CN103409569B - Method for controlling airflow distribution by utilizing furnace top infrared image - Google Patents
Method for controlling airflow distribution by utilizing furnace top infrared image Download PDFInfo
- Publication number
- CN103409569B CN103409569B CN201310384780.XA CN201310384780A CN103409569B CN 103409569 B CN103409569 B CN 103409569B CN 201310384780 A CN201310384780 A CN 201310384780A CN 103409569 B CN103409569 B CN 103409569B
- Authority
- CN
- China
- Prior art keywords
- image
- blast furnace
- furnace roof
- flow
- air
- 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
Abstract
The invention relates to a method for controlling airflow distribution by utilizing a furnace top infrared image. the method comprises the following steps of: (1), obtaining and saving the furnace top infrared image; (2), identifying the mode of the furnace top infrared image; (3), establishing the corresponding relationship among centre airflow distribution, edge airflow distribution and image features; (4), judging current airflow distribution by utilizing a real-time image; (5), according to the conditions of the blast furnace, judging whether airflow distribution of a blast furnace needs to be adjusted, if necessary, adjusting airflow distribution, and carrying out the step (6); and (6), distributing materials of the blast furnace, wherein the main purpose is to control airflow change; in addition, according to the purpose of adjustment, a proper material distribution matrix is executed through a furnace top material distribution PLC (Programmable Logic Controller), so that the purpose of adjusting the furnace top coal airflow distribution of the blast furnace is reached. According to the method, control of furnace top airflow distribution is convenient; furthermore, the blast furnace can run more stably.
Description
Technical field
The present invention relates to blast furnace roof infrared imagery technique field, refer to a kind of method utilizing furnace roof infrared image to control air flow method particularly.
Technical background
Guarantee the stable of blast furnace and direct motion, just need the operation adjusting blast furnace according to the change of the working of a furnace in time, the upper and lower part of the to the effect that blast furnace of blast furnace operating is adjusted, and upper blast furnace adjustment has mainly been come by blast furnace material distribution adjustment air flow method.
At present, a lot of blast furnace needs by means of furnace roof Infrared Image Information when carrying out cloth (furnace charge layout) and adjusting, furnace roof infrared image is along with the change of the working of a furnace is changing at any time, people is difficult to the bulk information processed at short notice with " noise ", if infrared image can be carried out necessary process, infrared image just can be utilized accurately to judge the distribution of blast furnace air-flow, thus for blast furnace cloth adjust foundation is accurately provided.Realize this target, need to solve following problem, a problem selects effective image: because blast furnace is that a quantity-produced is huge to flow reactor, furnace roof infrared information is continually varying, needs automatically to select representational air flow method figure from continually varying furnace roof infrared image.Another problem is process infrared image: process the furnace roof infrared image selected, and becomes exponentiate, can for blast furnace operating personnel parameter information directly.Another problem is application Infrared Image Information: according to the requirement of blast furnace operating, by the rule changed between accumulation furnace roof blast furnace operating parameter and furnace roof infrared image, the technology of Land use models identification, the changes in distribution of blast furnace air-flow can be assessed at any time, and automatically select suitable cloth pattern with reference to former blast furnace adjustment experience, the closed loop really realizing cloth in blast furnace operating is adjusted.
Following several image processing method is generally had at present for furnace roof infrared image, a kind of is directly carry out Continuous Play to the image obtained, retain relevant video recording if desired, one carries out coloring treatment to image, by other image-region of different grey-scale is composed with distinct colors, different temperature ranges is represented with different colors, by the conversion of this gray scale and color, the temperature distributing characteristic of blast furnace roof can be shown more intuitively, because furnace roof image is in consecutive variations process, this simple image conversion is difficult to reflection change in process information, real temperature distribution information can not be obtained.Document " Chen Lingkun, " China's steel industry nd Annual Meeting collection ", 2007, Chengdu " in inquired into a kind of technology that image continuous-changing features is identified, the method with per hour be a time period, within every 6 seconds, catch piece image, per hourly obtain 600 width images, by adding up the light and shade distribution characteristics of 600 width images, the feature of image change in different time sections is expressed with curve, utilize this feature can reflect the feature of variations in flow, by contrasting the difference of image continuous-changing features in two time periods, just can assess the trend of furnace roof variations in flow, adjust for cloth and foundation is provided.Actual furnace roof infrared image is a kind of synthetic image, every piece image reflects the variations in flow of blast furnace center and peripheral all simultaneously, but, above-mentioned treatment process processes entire image, do not distinguish edge and central gas stream to the contribution of Computer image genration, be difficult to the relative change really reflecting center and change air-flow.
The method of existing process furnace roof infrared image, no matter be direct staining method, or is all difficult to the consecutive variations situation of actual response center, edge airflow to the pattern-recongnition method of entire image.
Summary of the invention
Object of the present invention is exactly to provide a kind of method utilizing furnace roof infrared image to control air flow method, the method achieves the accurate control of operation of blast furnace state.
For realizing this object, the furnace roof infrared image that utilizes designed by the present invention controls the method for air flow method, and it is characterized in that, it comprises the steps:
Step 1: by being arranged on the infrared temperature distributed image of the noctovisor picked-up furnace roof charge level of blast furnace roof, and the infrared temperature distributed image of the furnace roof charge level of picked-up is kept in the computer of infrared image supervisory system;
Step 2: feature extraction process is carried out to the infrared temperature distribution of furnace roof charge level, thus obtain circular image corresponding to blast furnace roof airflow state, the brightness of image of principle corresponding to the large blast furnace temperature area of air-flow of process is high, the brightness of image corresponding to low furnace temperature region that air-flow is little is low, then circular image is divided into three parts of same core, in the middle of described circular image, roundlet represents blast furnace roof central gas stream, in the middle of circular image, endless belt represents the air-flow endless belt that in the middle of blast furnace roof, ventilation property is poor, and circular image outer shroud represents the air-flow at blast furnace roof edge;
Step 3: filtering process is carried out to circular image, the noise superposed to circular image with a large amount of dusts removed owing to existing in blast furnace gas;
Step 4: after filtering process, obtain the consecutive variations image of for some time, then in above-mentioned continually varying image, respectively above-mentioned circular image is carried out to the process of grey level histogram with three parts of core, again all circular image are all carried out to the pattern recognition of image with three subregions of core, the characteristic image in each region is found out, to form three typical image storehouses, the change procedure that the typical image in typical image storehouse reflects blast furnace roof central gas stream, middle endless belt air-flow, edge airflow grow from weak to strong; The acquisition process of above-mentioned characteristic image is: the process first each width circular image being carried out to histogram distribution, then the classification samples storehouse be made up of above-mentioned all histograms is set up, then be standard with histogram distribution, in all histograms processed, find out histogrammic distribution pattern, these images corresponding to histogram distribution pattern are exactly characteristic image;
Step 5: the brightness of image corresponding to the blast furnace temperature area that air-flow is large is high, the technology characteristics that the brightness of image corresponding to low furnace temperature region that air-flow is little is low, set up blast furnace roof central gas stream, blast furnace roof middle air flow endless belt, corresponding relation between blast furnace roof edge airflow and typical image: in fact the image in typical image storehouse corresponds to the air-flow of varying strength, the distribution pattern of the N analogous column diagram obtained in step 4, represent the picture pattern that the brightness of N class strengthens gradually, the picture pattern that the brightness of above-mentioned N class strengthens gradually is as the foundation of carrying out Images Classification;
Step 6: the picture pattern strengthened gradually using the brightness of above-mentioned N class is as standard, the all images gathered in the current M time are classified, obtain the quantity of every class image in endless belt, blast furnace roof edge in the middle of blast furnace roof center, blast furnace roof respectively, then be X-coordinate with brightness, with the quantity of each class image for ordinate zou, obtain a curve, this curve represents the frequency that in the M time, dissimilar image occurs, this frequency reflects the difference of variations in flow, and the curve in different time sections just corresponds to the variations in flow distribution in the corresponding time period;
Step 7: whether normally judge whether the blast furnace air-flow change profile needing to obtain in set-up procedure 6 according to operation of blast furnace situation, if operation of blast furnace situation is normal, then maintains existing operating duty, if operation of blast furnace situation is abnormal, then enters step 8;
Step 8: arrange PLC(Programmable Logic Controller at furnace roof furnace charge, programmable logic controller) the corresponding burden distribution matrix of middle selection, blast furnace air-flow change profile is adjusted, makes operation of blast furnace situation reach normal.
Further, whether described operation of blast furnace situation is normal is judged by slag crust degree of stability, blanking situation and pressure surge situation.
Further, in the process adjusted blast furnace air-flow change profile in described step 8, the real-time change distribution of blast furnace air-flow is obtained by step 1 ~ 6.
Further, in described step 1, the infrared temperature distributed image of furnace roof charge level is BMP(Bitmap, image text form) bitmap format, resolving power is 726*540.
Further, bayes method is adopted to carry out filtering process to circular image in step 3.
Further, in described step 5 and step 6, histogrammic distribution pattern is 25 or 36 classes.
Further, in described step 6, the M time is one hour.
Further, described slag crust degree of stability is weighed by the fluctuation of cooling stave temperature, when the fluctuation range of cooling stave temperature be error between 80 ~ 120 DEG C, every batch blanking interval within 1 minute and pressure surge be less than above-mentioned three conditions of 0.006Mpa meet simultaneously time, corresponding operation of blast furnace situation is normal, when above-mentioned three conditions have one not meet, corresponding operation of blast furnace situation has fluctuation.
The shortcoming processed entire image before instant invention overcomes, differentiation edge and central gas stream, to the contribution of Computer image genration, really reflect the relative change of center and change air-flow.The image in the image of real-time grasp shoot and image library is utilized to mate, air flow method feature in for some time can be obtained, this air flow method feature is expressed with the form quantized and is adjusted for blast furnace material distribution, facilitate the control of furnace roof air flow method, and make operation of blast furnace more stable.
Accompanying drawing explanation
Fig. 1 is the fundamental diagram of the furnace roof infrared image measuring system that the present invention is based on.
Embodiment
Below in conjunction with specific embodiment, the present invention is described in further detail:
Blast furnace situation: the blast furnace of enforcement is 3200m
3, furnace roof is provided with furnace roof infrared image detection module, infrared image processing module, gas flow optimized module (cloth PLC), and wherein furnace roof infrared image detection module has the feature that can detect low temperature distribution, and easy to maintenance, Performance comparision is superior.System hardware adopts import industrial computer and high-quality large color screen watch-dog, what run in computer is high-quality import industrial control computer mainboard, various special functional interface board is the plug and play interface model of the computer compatibility that particular design specialization manufactures, wherein thermometric is touched high-speed cruising on plate and multinomial functional operation formula, and the on-site data gathering equipment of special specialized particular design manufacture has high-precision accuracy of detection and bionical environmental adaptation and self-holding ability.Camera gun and blast furnace adopt to seal and install and mode of connection, are provided with high withstand voltage stainless steel draught excluder, reliably can close open holes when installing or overhaul, and occur to reveal or breathe freely to avoid blast furnace.Furnace roof camera system technical feature refers to: sweep velocity: 50/s; Minimal illumination: 0LUX/1.2F; Operating distance: 0.2m ~ 10m; Video recording mode: instant on-line automatic video recording; There is playback; Adopt 12VDC voltage of supply, directly provided by computer; Gaseous tension is 0.10MPa; Gas consumption <3m
3/ h, adjustable; Adopt all steel seal cartridge; Construction opening is of a size of 12cm.
Above-mentioned blast furnace utilizes furnace roof infrared image to control the method for air flow method, mainly comprises the steps:
Step 1: by being arranged on the infrared temperature distributed image of the noctovisor picked-up furnace roof charge level of blast furnace roof, and the infrared temperature distributed image of the furnace roof charge level of picked-up is kept in the computer of infrared image supervisory system;
Step 2: feature extraction process is carried out to the infrared temperature distributed image of furnace roof charge level, thus obtain circular image corresponding to blast furnace roof airflow state, the brightness of image of principle corresponding to the large blast furnace temperature area of air-flow of process is high, the brightness of image corresponding to low furnace temperature region that air-flow is little is low, then circular image is divided into three parts of same core, in the middle of described circular image, roundlet represents blast furnace roof central gas stream, in the middle of circular image, endless belt represents blast furnace roof middle air flow endless belt, and circular image outer shroud represents the air-flow at blast furnace roof edge;
Step 3: filtering process is carried out to circular image, the noise superposed to circular image with a large amount of dusts removed owing to existing in blast furnace gas;
Step 4: after filtering process, obtain the consecutive variations image of for some time, then in above-mentioned continually varying image, respectively above-mentioned circular image is carried out to the process of grey level histogram with three parts of core, again all circular image are all carried out to the pattern recognition of image with three subregions of core, the characteristic image in each region is found out, to form three typical image storehouses, the change procedure that the typical image in typical image storehouse reflects blast furnace roof central gas stream, middle endless belt air-flow, edge airflow grow from weak to strong; The acquisition process of above-mentioned characteristic image is: the process first each width circular image being carried out to histogram distribution, then the classification samples storehouse be made up of above-mentioned all histograms is set up, then be standard with histogram distribution, in all histograms processed, find out histogrammic distribution pattern, these images corresponding to histogram distribution pattern are exactly characteristic image;
Step 5: the brightness of image corresponding to the blast furnace temperature area that air-flow is large is high, the technology characteristics that the brightness of image corresponding to low furnace temperature region that air-flow is little is low, set up blast furnace roof central gas stream, blast furnace roof middle air flow endless belt, corresponding relation between blast furnace roof edge airflow and typical image: in fact the image in typical image storehouse corresponds to the air-flow of varying strength, the distribution pattern of the N analogous column diagram obtained in step 4, represent the picture pattern that the brightness of N class strengthens gradually, the picture pattern that the brightness of above-mentioned N class strengthens gradually is as the foundation of carrying out Images Classification;
Step 6: the picture pattern strengthened gradually using the brightness of above-mentioned N class is as standard, the all images gathered in the current M time are classified, obtain the quantity of every class image in endless belt, blast furnace roof edge in the middle of blast furnace roof center, blast furnace roof respectively, then be X-coordinate with brightness, with the quantity of each class image for ordinate zou, obtain a curve, this curve represents the frequency that in the M time, dissimilar image occurs, this frequency reflects the difference of variations in flow, and the curve in different time sections just corresponds to the variations in flow distribution in the corresponding time period;
Step 7: whether normally judge whether the blast furnace air-flow change profile needing to obtain in set-up procedure 6 according to operation of blast furnace situation, if operation of blast furnace situation is normal, then maintains existing operating duty, if operation of blast furnace situation is abnormal, then enters step 8;
Step 8: arrange in PLC at furnace roof furnace charge and select corresponding cloth (furnace charge layout) matrix, blast furnace air-flow change profile is adjusted, makes operation of blast furnace situation reach normal.
Above-mentioned steps 7 and 8 is the control mode of existing routine.
In technique scheme, whether described operation of blast furnace situation is normal is judged by slag crust degree of stability, blanking situation and pressure surge situation.Described slag crust degree of stability is weighed by the fluctuation of cooling stave temperature, when the fluctuation range of cooling stave temperature be error between 80 ~ 120 DEG C, every batch blanking interval within 1 minute and pressure surge be less than above-mentioned three conditions of 0.006Mpa meet simultaneously time, corresponding operation of blast furnace situation is normal, when above-mentioned three conditions have one not meet, corresponding operation of blast furnace situation has fluctuation.
In technique scheme, in the process adjusted blast furnace air-flow change profile in described step 8, the real-time change distribution of blast furnace air-flow is obtained by step 1 ~ 6.
In technique scheme, be saving resource, improve process frequency, in described step 1, the infrared temperature distributed image of furnace roof charge level is BMP bitmap format, and resolving power is 726*540.Image adopts the principle with grabbing with process to obtain.
In technique scheme, bayes method in step 3, is adopted to carry out filtering process to circular image.Due to blast furnace internal work bad environments, infrared image change is very fast, charge level temperature just changed to 500 ~ 700 DEG C from 20 ~ 30 DEG C in 2 ~ 3 minutes, the infrared image change of corresponding acquisition is also than comparatively fast, simultaneously owing to there is a large amount of dust in blast furnace gas, camera head is easy to assemble dust in the edge section of camera lens, and this has just been equivalent to noise to imaging importing, must be removed by the noise of image before image procossing.
In technique scheme, in described step 5 and step 6, histogrammic distribution pattern is 25 or 36 classes.
In technique scheme, in described step 6, the M time is one hour.
In the step 1 of technique scheme, by installing special image pick-up card on computers, writing special image acquisition procedure, just frequency on demand can arrest infrared image, image temporarily can be kept in the computer of infrared image supervisory system.
In technique scheme, described employing bayes method to the method detailed that circular image carries out filtering process is: image is regarded as a random field, the prior probability distribution P (y/x) of known degraded image, adopts maximum likelihood estimation (maximum likehood, ML):
Realize the recovery of image.
The concrete grammar setting up typical image storehouse in the step 4 of technique scheme is:
S1: initialize
Classification number K is set, the cluster centre initialize of each classification:
{z
1(l),z
2(l),…,z
k(l)}
Wherein, z
jl () represents the cluster centre value of the l time iteration, j is any one numerical value in 1 ~ k.Initial value can be arbitrary, is usually arranged to front K value of sample vector.
S2: sample divides
Divide all sample vector, make each sample vector x
(p)be associated with one of in K class, its division condition is:
X
(p)∈ S
j(l), if || x
(p)-z
j(l) || <||x
(p)-z
i(l) ||
Then for all i=1,2 ..., K, i ≠ j, wherein, S
jthe entirety of classification j when () represents the l time iteration l.
S3: calculate new cluster centre
Be used in all member set of the new class set up in S2, recalculate the central position of each class, the distance sum from each vector classification to new cluster centre can be made minimum.Here J
jrepresent the distance sum of each vector to new cluster centre:
Z
j(l+1) be make the minimized all sample S of above-mentioned formula
jthe mean value of (l).The following formula of new cluster centre calculates:
Wherein, N
jbelong to S in S2
jthe quantity of sample vector.
S4: check convergence
Determine when convergence situation changes its position again without any cluster centre in S3, mathematically can be expressed as:
z
j(l+1)=z
j(l),j=1,2,…,K
If meet above-mentioned formula, so just restrain.Otherwise return S2 and continue iteration.
Here the learning sample adopted is 1200 (random sampling), and K value is 25.After cluster, furnace roof infrared image can be divided into 25 classifications more clearlyly.
In the step 6 of technique scheme, all images gathered in current 1 hour have 600 width.
In the step 8 of technique scheme, if slag crust is seriously bonding, blanking is sometimes fast and sometimes slow, there is little lubricating substance, at this moment edge airflow, central gas stream are generally more chaotic, more weak, at this moment need loose center and peripheral, keep blast furnace two strands of air flow method, if the burden distribution matrix of normal running is
(C represents coke here, and O represents ore, ↓ represent charging, position, numeral cloth angle above, numeral is below in the cloth number of rings of position, corresponding cloth angle), then can be by above-mentioned adjustment of matrix
this matrix and
compare, add a ring coke in coke 9 bugle position, subtract a ring ore in 8 bugle positions simultaneously; If slag crust comes off comparatively large, or cooling stave lip temperature is higher, edge airflow is more prosperous, at this moment needs to increase the weight of edge, suitably decontrols center, then can be by above-mentioned adjustment of matrix
this matrix and
compare, subtract a ring coke in coke 8 bugle position, add a ring ore in 8 bugle positions simultaneously; If blanking is steady, lip temperature is stable, moderate, and central gas stream is also moderate, at this moment can maintain current operation system.Air-flow is now completed by the method for step 1 ~ 6 respectively.
The content that specification sheets is not described in detail belongs to the known prior art of professional and technical personnel in the field.
Claims (5)
1. utilize furnace roof infrared image to control a method for air flow method, it is characterized in that, it comprises the steps:
Step 1: by being arranged on the infrared temperature distributed image of the noctovisor picked-up furnace roof charge level of blast furnace roof, and the infrared temperature distributed image of the furnace roof charge level of picked-up is kept in the computer of infrared image supervisory system;
Step 2: feature extraction process is carried out to the infrared temperature distribution of furnace roof charge level, thus obtain circular image corresponding to blast furnace roof airflow state, the brightness of image of principle corresponding to the large blast furnace temperature area of air-flow of process is high, the brightness of image corresponding to low furnace temperature region that air-flow is little is low, then circular image is divided into three parts of same core, in the middle of described circular image, roundlet represents blast furnace roof central gas stream, in the middle of circular image, endless belt represents blast furnace roof middle air flow endless belt, and circular image outer shroud represents the air-flow at blast furnace roof edge;
Step 3: filtering process is carried out to circular image, the noise superposed to circular image with a large amount of dusts removed owing to existing in blast furnace gas;
Step 4: after filtering process, obtain the consecutive variations image of for some time, then in above-mentioned continually varying image, respectively above-mentioned circular image is carried out to the process of grey level histogram with three parts of core, again all circular image are all carried out to the pattern recognition of image with three subregions of core, the characteristic image in each region is found out, to form three typical image storehouses, the change procedure that the typical image in typical image storehouse reflects blast furnace roof central gas stream, middle endless belt air-flow, edge airflow grow from weak to strong; The acquisition process of above-mentioned characteristic image is: the process first each width circular image being carried out to histogram distribution, then the classification samples storehouse be made up of above-mentioned all histograms is set up, then be standard with histogram distribution, in all histograms processed, find out histogrammic distribution pattern, these images corresponding to histogram distribution pattern are exactly characteristic image;
Step 5: the brightness of image corresponding to the blast furnace temperature area that air-flow is large is high, the technology characteristics that the brightness of image corresponding to low furnace temperature region that air-flow is little is low, set up blast furnace roof central gas stream, blast furnace roof middle air flow endless belt, corresponding relation between blast furnace roof edge airflow and typical image: in fact the image in typical image storehouse corresponds to the air-flow of varying strength, the distribution pattern of the N analogous column diagram obtained in step 4, represent the picture pattern that the brightness of N class strengthens gradually, the picture pattern that the brightness of above-mentioned N class strengthens gradually is as the foundation of carrying out Images Classification;
Step 6: the picture pattern strengthened gradually using the brightness of above-mentioned N class is as standard, the all images gathered in the current M time are classified, obtain the quantity of every class image in endless belt, blast furnace roof edge in the middle of blast furnace roof center, blast furnace roof respectively, then be X-coordinate with brightness, with the quantity of each class image for ordinate zou, obtain a curve, this curve represents the frequency that in the M time, dissimilar image occurs, this frequency reflects the difference of variations in flow, and the curve in different time sections just corresponds to the variations in flow distribution in the corresponding time period;
Step 7: whether normally judge whether the blast furnace air-flow change profile needing to obtain in set-up procedure 6 according to operation of blast furnace situation, if operation of blast furnace situation is normal, then maintains existing operating duty, if operation of blast furnace situation is abnormal, then enters step 8;
Step 8: arrange in PLC at furnace roof furnace charge and select corresponding burden distribution matrix, blast furnace air-flow change profile is adjusted, makes operation of blast furnace situation reach normal;
Whether described operation of blast furnace situation is normal is judged by slag crust degree of stability, blanking situation and pressure surge situation;
Described slag crust degree of stability is weighed by the fluctuation of cooling stave temperature, when the fluctuation range of cooling stave temperature be error between 80 ~ 120 DEG C, every batch blanking interval within 1 minute and pressure surge be less than above-mentioned three conditions of 0.006MPa meet simultaneously time, corresponding operation of blast furnace situation is normal, when above-mentioned three conditions have one not meet, corresponding operation of blast furnace situation has fluctuation;
Bayes method is adopted to carry out filtering process to circular image in step 3;
Described employing bayes method to the method detailed that circular image carries out filtering process is: image is regarded as a random field, the prior probability distribution P (y/x) of known degraded image, adopt maximum likelihood estimation (maximum likehood, ML):
(wherein
corresponding maximum probability) (1)
Realize the recovery of image;
The concrete grammar setting up typical image storehouse in the step 4 of technique scheme is:
S1: initialize
Classification number K is set, the cluster centre initialize of each classification:
{z
1(l),z
2(l),…,z
k(l)}
Wherein, z
jl () represents the cluster centre value of the l time iteration, j is any one numerical value in 1 ~ k, and initial value can be arbitrary, is usually arranged to front K value of sample vector;
S2: sample divides
Divide all sample vector, make each sample vector x
(p)be associated with one of in K class, its division condition is:
X
(p)∈ S
j(l), if || x
(p)-z
j(l) || < || x
(p)-z
i(l) ||
Then for all i=1,2 ..., K, i ≠ j, wherein, S
jthe entirety of classification j when () represents the l time iteration l;
S3: calculate new cluster centre
Be used in all member set of the new class set up in S2, recalculate the central position of each class, the distance sum from each vector classification to new cluster centre can be made minimum, J here
jrepresent the distance sum of each vector to new cluster centre:
Z
j(l+1) be make the minimized all sample S of above-mentioned formula
jthe mean value of (l), the following formula of new cluster centre calculates:
Wherein, N
jbelong to S in S2
jthe quantity of sample vector;
S4: check convergence
Determine when convergence situation changes its position again without any cluster centre in S3, mathematically can be expressed as:
z
j(l+1)＝z
j(l),j＝1,2,…,K
If meet above-mentioned formula, so just restrain, otherwise returned S2 continuation iteration.
2. the furnace roof infrared image that utilizes according to claim 1 controls the method for air flow method, it is characterized in that: in the process adjusted blast furnace air-flow change profile in described step 8, the real-time change distribution of blast furnace air-flow is obtained by step 1 ~ 6.
3. the method utilizing furnace roof infrared image to control air flow method according to claim 1, it is characterized in that: in described step 1, the infrared temperature distributed image of furnace roof charge level is BMP bitmap format, resolving power is 726*540.
4. the furnace roof infrared image that utilizes according to claim 1 controls the method for air flow method, it is characterized in that: in described step 5 and step 6, histogrammic distribution pattern is 25 or 36 classes.
5. the method utilizing furnace roof infrared image to control air flow method according to claim 1, is characterized in that: in described step 6, the M time is one hour.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310384780.XA CN103409569B (en) | 2013-08-29 | 2013-08-29 | Method for controlling airflow distribution by utilizing furnace top infrared image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310384780.XA CN103409569B (en) | 2013-08-29 | 2013-08-29 | Method for controlling airflow distribution by utilizing furnace top infrared image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103409569A CN103409569A (en) | 2013-11-27 |
| CN103409569B true CN103409569B (en) | 2015-04-08 |
Family
ID=49602610
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310384780.XA Active CN103409569B (en) | 2013-08-29 | 2013-08-29 | Method for controlling airflow distribution by utilizing furnace top infrared image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103409569B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9212851B2 (en) * | 2014-02-10 | 2015-12-15 | Honeywell International Inc. | Multi resolution, hierarchical radiance field estimation |
| CN109022650B (en) * | 2018-09-29 | 2020-06-16 | 武汉钢铁有限公司 | Method for identifying distribution of upper gas flow of blast furnace |
| CN109852748B (en) * | 2019-02-27 | 2020-05-08 | 内蒙古科技大学 | Method for monitoring development process of gas flow in distribution period of blast furnace and predicting gas utilization rate |
| CN112347154A (en) * | 2020-10-23 | 2021-02-09 | 中冶南方工程技术有限公司 | Slag crust shedding index calculation method, terminal equipment and storage medium |
| CN114581785B (en) * | 2022-05-09 | 2022-07-22 | 中国科学院自动化研究所 | Multi-source heterogeneous data fusion blast furnace gas flow distribution state identification method and system |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS6013006A (en) * | 1983-07-01 | 1985-01-23 | Nippon Steel Corp | Method for controlling distribution of blast furnace charge |
| CN1844409A (en) * | 2006-03-14 | 2006-10-11 | 中南大学 | Detection method and system for temperature field and charge lever distribution based on infrared imagery |
| CN101881955B (en) * | 2009-05-06 | 2014-05-07 | 鞍钢股份有限公司 | Method for evaluating conditions of blast furnace |
-
2013
- 2013-08-29 CN CN201310384780.XA patent/CN103409569B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN103409569A (en) | 2013-11-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103409569B (en) | Method for controlling airflow distribution by utilizing furnace top infrared image | |
| US11765321B2 (en) | Intelligent video surveillance system and method | |
| CN110070074B (en) | Method for constructing pedestrian detection model | |
| CN103226088B (en) | Particulate counting method | |
| CN101456501B (en) | Method and apparatus for controlling elevator button | |
| CN109886994B (en) | Self-adaptive occlusion detection system and method in video tracking | |
| CN201084378Y (en) | An advertisement machine and network advertisement machine system | |
| CN109062196A (en) | The blast furnace process monitoring of integrated PCA-ICA a kind of and method for diagnosing faults | |
| CN113324864B (en) | Pantograph carbon slide plate abrasion detection method based on deep learning target detection | |
| CN104574389A (en) | Battery piece chromatism selection control method based on color machine vision | |
| CN105279773A (en) | TLD framework based modified video tracking optimization method | |
| CN105002321A (en) | Method for processing coal gas flow center dynamic tracking and monitoring coal gas utilization rate | |
| CN105741324A (en) | Moving object detection identification and tracking method on moving platform | |
| He et al. | Nonparametric density estimation of froth colour texture distribution for monitoring sulphur flotation process | |
| CN109033511B (en) | A kind of quality coal in cement kiln systems heat consumption analysis method of combined data driving and data mining | |
| CN104539937A (en) | Stain detection method and device for camera module | |
| CN112257590B (en) | Automatic detection method, system and storage medium for working state of blast furnace tap hole | |
| CN109189693A (en) | The method and SSD that a kind of pair of LBA information is predicted | |
| CN115841624A (en) | Blast furnace gas flow distribution identification method based on infrared image | |
| CN109598200B (en) | Intelligent image identification system and method for molten iron tank number | |
| CN106191350A (en) | Bottom house air port working condition appraisal procedure based on fixed point radar | |
| CN108399416A (en) | A kind of substation's pointer instrument automatic identifying method having circular arc boundary | |
| CN103578121B (en) | Method for testing motion based on shared Gauss model under disturbed motion environment | |
| CN103020635A (en) | Method and system for detecting sintering belt temperature in rotary kiln | |
| CN105195438A (en) | Embedded type automatic pearl sorting device and method based on image recognition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20170718 Address after: 430083, Hubei Wuhan Qingshan District Factory No. 2 Gate joint stock company organs Patentee after: Wuhan iron and Steel Company Limited Address before: 430080 Wuchang, Hubei Friendship Road, No. A, block, floor 999, 15 Patentee before: Wuhan Iron & Steel (Group) Corp. |