of the operation of the blast furnace directly relates to the benefit of the enterprises, and whether the blast furnace can operate with high efficiency and long service life is more and more valued by iron and steel enterprises of various countries.
Disclosure of Invention
The invention solves the problem that the damage degree of the blast furnace still needs to be intelligently identified in the prior art, and provides the blast furnace damage analysis method based on artificial intelligence and image processing, which has low maintenance cost and high intelligent identification degree.
The technical scheme of the invention is that the blast furnace damage analysis method based on artificial intelligence and image processing comprises the following steps: the method comprises the following steps:
step one, installing a high-temperature-resistant infrared camera at a furnace mouth to collect images in a furnace, and installing a plurality of high-temperature-resistant CCD cameras beside a blast furnace to collect images of a furnace surface;
step two, sending the infrared image acquired by the infrared camera into a trained semantic perception network to acquire a gas flow area at the edge of the furnace;
extracting temperature distribution characteristic data of the edge gas flow to analyze the corrosion and damage condition of the furnace lining;
step four, extracting the surface characteristics of the furnace shell, compensating edge gas flow characteristic data, and analyzing the corrosion loss degree of the furnace lining by combining various characteristics;
extracting a crack characteristic vector on the surface of the furnace body through an image processing technology;
and step six, constructing a blast furnace damage degree analysis model by combining the characteristics of the blast furnace lining corrosion degree and the blast furnace surface crack degree.
Preferably, the furnace mouth high temperature resistant infrared camera is positioned at the furnace mouth edge in the first step, and infrared image data in the furnace is collected; meanwhile, a plurality of industrial CCD cameras are arranged around the blast furnace, the heights of the cameras are positioned in the middle of the furnace body, the positions of the cameras are adjusted and fixed, and the shooting ranges between the adjacent cameras are partially overlapped.
Preferably, in the second step, a semantic perception network is adopted to segment and detect the infrared image in the furnace, firstly, label data are manually made, pixels in the edge gas flow area are marked as 1, pixels in other areas are marked as 0, and the marking of the pixels is required to be performed according to human experience; and then inputting the image data and the label data into a network model, and performing iterative training on the network by adopting a cross entropy loss function.
Preferably, after the semantic perception effect map in the furnace is obtained in the third step, the semantic perception effect map is used as a mask to obtain an IR image of the edge gas flow, the temperature distribution condition of the edge gas flow is obtained, and the average value of the temperature distribution of the edge gas flow area is calculatedAs characteristic data for analyzing the erosion degree of the furnace lining of the blast furnace.
Preferably, after denoising and deblurring the image acquired by the camera in the fourth step, setting an ROI area 1 as a furnace body furnace belly part in the image for each frame of image, and taking the furnace belly redness of the furnace body as characteristic data of furnace lining erosion analysis; the specific process for evaluating the redness of the furnace belly comprises the following steps: firstly, channel separation is carried out on the extracted ROI area 1, R channel of the ROI area 1 is obtained, the channel component is calculated and recorded as R n An evaluation model of the redness of the blast furnace shell is constructed and analyzed, wherein the evaluation model represents the R channel component value of the ROI area 1 in the nth frame image:
wherein N is the selected number of frames, R n An R-channel component value representing ROI area 1 in the nth frame image,representing the redness degree of the furnace shell, wherein the larger the average value of R channel components of the ROI area 1 in the furnace shell image is, the higher the redness degree of the furnace shell is; analyzing and estimating the corrosion degree of the blast furnace lining based on the characteristic data, and constructing a lining corrosion degree analysis model to obtain the corrosion condition of the blast furnace lining, wherein the function expression of the blast furnace corrosion degree analysis model is as follows:
wherein K is an adjustable parameter of the model, K is set to be 10, and C is the erosion degree of the lining in the furnace.
In the fifth step, firstly, the preprocessed RGB image of the surface of the furnace body is segmented, a pixel threshold segmentation method is used for segmenting a defect area, and then the crack duty ratio feature, the crack expansion feature and the crack width feature in the crack defect degree of the blast furnace body are evaluated:
the crack occupation ratio characteristic acquisition process comprises the following steps: acquiring the width and the area of the crack in the image according to the crack segmentation binary image, and marking as w m 、s m Acquiring the minimum circumscribed rectangle of the crack area in an image processing mode, and marking the length and width of the minimum circumscribed rectangle as L m 、W m Constructing a crack ratio calculation model for calculating the crack ratio in the image:
the crack growth characteristic analysis model expression is:
τ in m Expansibility of the mth crack; according to width characteristics, crack occupation ratio characteristics and expansibility characteristics of cracks on the blast furnace surface, constructing
The crack severity estimation model has the following specific function expression:
wherein D is m For the severity of the mth crack, for the overall analysis of the crack severity index of the blast furnace table, all the cracks are summed up to obtain a judgmentDetermining a final model of the crack severity index of the whole blast furnace surface:
wherein m represents the number of cracks, D is the crack severity index of the whole surface of the blast furnace shell, and the larger the judging function value is, the more serious the cracks of the blast furnace surface are considered.
Preferably, in the sixth step, the blast furnace lining erosion degree and the blast furnace surface crack severity are combined, and the blast furnace damage degree analysis model is constructed as follows:
wherein, the weight values of alpha and beta are alpha+beta=1, and the weight value distribution implementation person can select the weight values by oneself, the invention sets that alpha=0.5, beta=0.5, delta represents the damage degree index of the blast furnace body, the larger the model function value is, the higher the damage degree of the blast furnace body is considered, and the preset damage degree threshold delta of the blast furnace body is set T When the damage degree index of the blast furnace body is higher than the set degree threshold, namely delta>δ T The damage degree of the blast furnace body is considered to be larger, and the system can make corresponding early warning maintenance prompts.
Compared with the prior art, the blast furnace damage analysis method based on artificial intelligence and image processing has the following advantages: by adopting an artificial intelligence method, the blast furnace damage degree evaluation index is analyzed through the blast furnace body and the condition in the furnace, so that the blast furnace state is conveniently controlled by operators of a smelting plant in real time, and the safety accidents caused by the condition of the blast furnace are avoided. The damage degree of the blast furnace is mainly analyzed by the self condition of the blast furnace body, and the mechanical scouring effect of furnace burden and the like in the furnace is not considered.
Aiming at the damage degree of the blast furnace, a mathematical model is constructed by combining the furnace lining, the furnace shell and the surface defect characteristics of the furnace body, and the damage degree of the blast furnace is analyzed based on the model. The method can reliably monitor the damage degree of the blast furnace in real time, greatly reduce overhaul investment, improve economic benefit, assist operation and maintenance of the blast furnace, facilitate workers to know the damage degree of the blast furnace in real time, timely take effective measures, effectively improve the production efficiency of the blast furnace and prolong the service life of the blast furnace.
Drawings
Fig. 1 is a system flow diagram of the present invention.
Detailed Description
The blast furnace damage analysis method based on artificial intelligence and image processing is further described below with reference to the accompanying drawings and the detailed description: for a better understanding of those skilled in the art, reference is made to fig. 1 for a description of the following description taken in conjunction with the accompanying examples and figures.
Firstly, installing a high-temperature-resistant infrared camera at a furnace mouth, collecting images in a furnace, and arranging a plurality of high-temperature-resistant CCD cameras beside the blast furnace for collecting images of a furnace surface of the blast furnace.
The method comprises the steps that an image in the furnace is collected through a high-temperature-resistant infrared camera at the furnace mouth, the camera is positioned at the edge of the furnace mouth, the position of the camera is adjusted before image collection so as to shoot the image in the whole furnace, and the camera is fixed after the position of the camera is adjusted, so that the image data of the infrared image in the furnace are collected. Meanwhile, a plurality of industrial CCD cameras are arranged around the blast furnace, the heights of the cameras are positioned in the middle of the furnace body, the positions of the cameras are adjusted and fixed, the phenomenon that the cameras shake in the acquisition process is avoided, and a small number of overlapping areas exist in the shooting range between the adjacent cameras, so that the complete surface image of the furnace body is acquired.
And step two, sending the infrared image acquired by the infrared camera into a trained semantic perception network to acquire a gas flow area at the edge of the furnace. In order to identify the gas flow area at the edge of the furnace and detect the distribution condition of the gas flow area, a semantic perception network is adopted to carry out segmentation detection on the infrared image in the furnace.
For training of a semantic perception network, firstly, label data are manually produced, pixels in an edge gas flow area are marked as 1, pixels in other areas are marked as 0, and the marking of the pixels is carried out according to human experience. And then inputting the image data and the label data into a network model, and performing iterative training on the network by adopting a cross entropy loss function.
The trained semantic perception network can acquire the edge gas flow area perception map from the infrared image in the furnace.
Step three, firstly extracting an edge gas flow area in the furnace, and acquiring the temperature distribution characteristics of the edge gas flow for analyzing the corrosion and damage conditions of the furnace lining by a subsequent system.
The distribution state of the edge gas flow directly influences the service condition of the furnace lining at the inner side of the blast furnace, when the gas flow at the edge of the furnace is excessively developed, the surface temperature of the furnace lining can be increased, the furnace lining is locally overheated and damaged, and the corrosion and the metal penetration of the furnace lining are aggravated. Thus, the erosion of the blast furnace lining will be analyzed by the in-furnace edge gas flow development profile.
After the semantic perception effect diagram in the furnace is obtained, the semantic perception effect diagram is used as a mask, an IR image of the edge gas flow is obtained based on the semantic perception effect diagram, and the temperature distribution condition of the edge gas flow is further obtained. Calculating the mean value of the temperature distribution in the edge gas flow regionAnd is used as a characteristic for analyzing the corrosion damage of the blast furnace lining.
Step four: in order to improve the accuracy of the system, the surface characteristics of the blast furnace shell are extracted, and the corrosion damage of the furnace lining in the blast furnace is further analyzed according to the redness of the furnace surface.
Because the reaction process in the furnace is complex, and the area boundary between the edge gas flow and the central gas flow in the furnace is not obvious enough, the extraction of the edge gas flow area is easy to cause errors, the corrosion loss of the furnace lining is not enough to be accurately judged only through the development state of the edge gas flow, and the furnace lining in the furnace is further analyzed by combining the condition of the furnace shell in order to improve the precision of detecting the corrosion of the furnace lining by the system.
Firstly, a plurality of high temperature resistant industrial cameras are adopted to collect surface images of a furnace body, the height of the cameras is positioned in the middle of the furnace body, the positions of the cameras are adjusted, so that the heights of the furnace belly of the furnace body and the cameras are approximately consistent, the situation that a blast furnace smelting environment is severe and a large amount of smoke dust exists is considered, so that a lot of noise is generated in the images, and denoising and blurring elimination treatment are carried out on the images collected by the cameras. Specific denoising and image blurring eliminating methods are numerous, and an operator can select the methods by himself without going into the description.
After the image is preprocessed, an ROI (region of interest) area 1 is set for each frame of image, and in consideration of the fact that in the using process of the blast furnace, the furnace bellies on the surface of the furnace body can generate redness to different degrees due to different conditions of the furnace lining in the blast furnace, therefore, the ROI area 1 is the furnace bellies in the image, and the redness of the furnace bellies of the furnace body is used as characteristic data for analyzing the corrosion damage of the furnace lining.
The specific process for evaluating the redness of the furnace belly comprises the following steps: firstly, channel separation is carried out on the extracted ROI area 1, R channel of the ROI area 1 is obtained, the channel component is calculated and recorded as R n An evaluation model of the redness of the blast furnace shell is constructed and analyzed, wherein the evaluation model represents the R channel component value of the ROI area 1 in the nth frame image:
wherein N is the selected number of frames, R n An R-channel component value representing ROI area 1 in the nth frame image,the greater the mean value of the R channel components of the ROI area 1 in the furnace shell image, which represents the redness degree of the furnace shell, the higher the redness degree of the furnace shell is considered.
The redness of the furnace shell and the temperature average value of the gas flow at the inner edge of the furnace can be obtained, the corrosion degree of the furnace lining of the blast furnace is analyzed and estimated based on the characteristic data, and a furnace lining corrosion degree analysis model is constructed to obtain the corrosion condition of the furnace lining of the blast furnace. The blast furnace erosion degree analysis model function expression is:
wherein K is an adjustable parameter of the model, K is set to be 10, and C is the corrosion loss of the lining in the furnace.
Step five: the surface condition of the furnace body is also an important characteristic for reflecting the damage degree of the blast furnace, an industrial camera is adopted to shoot an image of the surface of the blast furnace, the crack characteristic of the surface of the furnace body is extracted through an image processing technology, and the crack severity index of the surface of the blast furnace is analyzed and used for subsequent analysis of the final damage condition of the blast furnace.
The main purpose is to analyze the damage degree of the blast furnace, and the blast furnace damage is mainly reflected in the blast furnace and the damage of the blast furnace surface, so that the expression condition of the blast furnace is judged and analyzed for comprehensively analyzing the damage condition of the blast furnace.
The blast furnace table analysis mainly analyzes the crack phenomenon of the blast furnace table. In order to analyze the conditions of the blast furnace surface, firstly, the RGB image of the pretreated furnace body surface is divided, and a defect area is divided for subsequent analysis of the crack degree. The crack region is segmented into pixel threshold segmentation methods, pixel threshold values are set for the blast furnace surface image, the crack region is segmented based on the pixel threshold values, an image segmentation algorithm is mature and is a conventional technology, a specific segmentation method implementation person selects the crack region by himself, and the crack region is not described in detail in a protection range.
So far, the crack area of each furnace surface can be obtained.
In order to evaluate the crack defect degree of the blast furnace body, the defect characteristics of the blast furnace body are extracted and are used for subsequent analysis and judgment of the crack severity degree of the blast furnace. The defect characteristics of the blast furnace surface mainly comprise: crack occupancy characteristics, crack propagation characteristics, and crack width characteristics.
The crack duty ratio characteristic acquisition process specifically comprises the following steps: acquiring the width and the area of the crack in the image according to the crack segmentation binary image, and marking as w m 、s m . The crack width is used as a width feature of the crack. Acquiring the minimum circumscribed rectangle of the crack area in an image processing mode, and marking the length and width of the minimum circumscribed rectangle as L m 、W m Constructing a crack ratio calculation model for calculating the crack ratio in the image:
the specific analysis process of the crack expansibility characteristics comprises the following steps: in order to embody the crack grade of the blast furnace body, accurately analyzing the severity of the blast furnace crack, constructing a crack expansion characteristic analysis model, and analyzing the possibility of the expansion of the blast furnace crack to the periphery, wherein the expression of the furnace surface crack expansion characteristic analysis model is as follows:
τ in m Is the expansibility of the mth crack.
And finally, evaluating and analyzing the severity of the crack according to the width characteristic (namely the crack width), the crack occupation characteristic and the expansibility characteristic of the crack on the blast furnace surface. Constructing a crack severity estimation model, wherein the specific function expression is as follows:
wherein D is m For the severity of the mth crack, for integrally analyzing the crack severity index of the blast furnace table, summing all the cracks to obtain a final model for judging the crack severity index of the whole blast furnace table:
wherein m represents the number of cracks, and D is an index of the severity of cracks on the surface of the whole blast furnace shell. The larger the judgment function value is, the more serious the blast furnace surface crack is considered.
Thus, the method can obtain the corrosion damage degree of the lining in the blast furnace and the severity degree of the surface crack of the blast furnace body.
Step six: and constructing a blast furnace damage degree analysis model by combining the blast furnace lining corrosion degree and the blast furnace surface crack severity, and evaluating the overall damage degree of the blast furnace based on the model.
Finally, the damage degree of the whole blast furnace body is analyzed and judged based on the blast furnace lining and the furnace expression condition, so that a worker can master the service condition of the blast furnace in real time. The blast furnace body damage degree analysis model is as follows:
wherein, the weight values of alpha and beta and alpha+beta=1 can be selected by an allocation implementer of the weight values, the weight values are set to alpha=0.5 and beta=0.5, delta represents the damage degree index of the blast furnace body, and the larger the model function value is, the higher the damage degree of the blast furnace body is considered.
So far, the practitioner can grasp the condition of the blast furnace body in real time based on the system, and in order to avoid the problems of safety problem, product quality reduction and the like in the smelting process caused by overlarge damage degree of the blast furnace body, the damage degree of the blast furnace body is preset with a degree threshold delta T . When the damage degree index of the blast furnace body is higher than the set degree threshold, namely delta>δ T The damage degree of the blast furnace body is considered to be larger, the system can give corresponding early warning prompt, prompt maintenance personnel to detect and maintain the blast furnace condition as soon as possible, and replace blast furnace parts if necessary, so as to prevent the problems of safety accidents and the like caused by the too high damage degree of the blast furnace.
In summary, the invention provides a blast furnace damage analysis system based on artificial intelligence and image processing. The method is characterized in that an image in the furnace is acquired through an anti-high temperature industrial camera at the furnace mouth, filtering treatment is carried out on the image, when the gas flow at the edge of the furnace is developed excessively, corrosion and metal penetration of furnace lining slag are aggravated, and local overheating and burning through of the furnace lining are easily caused, so that the method is combined with the characteristic vector of the gas flow at the edge of the furnace and the characteristic of redness degree of the furnace shell, and the corrosion and damage degree of the furnace lining are analyzed. Meanwhile, a multi-camera is adopted to collect a blast furnace surface image outside the blast furnace, the RGB image of the blast furnace surface is processed, the surface defect characteristics of the furnace body are extracted, and the crack degree of the furnace body is analyzed. And constructing a mathematical model based on the crack degree characteristics of the furnace surface and the corrosion damage degree of the furnace lining, detecting the damage degree of the blast furnace in real time, and providing data analysis for the operation and maintenance of the blast furnace so as to prompt the staff to detect and maintain the blast furnace correspondingly in time and prolong the service life of the blast furnace.
Claims (6)
1. A blast furnace damage analysis method based on artificial intelligence and image processing is characterized in that: the method comprises the following steps:
step one, installing a high-temperature-resistant infrared camera at a furnace mouth to collect infrared images in a furnace, and installing a plurality of high-temperature-resistant CCD cameras beside a blast furnace to collect RGB images on the surface of the furnace body;
step two, sending the infrared image in the furnace collected by the high-temperature-resistant infrared camera into a trained semantic perception network to obtain a gas flow area at the edge of the inside of the furnace;
extracting temperature distribution characteristic data of a gas flow area at the inner edge of the furnace;
extracting furnace shell surface features from the RGB image of the surface of the furnace body, compensating temperature distribution feature data of an edge gas flow area, and fusing various features to analyze the corrosion degree of the blast furnace lining;
fifthly, determining the severity of cracks on the surface of the blast furnace of the RGB image on the surface of the furnace body through an image processing technology;
step six, constructing a blast furnace damage degree analysis model by combining the blast furnace lining damage degree and the blast furnace surface crack severity degree;
in the fifth step, firstly, the preprocessed RGB image of the surface of the furnace body is segmented, a crack area is segmented by a threshold segmentation method, and then crack duty ratio characteristics, crack expansion characteristics and crack width characteristics in the crack severity of the surface of the blast furnace are evaluated:
the crack occupation ratio characteristic acquisition process comprises the following steps: determining the width and the area of the crack in the RGB image of the furnace body surface according to a crack segmentation binary image obtained after threshold segmentation of the RGB image of the furnace body surface, and marking as w m Sum s m Acquiring the minimum circumscribed rectangle of the crack area by an image processing mode, and respectively recording the length and the width of the minimum circumscribed rectangleIs L m And W is m Constructing a crack ratio calculation model for calculating the crack ratio in the image:
the crack growth characteristic analysis model expression is:
τ in m Expansibility of the mth crack;
according to the crack width characteristic, the crack duty ratio characteristic and the crack expansion characteristic in the crack severity of the blast furnace surface, a crack severity estimation model is constructed, and the specific function expression is as follows:
wherein D is m For the severity of the mth crack, for the overall analysis of the severity of the blast furnace surface crack, summing the severity of all cracks to obtain a final model for determining the severity of the entire blast furnace surface crack:
wherein m represents the number of cracks, and the larger the D value is, the more serious the cracks on the blast furnace surface are.
2. The blast furnace damage analysis method based on artificial intelligence and image processing according to claim 1, wherein: the high-temperature-resistant infrared camera in the first step is positioned at the edge of the furnace mouth and is used for collecting infrared image data in the furnace; meanwhile, a plurality of industrial CCD cameras are arranged around the blast furnace, the heights of the cameras are positioned in the middle of the furnace body, the positions of the cameras are adjusted and fixed, and the shooting ranges between the adjacent cameras are partially overlapped.
3. The blast furnace damage analysis method based on artificial intelligence and image processing according to claim 1, wherein: in the second step, a semantic perception network is adopted to carry out segmentation detection on the infrared image in the furnace, firstly, label data are manually manufactured, the pixels of the edge gas flow area are marked as 1, and the pixels of other areas are marked as 0; and then inputting the image data and the label data into a network model, and performing iterative training on the network by adopting a cross entropy loss function.
4. The blast furnace damage analysis method based on artificial intelligence and image processing according to claim 1, wherein: in the third step, after the semantic perception effect diagram in the furnace is obtained, the semantic perception effect diagram is used as a mask, an IR image of the gas flow area at the edge of the furnace is obtained, the temperature distribution condition of the gas flow area at the edge of the furnace is obtained, and the average value of the temperature distribution of the gas flow area at the edge of the furnace is calculatedAs characteristic data for analyzing the degree of erosion of the lining of a blast furnace.
5. The blast furnace damage analysis method based on artificial intelligence and image processing according to claim 1, wherein: in the fourth step, after denoising and blurring elimination are carried out on the RGB image of the surface of the furnace body acquired by the high-temperature-resistant CCD camera, the furnace belly part in each frame of image is set as an ROI area, and the redness degree of the furnace belly of the furnace body is used as characteristic data for analyzing the erosion degree of the furnace lining; the specific process for evaluating the redness degree of the furnace belly comprises the following steps: firstly, channel separation is carried out on the extracted ROI area, R channels of the ROI area are obtained, channel components are calculated and recorded as R n R channel component values representing the ROI area in the nth frame image are used for constructing an evaluation model of the redness degree of the blast furnace shell, and analyzing the redness degree of the blast furnace shell:
wherein N is the selected number of frames,representing the redness degree of the blast furnace shell, wherein the larger the average value of R channel components of the ROI area in the blast furnace shell image is, the higher the redness degree of the furnace shell is; based on the redness degree of the furnace shell, the corrosion degree of the blast furnace lining is analyzed and estimated, a blast furnace lining corrosion degree analysis model is constructed, the corrosion condition of the blast furnace lining is obtained, and the function expression of the blast furnace lining corrosion degree analysis model is as follows:
wherein K is an adjustable parameter of the model, and C is the corrosion loss degree of the blast furnace lining.
6. The blast furnace damage analysis method based on artificial intelligence and image processing according to claim 5, wherein: in the step six, the corrosion degree of the blast furnace lining and the crack severity degree of the blast furnace surface are combined, and an index analysis model of the damage degree of the blast furnace body is constructed as follows:
wherein α and β are weight values and α+β=1, α=0.5, β=0.5, δ represents a damage degree index of the blast furnace body, and the larger the δ value is, the higher the damage degree of the blast furnace body is considered, and a preset damage degree threshold δ of the blast furnace body is set T When the damage degree index of the blast furnace body is higher than the preset degree threshold, namely delta>δ T The damage degree of the blast furnace body is large, and the system makes corresponding early warning maintenance prompts.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110356247.7A CN113034480B (en) | 2021-04-01 | 2021-04-01 | Blast furnace damage analysis method based on artificial intelligence and image processing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110356247.7A CN113034480B (en) | 2021-04-01 | 2021-04-01 | Blast furnace damage analysis method based on artificial intelligence and image processing |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113034480A CN113034480A (en) | 2021-06-25 |
CN113034480B true CN113034480B (en) | 2023-12-19 |
Family
ID=76454048
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110356247.7A Active CN113034480B (en) | 2021-04-01 | 2021-04-01 | Blast furnace damage analysis method based on artificial intelligence and image processing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113034480B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113657193A (en) * | 2021-07-27 | 2021-11-16 | 中铁工程装备集团有限公司 | Segment damage detection method and system based on computer vision and shield machine |
CN114507760A (en) * | 2022-02-28 | 2022-05-17 | 马鞍山钢铁股份有限公司 | Method and system for detecting damage condition of blast furnace lining |
CN114998326B (en) * | 2022-07-19 | 2023-03-07 | 江苏瑞立环保工程股份有限公司 | Method and system for detecting distribution condition of gas flow based on artificial intelligence |
JP2024047865A (en) * | 2022-09-27 | 2024-04-08 | Jfeスチール株式会社 | Gutter management system and gutter management method |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0674977A1 (en) * | 1994-03-29 | 1995-10-04 | General Electric Company | Maintenance system |
CN107749058A (en) * | 2017-10-23 | 2018-03-02 | 中南大学 | A kind of machine vision detection method and system of boiler tubing surface defect |
JP2018148201A (en) * | 2017-03-03 | 2018-09-20 | 株式会社Screenホールディングス | Heat treatment method and heat treatment apparatus |
CN108998608A (en) * | 2018-07-24 | 2018-12-14 | 中南大学 | A kind of blast furnace iron notch molten iron temperature measurement method and system based on infrared machine vision |
CN109852748A (en) * | 2019-02-27 | 2019-06-07 | 内蒙古科技大学 | Monitor the Gas Flow development process of blast furnace material distribution period and prediction gas utilization rate method |
CN110762540A (en) * | 2019-10-22 | 2020-02-07 | 华南理工大学 | Combustion decision method and automatic combustion control method and system for garbage incinerator |
CN112200782A (en) * | 2020-09-30 | 2021-01-08 | 邵传宏 | Photovoltaic cell panel damage risk degree detection method based on artificial intelligence |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018020558A1 (en) * | 2016-07-25 | 2018-02-01 | オリンパス株式会社 | Image processing device, image processing method, and program |
-
2021
- 2021-04-01 CN CN202110356247.7A patent/CN113034480B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0674977A1 (en) * | 1994-03-29 | 1995-10-04 | General Electric Company | Maintenance system |
JP2018148201A (en) * | 2017-03-03 | 2018-09-20 | 株式会社Screenホールディングス | Heat treatment method and heat treatment apparatus |
CN107749058A (en) * | 2017-10-23 | 2018-03-02 | 中南大学 | A kind of machine vision detection method and system of boiler tubing surface defect |
CN108998608A (en) * | 2018-07-24 | 2018-12-14 | 中南大学 | A kind of blast furnace iron notch molten iron temperature measurement method and system based on infrared machine vision |
CN109852748A (en) * | 2019-02-27 | 2019-06-07 | 内蒙古科技大学 | Monitor the Gas Flow development process of blast furnace material distribution period and prediction gas utilization rate method |
CN110762540A (en) * | 2019-10-22 | 2020-02-07 | 华南理工大学 | Combustion decision method and automatic combustion control method and system for garbage incinerator |
CN112200782A (en) * | 2020-09-30 | 2021-01-08 | 邵传宏 | Photovoltaic cell panel damage risk degree detection method based on artificial intelligence |
Non-Patent Citations (1)
Title |
---|
基于红外热图像信息的铜转炉炉衬蚀损状态在线测量系统研究;匡以顺;《中国优秀硕士学位论文全文数据库》(第09期);摘要、第21-55页 * |
Also Published As
Publication number | Publication date |
---|---|
CN113034480A (en) | 2021-06-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113034480B (en) | Blast furnace damage analysis method based on artificial intelligence and image processing | |
WO2020098299A1 (en) | Method and system for detecting flow rate of high-temperature molten fluid | |
CN113989279B (en) | Plastic film quality detection method based on artificial intelligence and image processing | |
CN109580657B (en) | Crack detection method in bridge quality detection | |
CN109115785B (en) | Casting polishing quality detection method and device and use method thereof | |
JP5645730B2 (en) | Method for detecting closed cracks on concrete surface | |
CN106991668B (en) | Evaluation method for pictures shot by skynet camera | |
CN110493574B (en) | Security monitoring visualization system based on streaming media and AI technology | |
CN101908142A (en) | Feature analysis-based video flame detecting method | |
CN104331521A (en) | Automatic power transformation equipment abnormity identification method based on image processing | |
CN103996203A (en) | Method and device for detecting whether face in image is sheltered | |
Diaz et al. | Rust detection using image processing via Matlab | |
CN106934801A (en) | A kind of fluorescentmagnetic particle(powder) defect inspection method based on Laws texture filterings | |
CN112270658A (en) | Elevator steel wire rope detection method based on machine vision | |
CN114332081B (en) | Textile surface abnormity determination method based on image processing | |
CN113177605A (en) | Scrap steel carriage grade judgment method based on video monitoring | |
CN113469974B (en) | Method and system for monitoring state of grate plate of pellet grate | |
CN110363749B (en) | Evaluation method for rusting degree of vibration damper based on image processing | |
CN117333825B (en) | Cable bridge monitoring method based on computer vision | |
CN110992629A (en) | Method for detecting static human body based on video monitoring | |
CN111402321A (en) | Method for target identification and volume estimation of solid particles in high-temperature molten pool | |
CN105872464B (en) | Factory smoke discharge automatic monitoring system based on vision signal | |
CN114758260A (en) | Construction site safety protection net detection method and system | |
CN113963291A (en) | Coal-fired working condition identification method based on video image quality and model modeling method thereof | |
CN113538393A (en) | Bar and wire stock blank bending detection method, device and equipment and readable storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
TA01 | Transfer of patent application right | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20231123 Address after: Unit 1104-B, No. 78, Lane 1688, Guoquan North Road, Yangpu District, Shanghai, 200082 Applicant after: Aide Linker (Shanghai) Digital Technology Co.,Ltd. Address before: 710075 10612, unit 1, building 6, Qidi Science Park, No.65, Keji 2nd Road, high tech Zone, Xi'an City, Shaanxi Province Applicant before: Xi'an Daofa Digital Information Technology Co.,Ltd. |
|
GR01 | Patent grant | ||
GR01 | Patent grant |