A machine vision-based converter smelting temperature monitoring method

Abstract

The application discloses a converter smelting temperature monitoring method based on machine vision, collects historical furnace data including flame information analysis system and online infrared temperature measuring device detected furnace mouth, tapping hole temperature-time control curve, classifies and combines according to the same or similar principle, and obtains the best control model of the furnace mouth and tapping hole temperature under each classification condition through fitting; the real-time monitoring temperature is compared with the model curve recommended temperature, the real-time temperature is corrected and dynamically adjusted, and the accurate control of the process and end point temperature is realized. The application accurately detects the real-time temperature by applying the high-precision temperature detection device; secondly, the application of the model reduces the occurrence of abnormal phenomena such as splashing, dry back, unreasonable temperature rising, guarantees the smooth, safe and efficient operation of the blowing process, improves the converter end point temperature hit rate, promotes the production of high-quality steel, and is favorable for realizing the intelligent control of the converter smelting.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of converter steelmaking process control, and particularly relates to a converter smelting temperature monitoring method based on machine vision. BACKGROUND

[0002] Converter steelmaking is one of the key links in steel production and has been widely applied and promoted. Stable control of the end point of converter smelting is an important guarantee for the quality and safety of converter production. The specific goals of end point control are: (1) the carbon content of molten steel should reach the target range required by the steel grade; (2) the phosphorus and sulfur contents in molten steel should be lower than the range required by the lower limit of the specification; (3) the tapping temperature should ensure the smooth progress of the next process; and (4) the appropriate oxidizing property of molten steel. The end point control is essentially the control of the converter blowing process, and the quality of the end point control directly affects the production rate, metal yield, production cost, and steel quality, etc. indicators, and is a very important link in the operation of the converter steelmaking process.

[0003] Molten steel temperature is an important parameter during converter smelting, and appropriate temperature is the primary condition for all steelmaking reactions in the molten pool. Factors affecting the molten steel temperature in production include molten iron composition, molten iron temperature, molten iron ratio, furnace age, charging and slagging, blowing lance position, etc. Molten steel temperature control mainly includes process temperature control and end point temperature control, and accurate control of the end point temperature directly affects the energy, alloy element yield, service life of the furnace lining, and quality of the finished steel in the smelting process. However, converter steelmaking is a complex physical and chemical process at high temperature, and it is impossible to monitor the temperature in the molten pool in real time online by direct temperature measurement. Existing converter steelmaking temperature control methods mainly rely on static models of reaction mechanism, material balance, and thermodynamics or rely on artificial experience evaluation. For example, CN117272788A discloses a method for predicting the temperature of the molten pool, which is based on the change of the oxygen lance inlet and outlet water temperature of the converter, and through the heat transfer principle and related calculations, combined with the BP neural network algorithm, the temperature of the converter at the end of the static model is predicted. The method mainly involves measuring the results of the sub-lance, the oxygen lance control height, the oxygen lance cooling water conditions, etc. to predict the temperature of the molten pool. CN111893237A discloses a method for predicting the carbon content and temperature of the molten pool in the whole process of converter steelmaking. The invention only performs function type data analysis on the basis of the raw material data, smelting process data, molten pool carbon content and temperature information data of a number of historical heats, and predicts the carbon content and temperature of the molten pool without considering the dynamic correction of real-time data and conditions. CN117553921A discloses a converter molten steel temperature prediction method, system, terminal and storage medium, which predicts the molten steel temperature by neural network calculation through converter smelting data and sub-lance measurement data, combined with the flame temperature collection of the furnace mouth.

[0004] The above prediction method lacks real-time correction and dynamic adjustment of the model, has limited control accuracy and is prone to deviation, leading to frequent abnormal conditions in the steelmaking process, and the deviation will directly affect the control of the converter endpoint temperature and carbon content, thereby affecting the overall rhythm of steelmaking, energy and alloy element yield, steel quality, and stability control of production line safety production, which is not conducive to the intelligent high-quality development of the steel metallurgical industry. SUMMARY

[0005] To solve the above problems, the present application provides a converter smelting temperature monitoring method based on machine vision, which obtains temperature-time control curves by setting a furnace mouth flame information analysis system and an online infrared temperature measuring device at the furnace mouth and the tapping hole respectively, then combines historical furnace data to establish furnace mouth and tapping hole detection temperature-time control curve models under various conditions; according to the same or similar classification principle, the model recommended curve is applied, the real-time temperature is compared with the curve temperature recommended by the model at this moment, and the furnace mouth and tapping hole temperature is timely corrected and dynamically adjusted to maintain the consistency of the actual temperature and the recommended temperature, effectively control the process and endpoint temperature, reduce the occurrence of abnormal phenomena such as splashing, dry back, and unreasonable temperature rise, ensure the smooth and efficient and reasonable operation of the blowing process, and greatly improve the control accuracy of the converter smelting endpoint temperature.

[0006] In order to achieve the above effects, the present application adopts the following technical scheme:

[0007] A converter smelting temperature monitoring method based on machine vision collects historical furnace data including furnace mouth and tapping hole temperature-time control curves detected by a flame information analysis system and an online infrared temperature measuring device, classifies and combines them according to the same or similar principle, and fits to obtain the best control model of furnace mouth and tapping hole temperature; then compare the real-time detection temperature with the model curve recommended temperature, dynamically correct and adjust the real-time temperature, and control the process and endpoint temperature.

[0008] The above scheme includes the following specific steps:

[0009] (1) Furnace mouth temperature detection, obtain furnace mouth temperature-time control curve T 炉口(t) ;

[0010] (2) Tapping hole temperature detection, obtain tapping hole temperature-time control curve T 出钢口(t) ;

[0011] (3) Collect converter smelting historical furnace information including T 炉口(t) and T 出钢口(t) , and establish a historical database;

[0012] (4) Preprocess the historical data, classify and combine them according to the same or similar principle, and establish a secondary database;

[0013] (5) Data fitting is performed on T 炉口(t) and T 出钢口(t) to obtain the optimal control models T 炉口(t)-推荐 and T 出钢口(t)-推荐 under different classification conditions.

[0014] (6) The optimal curve recommended by the model is applied according to the same or similar principles, the real-time furnace mouth and tapping mouth temperatures are compared with the temperatures recommended by the model at the moment, the actual temperature is adjusted in time to maintain the consistency of the actual temperature and the recommended temperature, and the process and the end point temperature are controlled.

[0015] (7) After the furnace period ends, the actual control curve T 炉口(t)-实际 and T 出钢口(t)-实际 is obtained, and all data and operation results are fed back to the model for self-learning.

[0016] Further, the furnace mouth temperature detection mode in step (1) is to set a furnace mouth flame information analysis system to accurately analyze and judge the flame image features and spectral features of the converter furnace mouth, form a series of furnace mouth temperatures corresponding to any moment on the time axis of the blowing process, and obtain the furnace mouth detection temperature-time control curve T 炉口(t) of the whole blowing process of the furnace period.

[0017] Further, the tapping mouth temperature detection mode in step (2) is to install an online infrared temperature measuring device in the tapping mouth to detect the temperature and change trend in the converter molten pool, form a series of tapping mouth temperatures corresponding to any moment on the time axis of the blowing process, and obtain the tapping mouth detection temperature-time control curve T 出钢口(t) of the whole blowing process of the furnace period.

[0018] Further, the historical furnace period information in step (3) includes the temperature and addition amount of the molten iron, the contents of C, Si, Mn, P and S in the molten iron, the addition amount of scrap steel, the addition amounts of lime, dolomite and iron ore, the end point temperature and the end point carbon content. The molten iron charging conditions, the components and addition amounts of the slag-making auxiliary materials determine the total heat, the initial temperature, the process temperature, the temperature rising speed, the end point temperature, and the temperatures at any moment of the furnace mouth and the tapping mouth during the process through the steelmaking material balance and the heat balance.

[0019] Further, the pretreatment in step (4) is to eliminate invalid, useless and large deviation data.

[0020] Further, the same or similar principle of step (4) is to set the data classification interval range, and the process of reclassifying the pretreated historical heat data, the classification interval is: the C content in molten iron is classified as an interval of every 0.01%, and the interval value is: 0.351%~0.361%, 0.361%~0.371%, 0.371%~0.381%, and so on; the Si and Mn contents in molten iron are classified as an interval of every 0.1%, and the interval value is: 0~0.1%, 0.1%~0.2%, 0.2%~0.3%, and so on; the P content in molten iron is classified as an interval of every 0.01%, and the interval value is: 0.071%~0.081%, 0.081%~0.091%, 0.091%~0.11%, and so on; the S content in molten iron is classified as an interval of every 0.01%, and the interval value is: 0.011%~0.021%, 0.021%~0.031%, 0.031%~0.041%, and so on; the scrap steel addition amount is classified as an interval of every 5000 kg; the lime, dolomite and iron ore addition amounts are classified as an interval of every 500 kg; and the blowing endpoint temperature is classified as an interval of every 5°C; and the endpoint values of the above classification intervals are all collected in the previous classification interval.

[0021] Further, the data fitting formula of step (5) is: T0=polyval(a,t), a=polyfit(tdata,Tdata,n);

[0022] Wherein, n is the highest order of the polynomial, tdata is the time array, and Tdata is the temperature array, which are both input in the form of array.

[0023] Further, the process temperature adjustment mode of step (6) is to adjust the blowing oxygen lance height or add iron ore. When the temperature rises faster, the oxygen lance height is appropriately increased to slow down the carbon-oxygen reaction speed and reduce the temperature rise, and iron ore can also be added for cooling to inhibit the temperature rise speed; when the temperature rises slowly, the oxygen lance height is appropriately reduced to promote the carbon-oxygen reaction speed and increase the temperature rise speed.

[0024] A computer program for realizing the above-mentioned machine vision-based converter smelting temperature monitoring method.

[0025] An information processing terminal for realizing the above-mentioned machine vision-based converter smelting temperature monitoring method.

[0026] A computer readable storage medium comprising instructions which, when executed on a computer, cause the computer to perform the above-mentioned machine vision-based converter smelting temperature monitoring method.

[0027] Advantages:

[0028] (1) By applying high-precision temperature detection devices such as furnace flame information analysis system and online infrared temperature measurement device, the present invention can achieve accurate monitoring of real-time temperature; secondly, by combining the optimal control curve recommended by the model, it can compare, predict and judge in real time, and adjust and correct the gun position height and feeding in a timely manner, thereby ensuring that the operation process is carried out according to the preset optimal heating trend, realizing precise control of process temperature, reducing the occurrence of abnormal phenomena such as splashing, re-drying and unreasonable heating, effectively ensuring the stable, safe, efficient and reasonable operation of the blowing process, and improving the hit rate of the converter end target temperature.

[0029] (2) The application of the model of the present invention can realize high-precision control of the endpoint temperature. The model can be applied to the production of different steel grades, promote the production of high-quality steel grades, effectively improve production efficiency and reduce production costs, and has significant economic benefits and broad prospects for promotion. It is conducive to the intelligent and high-quality development of the iron and steel metallurgy industry. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the network structure for a machine vision-based converter smelting temperature monitoring method.

[0031] Figure 2 Example 1: Furnace mouth temperature-time control curve T 炉口(t) T 炉口(t)-推荐 and T 炉口(t)-实际 Trend chart;

[0032] Figure 3 The temperature-time control curve T at the tapping outlet of Example 1 出钢口(t) T 出钢口(t)-推荐 and T 出钢口(t)-实际 Trend chart;

[0033] Figure 4 Example 2: Furnace mouth temperature-time control curve T 炉口(t) T 炉口(t)-推荐 and T 炉口(t)-实际 Trend chart;

[0034] Figure 5 The temperature-time control curve T at the tapping outlet in Example 2 出钢口(t) T 出钢口(t)-推荐 and T 出钢口(t)-实际 The trend chart. Detailed Implementation

[0035] With reference to the contents in the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0036] (1) Furnace mouth temperature detection: A furnace mouth flame information analysis system is arranged at the furnace mouth to analyze and accurately determine the image features and spectral features of the converter furnace mouth flame, to obtain a series of furnace mouth temperatures corresponding to any moment on the time axis of the blowing process, and to obtain a furnace mouth detection temperature-time curve T 炉口 (t) of the blowing process control;

[0037] (2) Tapping hole temperature detection: An online infrared temperature measuring device is installed in the tapping hole to timely detect the temperature and change trend in the converter molten pool, to form a series of tapping hole temperatures corresponding to any moment on the time axis of the blowing process, and to obtain a tapping hole detection temperature-time curve T 出钢口 (t) of the blowing process control for the furnace;

[0038] (3) Data collection is performed on the molten iron temperature and addition amount of the historical furnace, the contents of C, Si, Mn, P and S in the molten iron, the addition amount of scrap steel, the addition amounts of lime, dolomite and iron ore, the end point temperature, the end point carbon content and the detection temperature-time curves T 炉口(t) and T 出钢口(t) of each furnace, and a historical database is established;

[0039] (4)Eliminate invalid, useless, large deviation historical data, according to the same or similar principle of classification combination, establish two-level database, classification interval: the C content in molten iron is taken as one classification interval every 0.01%, interval value is: 0.351%~0.361%, 0.361%~0.371%, 0.371%~0.381%, and so on; the Si, Mn content in molten iron is taken as one classification interval every 0.1%, interval value is: 0~0.1%, 0.1%~0.2%, 0.2%~0.3%, and so on; the P content in molten iron is taken as one classification interval every 0.01%, interval value is: 0.071%~0.081%, 0.081%~0.091%, 0.091%~0.11%, and so on; the S content in molten iron is taken as one classification interval every 0.01%, interval value is: 0.011%~0.021%, 0.021%~0.031%, 0.031%~0.041%, and so on; the scrap steel adding amount is taken as one classification interval every 5000kg; the lime, dolomite and iron ore adding amount is taken as one classification interval every 500kg; the blowing end point temperature is taken as one classification interval every 5℃; the end point value of above classification interval is all collected in the last classification interval;

[0040] (5)Data fitting is carried out on T 炉口(t) and T 出钢口(t) in the two-level database, the best control model T 炉口(t)-推荐 and T 出钢口(t)-推荐 are obtained, and the model is loaded into the computer;

[0041] The fitting formula of the furnace mouth temperature is: T 0-炉口 =polyval(a,t -炉口 ), a=polyfit(t -炉口 data,T -炉口 data,n);

[0042] n is the highest order of the polynomial, t -炉口 data is the furnace mouth time array, and T -炉口 data is the furnace mouth temperature array, which are input in the form of array;

[0043] The fitting formula of the tapping hole temperature is: T 0-出钢口 =polyval(a,t -出钢口 ), a=polyfit(t -出钢口 data,T -出钢口 data,n);

[0044] n is the highest order of the polynomial, t -出钢口 data is the tapping hole time array, and T -出钢口Data is the tapping temperature array, which is input in the form of array.

[0045] (6) According to the same or similar principles, the best curve T recommended by the model is applied 炉口(t)-推荐 and T 出钢口(t)-推荐 The real-time temperature detected by the furnace mouth flame information analysis system and the on-line infrared temperature measuring device of the tapping port is compared with the temperature value recommended by the model curve at this moment, and the real-time temperature is adjusted in time (when the temperature rises too fast, appropriately increase the oxygen lance height to slow down the carbon-oxygen reaction speed, reduce the temperature rising temperature, and also add iron ore to cool down to inhibit the temperature rising speed. When the temperature rises too slowly, appropriately reduce the oxygen lance height to promote the carbon-oxygen reaction speed and improve the temperature rising speed), so as to keep the consistency of the actual temperature and the recommended temperature, and realize the precise control of the process and the end temperature.

[0046] (7) After the end of the furnace, the actual control curve T 炉口(t)-实际 and T 出钢口(t)-实际 are obtained, and all data and operation results are fed back to the model for self-learning.

[0047] A computer program for realizing the above-mentioned machine vision-based temperature monitoring method of converter smelting.

[0048] An information processing terminal for realizing the above-mentioned machine vision-based temperature monitoring method of converter smelting.

[0049] A computer readable storage medium comprising instructions which, when executed on a computer, cause the computer to perform the above-mentioned machine vision-based temperature monitoring method of converter smelting.

[0050] The network structure diagram of the machine vision-based converter smelting temperature monitoring model control method of the application is shown in Figure 1 .

[0051] Example 1

[0052] The model obtained by the application is applied to the converter smelting of the present furnace, and the end point target value of the converter smelting is: the end point temperature is 1636℃, and the end point carbon is 0.081%. The specific operation steps are as follows:

[0053] (1) Converter smelting: the temperature of the molten iron entering the converter is 1386℃, the adding amount of scrap steel and molten iron is 180t and 51t respectively, among which the composition of the molten iron is C: 4.40%, Si: 0.65%, Mn: 0.35%, P: 0.082%, S: 0.028%; the adding amount of main slagging materials and alloy is: lime 30kg / t, dolomite 9kg / t, and ore 10kg / t; oxygen consumption is 49m 3 / t.

[0054] (2) input the charging condition and smelting condition in step (1) into the computer, and select the optimal control curve T recommended by the model suitable for the present heat 炉口(t)-推荐 and T 出钢口(t)-推荐 , as shown in Figure 2 , 3 ;

[0055] (3) compare the real-time temperature obtained by the present heat flame information analysis system and the on-line infrared temperature measuring device at the tapping hole with the curve temperature recommended by the model at the moment, and dynamically correct and adjust the process temperature in time, appropriately increase the oxygen lance height to slow down the carbon-oxygen reaction speed and reduce the temperature when the temperature rises too fast, and also add iron ore to cool down to inhibit the temperature rising speed; appropriately reduce the oxygen lance height to promote the carbon-oxygen reaction speed and increase the temperature rising speed when the temperature rises too slowly; keep the consistency of the actual temperature and the recommended temperature, realize the precise control of the end point temperature, and the end point temperature is 1635℃ and the end point carbon content is 0.080%;

[0056] (4) after the present heat is finished, feedback the actual control curve T 炉口(t)-实际 and T 出钢口(t)-实际 , all data and operation results to the model for self-learning, and the curve results are shown in Figure 2 , 3 .

[0057] The present heat converter smelting blowing process is stable, and there is no spatter and dry back phenomenon, the end point temperature is 1635℃, the end point carbon content is 0.080%, the end point is hit once, the end point temperature error is 1℃, the end point carbon error is 0.001%, and the end point temperature and the end point carbon content are all qualified. The converter end point temperature monitoring model based on machine vision can realize the real-time correction of the process temperature, and achieve the high-precision control of the end point temperature.

[0058] Example 2

[0059] The model obtained by the present application is applied to the present heat converter smelting, and the converter smelting end point target value is: the end point temperature is 1645℃ and the end point carbon is 0.076%, and the specific steps are as follows:

[0060] (1) converter smelting: the charging hot metal temperature is 1372℃, the scrap steel and hot metal adding amount is 180t and 51t respectively, the hot metal composition is C: 4.28%, Si: 0.58%, Mn: 0.42%, P: 0.072%, and S: 0.032%; the main slagging material and alloy adding amount is: lime 30kg / t, dolomite 8.2kg / t, and iron ore 9.6kg / t; the oxygen consumption is 49.2m 3 / t;

[0061] (2) input the furnace charging conditions and smelting conditions in step (1) into the computer, and select the optimal control curve T 炉口(t)-推荐 and T 出钢口(t)-推荐 recommended by the model for the present furnace Figure 4 , 5 ;

[0062] (3) compare the real-time temperature obtained by the present furnace mouth flame information analysis system and the tapping hole online infrared temperature measuring device with the temperature value of the curve recommended by the model at the moment, dynamically correct and adjust the process temperature in time, appropriately increase the oxygen lance height to slow down the carbon-oxygen reaction speed and reduce the temperature when the temperature rises too fast, and also add iron ore for cooling to inhibit the temperature rising speed; appropriately reduce the oxygen lance height to promote the carbon-oxygen reaction speed and increase the temperature rising speed when the temperature rises too slowly; keep the consistency of the actual temperature and the recommended temperature, realize the precise control of the end point temperature, and the end point temperature is 1646℃ and the end point carbon content is 0.077%;

[0063] (4) after the present furnace is finished, the actual control curve T 炉口(t)-实际 and T 出钢口(t)-实际 , all data and operation results are fed back to the model for self-learning, and the curve results are shown in Figure 4 , 5 .

[0064] The present converter smelting blowing process is stable, and there is no spatter and dry return phenomenon, the end point temperature is 1646℃, the end point carbon content is 0.077%, the end point is hit once, the end point temperature error is 1℃, the end point carbon error is 0.001%, and the end point temperature and the end point carbon content are all qualified. The converter end point temperature monitoring model based on machine vision of the present application effectively ensures the stable, efficient and reasonable operation of the blowing process, and improves the converter end point target temperature hit rate.

[0065] From the above, it can be seen from Figures 2-5 that the recommended curve T 炉口(t)-推荐 and T 出钢口(t)-推荐 of the present model are highly fitted with T 炉口 (t) and T 出钢口 (t) obtained by the furnace mouth flame information analysis system and the tapping hole online infrared temperature measuring device, and the actual control curve T 炉口(t)-实际 and T 出钢口(t)-实际 , which fully proves the rationality and effectiveness of the converter smelting temperature monitoring method based on machine vision obtained by the present application.

Claims

1. A method for monitoring converter smelting temperature based on machine vision, characterized in that, Historical furnace data, including furnace mouth and tapping outlet temperature-time control curves detected by a flame information analysis system and an online infrared temperature measurement device, are collected and classified according to the principle of similarity. The optimal control model for furnace mouth and tapping outlet temperatures is then fitted. The real-time detected temperature is then compared with the temperature recommended by the model curve to perform dynamic correction and adjustment of the real-time temperature, controlling the process and final temperature. The method includes the following specific steps: (1) Furnace mouth temperature detection to obtain furnace mouth temperature-time control curve T 炉口（t） ; (2) Detect the tapping outlet temperature and obtain the tapping outlet temperature-time control curve T. 出钢口（t） ; (3) Data collection includes T 炉口（t） and T 出钢口（t） Establish a historical database of converter smelting history; (4) Preprocess the historical data, classify and combine it according to the same or similar principles, and establish a secondary database; (5) For T in the secondary database 炉口（t） and T 出钢口（t） Perform data fitting to obtain the optimal control model T under different classification conditions. 炉口（t）-推荐 and T 出钢口（t）-推荐 ; (6) Apply the optimal curve recommended by the model based on the principle of similarity or similarity, compare the real-time furnace mouth and tapping mouth temperature with the temperature recommended by the model at that moment, adjust the actual temperature in a timely manner, maintain the consistency between the actual temperature and the recommended temperature, and control the process and end temperature. (7) After the furnace cycle is completed, the actual control curve T is obtained. 炉口（t）-实际 and T 出钢口（t）-实际 All data and calculation results are fed back to the model for self-learning; The furnace mouth temperature detection method in step (1) is to set up a furnace mouth flame information analysis system to accurately analyze and judge the image features and spectral features of the converter furnace mouth flame, form a series of furnace mouth temperatures corresponding to any moment on the time axis of the blowing process, and obtain the furnace mouth detection temperature-time control curve T for the entire blowing process of this furnace. 炉口（t） ; The method for detecting the tapping temperature in step (2) is to install an online infrared thermometer inside the tapping outlet to detect the temperature and its trend in the converter molten pool, forming a series of tapping temperatures corresponding to any moment on the time axis of the blowing process, and obtaining the tapping temperature-time control curve T for the entire blowing process of this heat. 出钢口（t） ; Step (4) preprocessing involves removing invalid, useless, or highly biased data; the principle of similarity or similarity involves setting data classification intervals and reclassifying the preprocessed historical furnace data. The classification intervals are as follows: the C content in the molten iron is divided into intervals of 0.01%, with interval values ​​of 0.351%–0.361%, 0.361%–0.371%, 0.371%–0.381%, and so on; the Si and Mn content in the molten iron is divided into intervals of 0.1%, with interval values ​​of 0%–0.1%. The following classifications apply: 0.1%–0.2%, 0.2%–0.3%, and so on; the phosphorus (P) content in the molten iron is categorized into intervals of 0.01%, with values ​​of 0.071%–0.081%, 0.081%–0.091%, 0.091%–0.11%, and so on; the sulfur (S) content in the molten iron is categorized into intervals of 0.01%, with values ​​of 0.011%–0.021%, 0.021%–0.031%, 0.031%–0.041%, and so on; the amount of scrap steel added is categorized into intervals of 5000 kg; the amount of lime, dolomite, and iron ore added is categorized into intervals of 500 kg; the final blowing temperature is categorized into intervals of 5°C; the endpoint values ​​of the above classification intervals are all aggregated into the previous classification interval. The data fitting formula in step (5) is: T0 = polyval(a,t), a = polyfit(tdata,Tdata,n); Where n is the highest order of the polynomial, tdata is the time array, and Tdata is the temperature array, both of which are input as arrays.

2. The converter smelting temperature monitoring method according to claim 1, characterized in that, The historical furnace information mentioned in step (3) includes the temperature and amount of molten iron added, the content of C, Si, Mn, P and S in the molten iron, the amount of scrap steel added, the amount of lime, dolomite and iron ore added, the final temperature and the final carbon content.

3. A computer program for implementing the machine vision-based converter smelting temperature monitoring method as described in any one of claims 1-2.

4. An information processing terminal for implementing the machine vision-based converter smelting temperature monitoring method as described in any one of claims 1-2.

5. A computer-readable storage medium comprising instructions, which, when executed on a computer, cause the computer to perform the machine vision-based converter smelting temperature monitoring method according to any one of claims 1-2.

Images