CN105956618B - Converter steelmaking blowing state identification system and method based on image dynamic and static characteristics - Google Patents
Converter steelmaking blowing state identification system and method based on image dynamic and static characteristics Download PDFInfo
- Publication number
- CN105956618B CN105956618B CN201610272037.9A CN201610272037A CN105956618B CN 105956618 B CN105956618 B CN 105956618B CN 201610272037 A CN201610272037 A CN 201610272037A CN 105956618 B CN105956618 B CN 105956618B
- Authority
- CN
- China
- Prior art keywords
- image
- flame
- color
- information
- pixel
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a converter steelmaking converting state recognition system and method based on image dynamic and static characteristics. Firstly, measuring and reserving pixels similar to flames to realize segmentation through an original image frame obtained by video input; and respectively extracting four types of image characteristics: and finally, classifying and identifying the input features through the established generalized regression neural network so as to achieve the functions of judging the converting stage and forecasting the end point. The invention can adapt to the transient steady-state transient problem of the flame, has the advantages of higher identification precision and low cost, can be used for steelmaking converters with different scales, improves the production efficiency and reduces the waste of raw materials.
Description
Technical Field
The invention relates to a converter steelmaking blowing state identification system and method based on image dynamic and static characteristics, and belongs to the technical field of metallurgy automation.
Background
The iron and steel industry is an important raw material industry supporting the economic development of China, and China is the largest iron and steel producing country in the world. The average share of the converter steel production in the total production of steel amounts to 70%. The end point control is a key operation in the later stage of the converter, means that the carbon content and the temperature of molten steel are controlled to meet the requirements of tapping, accurate and real-time judgment of the converter blowing end point is one of the difficulties in the steel industry, and accurate end point prediction has important significance for improving the production efficiency of a steel mill, reducing the waste of energy and raw materials and improving the quality of steel products.
The method comprises the steps of conducting blowing to about 85% in the main blowing stage of blowing, adjusting a secondary blowing strategy by a method for measuring carbon content and temperature value, using sublance detection and empirical judgment to be the most common data detection method, immersing a sublance into a molten pool for temperature measurement and carbon taking, or enabling a worker and a master worker to perform sampling detection by visual inspection and turning over, and adjusting the oxygen blowing amount and the addition amount of converter raw materials according to measured data. For the terminal judgment in the later stage of blowing, some steel enterprises adopt a photoelectric detection method for judging the terminal by the change condition of infrared laser penetrating through furnace gas, a gas analysis method for measuring the chemical components of the furnace gas and the like, detection equipment used in the methods needs to work in high-temperature and corrosive environments for a long time, the gas calibration period is short, a sampling head is frequently replaced, the use and maintenance cost of the equipment is high, and the method is difficult to popularize and use in the steel converter industry.
The sublance control has higher detection precision, but is generally used in a converter with more than 120t, which is difficult to meet the current situation of China mainly in medium and small steel mills and can not realize continuous measurement. The intelligent end point judgment method is characterized in that molten steel detection data are used as model input vectors, target molten steel components, temperature and oxygen blowing amount are used as output vectors, an end point prediction system based on models such as ELM, CBR and GMDH is established, the intelligent end point prediction method mostly utilizes actually acquired blowing data in a steelmaking process, and has better real-time performance in principle, but the problems of cost increase or untimely data acquisition and the like caused by accurate data acquisition exist in the existing methods.
With the rapid development of digital image processing technology and the improvement of computer processing performance, image recognition based on manual fire observation is rapidly developed and applied for converter endpoint judgment. For example, the existing gray level co-occurrence matrix method, the color mean value method, etc. The existing method obtains certain effect in describing static characteristics of flame, but ignores dynamic information of flame, and practical observation experience also shows that the dynamic characteristics of flame can reflect the intensity of different stages of blowing, and can be used as the reflection of carbon oxidation rate. Therefore, the static and dynamic characteristics of the flame are fused to accurately identify the converting end point, and the method has important practical value and significance for improving the production efficiency and reducing the waste of raw materials.
Disclosure of Invention
In order to solve the problems of inaccuracy, multiple blowing supplement times, low efficiency, raw material waste and the like caused by judging the end point by manual fire observation, the invention provides a converter steelmaking blowing state identification system and method based on image dynamic and static characteristics, and solves the problems that the cost of equipment for flue gas analysis, photoelectric analysis and the like is high, the dynamic characteristics are not considered in the traditional flame image identification, the identification rate is influenced and the like.
The invention is realized by the following technical scheme: the utility model provides a converter steelmaking converting state identification system based on image sound attitude characteristic, includes image acquisition module, color image flame region segmentation module, color flame image sound attitude characteristic express with describe the module, converting state classification module and host computer monitoring module based on image characteristic:
the image acquisition module is used for shooting the flame condition in real time, storing and transmitting the shot video signal to the color image flame area segmentation module;
the color image flame area segmentation module is used for segmenting the acquired video signal into separate color image frames and segmenting a flame background part and a flame body part in the color image;
the color flame image dynamic and static characteristic representation and description module is used for acquiring brightness information, chrominance information, texture information and dynamic information of an optical flow field of the flame body part obtained by segmentation;
the converting state classification module based on image characteristics is used for establishing a recognition model by taking various information collected by the dynamic and static characteristic representation and description module of the color flame image as input, and after setting an output state, collecting a data sample set to train the established recognition model; further judging the output state of the newly acquired input content, and sending the output state of the last stage of blowing entering to an upper computer monitoring module;
the upper computer monitoring module is used for sending out signal prompts.
The image acquisition module further comprises a camera, a memory and a signal transmitter.
The invention also aims to provide a converter steelmaking blowing state identification method based on image dynamic and static characteristics, which comprises the following steps:
(1) installing a color camera, keeping the position and the angle of the color camera consistent with those of a manual fire watching, and fixing parameters such as the position, the focal length and the like of the camera; the purpose of fixing the camera is to ensure that the shot color image has feature comparability between frames and accuracy of dynamic feature description; the size of the image is set to be equivalent to the perception capability of human eyes, the undersize of the image influences the effect of feature extraction, and the oversize of the image influences the algorithm processing speed; the real-time video signal collected by the color camera is connected with the processor through the signal collecting card;
(2) the processor divides the collected video signal into separate color image frames, and divides the flame background and flame body in the color image, i.e. converting RGB color space into uniform L a b space, and converting red (255,0,0) into RLabConversion of yellow (255, 0) to YLabWhite (255 ) is converted to WLabLet P be a pixel in the converted image and D be the Euclidean distance between pixels, according to the formula
The segmentation strategy is then:
in the formula, DPIThe color difference value between one pixel of the image after color space conversion and a reference point pixel is obtained; i is a pixel after the image is converted from RGB color to Lab color, and the pixel has three components, namely L, a and b; t is a threshold value for controlling the similarity;
then will be
The part less than T is used as a flame body, and the other parts are used as flame backgrounds for segmentation;
(3) using the flame body part obtained in the step (2) for collecting brightness information, chrominance information, texture information and dynamic information of an optical flow field:
collecting brightness information B:
if I (I, j) is the divided flame region pixel, the brightness characteristic B is (Σ I (I, j))/(Count), where B is the brightness characteristic, Count is the number of region pixels, and I, j is the position of the pixel, i.e., the coordinate in the image;
collecting chrominance information S:
the three-order moment of the color of each component in RGB space can effectively reflect the oxidation sequence of bath elements in the blowing period, and the three-order moment under the single component i is
In the formula, pi,jThe probability of the occurrence of a pixel with the gray level j in the ith color channel component of the color image is shown, i is the ith component representing the color image, j is the gray level representing the image, si is the color moment characteristic of the image, and mui is a related parameter;
collecting texture information ASM:
describing static texture complexity by using a gray difference statistical method, wherein (x, y) is set as a point in an image, and a gray difference value between the point (x + delta x, y + delta y) and the point (x + delta x, y + delta y) is set as gΔ(x, y) -g (x + Δ x, y + Δ y), assuming that all possible values of the gray-scale difference are m-level, moving the point (x, y) within a given flame image area, accumulating gΔThe number of times (x, y) takes different values is given as gΔHistogram of (x, y), g is known from the histogramΔThe probability of the value of (x, y) is pΔ(i)；
The uniformity degree of image gray distribution is reflected by extracting the second moment ASM in the angle direction, and if the difference of the gray values of the similar pixels is larger, the ASM value is larger, so that the texture is rougher;
the texture characteristics are related to the oxidation degree of chemical elements in a molten pool, and the more violent combustion, the higher the roughness of the texture;
collecting dynamic information Ent of the optical flow field:
in order to describe flame dynamic information in a converting process, particularly flame stroboscopic characteristics in the later period, firstly, a description model of flame dynamic change needs to be established, a flickering process is established by adopting an optical flow field, the basic assumption of optical flow calculation is that micro motion and brightness are constant, I (x, y, t) ═ I (x + dx, y + dy, t + dt) is obtained, points with constant pixel brightness are interconnected in a limited area between adjacent frames, and an interframe optical flow graph F is constructed; f records the intensity of the flame in the continuous change process, and the characteristic representation and description of F can reflect information such as stroboflash and the like;
wherein x, y refer to coordinates of points in the image, t is a time axis of the continuous image, because the video is composed of images on the continuous time axis, dx represents displacement of x axis, dy represents displacement of y axis, dt represents displacement of t axis, and I is an image frame of the video stream when the time is t moment;
for an interregional light flow graph, a gray level co-occurrence matrix is adopted to describe the characteristics of the interregional light flow graph, the direction in the calculation process is 0 degrees, 45 degrees, 90 degrees, 135 degrees and the step length step is 1 so as to adapt to the characteristic of the random micro-texture of flame, M is the obtained gray level co-occurrence matrix, the characteristics of the interregional light flow graph are described by adopting an entropy value, and the interregional light flow graph is calculated
Dynamic information of the flame;
wherein i, j is the coordinate of a pixel in the image;
(4) establishing an identification model by adopting a generalized regression neural network, taking the existing characteristic vector V (B, S, ASM, Ent) collected in the step (3) as input, and taking the state number of the converting as output: "1" represents the initial stage, "2" represents the intermediate stage, and "3" represents the final stage; the established generalized regression neural network is four-input and single-output, the hidden layer neuron excitation function is a Gaussian function, and the output is a linear mapping function; correspondingly listing the brightness information, the chrominance information, the texture information and the dynamic information of the optical flow field of the known flame body with the output data thereof, and establishing a data sample set; training the established recognition model, and classifying output data in the training according to state numbers;
(5) and (3) processing the real-time video signals acquired in the step (1) in the steps (2) and (3), inputting the processed real-time video signals as feature vectors into the recognition model trained in the step (4), wherein the output is recognition of the blowing state, and when the output number of the blowing state is 3, a signal prompt is sent out when the blowing enters the terminal stage.
The point (x + Δ x, y + Δ y) of the step (3) is a point around the value pixel (x, y) at a distance Δ x and Δ y from the value pixel (x, y).
The invention has the beneficial effects that: aiming at the defects of the existing converter blowing state identification method based on flame image identification, the invention provides a state identification scheme fusing the characteristics of dynamic and static images; the device is particularly suitable for the converter below 120t or the test environments such as smoke, sublance and the like due to limited use. The invention has the advantages that the dynamic characteristics of the flame are extracted and the static characteristics are combined to identify the converting state, the invention can adapt to the transient steady-state instantaneous problem of the flame, has the advantages of higher identification precision and low cost, can be used for steelmaking converters with different scales, improves the production efficiency and reduces the waste of raw materials. The invention solves the problems of inaccurate terminal judgment in the existing converter steelmaking process, high use and maintenance cost caused by adopting methods such as flue gas analysis, photoelectric analysis, sublance analysis and the like, and also solves the problems of large characteristic change and low accuracy rate caused by single static image analysis in the existing flame image identification method.
Drawings
FIG. 1 is a schematic structural diagram of a converter steelmaking blowing state identification system based on image dynamic and static characteristics according to the present invention;
FIG. 2 is a schematic flow chart of the method for identifying the converter steelmaking blowing state based on the image dynamic and static characteristics.
Detailed description of the preferred embodiments
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings. The embodiments described with reference to the drawings are illustrative only and should not be construed as limiting the invention.
As shown in figure 1, the converter steelmaking converting state recognition system based on image dynamic and static characteristics comprises an image acquisition module 1, a color image flame region segmentation module 2, a color flame image dynamic and static characteristic representation and description module 3, a converting state classification module 4 based on image characteristics and an upper computer monitoring module 5:
the image acquisition module 1 is used for shooting flame conditions in real time, storing and transmitting shot video signals to the color image flame area segmentation module; the image acquisition module also comprises a camera, a memory and a signal transmitter;
the color image flame area segmentation module 2 is used for segmenting the acquired video signal into separate color image frames and segmenting a flame background part and a flame body part in the color image;
the color flame image dynamic and static characteristic representation and description module 3 is used for acquiring brightness information, chrominance information, texture information and dynamic information of an optical flow field of the flame body part obtained by segmentation;
the converting state classification module 4 based on image characteristics is used for establishing a recognition model by taking various information collected by the dynamic and static characteristic representation and description module of the color flame image as input, and after setting an output state, collecting a data sample set to train the established recognition model; further judging the output state of the newly acquired input content, and sending the output state of the last stage of blowing entering to an upper computer monitoring module;
the upper computer monitoring module 5 is used for sending out signal prompts.
Referring to fig. 2, the method for identifying the converter steelmaking blowing state based on the image dynamic and static characteristics comprises the following steps:
(1) installing a color camera, keeping the position and the angle of the color camera consistent with those of a manual fire watching, and fixing parameters such as the position, the focal length and the like of the camera; the purpose of fixing the camera is to ensure that the shot color image has feature comparability between frames and accuracy of dynamic feature description; the size of the image is set to be equivalent to the perception capability of human eyes, the undersize of the image influences the effect of feature extraction, and the oversize of the image influences the algorithm processing speed; the real-time video signal collected by the color camera is connected with the processor through the signal collecting card;
(2) the processor divides the collected video signal into separate color image frames, and divides the flame background and flame body in the color image, i.e. converting RGB color space into uniform L a b space, and converting red (255,0,0) into RLabConversion of yellow (255, 0) to YLabWhite (255 ) is converted to WLabLet P be a pixel in the converted image and D be the Euclidean distance between pixels, according to the formula
The segmentation strategy is then:
in the formula, DPIThe color difference value between one pixel of the image after color space conversion and a reference point pixel is obtained; i is a pixel after the image is converted from RGB color to Lab color, and the pixel has three components, namely L, a and b; t is a threshold value for controlling the similarity;
then will be
The part less than T is used as a flame body, and the other parts are used as flame backgrounds for segmentation;
(3) using the flame body part obtained in the step (2) for collecting brightness information, chrominance information, texture information and dynamic information of an optical flow field:
collecting brightness information B:
if I (I, j) is the divided flame region pixel, the brightness characteristic B is (Σ I (I, j))/(Count), where B is the brightness characteristic, Count is the number of region pixels, and I, j is the position of the pixel, i.e., the coordinate in the image;
collecting chrominance information S:
the three-order moment of the color of each component in RGB space can effectively reflect the oxidation sequence of bath elements in the blowing period, and the three-order moment under the single component i is
In the formula, pi,jThe probability of the occurrence of a pixel with the gray level j in the ith color channel component of the color image is shown, i is the ith component representing the color image, j is the gray level representing the image, si is the color moment characteristic of the image, and mui is a related parameter;
collecting texture information ASM:
describing static texture complexity by using a gray difference statistical method, wherein (x, y) is set as a point in an image, and a gray difference value between the point (x + delta x, y + delta y) and the point (x + delta x, y + delta y) is set as gΔ(x, y) -g (x + Δ x, y + Δ y), assuming that all possible values of the gray-scale difference are m-level, moving the point (x, y) within a given flame image area, accumulating gΔThe number of times (x, y) takes different values is given as gΔHistogram of (x, y), g is known from the histogramΔThe probability of the value of (x, y) is pΔ(i) (ii) a Wherein the point (x + Δ x, y + Δ y) is a point around the value pixel (x, y) at a distance Δ x and Δ y from the value pixel (x, y);
the uniformity degree of image gray distribution is reflected by extracting the second moment ASM in the angle direction, and if the difference of the gray values of the similar pixels is larger, the ASM value is larger, so that the texture is rougher;
the texture characteristics are related to the oxidation degree of chemical elements in a molten pool, and the more violent combustion, the higher the roughness of the texture;
collecting dynamic information Ent of the optical flow field:
in order to describe flame dynamic information in a converting process, particularly flame stroboscopic characteristics in the later period, firstly, a description model of flame dynamic change needs to be established, a flickering process is established by adopting an optical flow field, the basic assumption of optical flow calculation is that micro motion and brightness are constant, I (x, y, t) ═ I (x + dx, y + dy, t + dt) is obtained, points with constant pixel brightness are interconnected in a limited area between adjacent frames, and an interframe optical flow graph F is constructed; f records the intensity of the flame in the continuous change process, and the characteristic representation and description of F can reflect information such as stroboflash and the like;
wherein x, y refer to coordinates of points in the image, t is a time axis of the continuous image, because the video is composed of images on the continuous time axis, dx represents displacement of x axis, dy represents displacement of y axis, dt represents displacement of t axis, and I is an image frame of the video stream when the time is t moment;
for an interregional light flow graph, a gray level co-occurrence matrix is adopted to describe the characteristics of the interregional light flow graph, the direction in the calculation process is 0 degrees, 45 degrees, 90 degrees, 135 degrees and the step length step is 1 so as to adapt to the characteristic of the random micro-texture of flame, M is the obtained gray level co-occurrence matrix, the characteristics of the interregional light flow graph are described by adopting an entropy value, and the interregional light flow graph is calculated
Dynamic information of the flame;
wherein i, j is the coordinate of a pixel in the image;
(4) establishing an identification model by adopting a generalized regression neural network, taking the existing characteristic vector V (B, S, ASM, Ent) collected in the step (3) as input, and taking the state number of the converting as output: "1" represents the initial stage, "2" represents the intermediate stage, and "3" represents the final stage; the established generalized regression neural network is four-input and single-output, the hidden layer neuron excitation function is a Gaussian function, and the output is a linear mapping function; correspondingly listing the brightness information, the chrominance information, the texture information and the optical flow field dynamic information of the known flame body and the output data thereof, and establishing an image video signal of not less than 10 heats as a data sample set; marking the blowing state of each image, training the established recognition model, and classifying output data in the training according to state numbers;
(5) and (3) processing the real-time video signals acquired in the step (1) in the steps (2) and (3), inputting the processed real-time video signals as feature vectors into the recognition model trained in the step (4), wherein the output is recognition of the blowing state, and when the output number of the blowing state is 3, a signal prompt is sent out when the blowing enters the terminal stage.
Claims (4)
1. The utility model provides a converter steelmaking converting state identification system based on image sound attitude characteristic which characterized in that includes image acquisition module, color image flame region segmentation module, color flame image sound attitude characteristic express with describe the module, converting state classification module and host computer monitoring module based on image characteristic:
the image acquisition module is used for shooting the flame condition in real time, storing and transmitting the shot video signal to the color image flame area segmentation module;
the color image flame area segmentation module is used for segmenting the acquired video signal into separate color image frames and segmenting a flame background part and a flame body part in the color image;
the color flame image dynamic and static characteristic representation and description module is used for acquiring brightness information, chrominance information, texture information and dynamic information of an optical flow field of the flame body part obtained by segmentation;
the converting state classification module based on image characteristics is used for establishing a recognition model by taking various information collected by the dynamic and static characteristic representation and description module of the color flame image as input, and after setting an output state, collecting a data sample set to train the established recognition model; further judging the output state of the newly acquired input content, and sending the output state of the last stage of blowing entering to an upper computer monitoring module;
the upper computer monitoring module is used for sending out signal prompts.
2. The system for identifying the steelmaking and blowing state of the converter based on the dynamic and static image characteristics as claimed in claim 1, wherein: the image acquisition module further comprises a camera, a memory and a signal transmitter.
3. A converter steelmaking blowing state identification method based on image dynamic and static characteristics is characterized by comprising the following steps:
(1) installing a color camera, keeping the position and the angle of the color camera consistent with those of a manual fire watching, and fixing the position and the focal length of the camera; the real-time video signal collected by the color camera is connected with the processor through the signal collecting card;
(2) the processor divides the collected video signal into separate color image frames, and divides the flame background and flame body in the color image, i.e. converting RGB color space into uniform L a b space, and converting red (255,0,0) into RLabConversion of yellow (255, 0) to YLabWhite (255 ) is converted to WLabLet P be a pixel in the converted image and D be the Euclidean distance between pixels, according to the formula
The segmentation strategy is then:
in the formula, DPIThe color difference value between one pixel of the image after color space conversion and a reference point pixel is obtained; i is a pixel after the image is converted from RGB color to Lab color, and the pixel has three components, namely L, a and b; t is a threshold value for controlling the similarity;
then will be
The other part of the flame body is used as a flame background for segmentation;
(3) using the flame body part obtained in the step (2) for collecting brightness information, chrominance information, texture information and dynamic information of an optical flow field:
collecting brightness information B:
if I (I, j) is the divided flame region pixel, the brightness characteristic B is (Σ I (I, j))/(Count), where B is the brightness characteristic, Count is the number of region pixels, and I, j is the position of the pixel, i.e., the coordinate in the image;
collecting chrominance information S:
the third moment at the simple component i is
In the formula, pi,jThe probability of the occurrence of a pixel with the gray level j in the ith color channel component of the color image is shown, i is the ith component representing the color image, j is the gray level representing the image, si is the color moment characteristic of the image, and mui is a related parameter;
collecting texture information ASM:
describing static texture complexity by using a gray difference statistical method, wherein (x, y) is set as a point in an image, and a gray difference value between the point (x + delta x, y + delta y) and the point (x + delta x, y + delta y) is set as gΔ(x, y) -g (x + Δ x, y + Δ y), assuming that all possible values of the gray-scale difference are m-level, moving the point (x, y) within a given flame image area, accumulating gΔThe number of times (x, y) takes different values is given as gΔHistogram of (x, y), g is known from the histogramΔThe probability of the value of (x, y) is pΔ(i)；
The uniformity degree of image gray distribution is reflected by extracting the second moment ASM in the angle direction, and if the difference of the gray values of the similar pixels is larger, the ASM value is larger, so that the texture is rougher;
the texture characteristics are related to the oxidation degree of chemical elements in a molten pool, and the more violent combustion, the higher the roughness of the texture;
collecting dynamic information Ent of the optical flow field:
establishing a description model of flame dynamic change, establishing a flicker process by adopting an optical flow field, obtaining I (x, y, t) which is I (x + dx, y + dy, t + dt) according to the basic assumption of optical flow calculation, interconnecting points with constant pixel brightness in a limited area between adjacent frames, and constructing an interframe optical flow graph F; f records the intensity of the flame in the continuous change process, and the stroboscopic information can be reflected by performing characteristic representation and description on F;
wherein x, y refer to coordinates of points in the image, t is a time axis of the continuous image, because the video is composed of images on the continuous time axis, dx represents displacement of x axis, dy represents displacement of y axis, dt represents displacement of t axis, and I is an image frame of the video stream when the time is t moment;
for an interregional light flow graph, a gray level co-occurrence matrix is adopted to describe the characteristics of the interregional light flow graph, the direction in the calculation process is 0 degrees, 45 degrees, 90 degrees, 135 degrees and the step length step is 1 so as to adapt to the characteristic of the random micro-texture of flame, M is the obtained gray level co-occurrence matrix, the characteristics of the interregional light flow graph are described by adopting an entropy value, and the interregional light flow graph is calculated
Dynamic information of the flame;
wherein i, j is the coordinate of a pixel in the image;
(4) establishing an identification model by adopting a generalized regression neural network, taking the existing characteristic vector V (B, S, ASM, Ent) collected in the step (3) as input, and taking the state number of the converting as output: "1" represents the initial stage, "2" represents the intermediate stage, and "3" represents the final stage; the established generalized regression neural network is four-input and single-output, the hidden layer neuron excitation function is a Gaussian function, and the output is a linear mapping function; correspondingly listing the brightness information, the chrominance information, the texture information and the dynamic information of the optical flow field of the known flame body with the output data thereof, and establishing a data sample set; training the established recognition model, and classifying output data in the training according to state numbers;
(5) and (3) processing the real-time video signals acquired in the step (1) in the steps (2) and (3), inputting the processed real-time video signals as feature vectors into the recognition model trained in the step (4), wherein the output is recognition of the blowing state, and when the output number of the blowing state is 3, a signal prompt is sent out when the blowing enters the terminal stage.
4. The method for identifying the steelmaking and blowing states of the converter based on the image dynamic and static characteristics as claimed in claim 3, wherein the method comprises the following steps: the point (x + Δ x, y + Δ y) of the step (3) is a point around the value pixel (x, y) at a distance Δ x and Δ y from the value pixel (x, y).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610272037.9A CN105956618B (en) | 2016-04-27 | 2016-04-27 | Converter steelmaking blowing state identification system and method based on image dynamic and static characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610272037.9A CN105956618B (en) | 2016-04-27 | 2016-04-27 | Converter steelmaking blowing state identification system and method based on image dynamic and static characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105956618A CN105956618A (en) | 2016-09-21 |
| CN105956618B true CN105956618B (en) | 2021-12-03 |
Family
ID=56916587
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610272037.9A Active CN105956618B (en) | 2016-04-27 | 2016-04-27 | Converter steelmaking blowing state identification system and method based on image dynamic and static characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105956618B (en) |
Families Citing this family (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109598200B (en) * | 2018-11-01 | 2023-03-03 | 云南昆钢电子信息科技有限公司 | Intelligent image identification system and method for molten iron tank number |
| CN111222360B (en) * | 2018-11-23 | 2023-10-31 | 隆基绿能科技股份有限公司 | Method, equipment and storage medium for detecting molten state of silicon material |
| CN110309973B (en) * | 2019-07-01 | 2023-06-02 | 中冶赛迪信息技术(重庆)有限公司 | Converter splash prediction method and system based on video intelligent algorithm |
| CN110378936B (en) * | 2019-07-30 | 2021-11-05 | 北京字节跳动网络技术有限公司 | Optical flow calculation method and device and electronic equipment |
| CN110796046B (en) * | 2019-10-17 | 2023-10-10 | 武汉科技大学 | Intelligent steel slag detection method and system based on convolutional neural network |
| CN111104856B (en) * | 2019-11-18 | 2023-06-30 | 中冶赛迪技术研究中心有限公司 | Converter smelting splash monitoring method, system, storage medium and equipment |
| CN111340116A (en) * | 2020-02-27 | 2020-06-26 | 中冶赛迪重庆信息技术有限公司 | Converter flame identification method and system, electronic equipment and medium |
| CN113592760B (en) * | 2020-04-30 | 2024-04-16 | 昆明理工大学 | Converter endpoint carbon content prediction method based on flame image texture features |
| CN111598905A (en) * | 2020-05-13 | 2020-08-28 | 云垦智能科技(上海)有限公司 | Method for identifying type of blast furnace flame by using image segmentation technology |
| CN111476244A (en) * | 2020-05-29 | 2020-07-31 | 宝钢湛江钢铁有限公司 | System and method for intelligently identifying number of scrap steel trough for converter |
| CN112981135B (en) * | 2021-02-06 | 2022-09-27 | 楚雄滇中有色金属有限责任公司 | Method for judging end point of slagging period of converter copper smelting |
| CN113158818B (en) * | 2021-03-29 | 2023-04-07 | 青岛海尔科技有限公司 | Method, device and equipment for identifying fake video |
| CN113033705B (en) * | 2021-04-22 | 2022-12-02 | 江西理工大学 | Intelligent judgment and verification method for copper converter blowing slagging period end point based on pattern recognition |
| CN113033704B (en) * | 2021-04-22 | 2023-11-07 | 江西理工大学 | Intelligent judging method and system for copper converter converting copper-making final point based on pattern recognition |
| CN115339879B (en) * | 2022-10-19 | 2023-03-31 | 昆明理工大学 | Intelligent conveying and tracking method and system for small long and square billets based on machine vision |
| CN115880490B (en) * | 2022-11-21 | 2023-10-27 | 广东石油化工学院 | Flame segmentation method based on isolated forest |
| CN116402813B (en) * | 2023-06-07 | 2023-08-04 | 江苏太湖锅炉股份有限公司 | Neural network-based copper converter converting copper-making period end point judging method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002074370A (en) * | 2000-08-28 | 2002-03-15 | Ntt Data Corp | System and method for monitoring based on moving image and computer readable recording medium |
| CN101393603A (en) * | 2008-10-09 | 2009-03-25 | 浙江大学 | Method for recognizing and detecting tunnel fire disaster flame |
| CN103116746A (en) * | 2013-03-08 | 2013-05-22 | 中国科学技术大学 | Video flame detecting method based on multi-feature fusion technology |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3810657B2 (en) * | 2001-07-31 | 2006-08-16 | 松下電器産業株式会社 | Moving object detection method and apparatus |
-
2016
- 2016-04-27 CN CN201610272037.9A patent/CN105956618B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002074370A (en) * | 2000-08-28 | 2002-03-15 | Ntt Data Corp | System and method for monitoring based on moving image and computer readable recording medium |
| CN101393603A (en) * | 2008-10-09 | 2009-03-25 | 浙江大学 | Method for recognizing and detecting tunnel fire disaster flame |
| CN103116746A (en) * | 2013-03-08 | 2013-05-22 | 中国科学技术大学 | Video flame detecting method based on multi-feature fusion technology |
Non-Patent Citations (1)
| Title |
|---|
| 《转炉炼钢吹炼数据预测中火焰图像多特征提取方法研究》;刘辉;《中国博士学位论文全文数据库 信息科技辑》;20121015(第10期);正文1-124页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105956618A (en) | 2016-09-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105956618B (en) | Converter steelmaking blowing state identification system and method based on image dynamic and static characteristics | |
| CN205845067U (en) | The pneumatic steelmaking blowing state recognition system of static nature is moved based on image | |
| CN112380952B (en) | Power equipment infrared image real-time detection and identification method based on artificial intelligence | |
| CN105678332B (en) | Converter steelmaking end point judgment method and system based on flame image CNN recognition modeling | |
| CN106886216B (en) | Robot automatic tracking method and system based on RGBD face detection | |
| CN109598200B (en) | Intelligent image identification system and method for molten iron tank number | |
| CN106483143A (en) | A kind of solar energy photovoltaic panel dust stratification on-Line Monitor Device and its detection method | |
| CN113436169A (en) | Industrial equipment surface crack detection method and system based on semi-supervised semantic segmentation | |
| CN102456142A (en) | Analysis method for smoke blackness based on computer vision | |
| CN111401246A (en) | Smoke concentration detection method, device, equipment and storage medium | |
| WO2024027009A1 (en) | Infrared thermal imaging defect inspection method and apparatus for substation insulator | |
| CN109086803A (en) | A kind of haze visibility detection system and method based on deep learning and the personalized factor | |
| CN114549955A (en) | Online intelligent identification method and system for blast furnace tapping endpoint | |
| CN114511519A (en) | Train bottom bolt loss detection method based on image processing | |
| CN110057820B (en) | Method, system and storage medium for on-line detection of chlorine-hydrogen ratio of hydrogen chloride synthesis furnace | |
| CN116824517A (en) | Substation operation and maintenance safety control system based on visualization | |
| CN109815957A (en) | A kind of character recognition method based on color image under complex background | |
| CN102968644A (en) | Method for predicting smelting finishing point of argon-oxygen refined iron alloy | |
| CN109389152B (en) | Refined identification method for power transmission line falling object | |
| CN108898080B (en) | Ridge line neighborhood evaluation model-based crack connection method | |
| CN111402223B (en) | Transformer substation defect problem detection method using transformer substation video image | |
| CN104165696A (en) | Material surface color feature on-line automatic detection method | |
| CN113487570A (en) | High-temperature continuous casting billet surface defect detection method based on improved yolov5x network model | |
| CN113643352A (en) | Natural icing on-line monitoring running wire image icing degree evaluation method | |
| CN112132088A (en) | Inspection point location missing inspection identification method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |