CN118016203A - LF refining temperature forecasting method based on mechanism model and XGBoost algorithm - Google Patents
LF refining temperature forecasting method based on mechanism model and XGBoost algorithm Download PDFInfo
- Publication number
- CN118016203A CN118016203A CN202410429459.7A CN202410429459A CN118016203A CN 118016203 A CN118016203 A CN 118016203A CN 202410429459 A CN202410429459 A CN 202410429459A CN 118016203 A CN118016203 A CN 118016203A
- Authority
- CN
- China
- Prior art keywords
- temperature
- slag
- heat
- refining
- molten steel
- 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.)
- Granted
Links
- 238000007670 refining Methods 0.000 title claims abstract description 76
- 230000007246 mechanism Effects 0.000 title claims abstract description 31
- 238000013277 forecasting method Methods 0.000 title claims abstract description 9
- 229910000831 Steel Inorganic materials 0.000 claims abstract description 120
- 239000010959 steel Substances 0.000 claims abstract description 120
- 238000000034 method Methods 0.000 claims abstract description 54
- 238000004519 manufacturing process Methods 0.000 claims abstract description 37
- 230000008569 process Effects 0.000 claims abstract description 33
- 238000010438 heat treatment Methods 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 13
- 238000012360 testing method Methods 0.000 claims abstract description 10
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 239000002893 slag Substances 0.000 claims description 89
- XKRFYHLGVUSROY-UHFFFAOYSA-N Argon Chemical compound [Ar] XKRFYHLGVUSROY-UHFFFAOYSA-N 0.000 claims description 88
- 229910045601 alloy Inorganic materials 0.000 claims description 64
- 239000000956 alloy Substances 0.000 claims description 64
- 229910052786 argon Inorganic materials 0.000 claims description 44
- 239000000779 smoke Substances 0.000 claims description 39
- 238000007664 blowing Methods 0.000 claims description 34
- 230000017525 heat dissipation Effects 0.000 claims description 20
- 230000004927 fusion Effects 0.000 claims description 17
- 238000000354 decomposition reaction Methods 0.000 claims description 15
- 238000003756 stirring Methods 0.000 claims description 14
- 238000009825 accumulation Methods 0.000 claims description 8
- 230000000630 rising effect Effects 0.000 claims description 8
- 239000000463 material Substances 0.000 claims description 7
- 239000007791 liquid phase Substances 0.000 claims description 6
- 238000007254 oxidation reaction Methods 0.000 claims description 6
- 239000007790 solid phase Substances 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 5
- 238000010586 diagram Methods 0.000 claims description 5
- 230000005855 radiation Effects 0.000 claims description 5
- 238000012546 transfer Methods 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 238000005275 alloying Methods 0.000 claims description 4
- 238000001514 detection method Methods 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 4
- 230000003647 oxidation Effects 0.000 claims description 4
- 238000012163 sequencing technique Methods 0.000 claims description 4
- 239000007787 solid Substances 0.000 claims description 4
- 239000000428 dust Substances 0.000 claims description 3
- 230000005611 electricity Effects 0.000 claims description 3
- 238000009529 body temperature measurement Methods 0.000 abstract description 7
- 238000004134 energy conservation Methods 0.000 abstract description 3
- 239000002994 raw material Substances 0.000 abstract description 3
- 229910052799 carbon Inorganic materials 0.000 description 7
- 230000000694 effects Effects 0.000 description 6
- 229910000519 Ferrosilicon Inorganic materials 0.000 description 4
- 235000019738 Limestone Nutrition 0.000 description 4
- 238000003723 Smelting Methods 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 4
- 238000013499 data model Methods 0.000 description 4
- 239000006028 limestone Substances 0.000 description 4
- 229910000604 Ferrochrome Inorganic materials 0.000 description 3
- 229910000616 Ferromanganese Inorganic materials 0.000 description 3
- WUKWITHWXAAZEY-UHFFFAOYSA-L calcium difluoride Chemical compound [F-].[F-].[Ca+2] WUKWITHWXAAZEY-UHFFFAOYSA-L 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 3
- 239000010436 fluorite Substances 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- DALUDRGQOYMVLD-UHFFFAOYSA-N iron manganese Chemical compound [Mn].[Fe] DALUDRGQOYMVLD-UHFFFAOYSA-N 0.000 description 3
- 238000009628 steelmaking Methods 0.000 description 3
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 2
- 235000008733 Citrus aurantifolia Nutrition 0.000 description 2
- 241000196324 Embryophyta Species 0.000 description 2
- 235000011941 Tilia x europaea Nutrition 0.000 description 2
- 229910052782 aluminium Inorganic materials 0.000 description 2
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000005266 casting Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000001816 cooling Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- -1 desulfurizing Substances 0.000 description 2
- 239000010459 dolomite Substances 0.000 description 2
- 229910000514 dolomite Inorganic materials 0.000 description 2
- 238000010891 electric arc Methods 0.000 description 2
- 238000009847 ladle furnace Methods 0.000 description 2
- 239000004571 lime Substances 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000004886 process control Methods 0.000 description 2
- HBMJWWWQQXIZIP-UHFFFAOYSA-N silicon carbide Chemical compound [Si+]#[C-] HBMJWWWQQXIZIP-UHFFFAOYSA-N 0.000 description 2
- 229910010271 silicon carbide Inorganic materials 0.000 description 2
- 229910052719 titanium Inorganic materials 0.000 description 2
- 239000010936 titanium Substances 0.000 description 2
- OYPRJOBELJOOCE-UHFFFAOYSA-N Calcium Chemical compound [Ca] OYPRJOBELJOOCE-UHFFFAOYSA-N 0.000 description 1
- VYZAMTAEIAYCRO-UHFFFAOYSA-N Chromium Chemical compound [Cr] VYZAMTAEIAYCRO-UHFFFAOYSA-N 0.000 description 1
- 229910001309 Ferromolybdenum Inorganic materials 0.000 description 1
- 240000007594 Oryza sativa Species 0.000 description 1
- 235000007164 Oryza sativa Nutrition 0.000 description 1
- 235000010627 Phaseolus vulgaris Nutrition 0.000 description 1
- 244000046052 Phaseolus vulgaris Species 0.000 description 1
- OAICVXFJPJFONN-UHFFFAOYSA-N Phosphorus Chemical compound [P] OAICVXFJPJFONN-UHFFFAOYSA-N 0.000 description 1
- CKUAXEQHGKSLHN-UHFFFAOYSA-N [C].[N] Chemical compound [C].[N] CKUAXEQHGKSLHN-UHFFFAOYSA-N 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000011575 calcium Substances 0.000 description 1
- 229910052791 calcium Inorganic materials 0.000 description 1
- 238000010000 carbonizing Methods 0.000 description 1
- 229910052804 chromium Inorganic materials 0.000 description 1
- 239000011651 chromium Substances 0.000 description 1
- 238000009749 continuous casting Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003009 desulfurizing effect Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000012535 impurity Substances 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- WPBNNNQJVZRUHP-UHFFFAOYSA-L manganese(2+);methyl n-[[2-(methoxycarbonylcarbamothioylamino)phenyl]carbamothioyl]carbamate;n-[2-(sulfidocarbothioylamino)ethyl]carbamodithioate Chemical compound [Mn+2].[S-]C(=S)NCCNC([S-])=S.COC(=O)NC(=S)NC1=CC=CC=C1NC(=S)NC(=O)OC WPBNNNQJVZRUHP-UHFFFAOYSA-L 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000002844 melting Methods 0.000 description 1
- 230000008018 melting Effects 0.000 description 1
- 238000005272 metallurgy Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 229910052757 nitrogen Inorganic materials 0.000 description 1
- 229910052698 phosphorus Inorganic materials 0.000 description 1
- 239000011574 phosphorus Substances 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000033764 rhythmic process Effects 0.000 description 1
- 235000009566 rice Nutrition 0.000 description 1
- 238000010079 rubber tapping Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/70—Machine learning, data mining or chemometrics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/10—Analysis or design of chemical reactions, syntheses or processes
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Biology (AREA)
- Medical Informatics (AREA)
- Crystallography & Structural Chemistry (AREA)
- Mathematical Physics (AREA)
- Analytical Chemistry (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Treatment Of Steel In Its Molten State (AREA)
Abstract
The invention discloses an LF refining temperature forecasting method based on a mechanism model and XGBoost algorithm, which comprises the following steps: acquiring process quality data in the LF refining production process; according to the energy conservation law, analyzing the energy income and expenditure conditions of the system, and constructing an LF refining energy conservation model; constructing an LF refining thermodynamic heating model according to thermodynamic laws; preprocessing data; calculating the temperature of molten steel according to the LF refining mechanism model; dividing a front section training set and a rear section training set and a testing set; constructing XGBoost models by using training sets; the model was evaluated. The invention has simple whole process, adopts a qualitative and quantitative combined method to analyze and predict the temperature of molten steel refined by different LF ladles by collecting raw material data, production process data and temperature measurement data of the whole process of the refining production process, is suitable for analyzing a large amount of nonlinear relation data in the LF refining production process, and can provide real-time early warning indication and operation guidance for production.
Description
Technical Field
The invention relates to the technical field of metallurgy, in particular to an LF refining temperature forecasting method based on a mechanism model and XGBoost algorithm.
Background
LF refining was developed in 1970 s, and has the functions of arc heating, removing impurities, desulfurizing, argon blowing and stirring, and the like, and is used for smelting high-grade high-quality steel. LF refining is a main process of steelmaking production, is an important ring in the steelmaking process, is positioned between a converter and continuous casting, and has the function of adjusting the production rhythm up and down. LF has the heating function, and can accurately adjust the components and the temperature of molten steel, thereby ensuring the proper superheat degree of the molten steel before casting and being very beneficial to improving the quality of casting blanks. The control precision and accuracy of molten steel temperature control, alloy addition and bottom argon blowing in steel production directly influence the quality of molten steel and the sequence of working procedures. In traditional steelmaking, an operator mainly predicts the temperature of molten steel according to experience, and only can judge whether arc heating or scrap steel adding and cooling are needed through multiple temperature measurement.
Therefore, the establishment of an accurate LF refining temperature prediction model has strong practical significance. As can be seen from research examples of domestic and foreign temperature prediction models, common modeling methods are divided into 3 types, namely experience modeling, mechanism modeling and data modeling. The empirical modeling is an empirical formula summarized by the experience of a first-line staff through a long-term on-site smelting process, is simple and convenient to operate, but ignores some potential factors, so that the accuracy of the empirical formula is influenced. The mechanism modeling mainly considers the heat change generated by the physicochemical reaction in the refining process, ladle heat dissipation, argon blowing stirring and power consumption in the smelting process, and derives a mathematical model by applying the theorem of formulas such as thermodynamics, chemical reaction dynamics and the like, but the physicochemical reactions in the refining link are mutually coupled, so that the deriving process is complex, and an ideal model is difficult to obtain. The data modeling is to find out various parameters affecting the molten steel temperature in the smelting process and the association degree between the parameters from production, process and equipment data, and the accuracy is higher than that of the empirical modeling and the mechanical modeling.
Today, where artificial intelligence is rapidly developed, data modeling is fast and efficient without relying on complex expertise and experience accumulation over time and month as compared with experience modeling and mechanism modeling, but the disadvantage of data modeling is that the interpretability is not strong, and as production equipment of a steel plant is updated, personnel are replaced, the process is optimized, the data is changed, and the data model originally based on old data inevitably loses accuracy. Therefore, the method has strong practical significance on how to construct an LF refining temperature prediction model with high accuracy, high efficiency and strong generalization capability.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides an LF refining temperature forecasting method based on a mechanism model and XGBoost algorithm.
In order to achieve the above purpose, the invention adopts the following technical scheme: an LF refining temperature forecasting method based on a mechanism model and XGBoost algorithm comprises the following steps:
s1: acquiring process quality data in the LF refining production process, wherein the process quality data comprise refining power consumption, components and corresponding amounts of alloy and slag, argon blowing flow and time, argon blowing pressure and LF entering temperature;
S2: according to the law of conservation of energy, the energy income and expenditure conditions of the system are analyzed, and the energy change in the molten steel refining process is deduced on the basis of the energy income and expenditure conditions, so that the following formula is satisfied:
Qsteel＝QE＋Qalloy－Qslag－QIn－Qshell－QAr－Qsurface－Qsmoke
Wherein: q steel is the temperature rising and heating of molten steel, KJ; q E is the arc energy fed into the molten bath, KJ; q alloy is energy generated by adding alloy, KJ; q slag is the energy lost by adding slag, KJ; q In is the heat accumulation of furnace lining, KJ; q shell is the heat dissipated by the furnace shell, KJ; q Ar is the heat lost by argon blowing stirring, KJ; q surface is the heat quantity, KJ, dissipated by the molten steel and the slag surface; q smoke is the heat taken away by the smoke and the KJ;
s3: according to thermodynamic law, the temperature rise of molten steel is analyzed and calculated, and the temperature rise formula is as follows:
,
Wherein: q steel is the temperature rising and heating of molten steel, KJ; c steel is the specific heat capacity of molten steel, KJ/(kg.K); c slag is the specific heat capacity of slag, KJ/(kg.K); m steel is the mass of molten steel and Kg; m slag is the mass of molten steel and Kg;
The temperature of molten steel at the time t is as follows:
,
Wherein: Is the temperature of the incoming molten steel and is at the temperature of DEG C; the temperature of molten steel at the time t is DEG C;
S4: preprocessing the data obtained in the step S1, and removing repeated items and null values in the data; the z-score method is utilized to normalize data, eliminate the influence of the variation size and the numerical value of dimension and variable, and the box diagram detection method is utilized to delete abnormal points;
S5: substituting the parameters including the refining electricity consumption, the argon blowing time and the alloy addition amount obtained in the step S4 into the step S2 to calculate the molten steel heating temperature, calculating the molten steel temperature at the time t through the formula of the step S3, and adding the molten steel temperature as a newly added parameter into the data set obtained in the step S4;
S6: dividing the optimized data set in the step S5 into a training set and a testing set;
s7: constructing XGBoost models of training set data in a python environment, establishing a relation between input features and output features, sequencing contribution degrees of the input features to the output features, and finally eliminating features with importance scores smaller than 0.005 on the target index;
S8: verifying the test set by using the built XGBoost model, taking a mean square error MSE and a determination coefficient R 2 as evaluation criteria, judging XGBoost that the model is evaluated to be qualified if MSE values and R 2 values are within a preset range, and adopting a qualified XGBoost model to predict LF refining temperature; otherwise, if the operation is not qualified, the operations from the step S4 to the step S8 are carried out again.
As a specific embodiment, Q E in step S2 is the arc energy fed into the molten bath, and the formula is decomposed into:
Wherein: q E is the arc energy fed into the molten bath, KJ; p a is arc power, KW; phi is a proportionality coefficient; t 0 is the discharge time, min.
As a specific embodiment, in step S2, Q alloy is energy generated by adding an alloy, and the formula is as follows:
,
wherein Q joxide is the oxidation exotherm of the alloy element j, which is calculated by the following formula:
Qjoxide＝ΔΗjoxideΜj（1－fj）
Wherein: ΔH joxide is the oxidation reaction heat of the alloy element j, KJ/Kg; f j is the yield of the alloy element,%;
,
Wherein: m j is the mass of the element j, kg; n m is the number of alloys containing the j element; c ij is the content of the element j in the alloy i,%; The addition amount of the alloy i is Kg;
Q jmelt is the heat of fusion of element j in the solid alloy, calculated from the following formula:
,
Wherein: c jsolid、cjfluid is the solid phase and liquid phase specific heat capacity of the alloy element j, KJ/(Kg.K); t jstart、Tjfluid The temperature of the alloy element j is the furnace inlet temperature, the liquidus temperature and the preset subcontracting temperature; ΔH jmelt latent heat of fusion of alloy element j, KJ/Kg;
Q jfuse is the heat of fusion of the alloy element j in the molten steel, and is calculated by the following formula:
Qjfuse＝ΔΗjfuseΜj
wherein: ΔH jfuse is the heat of fusion, KJ/Kg, of the alloying element j.
As a specific embodiment, Q slag in step S2 is energy generated by adding an alloy, and the formula decomposition is as follows:
,
Wherein: n slag is the number of slag materials; 、 And The temperature of the slag charge entering the furnace, the liquidus temperature and the preset subcontracting temperature are respectively set at the temperature of DEG C;、 the specific heat capacities of solid phase and liquid phase of slag charge are KJ/(Kg.K); The mass of the slag charge i is Kg; ΔH imelt is the latent heat of fusion of the slag, KJ/Kg. The thermal effect of common slag materials such as lime, fluorite and refining slag is calculated to obtain the thermal effect of the slag materials with unit mass at 1600 ℃.
As a specific embodiment, in step S2, Q In is the heat accumulation of the furnace lining, and the formula decomposition is as follows:
QIn＝cInWIn(TIn－TInstart)
Wherein: c In is the specific heat capacity of the furnace lining, KJ/(Kg.K); w In is the mass of the furnace lining, kg; t In is the temperature of the furnace lining and DEG C; t Instart is the initial temperature of the furnace lining, DEG C.
As a specific embodiment, Q shell in step S2 is the heat dissipated from the furnace shell, and the formula decomposition is as follows:
,
Wherein: q shell is the heat dissipated by the furnace shell, KJ; k is a coefficient, and 4.88 is taken; epsilon is the surface blackness of the furnace body and is 0.95; t shell is the temperature of the outer surface of the furnace body and is in DEG C; t 0 is the ambient temperature of the workshop and is in DEG C; alpha shell is the convective heat transfer coefficient between the outer surface of the furnace body and the workshop environment; f is the outer surface area of the furnace body, m 2;t1 is the heat dissipation time of the furnace body, and h;
if no transverse airflow flows in the workshop, the convection heat exchange coefficient between the outer surface of the furnace body and the workshop environment is calculated according to the following formula:
αshell＝k1(Tshell－T0)0.25
Wherein: k 1 is a coefficient, k 1 =2.8 when the heat radiation face is upward; when vertical, k 1 =2.2; downward, k 1 =1.5.
As a specific embodiment, Q Ar in step S2 is the heat lost by argon blowing stirring, and the formula decomposition is as follows:
,
Wherein: c p is the specific heat of argon, J/Nm 3·℃;VAr is the argon blowing amount, nm 3;TAr is the initial temperature of the blown argon, and the temperature is lower than the initial temperature; the temperature is preset for subcontracting, and the temperature is lower than the temperature.
In a specific embodiment, in step S2, Q surface is the heat dissipated from the molten steel and slag surface, and the formula is as follows:
，
Wherein: the heat dissipation capacity of the exposed surface of the molten steel in the argon blowing process is obtained; The heat dissipation capacity of the slag layer;
Wherein: ,
Wherein: a steel is the exposed area of molten steel, m 2;εsteel is the surface blackness of the molten steel, and 0.4 is taken; sigma is a Stefan-Boltzmann constant, and 5.67 x 10 -8W/(m2·K4);Ta is taken as the ambient temperature at DEG C; t 2 is the bare heat dissipation time of molten steel, min;
,
Wherein: a slag is the area of slag, m 2;Cslag is the heat loss coefficient, and 0.6 is taken; epsilon slag is the blackness of the slag surface, and 0.8 is taken; sigma is a Stefan Boltzmann constant, and 5.67 x 10 -8W/(m2·K4);t3 is slag surface heat dissipation time, min.
As a specific embodiment, Q smoke in step S2 is heat taken away by the smoke, and the formula is as follows:
,
Wherein: w smoke is the mass of the smoke and Kg; c smoke is the specific heat capacity of the smoke and dust, KJ/(Kg.K); t smoke is the temperature of the smoke and the temperature is DEG C; t 0 is the ambient temperature of the workshop and is in DEG C; is the fusion latent heat of smoke and KJ/Kg.
In a specific embodiment, in step S8, the specific formulas of the mean square error MSE and the decision coefficient R 2 are as follows:
，
，
Wherein: n is the number of sample predictions; Is a model predictive value; y i is the actual value; /(I) Is the average value; r 2 is a determination coefficient; MSE is the mean square error.
As a specific embodiment, in step S8, the XGBoost model is set as the tree model gbtree, in which:
the maximum depth max-depth of the tree is set to 6;
The learning rate cta is set to 0.222 in the model;
gbtree the number of classifiers is set to 450.
As a specific embodiment, the MSE value is set to less than 50, the R 2 value is set to greater than 0.92, and the ratio of predicted to actual values within + -5 ℃ is greater than 90%.
Due to the application of the technical scheme, compared with the prior art, the invention has the following advantages:
1) According to the invention, LF refining process equipment, raw materials, energy consumption and production data are fully utilized, a mechanism model is built according to thermodynamic laws, and the requirement on single production data quantity is reduced;
2) The temperature prediction model is built through a machine learning algorithm, and the convergence and accuracy of LF refining temperature prediction are further improved by combining a mechanism model prediction result;
3) According to the invention, the mechanism model and the data model are combined for mixed modeling, the calculation result of the mechanism model is used as the training value or the input parameter of the data model, and the mixed model combines the respective advantages of the mechanism model and the data model, so that the established model has higher calculation precision, stronger generalization capability and better data convergence, thereby providing more accurate early warning and guidance for production operation, and further improving the quality of steel and the stability of the production and manufacturing process.
Drawings
FIG. 1 is a flow chart of an LF refining temperature forecasting method based on a mechanism model and XGBoost algorithm;
FIG. 2 is a schematic diagram of feature importance parameter ordering;
FIG. 3 is a graph showing the comparison of measured values and predicted values.
Detailed Description
The technical scheme of the invention is further described below with reference to the accompanying drawings and specific embodiments.
The invention provides an LF refining temperature forecasting method based on a mechanism model and XGBoost algorithm. The ladle refining production process comprises a plurality of process sections such as slag making, alloy adding, argon blowing stirring, electric arc heating, temperature measuring sampling and the like, equipment for the ladle refining production process comprises a ladle furnace, a ladle trolley, a ladle cover, an argon station, an alloy bin, a wire feeder and the like, and a system for the ladle refining production process comprises a process control system and a production management system.
Referring to fig. 1, the LF refining temperature prediction method specifically includes the following steps:
And collecting process quality data in the LF refining production process, wherein the process quality data comprise ladle refining production process data, production plan data, temperature measurement data and basic data.
The production process data refer to data related to each process interval in the whole ladle refining production process, the production process data are acquired by a process control system, and the production process data comprise but are not limited to slag formation data, alloy addition data, argon blowing stirring data and arc heating data. The slag formation data, the alloy addition data, the argon blowing stirring data and the electric arc heating data all comprise a plurality of characteristics, and specifically, the slag formation data comprises a furnace number, a dolomite component, a dolomite addition amount, a limestone component, a limestone addition amount, a refining slag component, a limestone addition amount, a silicon carbide component, a silicon carbide addition amount, an accelerator component, an accelerator addition amount, a slag formation temperature and the like; the alloy addition data comprise high-purity ferrosilicon components, high-purity ferrosilicon addition, low-titanium high-carbon ferrochrome components, low-titanium high-carbon ferrochrome addition, medium-carbon ferromanganese components, medium-carbon ferromanganese addition and the like; the argon blowing stirring data comprise argon blowing total flow, argon flow with obvious fluctuation on a steel slag interface, argon average flow, argon blowing time, argon blowing temperature, argon pressure and the like; the arc heating data comprises submerged arc depth, power supply duration, transformer capacity, electrode diameter, power supply efficiency and the like.
The production plan data comprises steel grade components, protocol standards, refining tonnage, process numbers and the like.
The temperature measurement data comprise thermometer numbers, temperature measurement times, LF incoming temperature and the like.
The basic data comprise ladle furnace age, ladle temperature, tapping time, furnace cover furnace age, steam recovery, ladle capacity, ladle number and the like.
In this embodiment, the data collection uses the furnace number as the process actual performance statistical information, that is, the data collection actual performance statistics of each furnace number are summarized into one record.
S2: construction of LF refining energy conservation model
According to the law of conservation of energy, the income and expenditure conditions of the energy of the system are analyzed, and the change of the energy in the refining process of molten steel is deduced on the basis, wherein the formula is as follows:
Qsteel＝QE＋Qalloy－Qslag－QIn－Qshell－QAr－Qsurface－Qsmoke
Wherein: q steel is the temperature rising and heating of molten steel, KJ; q E is the arc energy fed into the molten bath, KJ; q alloy is energy generated by adding alloy, KJ; q slag is the energy lost by adding slag, KJ; q In is the heat accumulation of furnace lining, KJ; q shell is the heat dissipated by the furnace shell, KJ; q Ar is the heat lost by argon blowing stirring, KJ; q surface is the heat quantity, KJ, dissipated by the molten steel and the slag surface; q smoke is the heat taken away by the smoke and the KJ.
Here, Q E is the arc energy fed into the puddle, the formula of which is broken down into:
,
Wherein: q E is the arc energy fed into the molten bath, KJ; p a is arc power, KW; phi is a proportionality coefficient; t 0 is the discharge time, min.
Q alloy is the energy generated by adding the alloy, and the formula is as follows:
,
wherein Q joxide is the oxidation exotherm of the alloy element j, which is calculated by the following formula:
Qjoxide＝ΔΗjoxideΜj（1－fj）
Wherein: ΔH joxide is the oxidation reaction heat of the alloy element j, KJ/Kg; f j is the yield of the alloy element,%, and the data are shown in table 1;
,
Wherein: m j is the mass of the element j, kg; n m is the number of alloys containing the j element; c ij is the content of the element j in the alloy i,%; The addition amount of the alloy i is Kg;
Q jmelt is the heat of fusion of element j in the solid alloy, calculated from the following formula:
,
Wherein: c jsolid、cjfluid is the solid phase and liquid phase specific heat capacity of the alloy element j, KJ/(Kg.K); t jstart、Tjfluid The temperature of the alloy element j is the furnace inlet temperature, the liquidus temperature and the preset subcontracting temperature; the melting potential heat of the DeltaH jmelt alloy element j, KJ/Kg, is shown in Table 1.
Q jfuse is the heat of fusion of the alloy element j in the molten steel, and is calculated by the following formula:
Qjfuse＝ΔΗjfuseΜj
Wherein: ΔH jfuse is the heat of fusion, KJ/Kg, of the alloying element j, and the data are shown in Table 1.
TABLE 1 alloy thermal Effect (25 ℃ C.)
Here, Q slag is the energy generated by adding the alloy, and its formula decomposition is as follows:
,
Wherein: n slag is the number of slag materials; 、 And The temperature of the slag charge entering the furnace, the liquidus temperature and the preset subcontracting temperature are respectively set at the temperature of DEG C;、 the specific heat capacities of solid phase and liquid phase of slag charge are KJ/(Kg.K); the mass of the slag charge i is Kg; ΔH imelt is the latent heat of fusion of the slag, KJ/Kg. The thermal effect of common slag materials such as lime, fluorite and refining slag is calculated to obtain the thermal effect of the slag materials with unit mass at 1600 ℃. As shown in table 2.
TABLE 2 thermal Effect of slag charge (25 ℃ C.)
In the step S2, Q In is the heat accumulation of the furnace lining, and the formula decomposition is as follows:
QIn＝cInWIn(TIn－TInstart)
Wherein: c In is the specific heat capacity of the furnace lining, KJ/(Kg.K); w In is the mass of the furnace lining, kg; t In is the temperature of the furnace lining and DEG C; t Instart is the initial temperature of the furnace lining, DEG C.
In step S2, Q shell is the heat dissipated from the furnace shell, and the formula decomposition is as follows:
,
Wherein: q shell is the heat dissipated by the furnace shell, KJ; k is a coefficient, and 4.88 is taken; epsilon is the surface blackness of the furnace body and is 0.95; t shell is the temperature of the outer surface of the furnace body and is in DEG C; t 0 is the ambient temperature of the workshop and is in DEG C; alpha shell is the convective heat transfer coefficient between the outer surface of the furnace body and the workshop environment; f is the outer surface area of the furnace body, m 2;t1 is the heat dissipation time of the furnace body, and h;
if no transverse airflow flows in the workshop, the convection heat exchange coefficient between the outer surface of the furnace body and the workshop environment is calculated according to the following formula:
αshell＝k1(Tshell－T0)0.25
Wherein: k 1 is a coefficient, k 1 =2.8 when the heat radiation face is upward; when vertical, k 1 =2.2; downward, k 1 =1.5.
In step S2, Q Ar is the heat lost by argon blowing stirring, and the formula decomposition is as follows:
,
Wherein: c p is the specific heat of argon, J/Nm 3·℃;VAr is the argon blowing amount, nm 3;TAr is the initial temperature of the blown argon, and the temperature is lower than the initial temperature; the temperature is preset for subcontracting, and the temperature is lower than the temperature.
The LF refining process always carries out argon blowing stirring, and the excessive argon blowing flow can cause the slag layer on the surface of molten steel to be blown open so as to expose part of molten steel in the atmosphere, and the heat dissipation of the exposed surface of molten steel in the argon blowing process is usedThe formula is as follows:
,
Wherein: KJ is the heat dissipation capacity of the exposed surface of the molten steel; a steel is the exposed area of molten steel, m 2;εsteel is the surface blackness of the molten steel, and 0.4 is taken; sigma is a Stefan-Boltzmann constant, and 5.67 x 10 -8W/(m2·K4);Ta is taken as the ambient temperature at DEG C; t 2 is the bare cooling time of molten steel, min.
The slag layer heat dissipation mainly comprises convection heat dissipation and radiation heat dissipation, and due to the fact that the slag temperature is high, radiation heat transfer plays a dominant role, convection heat transfer is negligible, and a slag layer heat dissipation formula is as follows:
,
Wherein: KJ is the heat dissipation capacity of the exposed surface of the molten steel; a slag is the area of slag, m 2;Cslag is the heat loss coefficient, and 0.6 is taken; epsilon slag is the blackness of the slag surface, and 0.8 is taken; sigma is a Stefan Boltzmann constant, and 5.67 x 10 -8W/(m2·K4);t3 is slag surface heat dissipation time, min.
Q surface is the heat dissipated by the molten steel and the slag surface, and the formula is as follows:
。
In step S2, Q smoke is the heat taken away by the smoke, and the formula decomposition is as follows:
,
Wherein: w smoke is the mass of the smoke and Kg; c smoke is the specific heat capacity of the smoke and dust, KJ/(Kg.K); t smoke is the temperature of the smoke and the temperature is DEG C; t 0 is the ambient temperature of the workshop and is in DEG C; is the fusion latent heat of smoke and KJ/Kg.
S3: construction of LF refining thermodynamic heating model
According to the thermodynamic law, the temperature rise of molten steel is analyzed and calculated, and the formula is as follows:
,
,
Wherein: q steel is the temperature rising and heating of molten steel, KJ; c steel is the specific heat capacity of molten steel, KJ/(kg.K); c slag is the specific heat capacity of slag, KJ/(kg.K); m steel is the mass of molten steel and Kg; m slag is the mass of molten steel and Kg; Is the temperature of the incoming molten steel and is at the temperature of DEG C; The temperature of molten steel at the time t is DEG C.
S4: data preprocessing
Preprocessing the data obtained in the step S1, and removing repeated items and null values in the data; the z-score method is utilized to normalize data, eliminate the influence of the variation size and the numerical value of dimension and variable, and the box diagram detection method is utilized to delete abnormal points;
substituting parameters including refining power consumption, argon blowing time, alloy addition amount and the like obtained in the step S4 into the LF refining mechanism model in the step S3, calculating the temperature of molten steel, and taking the temperature of molten steel as a newly added parameter to be listed in the data set obtained in the step S4;
dividing the data set obtained in the step S5 into a training set and a testing set;
Constructing a model by applying XGBoost to training set data in a python environment, establishing a relation between input features and output features, sequencing the contribution degree of the input features to the output features, and finally eliminating features with the importance score of the target index less than 0.005, which is shown in fig. 2;
the XGBoost model was set as follows:
boost is set to tree model gbtree;
the maximum depth max-depth of the tree is set to 6;
The learning rate cta is set to 0.222 in the model;
gbtree the classifier number was set to 450;
And verifying the test set by using the built XGBoost model, taking a mean square error MSE and a decision coefficient R 2 as evaluation criteria, judging XGBoost that the model is evaluated to be qualified if the MSE value and the decision coefficient R 2 value are within a preset range, and predicting the LF refining temperature by adopting the qualified model, otherwise, failing to be qualified, and carrying out the operations of the steps S4 to S8 again, wherein the specific formula is as follows:
，
，
Wherein: n is the number of sample predictions; Is a model predictive value; y i is the actual value; /(I) Is the average value; r 2 is a determination coefficient; MSE is the mean square error.
The MSE value setting range in the LF refining temperature is smaller than 50, the R 2 value setting range of the model is larger than 0.92, the ratio of the predicted value to the actual value within +/-5 ℃ is larger than 90%, and the LF refining temperature is predicted after the model is evaluated to be qualified, as shown in figure 3.
Example 1
In this example, the LF refining production data of Y steel mill 2023 is taken as an example, and the intelligent LF temperature prediction is carried out, and the specific steps are as follows:
S1: acquiring process quality data in the LF refining production process of the Y steel plant 2023 in 1 month to 12 months, wherein the process quality data comprise ladle refining production process data, production plan data, temperature measurement data and basic data, and the data comprise 1720 groups; s2: according to the law of conservation of energy, the income and expenditure conditions of the energy of the system are analyzed, and the change of the energy in the refining process of molten steel is deduced on the basis, wherein the formula is as follows:
Qsteel＝QE＋Qalloy－Qslag－QIn－Qshell－QAr－Qsurface－Qsmoke
Wherein: q steel is the temperature rising and heating of molten steel, KJ; q E is the arc energy fed into the molten bath, KJ; q alloy is energy generated by adding alloy, KJ; q slag is the energy lost by adding slag, KJ; q In is the heat accumulation of furnace lining, KJ; q shell is the heat dissipated by the furnace shell, KJ; q Ar is the heat lost by argon blowing stirring, KJ; q surface is the heat quantity, KJ, dissipated by the molten steel and the slag surface; q smoke is the heat taken away by the smoke and the KJ;
s3: according to thermodynamic law, the temperature rise of molten steel is analyzed and calculated, and the temperature rise formula is as follows:
,
Wherein: q steel is the temperature rising and heating of molten steel, KJ; c steel is the specific heat capacity of molten steel, KJ/(kg.K); c slag is the specific heat capacity of slag, KJ/(kg.K); m steel is the mass of molten steel and Kg; m slag is the mass of molten steel and Kg;
The temperature of molten steel at the time t is as follows:
,
Wherein: Is the temperature of the incoming molten steel and is at the temperature of DEG C; the temperature of molten steel at the time t is DEG C;
S4: preprocessing the data obtained in the step S1, and removing repeated items and null values in the data; the z-score method is utilized to normalize the data, the influence of the variation size and the numerical value size of the dimension and the variable is eliminated, the box diagram detection method is utilized to delete abnormal points, and 1525 groups of data are remained;
S5: substituting parameters including refining power consumption, argon blowing time, alloy addition amount and the like obtained in the step S4 into the LF refining mechanism model in the step S3, calculating the temperature of molten steel, and taking the temperature of molten steel as a newly added parameter to be listed in the data set obtained in the step S4;
s6: dividing the data set obtained in the step S5 into a training set and a test set, and randomly dividing the preprocessed data into a 1300 group of training data and a 225 group of test data;
S7: constructing a model by applying XGBoost to training set data in a python environment, establishing a relation between input features and output features, sequencing the contribution degree of the input features to the output features, finally eliminating features with the importance score of the target index less than 0.005, and finally selecting feature parameters, wherein the method comprises the following steps: carbonizing rice hulls, molten steel temperature, solid pure calcium wire, refining electricity consumption, inbound carbon, inbound CEQ, refining time, molten steel weight, ferrosilicon alloy, ferromolybdenum alloy, aluminum bean, soft blowing time, inbound temperature, argon consumption, inbound chromium, inbound manganese, inbound phosphorus, medium carbon ferromanganese, ladle number, limestone block ash, aluminum wire, low nitrogen carburant, fluorite, ladle condition, low nitrogen carbon wire, white slag time, non-deoxidizing molten steel sampler, high-purity ferrosilicon and high-carbon ferrochrome;
the XGBoost model was set as follows:
boost is set to tree model gbtree;
the maximum depth max-depth of the tree is set to 6;
The learning rate cta is set to 0.222 in the model;
gbtree the classifier number was set to 450;
S8: verifying the test set by using the built XGBoost model, taking a mean square error MSE and a decision coefficient R2 as evaluation standards, and calculating to obtain the MSE of 45.19 and the R2 of 0.9286 when the number of classifiers of the XGBoost model is set to 450, wherein the ratio of a predicted value to an actual value is 91.88% within +/-5 ℃; the MSE value is smaller than the set value 50, meanwhile, R2 is larger than the set value 0.92, the difference between the predicted value and the actual value is larger than 90% within +/-5 ℃, the model is judged to be qualified, and the LF refining temperature is predicted after the model is judged to be qualified.
The invention predicts the LF refining temperature based on a metallurgical mechanism and XGBoost algorithm, collects raw material data, production process data and temperature measurement data of the whole process of the refining production process based on the metallurgical mechanism, big data and machine learning technology, adopts a qualitative and quantitative combined method to analyze and predict the molten steel temperature in different LF ladle refining, is suitable for analyzing a large amount of nonlinear relation data in the LF refining production process, and can provide real-time early warning indication and operation guidance for production through an intelligent LF refining temperature prediction model.
The above embodiments are provided to illustrate the technical concept and features of the present invention and are intended to enable those skilled in the art to understand the content of the present invention and implement the same, and are not intended to limit the scope of the present invention. All equivalent changes or modifications made in accordance with the spirit of the present invention should be construed to be included in the scope of the present invention.
Claims (10)
1. The LF refining temperature forecasting method based on the mechanism model and XGBoost algorithm is characterized by comprising the following steps:
s1: acquiring process quality data in the LF refining production process, wherein the process quality data comprise refining power consumption, components and corresponding amounts of alloy and slag, argon blowing flow and time, argon blowing pressure and LF entering temperature;
S2: according to the law of conservation of energy, the energy income and expenditure conditions of the system are analyzed, and the energy change in the molten steel refining process is deduced on the basis of the energy income and expenditure conditions, so that the following formula is satisfied:
Qsteel＝QE＋Qalloy－Qslag－QIn－Qshell－QAr－Qsurface－Qsmoke
Wherein: q steel is the temperature rising and heating of molten steel; q E is the arc energy fed into the puddle; q alloy is the energy generated by adding the alloy; q slag is the energy lost by adding slag; q In is the heat accumulation of the furnace lining; q shell is the heat dissipated by the furnace shell; q Ar is the heat lost by argon blowing stirring; q surface is the heat dissipated by the molten steel and the slag surface; q smoke is the heat taken away by the smoke dust;
s3: according to thermodynamic law, the temperature rise of molten steel is analyzed and calculated, and the temperature rise formula is as follows:
,
Wherein: q steel is the temperature rising and heating of molten steel; c steel is the specific heat capacity of the molten steel; c slag is the specific heat capacity of the slag; m steel is the mass of molten steel; m slag is the mass of molten steel;
The temperature of molten steel at the time t is as follows:
S4: preprocessing the data obtained in the step S1, and removing repeated items and null values in the data; the z-score method is utilized to normalize data, eliminate the influence of the variation size and the numerical value of dimension and variable, and the box diagram detection method is utilized to delete abnormal points;
S5: substituting the parameters including the refining electricity consumption, the argon blowing time and the alloy addition amount obtained in the step S4 into the step S2 to calculate the molten steel heating temperature, calculating the molten steel temperature at the time t through the formula of the step S3, and adding the molten steel temperature as a newly added parameter into the data set obtained in the step S4;
S6: dividing the optimized data set in the step S5 into a training set and a testing set;
s7: constructing XGBoost models of training set data in a python environment, establishing a relation between input features and output features, sequencing contribution degrees of the input features to the output features, and finally eliminating features with importance scores smaller than 0.005 on the target index;
S8: verifying the test set by using the built XGBoost model, taking a mean square error MSE and a determination coefficient R 2 as evaluation criteria, judging XGBoost that the model is evaluated to be qualified if MSE values and R 2 values are within a preset range, and adopting a qualified XGBoost model to predict LF refining temperature; otherwise, if the operation is not qualified, the operations from the step S4 to the step S8 are carried out again.
2. The LF refining temperature prediction method based on a mechanism model and XGBoost algorithm as defined in claim 1, wherein in step S2, Q E is arc energy fed into the molten pool, and the formula is decomposed into:
,
Wherein: q E is the arc energy fed into the puddle; p a is arc power; phi is a proportionality coefficient; t 0 is the discharge time.
3. The LF refining temperature prediction method based on a mechanism model and XGBoost algorithm as claimed in claim 1, wherein in step S2, Q alloy is energy generated by adding an alloy, and the formula is:
,
wherein Q joxide is the oxidation exotherm of the alloy element j, which is calculated by the following formula:
Qjoxide＝ΔΗjoxideΜj（1－fj）
wherein: ΔH joxide is the heat of oxidation of alloy element j; f j is the yield of the alloy element;
,
wherein: m j is the mass of element j; n m is the number of alloys containing the j element; c ij is the content of element j in alloy i; the addition amount of the alloy i;
Q jmelt is the heat of fusion of element j in the solid alloy, calculated from the following formula:
,
Wherein: c jsolid、cjfluid is the solid phase and liquid phase specific heat capacities of the alloy elements j respectively; t jstart、Tjfluid The furnace charging temperature, the liquidus temperature and the preset subcontracting temperature of the alloy element j are respectively; the latent heat of fusion of Δh jmelt alloying element j;
Q jfuse is the heat of fusion of the alloy element j in the molten steel, and is calculated by the following formula:
Qjfuse＝ΔΗjfuseΜj
Wherein: ΔH jfuse is the heat of fusion of the alloying element j.
4. The LF refining temperature prediction method based on a mechanism model and XGBoost algorithm as claimed in claim 1, wherein in step S2, Q slag is energy generated by adding an alloy, and the formula decomposition is as follows:
,
Wherein: n slag is the number of slag materials; 、 And The method comprises the steps of respectively charging a slag charge, a liquidus temperature and a preset subcontracting temperature;、 the specific heat capacities of the solid phase and the liquid phase of the slag charge are respectively; The mass of the slag charge i; ΔH imelt is the latent heat of fusion of the slag.
5. The LF refining temperature prediction method based on the mechanism model and XGBoost algorithm according to claim 1, wherein in step S2Q In is the heat accumulation of the furnace lining, and the formula decomposition is as follows:
QIn＝cInWIn(TIn－TInstart)
Wherein: c In is the specific heat capacity of the furnace lining; w In is the mass of the furnace lining; t In is the temperature of the furnace lining; t Instart is the starting temperature of the furnace lining.
6. The LF refining temperature prediction method based on a mechanism model and XGBoost algorithm according to claim 1, wherein in step S2Q shell is heat dissipated from the furnace shell, and the formula decomposition is as follows:
,
Wherein: q shell is the heat dissipated by the furnace shell; k is a coefficient, and 4.88 is taken; epsilon is the surface blackness of the furnace body and is 0.95; t shell is the temperature of the outer surface of the furnace body; t 0 is the ambient temperature of the workshop; alpha shell is the convective heat transfer coefficient between the outer surface of the furnace body and the workshop environment; f is the external surface area of the furnace body; t 1 is the heat dissipation time of the furnace body;
if no transverse airflow flows in the workshop, the convection heat exchange coefficient between the outer surface of the furnace body and the workshop environment is calculated according to the following formula:
αshell＝k1(Tshell－T0)0.25
Wherein: k 1 is a coefficient, k 1 =2.8 when the heat radiation face is upward; when vertical, k 1 =2.2; downward, k 1 =1.5.
7. The LF refining temperature prediction method based on a mechanism model and XGBoost algorithm according to claim 1, wherein in step S2, Q Ar is heat lost by argon blowing stirring, and the formula decomposition is as follows:
,
Wherein: c p is the specific heat of argon; v Ar is argon blowing amount; t Ar is the initial temperature at which argon is blown in.
8. The LF refining temperature prediction method based on a mechanism model and XGBoost algorithm as claimed in claim 1, wherein in step S2Q surface is heat dissipated from the molten steel and slag surface, and the formula decomposition is as follows:
,
Wherein: the heat dissipation capacity of the exposed surface of the molten steel in the argon blowing process is obtained; The heat dissipation capacity of the slag layer;
Wherein: ,
Wherein: a steel is the exposed area of molten steel; epsilon steel is the surface blackness of molten steel, and 0.4 is taken; sigma is a stefin-boltzmann constant, taking 5.67 x 10 -8W/(m2·K4);Ta as the ambient temperature; t 2 is the bare heat dissipation time of molten steel;
,
Wherein: a slag is the area of slag; c slag is the heat loss coefficient, and 0.6 is taken; epsilon slag is the blackness of the slag surface, and 0.8 is taken; sigma is a stefin-boltzmann constant, and 5.67×10 -8W/(m2·K4);t3 is slag surface heat dissipation time.
9. The LF refining temperature prediction method based on a mechanism model and XGBoost algorithm according to claim 1, wherein in step S2, Q smoke is heat taken away by smoke in step S2, and the formula decomposition is as follows:
,
Wherein: w smoke is the mass of smoke; c smoke is the specific heat capacity of the smoke; t smoke is the temperature of the smoke; t 0 is the ambient temperature of the workshop; is the latent heat of fusion of the smoke.
10. The LF refining temperature prediction method based on a mechanism model and XGBoost algorithm as claimed in claim 1, wherein in step S8, the XGBoost model is set as a tree model gbtree, in which:
the maximum depth max-depth of the tree is set to 6;
The learning rate cta is set to 0.222 in the model;
gbtree the classifier number was set to 450;
The MSE value is set to less than 50, the R 2 value is set to greater than 0.92, and the ratio of the predicted value to the actual value within + -5 ℃ is greater than 90%.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410429459.7A CN118016203B (en) | 2024-04-10 | 2024-04-10 | LF refining temperature forecasting method based on mechanism model and XGBoost algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410429459.7A CN118016203B (en) | 2024-04-10 | 2024-04-10 | LF refining temperature forecasting method based on mechanism model and XGBoost algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118016203A true CN118016203A (en) | 2024-05-10 |
| CN118016203B CN118016203B (en) | 2024-06-14 |
Family
ID=90944940
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410429459.7A Active CN118016203B (en) | 2024-04-10 | 2024-04-10 | LF refining temperature forecasting method based on mechanism model and XGBoost algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118016203B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118504364A (en) * | 2024-07-18 | 2024-08-16 | 苏州方兴信息技术有限公司 | Method for predicting converter end temperature based on XGBoost and finite element combined driving |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090031860A1 (en) * | 2004-11-12 | 2009-02-05 | Johann Reichel | Production of Stainless Steel of AISI 4XX Grade Ferritic Steel in an Aod Converter |
| CN115114968A (en) * | 2020-10-09 | 2022-09-27 | 武汉科技大学 | Intelligent casting blank crack prediction method based on random forest |
| CN115829938A (en) * | 2022-11-07 | 2023-03-21 | 中国钢研科技集团有限公司 | Method for predicting gray wolf and NGboost for surface defects of casting blank genetic hot-rolled strip steel |
| CN116469481A (en) * | 2023-06-19 | 2023-07-21 | 苏州方兴信息技术有限公司 | LF refined molten steel composition forecasting method based on XGBoost algorithm |
| CN117494531A (en) * | 2023-12-29 | 2024-02-02 | 苏州方兴信息技术有限公司 | Medium carbon steel decarburization depth prediction method based on finite element and XGBoost algorithm |
-
2024
- 2024-04-10 CN CN202410429459.7A patent/CN118016203B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090031860A1 (en) * | 2004-11-12 | 2009-02-05 | Johann Reichel | Production of Stainless Steel of AISI 4XX Grade Ferritic Steel in an Aod Converter |
| CN115114968A (en) * | 2020-10-09 | 2022-09-27 | 武汉科技大学 | Intelligent casting blank crack prediction method based on random forest |
| CN115829938A (en) * | 2022-11-07 | 2023-03-21 | 中国钢研科技集团有限公司 | Method for predicting gray wolf and NGboost for surface defects of casting blank genetic hot-rolled strip steel |
| CN116469481A (en) * | 2023-06-19 | 2023-07-21 | 苏州方兴信息技术有限公司 | LF refined molten steel composition forecasting method based on XGBoost algorithm |
| CN117494531A (en) * | 2023-12-29 | 2024-02-02 | 苏州方兴信息技术有限公司 | Medium carbon steel decarburization depth prediction method based on finite element and XGBoost algorithm |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118504364A (en) * | 2024-07-18 | 2024-08-16 | 苏州方兴信息技术有限公司 | Method for predicting converter end temperature based on XGBoost and finite element combined driving |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118016203B (en) | 2024-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN118016203B (en) | LF refining temperature forecasting method based on mechanism model and XGBoost algorithm | |
| CN108676955B (en) | A kind of BOF Steelmaking Endpoint carbon content and temprature control method | |
| CN108342540B (en) | A kind of steel mill RH refining equipment automatic control system | |
| CN103388054A (en) | System and method for on-line control of molten steel temperature in LF refining | |
| CA2894813C (en) | Method and device for predicting, controlling and/or regulating steelworks processes | |
| CN109359723A (en) | Based on the converter terminal manganese content prediction technique for improving regularization extreme learning machine | |
| CN108647407A (en) | A kind of pneumatic steelmaking flue gas analysis carbon determination method | |
| CN115522012B (en) | Method for controlling TSC stage oxygen blowing amount of large converter | |
| CN113962050A (en) | Oxygen scheduling calculation method combining production consumption prediction and pipe network calculation | |
| CN116469481B (en) | LF refined molten steel composition forecasting method based on XGBoost algorithm | |
| JPWO2021014923A1 (en) | Process control method, operation guidance method, blast furnace operation method, hot metal manufacturing method and process control device | |
| Fei et al. | Prediction of oxygen-blowing volume in BOF steelmaking process based on BP neural network and incremental learning | |
| Xin et al. | Modeling of LF refining process: A review | |
| CN110373508A (en) | Cohesive zone recognition methods, device and electronic equipment | |
| CN108491679A (en) | A kind of steel Preparation Method and system based on gene pool | |
| CN116911057A (en) | Real-time prediction method for temperature of molten pool in converter steelmaking converting process | |
| Larsen et al. | Intelligent control of cupola melting | |
| CN116663774A (en) | Converter process evaluation method and system based on digital twin | |
| CN110059940A (en) | A kind of steel smelting-continuous casting interface linking power-economizing method and system | |
| CN114381569B (en) | Converter system gas recovery optimization comprehensive evaluation method | |
| KR101246436B1 (en) | Prediction method for product measuring of pig iron | |
| Ksiazek et al. | Measurement of metal temperature during tapping of an industrial FeSi furnace | |
| JP6516906B1 (en) | Wind blow calculation method, blow blow calculation program | |
| JPH11335710A (en) | Method for predicting furnace heat in blast furnace | |
| Huang et al. | Greedy kernel components acting on ANFIS to predict BOF steelmaking endpoint |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |