CN106521066A - Blast furnace burden particle size monitoring system and distributed data on-line analysis method - Google Patents
Blast furnace burden particle size monitoring system and distributed data on-line analysis method Download PDFInfo
- Publication number
- CN106521066A CN106521066A CN201611206265.2A CN201611206265A CN106521066A CN 106521066 A CN106521066 A CN 106521066A CN 201611206265 A CN201611206265 A CN 201611206265A CN 106521066 A CN106521066 A CN 106521066A
- Authority
- CN
- China
- Prior art keywords
- image
- pixel
- material block
- blast furnace
- value
- 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.)
- Pending
Links
Classifications
-
- 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
-
- C—CHEMISTRY; METALLURGY
- C21—METALLURGY OF IRON
- C21B—MANUFACTURE OF IRON OR STEEL
- C21B5/00—Making pig-iron in the blast furnace
- C21B5/006—Automatically controlling the process
Landscapes
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Manufacturing & Machinery (AREA)
- Materials Engineering (AREA)
- Metallurgy (AREA)
- Organic Chemistry (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a blast furnace burden particle size monitoring system and a particle size distributed data on-line analysis method. Imaging equipment of the system comprises at least one industrial video camera. Lighting sources are obliquely arranged downwards in the opposite directions at the positions lower than the imaging equipment by 0.4-0.5m so that low-angle lighting can be realized. The luminous flux is 40000+/-5%LM, the color temperature is 5500-6000K, and accordingly field images can be collected clearly. By matching the particle size distributed data on-line analysis method, the images collected at site are sequentially subjected to denoising, balancing, smoothing, reinforcement, sharpening, self-adaption threshold value binaryzation and corrosion and dilation operation, so that the particle size and lump number information of the burden fed into a blast furnace is obtained. The monitoring system is adopted to be matched with the image analysis method so that particle size measurement can be realized, and the accuracy and speed are high. The particle size and lump number information and other information are automatically acquired in an online mode in real time, the varieties of materials are matched automatically, and manpower and material resources are effectively reduced.
Description
Technical field
The present invention relates to video monitoring on-line analysis technical field, more particularly to a kind of blast furnace burden granularity monitoring system and
Distributed data on-line analysis.
Background technology
In the blast furnace ironmaking production process of metallurgy industry, blast furnace furnace charge, especially sintering deposit and coke enter stokehold all
Physics and chemical analysis need to be carried out.Wherein, the distribution situation of furnace size, is exactly a very important parameter.According to blast furnace
Smelting process, has strict requirements to furnace charge, for example, enter the furnace charge that furnace size is uniform, furnace size is less than normal, add
In powder to screen out and furnace charge in objectionable impurities content to wait less.Varigrained crude fuel loads blast furnace, and furnace charge has
Filling effect, reduces the space of furnace charge, poor air permeability.And epigranular can then improve furnace charge air permeability, ore is improved indirect
Reduction degree.This energy-conservation to blast fumance, high yield have positive meaning.
At present, the grain size analysis of general blast furnace burden, is using the easy mechanical means of tradition mostly.
For example, using sample spoon or bucket-shaped container, sample at regular intervals once, deliver to screening machine classification, what is separated is each
Grade is sent into weighing funnel and is weighed, by the distribution situation for calculating furnace size size.As can be seen here, using traditional inspection
Proved recipe method carries out the grain size analysis of furnace charge, not only loaded down with trivial details time-consuming and extremely inefficient, it is impossible to meet wanting for the big production of modernization
Ask.Therefore, be badly in need of realizing it is online, in real time, efficiently detection mode.
With developing rapidly for industrial video technology and computer graphics disposal technology, be realize it is online, in real time, efficiently
Detection is there is provided possible.The present invention is exactly to utilize digital image processing techniques, by carrying out image to the furnace charge on belt conveyor
Collection, using analysis and process to video big data, calculates the size and distribution situation of furnace size, the person that makes blast furnace operating
Furnace size information is obtained quickly and accurately.The device has it is accurate, quick and efficient the characteristics of, can substitute it is existing fall
Characterization processes, have important function to instructing blast furnace operating and improving technology controlling and process level afterwards.
The content of the invention
It is an object of the invention to provide one kind can realize that real-time online is monitored to blast furnace burden particle size distribution data
With the blast furnace burden granularity monitoring system for processing.
It is a further object of the present invention to provide a kind of blast furnace burden image to video acquisition carries out quick fmer-granularity point
The blast furnace burden particle size distribution data on-line analysis of analysis
For this purpose, technical solution of the present invention is as follows：
A kind of blast furnace burden granularity monitoring system, including the imaging device being arranged on above conveying equipment and be located at it is described into
As the lighting source of equipment both sides；The lighting source less than at the 0.4～0.5m of imaging device to being tilted towards dividing into
Put；40000 ± the 5%LM of luminous flux of the lighting source, colour temperature are 5500～6000K.
Further, the lighting source is LED/light source, and which is arranged on the symmetrical low angle polishing in imaging device both sides；Institute
It is 30～60 ° to state the angle between the beam center of lighting source and vertical direction.
Further, the imaging device includes at least one industrial camera, between the adjacent industrial camera
Level interval is 0.5～1m；Camera lens, the industrial camera and the camera lens are being installed per industrial camera front end described in platform
Arrange vertically downward.
Further, gas sweeping type protective jacket is also set with the outside of the imaging device, it is ensured that imaging device and illumination
Light source stable normal work in the environment of high dust；Wherein, blowing medium is compressed air.
Further, the system also includes being that imaging device and lighting device provide the field control case of stabilized power source and set
Put in the indoor control system of control；The control system includes analyzing and processing the industrial computer of video image, is used for
Show real-time image acquisition and process after data display device, for control field control case, industrial computer and show dress
Indoor control cabinet, acoustic-optic alarm and the signal conversion being arranged between industrial computer and acoustic-optic alarm put
Device.
A kind of blast furnace burden particle size distribution data on-line analysis, comprise the steps:
S1, IMAQ is carried out by imaging device to the blast furnace burden on conveying equipment, obtain original image；
S2, the original image to the acquisition of Jing steps S1 carry out denoising, balanced and smooth pretreatment successively；
S3, image enhaucament and image sharpening are carried out successively to the image Jing after the process of step S2 process；
S4, to Jing step S3 process after image using adaptive threshold binarization method obtain can recognize that image
In all of discrete material block image；
S5, the binary image that Jing step S4 process is obtained is corroded and dilation operation process；
S6, the quantity to the material block in the image that obtains Jing after the process of step S5 judge with the presence or absence of material block and
The species of material block；
S7, the image Jing after the process of step S5 is carried out by material block carries out the calculating of granularity, particle mean size and quantity；
S8, after judging that material block passes through completely by the gross, the whole original images of the collection during this are repeated on
The processing procedure of step S2～S7 is stated, and calculates the particle mean size and size distribution of material block by the gross.
Wherein, the image denoising of original image is processed, image equalization is processed and the concrete grammar of picture smooth treatment is：
The method that S201, described image denoising adopt adaptive median filter；
S202, described image equilibrium treatment are according to formula：
Carry out, wherein, rkIn being original image
K-th gray level, skIt is rkGray value after transformed, T (rk) for change function, Pr(rk) for gray level in original image be rk
Pixel occur probability, njFor the pixel count of kth level gray scale, N is the sum of pixel in piece image；
S203, described image smoothing processing are according to formula：Carry out, wherein, (x, y) is
The coordinate of image, g (x, y) are the gray values of image slices vegetarian refreshments (x, y) after smoothing, and σ is the width of Gaussian filter.
The concrete grammar that image enhaucament and image sharpening process is carried out to the image of step S2 process is：
S301, image enhancement processing are according to formula：G (x, y)=× factor+orig is carried out | f (x, y)-mean |, its
In, f (x, y) is the gray value of original image pixels point (x, y), and g (x, y) is the ash of image slices vegetarian refreshments (x, y) after smoothing
Angle value, mean are the average gray values of entire image, and factor is to strengthen coefficient, and orig is penalty coefficient；
S302, image sharpening process are according to formula：Carry out, its
In, g (x, y) is the gray value of the pixel (x, y) after image enhaucament,
G [f (x, y)]=| f (x, y)-f (x+1, y+1) |+| f (x+1, y)-f (x, y+1) | it is the ladder at pixel (x, y) place
Angle value, f (x, y) are gray value of the original image at pixel (x, y) place, and A is image sharpening coefficient, and T is image sharpening threshold value；
When G [f (x, y)] is more than or equal to threshold value T, assert edge of the pixel in image, image sharpening system is added to result
Number A, so that edge brightens；When G [f (x, y)] is less than threshold value, assert that the pixel is similar pixel.
Image Jing after the process of step S3 is included entering as follows successively using the binarization method of adaptive threshold
Row is calculated：
N1+N2=M × N；
ω1+ω2=1；
μ=μ1×ω1+μ2×ω2；
G=ω1×(μ-μ1)2+ω2×(μ-μ2)2=ω1×ω2×(μ1-μ2)2；
Threshold=maxJ=0,1.....255gj；
Wherein, M × N is the size of image, N1It is number of pixels of the gray value less than threshold value, ω in image1It is that gray value is little
The ratio of entire image, μ are accounted in the number of pixels of threshold value1Gray value less than threshold value pixel average gray, N2In being image
Number of pixels of the gray value more than threshold value, ω2It is that number of pixels in image more than threshold value accounts for the ratio of entire image, μ2It is ash
Average gray of the angle value more than the pixel of threshold value, g is inter-class variance, threshold be between the class for traveling through gray value 0～255 most
Big variance, i.e. self-adaption binaryzation threshold value, f (x, y) is gray value of the original image in pixel (x, y), g (x, y) be through
Gray value of the image after self-adaption binaryzation process in pixel (x, y).
The binary image that Jing step S4 process is obtained is corroded and dilation operation process concrete process step bag
Include：
In morphological image, " the spectral window that function as in signal transacting of the structural element in morphological transformation
Mouthful ".With B (x) representative structure elements, erosion operation process is carried out to the every bit x in working space E；
Erosion operation presses formula：
Carry out；It is exactly making B be contained in all of E after structural element B translations to the result that E carries out erosion operation with B (x)
The set that point is constituted；
Dilation operation presses formula：
Carry out, be exactly the common factor non-NULL that B and E are made after structural element B translations to the result that E carries out dilation operation with B (y)
Point constitute set.
The quantity of the material block in the image that obtains Jing after the process of step S5 is judged with the presence or absence of material block and thing
The species of material block；Specifically, the species of blast furnace burden is generally divided into two kinds：Coke and ore；Wherein, ore is by sintering
Various ore compositions such as ore deposit, lump ore, pellet, thus coke and two class material of ore are visually by the big of material block
It is little just to distinguish；When being judged, as the image of imaging device collection is respectively provided with identical field range, therefore can be with
Judged according to the number of material block in every image, i.e., the number of the material block in the range of the bigger same field of view of material block is more
It is few, conversely, the number of the material block in the range of the less same field of view of material block is more；Concrete determination methods are as follows：
Two kinds of material variety judgment thresholds are set：Coke decision threshold YJ(block) and ore decision threshold YK(block), it is assumed that when
Material in front picture is Y (block), then：
As Y ＜ YJ, then no material block in image, image of the image for conveyer belt；
Work as YJ≤Y≤YK, then the material block that IMAQ is arrived is coke；
As Y ＞ YK, then the material block that IMAQ is arrived is ore；
Material variety judgement is carried out to every image Jing after the process of step S5 based on above-mentioned determination methods, after the completion of judgement
Following step S7 and step S8 will be carried out, and classification storage will be carried out to image.
In step S7,1) in image the pixel value size of material block diameter according to formula：Carry out；Thing
The actual diameter size of material block is according to formula：Carry out；Wherein, DxIt is that the material block for calculating in the picture is straight
The pixel value size of footpath size, SxIt is the pixel value size of the external area of a circle of minimal closure of the material block for calculating in the picture, Ds
It is the actual diameter size dimension of material block, FsIt is the actual laterally visual field of image captured by camera, fxIt is the horizontal resolution of image
Rate；2) computational methods of material block particle mean size l press formula：Carry out, wherein, K is one
The number of effective material block in image；riIt is the actual diameter size dimension of i-th piece of material block；tempriIt is i-th piece of material block
The average of the grain size intervals corresponding to actual diameter, it is assumed that grain size intervals are (D1,D2), then the average of the grain size intervals is：
In step S8, formula is pressed in the particle mean size calculating to material block by the gross：Carry out, wherein, L is the time
The particle mean size of material in section；N is the number of the picture that material is included in the time period；liIt is the flat of material in the i-th pictures
Equal granularity；
The average particle size distribution of material block by the gross is calculated successively by following formula：WithCarry out, wherein, S is the summation of whole material areas in the time period, SiIt is correspondence granularity point
The summation of the material area in cloth interval, N2It is the interval division number of size distribution, sjIt is that correspondence size distribution is interval interior each
The area of block material, N1Be correspondence size distribution it is interval in material block quantity, periIt is the percentage shared by size distribution interval
Than.
Description of the drawings
Fig. 1 is the structural representation of the blast furnace burden granularity monitoring system of the present invention；
Fig. 2 is that the imaging device and lighting source of the blast furnace burden granularity monitoring system of the present invention are arranged on conveying equipment
Structural representation；
Side structure schematic diagrams of the Fig. 3 for Fig. 2；
Fig. 4 is the original image of imaging device collection in the embodiment of the present invention；
Fig. 5 is the image in the embodiment of the present invention Jing after the process of step S3；
Fig. 6 is the image in the embodiment of the present invention Jing after the process of step S4；
Fig. 7 is the image in the embodiment of the present invention Jing after the process of step S5；
Fig. 8 is the minimum circumscribed circle calculating schematic diagram in step S6 in the embodiment of the present invention to particle.
Specific embodiment
Below in conjunction with the accompanying drawings and specific embodiment the present invention is described further, but following embodiments are absolutely not to this
It is bright to have any restriction.
Embodiment
As shown in figure 1, the blast furnace burden granularity monitoring system arranges imaging device 1 at the scene, lighting source 2, scene
Control cabinet 3, and it is arranged on the indoor industrial computer 4 of control, display device 5, indoor control cabinet 6,7 harmony of signal adapter
Light alarm device 8.
As shown in figures 2-3, imaging device 1 and lighting source 2 are arranged on above blast furnace burden conveying equipment；Blast furnace burden
It is delivered in State of Blast Furnace using feed belt, imaging device 1 is arranged on the surface of feed belt, and lighting source 2 is less than institute
State and be arranged oppositely at imaging device 0.4m, it is 45 ° that its angle of inclination is 45 °, i.e. lighting angle；Specifically, the lighting source 2
From HQ-SD-W400F floodlights, which meets 40000 ± 5%LM of luminous flux, and colour temperature is the requirement of 5500～6000K.
As the width of the feed belt is 1.6m, therefore imaging device 1 is made up of two industrial cameras, and two industrial
Position for video camera is in same horizontal line and spacing is 0.8m, makes every video camera of two industrial cameras correspond to feed belt respectively
Zones of different carries out IMAQ, while ensureing the width of the image-capture field in feed belt of two industrial cameras
It is upper to cover whole feed belt.Vertically lower section material 9 is arranged for two industrial cameras；Install every industrial camera front end
There is camera lens.As blast fumance condition is poor, under open environment, dust is more, therefore on the outside of every industrial camera and camera lens
Gas sweeping type protective jacket is also set with, is persistently purged as blowing medium using compressed air, it is ensured that camera lens
It is clean.
Imaging device 1 and lighting device 2 are connected with external source of the gas and power supply by field control case 3, by scene control
Case processed 3 is used to be that imaging device 1 and lighting device 2 provide stabilized power source and source of the gas；Meanwhile, field control case 3 is controlled with being located at
Indoor indoor control cabinet 6 is connected by optical fiber, correspondingly, is provided with termination box and network optical transmitter and receiver in indoor control cabinet 6；
Indoor control cabinet 6, industrial computer 4 and display device 5 pass sequentially through cable connection；Indoor control cabinet 6 receives existing
The image information of control cabinet 3 transmission is simultaneously sent to industrial computer 4 and carries out Image Information Processing, obtains the granularity and block of material
Number information the actual material variety of Auto-matching；5 real-time displaying scene furnace charge of display device transports image information and industry calculating
The data message result that the process of machine 4 is obtained；Acoustic-optic alarm 8 is connected with industrial computer 4 by signal adapter 7, works as skin
When appearance and size with occurring material on machine is more than Process Planning definite value, warning device can send alarm signal automatically.
Wherein, industrial computer is as follows to concrete grammar that the original image that imaging device is gathered is processed.
A kind of blast furnace burden particle size distribution data on-line analysis, comprise the steps：
S1, the blast furnace burden on conveying equipment is carried out by the imaging device in above-mentioned blast furnace burden granularity monitoring system
IMAQ, obtains original image, as shown in Figure 4.
Wherein, in order to coordinate the image processing speed of industrial computer in blast furnace burden granularity monitoring system, imaging device
Picture-taken frequency be 350ms/ time.
S2, the original image to the acquisition of Jing steps S1 carry out the pretreatment of image denoising, image equalization and image smoothing successively；
Concrete process step includes：
S201, image denoising process are processed using the method for adaptive median filter, are calculated including two steps：
The first step：Calculate A1And A2Numerical value：
A1=zmed-zminFormula (1)
A2=zmed-zmaxFormula (2)
If A1＞ 0 and A2＜ 0, then carry out second step calculating,
Otherwise, increase the size of filter window；
If window size≤the S after increasemax, then repeatedly the first step is calculated,
Z is exported otherwisemed；
Second step：Calculate B1And B2Numerical value：
B1=zxy-zminFormula (3)
B2=zxy-zmaxFormula (4)
If B1＞ 0 and B2＜ 0, then export zxy, otherwise export zmed；
Wherein, ZminFor SxyThe minimum of a value of middle gray level, ZmaxFor SxyThe maximum of middle gray level, ZmedFor SxyMiddle gray level
Intermediate valueZxyIt is the gray level on coordinate (x, y), SmaxFor SxyThe maximum chi of permission
It is very little, B1=zxy-zmin, B2=zxy-zmax；Real image is solved by image denoising process causes image matter due to noise jamming
The problem that amount declines, effectively improves picture quality, increases signal to noise ratio, preferably embodies the information entrained by original image, to the greatest extent
The details of the reservation original image more than possible, is that follow-up image procossing is laid a good foundation；
S202, image equalization process are according to formula：
Carry out, wherein, rkIt is k-th gray level in original image, skIt is rkGray value after transformed, T (rk) are change
Change function, Pr(rk) for gray level in original image be rkPixel occur probability, njFor the pixel count of kth level gray scale, N is
The sum of pixel in piece image；It is that the pixel grey scale to original image does to the purpose that original image carries out image equalization process
Mapping transformation, is uniformly distributed the probability density of the gradation of image after conversion, increases the dynamic range of gradation of image, reaches raising
The purpose of the contrast of image；
S203, picture smooth treatment adopt Gaussian smoothing function, according to formula：
Carry out, wherein, (x, y) is the coordinate of image, and g (x, y) is the gray scale of image slices vegetarian refreshments (x, y) after smoothing
Value, σ is Gaussian filter width (the bigger smoothness of σ is better)；The purpose of the picture smooth treatment is to extract effective mesh
Before mark remove image in some little details, bridge joint straight line or curve gap, suppress image in interference high frequency into
Point, the gentle gradual change of brightness of image is made, is reduced mutation gradient, is improved picture quality.
S3, image enhaucament and image sharpening are carried out successively to the image Jing after the process of step S2 process；Concrete process step
Including：
S301, image enhancement processing are according to formula：
G (x, y)=| f (x, y)-mean | × factor+orig formulas (7)
Carry out, wherein, f (x, y) is the gray value of original image pixels point (x, y), and g (x, y) is the image after smooth
The gray value of pixel (x, y), mean are the average gray values of entire image, and factor is to strengthen coefficient, and orig is compensation system
Number；The purpose for carrying out image enhancement processing is the contrast for adjusting image so that in image, dark region is enhanced, bright area
Domain is weakened, and darker or brighter region be conditioned it is more obvious, be finally reached balance brightness of image and contrast mesh
's.
S302, image sharpening process are according to formula：
Carry out, wherein, g (x, y) is the gray value of the pixel (x, y) after image enhaucament,
G [f (x, y)]=| f (x, y)-f (x+1, y+1) |+| f (x+1, y)-f (x, y+1) |, i.e. pixel (x, y) place
Grad, f (x, y) are gray value of the original image at pixel (x, y) place, and A is image sharpening coefficient, and T is image sharpening threshold
Value.Edge of the pixel in image is thought when G [f (x, y)] is more than or equal to threshold value T, image sharpening is added to result
Coefficient A, makes image border brighten；The pixel is thought when G [f (x, y)] is less than threshold value for similar pixel, that is, be all object or
It is all background；The border of image using this processing method both blast, while and remain the original state of image background, than
Traditional Grads Sharp has more preferable enhancing effect and applicability；Above-mentioned image sharpening is processed and eliminates image smoothing pretreatment institute
The ill-defined impact for bringing so that the edge contour of image is apparent from, it is easy to recognize, i.e., as shown in Figure 5.
S4, to Jing step S3 process after image using adaptive threshold binarization method obtain can recognize that image
In all of discrete material block image, i.e., be uneven using surface of material thus must between each material block under illumination condition
The characteristics of so having shadow region to produce, carries out image procossing, obtains chequered with black and white binary image as shown in Figure 6 (wherein,
White is material block, and black is gap between material block and material block)；
By the binarization method of adaptive threshold, i.e., according to the gamma characteristic of image, image is divided into into background (material)
With target (gap) two parts；When the inter-class variance between background and target is bigger, the two-part difference of pie graph picture is illustrated
It is bigger, when partial target mistake is divided into background or part background mistake is divided into target and can all cause two-part variance to diminish, therefore
The segmentation for making inter-class variance maximum means that the probability of misclassification is minimum；Concrete process step for (9) according to the following equation～
(16) calculated successively：
N1+N2=M × N formulas (11)
ω1+ω2=1 formula (12)
μ=μ1×ω1+μ2×ω2Formula (13)
G=ω1×(μ-μ1)2+ω2×(μ-μ2)2=ω1×ω2×(μ1-μ2)2Formula (14)
Threshold=maxJ=0,1.....255 gjFormula (15)
Wherein, M × N is the size of image, N1It is number of pixels of the gray value less than threshold value, ω in image1It is that gray value is little
The ratio of entire image, μ are accounted in the number of pixels of threshold value1Gray value less than threshold value pixel average gray, N2In being image
Number of pixels of the gray value more than threshold value, ω2It is that number of pixels in image more than threshold value accounts for the ratio of entire image, μ2It is ash
Average gray of the angle value more than the pixel of threshold value, g is inter-class variance, threshold be between the class for traveling through gray value 0～255 most
Big variance, i.e. self-adaption binaryzation threshold value, f (x, y) is gray value of the original image in pixel (x, y), g (x, y) be through
Gray value of the image after self-adaption binaryzation process in pixel (x, y).
S5, the binary image that Jing step S4 process is obtained is corroded and dilation operation process, i.e., carried using edge
Take technology to separate the discrete material block for overlaping, and the opening operation by morphological image and closed operation separation of material
Border, then removes the less interference of area by area filtration method big with what miscellaneous point, identification and removal were connected with image boundary
Area white space, obtains all effectively discrete material block image, as shown in Figure 7；Concrete process step includes：
In morphological image, " the spectral window that function as in signal transacting of the structural element in morphological transformation
Mouthful ".With B (x) representative structure elements, erosion operation process is carried out to the every bit x in working space E；
Erosion operation presses formula：
Carry out；It is exactly making B be contained in all of E after structural element B translations to the result that E carries out erosion operation with B (x)
The set that point is constituted；
Dilation operation presses formula：
Carry out, be exactly the common factor non-NULL that B and E are made after structural element B translations to the result that E carries out dilation operation with B (y)
Point constitute set.The above-mentioned process for expanding afterwards of first corroding is referred to as opening operation, and the process for first expanding post-etching is referred to as closed operation.
Opening operation is first carried out in this step S5 and tiny debris is eliminated with enough, in very thin place's separation of material, smooth the border of larger thing block；
Closed operation is carried out again with minuscule hole in fill material, the neighbouring thing block of connection and smooth material border.
S6, the quantity to the material block in the image that obtains Jing after the process of step S5 judge with the presence or absence of material block and
The species of material block；Specifically, the species of blast furnace burden is generally divided into two kinds：Coke and ore；Wherein, ore is by sintering
Various ore compositions such as ore deposit, lump ore, pellet, thus coke and two class material of ore are visually by the big of material block
It is little just to distinguish；When being judged, as the image of imaging device collection is respectively provided with identical field range, therefore can be with
Judged according to the number of material block in every image, i.e., the number of the material block in the range of the bigger same field of view of material block is more
It is few, conversely, the number of the material block in the range of the less same field of view of material block is more；Concrete determination methods are as follows：
Two kinds of material variety judgment thresholds are set：Coke decision threshold YJ(block) and ore decision threshold YK(block), it is assumed that when
Material in front picture is Y (block), then：
As Y ＜ YJ, then no material block in image, image of the image for conveyer belt；
Work as YJ≤Y≤YK, then the material block that IMAQ is arrived is coke；
As Y ＞ YK, then the material block that IMAQ is arrived is ore；
Material variety judgement is carried out to every image Jing after the process of step S5 based on above-mentioned determination methods, after the completion of judgement
Following step S7 and step S8 will be carried out, and classification storage will be carried out to image.
S7, the calculating that granularity, particle mean size and block number are carried out to the material block in the image Jing after the process of step S5；Tool
Body computational methods are respectively：
The Granular Computing Methods of S701, material block：By the area to all effectively discrete material blocks in image, by face
Product conversion calculates effective diameter；
Circular is：In acquisition image after each material block edge, find out and can cover material block boundary maximum
The minimal closure circumscribed circle of extended region whole pixel, the minimal closure circumscribed circle should contain material block boundary maximum extension
Region whole pixel；As the diameter D of minimum circumscribed circle is directly proportional to the size of particle, by actual photographed scope and phase
The conversion of machine pixel ratio, you can calculate the actual diameter size of the material block, as shown in Figure 8；Specifically, material block diameter
Pixel value size is according to formula：
Carry out；The actual diameter size of material block is according to formula：
Carry out；Wherein, DxIt is the pixel value size of the material block diameter for calculating in the picture, SxIt is to count in the picture
The pixel value size of the external area of a circle of minimal closure of the material block of calculation, DsIt is the actual diameter size dimension of material block, FsIt is phase
The actual laterally visual field of image captured by machine, fxIt is the lateral resolution of image；
The computational methods of S702, particle mean size：
Jing after the diameter that previous step obtains each piece of material block, with reference to coefficient correlation π according to formula：
Carry out, calculate particle mean size l of whole material blocks in each image, i.e., all materials block in every single image
Particle mean size；Wherein, K is the number of effective material block in an image；riIt is the actual diameter size chi of i-th piece of material block
It is very little；tempriBe i-th piece of material block actual diameter corresponding to grain size intervals average, it is assumed that grain size intervals be (D1,D2),
The average of so grain size intervals is：
S703, material number of blocks computational methods：
During step S601 is carried out, after the edge of a material block in each pair image is obtained, one is carried out
Secondary counting operation, then by being circulated counting to each material block edge acquisition process in whole image, that is, obtain per
The sum of the whole material blocks in picture.
S8, the particle mean size of the block of material by the gross that produces in balling disk (-sc) and size distribution are calculated：
After batch materials enter blast furnace by conveyer belt completely, all original image weights that the batch materials correspondence is obtained
The particle mean size and size distribution of the batch materials block are calculated after the processing procedure of multiple above-mentioned steps S2～S7；Specifically,
Judge whether that material is all by conveyer belt by the gross using the decision method of step S6；Specifically, when in the image of collected by camera
When there are continuous more than 8 to be judged as belt, then assert that material, all by conveyer belt, proceeds by the corresponding of step S8 by the gross
Calculate.
Circular is as follows：
The particle mean size of S801, by the gross material block is calculated：
Averaged by material block particle mean size l of all original images to collecting in certain period of time, you can
Obtain particle mean size L of material block in the time period.Concrete calculating presses formula：
Carry out；Wherein, L is the particle mean size of material in the time period；N is opening for the picture that material is included in the time period
Number；liIt is the particle mean size of material in the i-th pictures.
S802, size distribution are calculated：
The area of each the material block in all original images in the above-mentioned time period is added to into its diameter correspondence successively
Size distribution it is interval in and counted respectively, finally by material area summation in the granularity each distributed area and this
The ratio of whole material area summations in time period, you can calculate the percentage shared by size distribution interval.It is concrete to calculate
Following formula are pressed successively：
Carry out；Wherein, S is the summation of whole material areas in the time period, SiIt is the thing in correspondence size distribution interval
The summation of charge level product, N2It is the interval division number of size distribution, sjIt is the face of the interval interior each block of material of correspondence size distribution
Product, N1Be correspondence size distribution it is interval in material block quantity, periIt is the percentage shared by size distribution interval；By right
The size distribution situation of material is calculated by the gross, can be understood the production distribution situation of material in balling disk (-sc) in real time, be instructed thing
The material producer adjusts production technology in time, improves balling-up qualification rate, is finally reached the purpose of improve production efficiency, reduces cost.
Foreign matter alarm threshold is preset with industrial computer, when the particle that industrial computer calculates above-mentioned material block it is average
When granularity exceedes foreign matter alarm threshold, industrial computer exports abnormal data to signal adapter, starts Jing after signal conversion
Acoustic-optic alarm, acoustic-optic alarm sound the alarm sound.
Claims (10)
1. a kind of blast furnace burden granularity monitoring system, it is characterised in that including the imaging device (1) being arranged on above conveying equipment
With the lighting source (2) positioned at the imaging device both sides；The lighting source (2) less than the imaging device 0.4～
To arranging diagonally downward at 0.5m；40000 ± the 5%LM of luminous flux of the lighting source (2), colour temperature are 5500～6000K.
2. blast furnace burden granularity monitoring system according to claim 1, it is characterised in that the light of the lighting source (2)
Angle between beam center and vertical direction is 30～60 °.
3. blast furnace burden granularity monitoring system according to claim 1, it is characterised in that the imaging device is included at least
One industrial camera, the level interval between the adjacent industrial camera are 0.5～1m；Per industrial camera described in platform
Front end is provided with camera lens, and the industrial camera and the camera lens are arranged vertically downward.
4. blast furnace burden granularity monitoring system according to claim 1, it is characterised in that also cover on the outside of the imaging device
Equipped with gas sweeping type protective jacket, blowing medium is compressed air.
5. the blast furnace burden granularity monitoring system according to any one of claims 1 to 3, it is characterised in that also include for
Imaging device (1) and lighting device (2) provide the field control case (3) of stabilized power source and are arranged on the indoor control system of control
System；The control system include analyzing and processing video image industrial computer (4), for show real-time image acquisition and
The display device (5) of data, the room for controlling field control case (3), industrial computer (4) and display device (5) after process
Interior control cabinet (6), acoustic-optic alarm (8) and the signal being arranged between industrial computer (4) and acoustic-optic alarm (8)
Converter (7).
6. a kind of blast furnace burden particle size distribution data on-line analysis, it is characterised in that comprise the steps：
S1, IMAQ is carried out by imaging device to the blast furnace burden on conveying equipment, obtain original image；
It is flat that S2, the original image to step S1 acquisition described in Jing carry out image denoising process, image equalization process and image successively
It is sliding to process；
S3, step S2 described in is processed after image carry out image enhaucament and image sharpening successively and process；
S4, to described in step S3 process after image using adaptive threshold binarization method obtain can recognize that image
In all of discrete material block image；
S5, the binary image that step S4 process described in Jing is obtained is corroded and dilation operation process；
S6, the quantity to the material block in the image that obtains Jing after the process of step S5 are judged with the presence or absence of material block and material
The species of block；
S7, the image Jing after the process of step S5 is carried out by material block carries out the calculating of granularity, particle mean size and quantity；
S8, after judging that material block passes through completely by the gross, the whole original images of the collection during this are repeated into above-mentioned step
The processing procedure of rapid S2～S7, and calculate the particle mean size and size distribution of material block by the gross.
7. blast furnace burden particle size distribution data on-line analysis according to claim 6, it is characterised in that to the original
The image denoising of beginning image is processed, image equalization is processed and the concrete grammar of picture smooth treatment is：
The method that S201, described image denoising adopt adaptive median filter；
S202, described image equilibrium treatment are according to formula：
Carry out, wherein, rkIt is the kth in original image
Individual gray level, skIt is rkGray value after transformed, T (rk) for change function, Pr(rk) for gray level in original image be rk's
The probability that pixel occurs, njFor the pixel count of kth level gray scale, N is the sum of pixel in piece image；
S203, described image smoothing processing are according to formula：Carry out, wherein, (x, y) is image
Coordinate, g (x, y) be through smoothing after image slices vegetarian refreshments (x, y) gray value, σ is the width of Gaussian filter.
8. blast furnace burden particle size distribution data on-line analysis according to claim 6, it is characterised in that to the step
The image that rapid S2 is processed carries out image enhaucament and the concrete grammar of image sharpening process is：
S301, image enhancement processing are according to formula：G (x, y)=× factor+orig is carried out | f (x, y)-mean |, wherein, f
(x, y) is the gray value of original image pixels point (x, y), and g (x, y) is the gray value of image slices vegetarian refreshments (x, y) after smoothing,
Mean is the average gray value of entire image, and factor is to strengthen coefficient, and orig is penalty coefficient；
S302, image sharpening process are according to formula：Carry out, wherein, g
(x, y) is the gray value of the pixel (x, y) after image enhaucament, G [f (x, y)]=| f (x, y)-f (x+1, y+1) |+| f
(x+1, y)-f (x, y+1) | it is the Grad at pixel (x, y) place, f (x, y) is ash of the original image at pixel (x, y) place
Angle value, A are image sharpening coefficients, and T is image sharpening threshold value；When G [f (x, y)] is more than or equal to threshold value T, the pixel is assert
Edge of the point in image, adds image sharpening coefficient A to result, so that edge brightens；When G [f (x, y)] is less than threshold value,
Assert that the pixel is similar pixel.
9. blast furnace burden particle size distribution data on-line analysis according to claim 6, it is characterised in that to described in Jing
The concrete processing method of step S4 is to be calculated as follows successively：
N1+N2=M × N；
ω1+ω2-1；
μ=μ1×ω1+μ2×ω2；
G=ω1×(μ-μ1)2+ω2×(μ-μ2)2=ω1×ω2×(μ1-μ2)2；
Threshold=maxj=0,1.....233 gj；
Wherein, M × N is the size of image, N1It is number of pixels of the gray value less than threshold value, ω in image1It is that gray value is less than threshold
The number of pixels of value accounts for the ratio of entire image, μ1Gray value less than threshold value pixel average gray, N2It is gray scale in image
Number of pixels of the value more than threshold value, ω2It is that number of pixels in image more than threshold value accounts for the ratio of entire image, μ2It is gray value
More than the average gray of the pixel of threshold value, g is inter-class variance, and threshold is most generous between the class for traveling through gray value 0～255
Difference, i.e. self-adaption binaryzation threshold value, f (x, y) is gray value of the original image in pixel (x, y), and g (x, y) is through adaptive
The image after binary conversion treatment is answered in the gray value of pixel (x, y).
10. blast furnace burden particle size distribution data on-line analysis according to claim 6, it is characterised in that step S7
In, 1) in image the pixel value size of material block diameter according to formula：Carry out；The actual diameter of material block
Size is according to formula：Carry out；Wherein, DxIt is the pixel value chi of the material block diameter for calculating in the picture
It is very little, SxIt is the pixel value size of the external area of a circle of minimal closure of the material block for calculating in the picture, DsIt is the actual straight of material block
Footpath size dimension, FsIt is the actual laterally visual field of image captured by camera, fxIt is the lateral resolution of image；2) material block is average
The computational methods of granularity l press formula：Carry out, wherein, K is effective material in an image
The number of block；riIt is the actual diameter size dimension of i-th piece of material block；tempriBe i-th piece of material block actual diameter corresponding to
Grain size intervals average, it is assumed that grain size intervals be (D1,D2), then the average of the grain size intervals
In step S8,1) particle mean size of material block by the gross is calculated and presses formula：Carry out, wherein, L is the time period
The particle mean size of interior material；N is the number of the picture that material is included in the time period；liIt is the average of material in the i-th pictures
Granularity；2) average particle size distribution of material block by the gross is calculated and press successively following formula：WithCarry out, wherein, S is the summation of whole material areas in the time period, SiIt is correspondence granularity point
The summation of the material area in cloth interval, N2It is the interval division number of size distribution, sjIt is that correspondence size distribution is interval interior each
The area of block material, N1Be correspondence size distribution it is interval in material block quantity, periIt is the percentage shared by size distribution interval
Than.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611206265.2A CN106521066A (en) | 2016-12-23 | 2016-12-23 | Blast furnace burden particle size monitoring system and distributed data on-line analysis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611206265.2A CN106521066A (en) | 2016-12-23 | 2016-12-23 | Blast furnace burden particle size monitoring system and distributed data on-line analysis method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106521066A true CN106521066A (en) | 2017-03-22 |
Family
ID=58337722
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611206265.2A Pending CN106521066A (en) | 2016-12-23 | 2016-12-23 | Blast furnace burden particle size monitoring system and distributed data on-line analysis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106521066A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108803516A (en) * | 2018-06-22 | 2018-11-13 | 李燕汝 | A kind of automatic intelligent appliance control system based on Internet of Things |
| CN108872245A (en) * | 2018-05-28 | 2018-11-23 | 李春莲 | A kind of intelligence dining table oily waste degradation method |
| CN109975301A (en) * | 2017-12-27 | 2019-07-05 | 宝山钢铁股份有限公司 | A kind of sieve breakage detection method and detection device for blast furnace |
| JP6566180B1 (en) * | 2018-04-03 | 2019-08-28 | Ｊｆｅスチール株式会社 | Particle size distribution measuring apparatus and particle size distribution measuring method |
| WO2019193971A1 (en) * | 2018-04-03 | 2019-10-10 | Ｊｆｅスチール株式会社 | Particle size distribution measurement apparatus and particle size distribution measurement method |
| CN110598577A (en) * | 2019-08-23 | 2019-12-20 | 山东钢铁股份有限公司 | Image processing and analyzing method for information mining of coke in residual iron of blast furnace hearth |
| CN111344420A (en) * | 2017-11-14 | 2020-06-26 | 株式会社Posco | Blast control device and method for blast furnace |
| CN112903722A (en) * | 2019-12-04 | 2021-06-04 | 中联重科股份有限公司 | Detection method and detection system for water content of sandstone aggregate and concrete production equipment |
| CN113148342A (en) * | 2021-03-13 | 2021-07-23 | 丁梦洁 | Mechanical control platform and method based on difference analysis |
| CN115436238A (en) * | 2022-09-22 | 2022-12-06 | 中冶南方工程技术有限公司 | Blast furnace raw fuel granularity online detection method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102827984A (en) * | 2012-09-06 | 2012-12-19 | 天津市三特电子有限公司 | Blast furnace burden granularity video detection device and blast furnace burden granularity video detection method |
| CN103136530A (en) * | 2013-02-04 | 2013-06-05 | 国核自仪系统工程有限公司 | Method for automatically recognizing target images in video images under complex industrial environment |
| CN103409572A (en) * | 2013-07-23 | 2013-11-27 | 天津市三特电子有限公司 | Apparatus and method used for radar online detection of surface shape of blast furnace burden surface |
| CN103509901A (en) * | 2012-06-25 | 2014-01-15 | 宝山钢铁股份有限公司 | Blast furnace raw fuel particle size detection method |
| CN104424458A (en) * | 2013-08-23 | 2015-03-18 | 希姆通信息技术（上海）有限公司 | Image identification method and device, network server, image identification equipment and system |
| CN105809646A (en) * | 2016-03-31 | 2016-07-27 | 北京林业大学 | Method and system for calculating pore parameters of frozen earth based on iteration best threshold method |
| CN206289269U (en) * | 2016-12-23 | 2017-06-30 | 天津市三特电子有限公司 | Blast furnace burden granularity monitoring system |
-
2016
- 2016-12-23 CN CN201611206265.2A patent/CN106521066A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103509901A (en) * | 2012-06-25 | 2014-01-15 | 宝山钢铁股份有限公司 | Blast furnace raw fuel particle size detection method |
| CN102827984A (en) * | 2012-09-06 | 2012-12-19 | 天津市三特电子有限公司 | Blast furnace burden granularity video detection device and blast furnace burden granularity video detection method |
| CN103136530A (en) * | 2013-02-04 | 2013-06-05 | 国核自仪系统工程有限公司 | Method for automatically recognizing target images in video images under complex industrial environment |
| CN103409572A (en) * | 2013-07-23 | 2013-11-27 | 天津市三特电子有限公司 | Apparatus and method used for radar online detection of surface shape of blast furnace burden surface |
| CN104424458A (en) * | 2013-08-23 | 2015-03-18 | 希姆通信息技术（上海）有限公司 | Image identification method and device, network server, image identification equipment and system |
| CN105809646A (en) * | 2016-03-31 | 2016-07-27 | 北京林业大学 | Method and system for calculating pore parameters of frozen earth based on iteration best threshold method |
| CN206289269U (en) * | 2016-12-23 | 2017-06-30 | 天津市三特电子有限公司 | Blast furnace burden granularity monitoring system |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111344420A (en) * | 2017-11-14 | 2020-06-26 | 株式会社Posco | Blast control device and method for blast furnace |
| CN109975301A (en) * | 2017-12-27 | 2019-07-05 | 宝山钢铁股份有限公司 | A kind of sieve breakage detection method and detection device for blast furnace |
| US11187637B2 (en) | 2018-04-03 | 2021-11-30 | Jfe Steel Corporation | Particle size distribution measurement apparatus and particle size distribution measurement method |
| JP6566180B1 (en) * | 2018-04-03 | 2019-08-28 | Ｊｆｅスチール株式会社 | Particle size distribution measuring apparatus and particle size distribution measuring method |
| WO2019193971A1 (en) * | 2018-04-03 | 2019-10-10 | Ｊｆｅスチール株式会社 | Particle size distribution measurement apparatus and particle size distribution measurement method |
| CN111954800A (en) * | 2018-04-03 | 2020-11-17 | 杰富意钢铁株式会社 | Particle size distribution measuring apparatus and particle size distribution measuring method |
| RU2750581C1 (en) * | 2018-04-03 | 2021-06-29 | ДжФЕ СТИЛ КОРПОРЕЙШН | Device for measuring particle size distribution and method for measuring particle size distribution |
| CN111954800B (en) * | 2018-04-03 | 2024-05-31 | 杰富意钢铁株式会社 | Particle size distribution measuring device and particle size distribution measuring method |
| CN108872245A (en) * | 2018-05-28 | 2018-11-23 | 李春莲 | A kind of intelligence dining table oily waste degradation method |
| CN108803516A (en) * | 2018-06-22 | 2018-11-13 | 李燕汝 | A kind of automatic intelligent appliance control system based on Internet of Things |
| CN110598577A (en) * | 2019-08-23 | 2019-12-20 | 山东钢铁股份有限公司 | Image processing and analyzing method for information mining of coke in residual iron of blast furnace hearth |
| CN112903722A (en) * | 2019-12-04 | 2021-06-04 | 中联重科股份有限公司 | Detection method and detection system for water content of sandstone aggregate and concrete production equipment |
| CN112903722B (en) * | 2019-12-04 | 2022-11-01 | 中联重科股份有限公司 | Detection method and detection system for water content of sandstone aggregate and concrete production equipment |
| CN113148342A (en) * | 2021-03-13 | 2021-07-23 | 丁梦洁 | Mechanical control platform and method based on difference analysis |
| CN115436238A (en) * | 2022-09-22 | 2022-12-06 | 中冶南方工程技术有限公司 | Blast furnace raw fuel granularity online detection method and system |
| CN115436238B (en) * | 2022-09-22 | 2023-11-21 | 中冶南方工程技术有限公司 | Blast furnace raw fuel granularity online detection method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106521066A (en) | Blast furnace burden particle size monitoring system and distributed data on-line analysis method | |
| CN107063946B (en) | Video green ball particle diameter instrument and green ball particle diameter state detection method thereof | |
| CN109598715B (en) | Material granularity online detection method based on machine vision | |
| CN111047555A (en) | Ore image granularity detection algorithm based on image processing technology | |
| CN104867159A (en) | Stain detection and classification method and device for sensor of digital camera | |
| CN109472261B (en) | Computer vision-based automatic monitoring method for grain storage quantity change of granary | |
| RU2376580C2 (en) | Method and device for determining parametres of particles and working performance of processor in bituminous coal and minerals processing system | |
| CN109785378B (en) | Online ore granularity detection equipment based on atlas image algorithm analysis technology | |
| CN109655466A (en) | A kind of spoil coal carrying rate online test method and device based on machine vision | |
| CN108719424A (en) | A kind of aquatic products sorting technique and system based on machine vision | |
| CN103364315B (en) | On-line detection method and detection device for particle size of sintered solid fuel | |
| CN100596330C (en) | Intelligent identification method of metallurgical mine phase | |
| CN113552028B (en) | Mixture visual analysis device, and method and system for detecting granularity composition of mixture | |
| CN103047165A (en) | Control method and control system for dust removal fan of smelting furnace | |
| O'Brien et al. | Coal characterisation by automated coal petrography☆ | |
| CN107314957A (en) | A kind of measuring method of rock fragmentation Size Distribution | |
| CN112419397B (en) | Ore granularity grading method and system based on image and deep neural network | |
| CN113887308A (en) | Blast furnace charge identification method, device and system based on image multivariate features | |
| CN115131353B (en) | Flat screen printing textile production abnormity identification and positioning method and system | |
| CN107220946B (en) | Real-time removing method for bad block degree images on rock conveyer belt | |
| CN112330653A (en) | Online ore granularity detection method based on image recognition technology | |
| CN112561875A (en) | Photovoltaic cell panel coarse grid detection method based on artificial intelligence | |
| CN111222360B (en) | Method, equipment and storage medium for detecting molten state of silicon material | |
| CN113902648A (en) | Coal gangue identification method based on MATLAB image processing technology | |
| Yang et al. | Application of highlight removal and multivariate image analysis to color measurement of flotation bubble images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170322 |