CN1603026A - Method for real-time estimating temperature of liquid steel in RH fining furnace - Google Patents
Method for real-time estimating temperature of liquid steel in RH fining furnace Download PDFInfo
- Publication number
- CN1603026A CN1603026A CNA031514367A CN03151436A CN1603026A CN 1603026 A CN1603026 A CN 1603026A CN A031514367 A CNA031514367 A CN A031514367A CN 03151436 A CN03151436 A CN 03151436A CN 1603026 A CN1603026 A CN 1603026A
- Authority
- CN
- China
- Prior art keywords
- refining furnace
- temperature
- steel
- liquid steel
- 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
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P10/00—Technologies related to metal processing
- Y02P10/25—Process efficiency
Landscapes
- Treatment Of Steel In Its Molten State (AREA)
Abstract
The invention has provides one RH refining furnace steel fluid temperature implementation forecast method, it can real-time, accurately forecasts the RH precision stove the steel fluid temperature, thus both enhanced time the fining processing conclusion temperature hit probability, and reduced has measured the warm number of times.The method including below step: (1) calculates may use the RH refining furnace steel fluid temperature change component which the metallurgy mechanism model or the production data quantification factor causes; (2) the RH refining furnace steel fluid temperature change component which will cause using the nerve network model computation natural temperature drop states the nerve network model input level the input variable to contain from processing starts the time to this time time-gap, the RH refining furnace the environment temperature and the steel fluid weight, the output level output variable RH which will cause for the natural temperature drop builds up the stove steel fluid temperature the change component;step (3) getherthe (1) and (2) calculates the change component which obtains this time the RH refining furnace steel fluid temperature opposite change quantity of starts the time.
Description
Technical field
The present invention relates to the production and the control field of metallurgical process, particularly a kind of real-time predicting method of RH Liquid Steel in Refining Furnace temperature.
Background technology
The RH refining furnace is the main technique equipment of high-grade steel grade refining treatment, the principal feature of this technology is by molten steel circulating in vacuum tank and ladle, and be aided with top rifle oxygen blast, realize the metallurgical functions such as impurity in decarburization, the degassing, adjustment molten steel temperature and composition, the removal molten steel.Fig. 1 shows a kind of synoptic diagram of typical R H refining furnace.As shown in Figure 1, vacuum system 1 is connected with RH vacuum tank 3, when the end opening of vacuum tank 3 immerses molten steel in the ladle 2 fully, will form a closed system.The port that joins with material alloying system 4 and oxygen rifle 5 also is set on the vacuum tank 3.When RH handles beginning, start vacuum system 1 exhaust and vacuumize.In treating processes subsequently, in order to reach every processing requirement, can add the alloy of certain kind and quantity by material alloying system 4, and by oxygen rifle 5 oxygen blast in molten steel.At last, when molten steel component and temperature reach target call, finish whole RH treating processes thereby make vacuum system 1 stop exhaust.
As the processing unit of realizing the middle operation of converter and continuous casting, the RH refining furnace has been born the task of adjusting liquid steel temperature, because liquid steel temperature is most important for the production that guarantees molten steel quality and follow-up continuous casting, therefore must predict the molten steel temperature in the RH refining furnace exactly.
In actual production at present, operator often use the molten steel temperature information during disposable thermocouple temperature measurement obtains whole RH processing.Because it is longer that RH handles the cycle, therefore need operator to carry out repeatedly thermometric, the loss that this certainly will increase thermopair has improved production cost.And this thermometric mode can only obtain the temperature of several time points in the entire treatment cycle, lack the continuous temperature information that whole RH handles the cycle, this makes operator often will or add cold material cooling by heating by O2 blowing repeatedly, just can hit the target temperature that will reach when processing finishes, thereby increase the production cost of refining treatment.
There has been people's research and utilization theoretical model prediction RH Liquid Steel in Refining Furnace method of temperature to replace or reduce the temperature survey of reality, these class methods rest on the pure physical computing mostly, or predict temperature drop speed, thereby forecast the temperature of RH Liquid Steel in Refining Furnace according to the heat transfer model that thermodynamics and heat balance theory are set up.But because the RH treatment process is complicated unusually, the factor that influences temperature variation is many, in order to obtain the real-time estimate value of molten steel temperature, existing Forecasting Methodology has been done a large amount of hypothesis and simplification to computation process, thereby the physical model and the truth that cause being set up differ bigger, in addition, and in order to reduce the complicacy that non-linear factor causes, existing Forecasting Methodology needs many non-linear factors are made linearization process, and the factor of unstable state is done the stable state processing.Above-mentioned these are simplified to handle has influenced the molten steel temperature precision of prediction greatly, and therefore up to the present, Shang Weiyou can reach the Forecasting Methodology that practical application in industry requires.
Summary of the invention
The purpose of this invention is to provide a kind of RH Liquid Steel in Refining Furnace temperature real-time predicting method, it can predict the molten steel temperature of RH refining furnace in real time, exactly, thereby has both improved the temperature hit rate when refining treatment finishes, and has reduced the thermometric number of times again.
Goal of the invention of the present invention realizes by following technical proposal:
A kind of RH Liquid Steel in Refining Furnace temperature real-time predicting method, the RH Liquid Steel in Refining Furnace temperature in arbitrary moment is with respect to the variable quantity of handling the zero hour during calculating RH and handle according to following steps:
(1) calculates the RH Liquid Steel in Refining Furnace temperature variation component that available metallurgy mechanism model or production data quantification factor cause;
(2) utilize neural network model to calculate the RH Liquid Steel in Refining Furnace temperature variation component that the nature temperature drop causes, the input variable of described neural network model input layer comprises from handling the timed interval, RH refining furnace ambient temperature and the molten steel weight in the zero hour to this moment, the RH Liquid Steel in Refining Furnace variation of temperature component that the output variable of output layer causes for the nature temperature drop;
(3) change component that step (1) and (2) are calculated is obtained the RH Liquid Steel in Refining Furnace temperature in this moment mutually with respect to the variable quantity of handling the zero hour.
Reasonable is that in above-mentioned RH Liquid Steel in Refining Furnace temperature real-time predicting method, the factor that described available metallurgy mechanism model or production data are described comprises kind and quantity and the blowing oxygen quantity that adds alloy in molten steel.
Reasonable is that in above-mentioned RH Liquid Steel in Refining Furnace temperature real-time predicting method, the factor that described available metallurgy mechanism model or production data are described comprises that stove cast finishes to the timed interval and the cold steel quantity of ladle bottom of this stove tapping beginning on the ladle.
Reasonable is that in above-mentioned RH Liquid Steel in Refining Furnace temperature real-time predicting method, described neural network model adopts three layers of back transfer (BP) network neuromechanism.Be more preferably, the excitation function of described three layers of BP network neuromechanism is a logarithm S shape transport function, and its structure is:
Forecasting Methodology of the present invention can accurately be forecast molten steel temperature, make operator effectively control treating processes according to the temperature forecast value, improve the temperature hit rate when treating processes finishes, reduced the thermometric number of times in the treating processes, significant for reducing production costs.
Description of drawings
By below in conjunction with the description of accompanying drawing to preferred embodiment of the present invention, can further understand purpose of the present invention, feature and advantage, wherein:
Fig. 1 is a kind of synoptic diagram of typical RH refining furnace.
Fig. 2 is the topological structure synoptic diagram of three layers of back transfer (BP) neural network of first embodiment of the invention employing.
Fig. 3 shows first embodiment and is applied to the forecasting accuracy evaluation result synoptic diagram that actual production obtains.
Embodiment
In Forecasting Methodology of the present invention, with the factor merger that influences molten steel temperature is two classes, one class is to carry out quantized based on metallurgy mechanism model or production data, the for example quantity of alloy, kind and blowing oxygen quantity etc., another kind of factor often has very complicated nonlinear characteristic, be difficult to accurately describe with physical model or thermodynamical model and production data, below we are referred to as nature temperature drop factor.For last class factor, calculate as long as adopt based on known metallurgy mechanism model or the production data algebraic model that draws of deriving, and for back one class factor, the present invention then sums up in the point that it in neural network model and avoids adopting physical model to be described, that is to say, input-output relation between the temperature variation that outer bound variable and natural temperature drop cause is set up in the study of a series of measured datas by making neural network.Can under the prerequisite that reduces computation complexity, improve precision of prediction thus, thereby reach the purpose of predicting RH Liquid Steel in Refining Furnace temperature in real time, exactly.
It is worthy of note, the contriver is through discovering, just can portray the temperature variation that nature temperature drop factor causes fully to the timed interval, RH refining furnace ambient temperature and the molten steel weight that will predict the moment zero hour as long as the input variable of neural network model comprises from the RH processing, precision can satisfy the needs of practical application fully.
Below further describe the preferred embodiment of Forecasting Methodology of the present invention by accompanying drawing.
First embodiment
Suppose that the RH treating processes betides in the RH refining furnace shown in Figure 1, for the sake of simplicity, will handle being set at 0 the zero hour, then constantly during t in the RH refining furnace molten steel temperature constantly be that 0 variable quantity f (t) can be expressed as relatively:
f(t)＝T
NATURAL(t)+T
ALLOY+T
KTB??(1)
Wherein, T
NATURAL(t) the RH Liquid Steel in Refining Furnace temperature variation component that causes for nature temperature drop factor, this change component need adopt below the neural network model that is further described is calculated, T
ALLOYBe kind that adds alloy and the RH Liquid Steel in Refining Furnace temperature variation component that quantity causes, T
KTBBe the RH Liquid Steel in Refining Furnace temperature variation component that blowing oxygen quantity causes, they can calculate according to known metallurgy mechanism model, also will be described specifically below.
In the present embodiment, adopt three layers of back transfer (BP) network neuromechanism to handle the change component T that nature temperature drop factor causes with topological structure shown in Figure 2
NATURAL(t), this network neuromechanism is a kind of multilayer feedforward neural network of error back propagation, can realize from being input to any nonlinear mapping of output.As shown in Figure 2, the topological structure of this neural network comprises three layers of input layer, hidden layer and output layers, wherein, input layer comprises three input nodes, corresponding respectively these three input variables of the timed interval, RH refining furnace ambient temperature and molten steel weight that extremely will predict the moment of handling from RH zero hour, output layer comprises 1 output node, the change component T that corresponding natural temperature drop factor causes
NATURAL(t); In addition, the excitation function in the neural network shown in Figure 2 adopts logarithm S shape (Sigmoid) transport function, and its structure is:
In order to set up the change curve of nature temperature drop, provide many group measured datas as learning sample to this neural network, every group of data comprise handles the zero hour to the timed interval, RH refining furnace ambient temperature and the molten steel weight that will predict the moment and the change component T that causes as the natural temperature drop factor of output variable as input variable from RH
NATURAL(t).Through study, this neural network will be set up certain input-output relation.When the prediction molten steel temperature,, can utilize input-output relation of having set up to calculate the change component T of nature temperature drop as long as concrete input variable is imported this neural network
NATURAL(t).
It is worthy of note, handle the change component T that nature temperature drop factor causes
NATURAL(t) used neural network is not limited to structure shown in Figure 2, excitation function also is not limited to the form shown in the formula (2), it will be apparent to one skilled in the art that, adopt neural network model to handle after nature temperature drop factor can effectively avoid computational complexity and improve the real-time, this key point of accuracy of temperature prediction recognizing, the neural network structure that adopts adequate types as the case may be is conspicuous, and therefore three layers of BP neural network that adopted with the alternative present embodiment of the neural network of other type all belong to and are equal to replacement.
Below describe alloy and add the molten steel temperature variable quantity T that causes
ALLOYMethod of calculation.In RH refining treatment process, the alloy of adding generally is divided into two classes according to the chemical action that is risen: common alloy and specific alloy.Common alloy is meant that only there be (promptly only playing alloying action) in the alloy of adding as the composition element in the molten steel, and itself does not participate in deoxidation.Specific alloy then is the composition element in the molten steel, plays the effect of reductor again.When the calculating alloy adds the molten steel temperature variable quantity that causes, at first should judge that the effect that adds alloy be as reductor or composition element according to adding the residing state of fashionable molten steel, or not only as reductor but also as the composition element.The temperature drop calculation formula of various alloys under different situations listed in table 1～3, and unit is ℃/kg/ton that promptly the add-on of every kilogram of alloy makes the temperature change amount of molten steel per ton.After definite temperature drop calculation formula, the caused molten steel temperature change component of adding that the add-on and the corresponding temperature drop calculation formula of molten steel weight substitution of every kind of alloy can be calculated every kind of alloy.At last, these temperature variation component additions are promptly obtained T
ALLOY
(1) specific alloy
Specific alloy mainly contains aluminium (Al), ferrosilicon (FeSi) and carbon dust (C-PW), when its during not only as reductor but also as the composition element, the temperature drop calculation formula is as shown in table 1.
Table 1
| The alloy that adds | Add purpose | Every kilogram of alloy make ton steel the temperature change amount (℃/kg/t) |
| ????Al | Deoxidation and alloying | 0.144×W Al/W STEEL+3.353×10 -2×Δ[0] |
| ????FeSi(77％) | Deoxidation and alloying | 0.81×W FeSi/W STEEL+2.32×10 -2×Δ[0] |
| ????C-PW | Deoxidation and alloying | -5.92×W C-PW/W STEEL+0.178×10 -2×Δ[0] |
If specific alloy only is used for the alloying of molten steel, then its temperature drop calculation formula is as shown in table 2.
Table 2
| The alloy that adds | Add purpose | Every kilogram of alloy make ton steel the temperature change amount (℃/kg/t) |
| ??Al | Alloying | 0.144×W Al/W STEEL |
| ??FeSi(77％) | Alloying | 0.81×W FeSi/W STEEL |
| ??C-PW | Alloying | -5.92×W C-PW/W STEEL |
W in table 1 and the table 2
STEELBe molten steel weight; W
Al, W
FeSi, W
C-PWBe respectively the weight of aluminium, ferrosilicon and the carbon dust of adding.The free oxygen concentration of Δ [O] for sloughing, unit is ppm.
(2) common alloy
Common alloy mainly contains cold material, ferromanganese, ferrotianium, ferro-boron, ferrophosphorus, ferrochrome, molybdenum-iron and ferro-niobium, and their effect is an alloying, and its temperature drop calculation formula is as shown in table 3.
Table 3
| The alloy that adds | Add purpose | Every kilogram of alloy make ton steel the temperature change amount (℃/kg/t) |
| ????H-Mn | Alloying | ????-2.26×W H-Mn/W STEEL |
| ????M-Mn | Alloying | ????-2.01×W M-Mn/W STEEL |
| ????L-Mn | Alloying | ????-2.0×W L-Mn/W STEEL |
| ????LCTi(25％) | Alloying | ????0.1×W LCTi/W STEEL |
| ????Fe-B | Alloying | ????-4.02×W Fe-B/W STEEL |
| ????Fe-P | Alloying | ????-3.53×W Fe-P/W STEEL |
| ????LCCr | Alloying | ????-2.0×W LCCr/W STEEL |
| ????Fe-Mo | Alloying | ????-1.21×W Fe-Mo/W STEEL |
| ????ScHP | Alloying | ????-2.1×W ScHP/W STEEL |
| ????FeSi(77％) | Alloying | ????0.81×W FeSi/W STEEL |
W in the table 3
STEELBe molten steel weight; W
H-Mn, W
M-Mn, W
L-Mn, W
L-CTi, W
Fe-B, W
LCCr, W
Fe-Mo, W
ScHP, W
FeSiBe respectively the weight of the various common alloys of adding.
In the RH treating processes, oxygen blast mainly contains two purposes: decarburization and intensification.For the molten steel of decarburization, oxygen blast is mainly used to heat up.For the Ultra-low carbon molten steel of not decarburization, handle the carrying out that the oxygen blast in early stage is mainly used to provide the needed oxygen of decarburization and promotes decarburizing reaction, handle the later stage because molten steel has been finished decarburization and become decarburization molten steel, therefore oxygen blown main effect is intensification.The molten steel temperature variable quantity T that oxygen blast causes under the both of these case is below described respectively
KTBMethod of calculation.
(1) oxygen blast of decarburization molten steel
When the decarburization molten steel is blown into oxygen, the metallic element in the molten steel will with oxygen generation chemical reaction, thereby these chemical reactions cause absorption or release of heat the variation of molten steel temperature.In order to determine this temperature variation, at first can determine to be blown into the utilization ratio of oxygen according to production data, promptly have the oxygen of how much quantity to participate in chemical reaction and every kind of oxygen proportion that chemical reaction consumes, the calorimeter that absorbs or discharge according to every kind of chemical reaction is calculated and is blown into the molten steel temperature change component T that oxygen causes then
KTB
(2) the not oxygen blast of decarburization molten steel
When the decarburization molten steel is not blown into oxygen, oxygen decarburization decarburizing reaction heat effect in the molten steel when not having oxygen blast is identical, therefore can not consider the temperature variation that oxygen decarburization causes, what need to consider only is to dissolve in to be blown into the influence of oxygen part to molten steel temperature in the molten steel.Dissolve in being blown into amount of oxygen and can characterizing in the molten steel with the free oxygen concentration in the molten steel, can define the oxygen that how much is blown into by the increasing amount of measuring the oxygen concn that dissociates in the molten steel and dissolve in molten steel, according to the free oxygen concentration increasing amount of production data acquisition and the corresponding relation of temperature variation, can calculate and be blown into the molten steel temperature change component T that oxygen causes then
KTB
Fig. 3 shows the foregoing description and is applied to the forecasting accuracy evaluation result that actual production obtains, and wherein, ordinate zou is the stove number, and X-coordinate is the absolute value of the difference of observed temperature and predicted temperature.As shown in Figure 3, in amounting to the RH refining treatment process of 52 stoves, the predicted temperature of 45 stoves and the error between the observed temperature are arranged less than 5 ℃, the error that 3 stoves are arranged is between 5～6 ℃, the error that 3 stoves are arranged is between 6～8 ℃, and the error that also has 1 stove is between 8～10 ℃.Consider that the molten steel temperature in the RH treating processes all will be about 1600 ℃, such precision of prediction is very high.
It is worthy of note; because alloy and oxygen blast can be adopted metallurgy mechanism model or production data to describe accurately to the influence of molten steel temperature and it will be apparent to one skilled in the art that method of calculation also are known; therefore above-mentioned alloy adds and the method for calculation of the molten steel temperature variable quantity that oxygen blast causes only are illustrative nature, should not constitute the qualification to the present invention's spirit and protection domain.
Second embodiment
Suppose that the RH treating processes betides in the RH refining furnace shown in Figure 1, for the sake of simplicity, will handle being set at 0 the zero hour, be with the difference of first embodiment, constantly during t in the RH refining furnace molten steel temperature constantly be that 0 variable quantity f (t) is expressed as relatively:
f(t)＝T
NATURAL(t)+T
ALLOY+T
KTB+T
LADLEB(t)??(3)
Wherein, T
NATURAL(t) the RH Liquid Steel in Refining Furnace temperature variation component that causes for nature temperature drop factor, this change component adopt the mode identical with first embodiment to calculate T
ALLOYBe kind that adds alloy and the RH Liquid Steel in Refining Furnace temperature variation component that quantity causes, T
KTBBe the RH Liquid Steel in Refining Furnace temperature variation component that blowing oxygen quantity causes, they also adopt the mode identical with first embodiment to calculate.T
LADLEB(t) the molten steel temperature variable quantity that causes for the cold steel state of the ladle state and the bag end below it will be appreciated that, they mainly are to handle early stage at RH to the influence of molten steel temperature, and compare with molten steel temperature, and influence degree is also less.Below the account form of ladle state and the bag end molten steel temperature variable quantity that causes of cold steel state is described in detail.
So-called ladle state is promptly gone up stove cast and is finished to the timed interval of this stove tapping beginning, when this timed interval is longer, handle the initial stage at RH, ladle will absorb more heat, therefore make that the reduction degree of molten steel temperature is bigger, otherwise ladle will absorb less heat, therefore make that the reduction degree of molten steel temperature is less.In order to simplify processing, the ladle state can be divided into 1～6 and amount to 6 grades, also be about to the timed interval and be divided into 6 segment limits, every grade or a temperature revisal of every section correspondence amount, and calculate the molten steel temperature variable quantity that causes according to following formula:
ΔT
1(t)＝T
B1×t/8(t≤8)?????(4)
Wherein, Δ T
1(t) be the t temperature of molten steel constantly, T
B1Be the temperature revisal amount corresponding with section with appropriate level, concrete numerical value determines that according to the condition of production table 4 shows an example.According to knowhow, the heat absorption of ladle mainly occurs in preceding 8 minutes that RH handles, therefore t≤8 minute here.
The cold steel state in the so-called bag end is promptly gone up the cold steel quantity of staying ladle bottom after stove cast finishes.It is apparent in view to handle behind the beginning several minutes molten steel circulation in the ladle from RH, continue to reach circulation evenly behind the several minutes, therefore wrap the cold steel in the end influence for the treatment of processes temperature drop is confined to during this period of time, generally speaking, be the 3rd～6 minute after the RH processing during this period of time.If cold steel quantity is more, more heat will be absorbed, so the reduction degree of molten steel temperature is bigger, otherwise the heat that cold steel absorbs is less, so the reduction degree of molten steel temperature is less.In order to simplify processing, cold steel state can be divided into A～B and amount to 5 grades, also be about to cold steel quantity and be divided into 5 segment limits, every grade or a temperature revisal of every section correspondence amount, and calculate the molten steel temperature variable quantity that causes according to following formula:
ΔT
2(t)＝T
B2(t-3)/3(3≤t≤6)??(5)
Wherein, Δ T
2(t) be the t temperature of molten steel constantly, T
B2Be the temperature revisal amount corresponding with section with appropriate level, concrete numerical value determines that according to the condition of production table 4 shows an example.Here the cold steel in the hypothesis bag end is confined to handle the 3rd～6 minute of beginning from RH to the influence for the treatment of processes temperature drop, i.e. 3≤t≤6 minute.
The temperature variation Δ T that cold steel state at the bottom of above-mentioned ladle state and the bag is caused
1(t) and Δ T
2(t) addition can obtain T
LADLEB(t).
Table 4
| The ladle state | The temperature revisal (℃) | The cold steel in the bag end | The temperature revisal (℃) |
| ?1 | 0 | ?A | ?0 |
| ?2 | 0 | ?B | -5 |
| ?3 | -3 | ?C | -8 |
| ?4 | -5 | ?D | -12 |
| ?5 | -10 | ?E | Abnormality processing |
| ?6 | -10 |
It is worthy of note, can adopt the metallurgy mechanism model or accurately determine the cold steel state etc. of being not limited at the bottom of the listed alloy adding of above-mentioned first and second embodiment, oxygen blast, ladle state and bag factor the molten steel temperature influence degree according to production data, because the diversity of practical application, here can't be exhaustive go out various factors, when for example after vacuum tank is shelved for a long time, coming into operation again, its influence factor to molten steel temperature is not done to discuss in detail in specification sheets, and therefore the description of above-mentioned first and second embodiment only is an illustrative nature.It will be apparent to one skilled in the art that; as long as just can obtain to adopt neural network model to handle the enlightenment of the temperature effect factor that metallurgy mechanism model or production data can't accurately describe by reading this specification sheets; as for how to determine accurately that according to metallurgy mechanism model or production data this class factor then is conspicuous thing to the influence degree of molten steel temperature, therefore the description of above-mentioned illustrative nature should not constitute the qualification to the present invention's spirit and protection domain.
Claims (6)
1. a RH Liquid Steel in Refining Furnace temperature real-time predicting method is characterized in that, the RH Liquid Steel in Refining Furnace temperature in arbitrary moment is with respect to the variable quantity of handling the zero hour during calculating RH and handle according to following steps:
(1) calculates the RH Liquid Steel in Refining Furnace temperature variation component that available metallurgy mechanism model or production data quantification factor cause;
(2) utilize neural network model to calculate the RH Liquid Steel in Refining Furnace temperature variation component that the nature temperature drop causes, the input variable of described neural network model input layer comprises from handling the timed interval, RH refining furnace ambient temperature and the molten steel weight in the zero hour to this moment, the RH Liquid Steel in Refining Furnace variation of temperature component that the output variable of output layer causes for the nature temperature drop;
(3) change component that step (1) and (2) are calculated is obtained the RH Liquid Steel in Refining Furnace temperature in this moment mutually with respect to the variable quantity of handling the zero hour.
2. RH Liquid Steel in Refining Furnace temperature real-time predicting method as claimed in claim 1 is characterized in that, the factor that described available metallurgy mechanism model or production data are described comprises kind and quantity and the blowing oxygen quantity that adds alloy in molten steel.
3. RH Liquid Steel in Refining Furnace temperature real-time predicting method as claimed in claim 1 or 2, it is characterized in that the factor that described available metallurgy mechanism model or production data are described comprises that stove cast finishes to the timed interval and the cold steel quantity of ladle bottom of this stove tapping beginning on the ladle.
4. RH Liquid Steel in Refining Furnace temperature real-time predicting method as claimed in claim 1 or 2 is characterized in that, described neural network model adopts three layers of back transfer (BP) network neuromechanism.
5. RH Liquid Steel in Refining Furnace temperature real-time predicting method as claimed in claim 3 is characterized in that, described neural network model adopts three layers of back transfer (BP) network neuromechanism.
6. RH Liquid Steel in Refining Furnace temperature real-time predicting method as claimed in claim 4 is characterized in that the excitation function of described three layers of BP network neuromechanism is a logarithm S shape transport function, and its structure is:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB031514367A CN1298458C (en) | 2003-09-29 | 2003-09-29 | Method for real-time estimating temperature of liquid steel in RH fining furnace |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB031514367A CN1298458C (en) | 2003-09-29 | 2003-09-29 | Method for real-time estimating temperature of liquid steel in RH fining furnace |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1603026A true CN1603026A (en) | 2005-04-06 |
| CN1298458C CN1298458C (en) | 2007-02-07 |
Family
ID=34659943
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB031514367A Expired - Lifetime CN1298458C (en) | 2003-09-29 | 2003-09-29 | Method for real-time estimating temperature of liquid steel in RH fining furnace |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1298458C (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101619378B (en) * | 2008-06-30 | 2011-07-27 | 鞍钢股份有限公司 | Molten steel deoxidation method |
| CN102277468A (en) * | 2010-06-12 | 2011-12-14 | 上海梅山钢铁股份有限公司 | Real-time forecasting method of LF refining furnace molten steel temperature |
| CN101320031B (en) * | 2008-05-27 | 2012-09-26 | 上海实达精密不锈钢有限公司 | Austenitic stainless steel accurate steel strip performance prediction model and cold rolling process planning thereof |
| CN102867220A (en) * | 2012-06-25 | 2013-01-09 | 攀钢集团研究院有限公司 | Method for forecasting temperature of refined molten steel in ladle refining furnace in real time |
| CN103276147A (en) * | 2013-06-06 | 2013-09-04 | 鞍钢股份有限公司 | Forecasting method for molten steel temperature in LF refining process |
| CN103305656A (en) * | 2013-03-27 | 2013-09-18 | 马钢(集团)控股有限公司 | IF steel RH vacuum decarburization process control method |
| CN103645633A (en) * | 2013-12-25 | 2014-03-19 | 中国科学院自动化研究所 | Furnace temperature self-learning control method of conversion furnace system |
| CN105652666A (en) * | 2016-03-09 | 2016-06-08 | 中南大学 | Large die forging press beam feeding speed predictive control method based on BP neural networks |
| CN107557535A (en) * | 2017-07-31 | 2018-01-09 | 唐山钢铁集团有限责任公司 | A kind of method for improving refining work and being precisely controlled liquid steel temperature |
| CN110490260A (en) * | 2019-08-22 | 2019-11-22 | 联峰钢铁（张家港）有限公司 | A kind of method and apparatus identifying iron packet sky packet temperature drop |
| CN110684882A (en) * | 2019-11-05 | 2020-01-14 | 山东钢铁集团日照有限公司 | Oxygen-containing steel RH vacuum carbon powder deoxidation cooling method |
| CN110850915A (en) * | 2018-08-21 | 2020-02-28 | 上海梅山钢铁股份有限公司 | Self-learning-based steelmaking molten steel process temperature control system and control method |
| CN111020118A (en) * | 2019-12-25 | 2020-04-17 | 武汉科技大学 | RH endpoint temperature prediction method based on particle swarm optimization case reasoning |
| CN114787394A (en) * | 2019-11-29 | 2022-07-22 | 杰富意钢铁株式会社 | Operating method for ladle refining treatment |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2907672B2 (en) * | 1993-03-12 | 1999-06-21 | 株式会社日立製作所 | Process adaptive control method and process control system |
| JP3352153B2 (en) * | 1993-06-17 | 2002-12-03 | 株式会社東芝 | Water distribution flow prediction device |
| CN1242848A (en) * | 1996-11-20 | 2000-01-26 | 罗伯特·J·詹恩阿罗尼 | Multi-kernel neural network concurrent learning, monitoring and forecasting system |
| US5832421A (en) * | 1996-12-13 | 1998-11-03 | Siemens Corporate Research, Inc. | Method for blade temperature estimation in a steam turbine |
-
2003
- 2003-09-29 CN CNB031514367A patent/CN1298458C/en not_active Expired - Lifetime
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101320031B (en) * | 2008-05-27 | 2012-09-26 | 上海实达精密不锈钢有限公司 | Austenitic stainless steel accurate steel strip performance prediction model and cold rolling process planning thereof |
| CN101619378B (en) * | 2008-06-30 | 2011-07-27 | 鞍钢股份有限公司 | Molten steel deoxidation method |
| CN102277468A (en) * | 2010-06-12 | 2011-12-14 | 上海梅山钢铁股份有限公司 | Real-time forecasting method of LF refining furnace molten steel temperature |
| CN102277468B (en) * | 2010-06-12 | 2013-04-24 | 上海梅山钢铁股份有限公司 | Real-time forecasting method of LF refining furnace molten steel temperature |
| CN102867220A (en) * | 2012-06-25 | 2013-01-09 | 攀钢集团研究院有限公司 | Method for forecasting temperature of refined molten steel in ladle refining furnace in real time |
| CN102867220B (en) * | 2012-06-25 | 2016-09-21 | 攀钢集团研究院有限公司 | A kind of method of real-time estimate ladle refining furnace refined molten steel temperature |
| CN103305656B (en) * | 2013-03-27 | 2016-05-18 | 马钢(集团)控股有限公司 | A kind of IF steel RH vacuum decarburization course control method for use |
| CN103305656A (en) * | 2013-03-27 | 2013-09-18 | 马钢(集团)控股有限公司 | IF steel RH vacuum decarburization process control method |
| CN103276147A (en) * | 2013-06-06 | 2013-09-04 | 鞍钢股份有限公司 | Forecasting method for molten steel temperature in LF refining process |
| CN103645633A (en) * | 2013-12-25 | 2014-03-19 | 中国科学院自动化研究所 | Furnace temperature self-learning control method of conversion furnace system |
| CN103645633B (en) * | 2013-12-25 | 2017-01-18 | 中国科学院自动化研究所 | Furnace temperature self-learning control method of conversion furnace system |
| CN105652666A (en) * | 2016-03-09 | 2016-06-08 | 中南大学 | Large die forging press beam feeding speed predictive control method based on BP neural networks |
| CN105652666B (en) * | 2016-03-09 | 2018-09-11 | 中南大学 | Large-scale drop press upper beam prediction of speed control method based on BP neural network |
| CN107557535A (en) * | 2017-07-31 | 2018-01-09 | 唐山钢铁集团有限责任公司 | A kind of method for improving refining work and being precisely controlled liquid steel temperature |
| CN110850915A (en) * | 2018-08-21 | 2020-02-28 | 上海梅山钢铁股份有限公司 | Self-learning-based steelmaking molten steel process temperature control system and control method |
| CN110490260A (en) * | 2019-08-22 | 2019-11-22 | 联峰钢铁（张家港）有限公司 | A kind of method and apparatus identifying iron packet sky packet temperature drop |
| CN110684882A (en) * | 2019-11-05 | 2020-01-14 | 山东钢铁集团日照有限公司 | Oxygen-containing steel RH vacuum carbon powder deoxidation cooling method |
| CN114787394A (en) * | 2019-11-29 | 2022-07-22 | 杰富意钢铁株式会社 | Operating method for ladle refining treatment |
| CN111020118A (en) * | 2019-12-25 | 2020-04-17 | 武汉科技大学 | RH endpoint temperature prediction method based on particle swarm optimization case reasoning |
| CN111020118B (en) * | 2019-12-25 | 2021-09-24 | 武汉科技大学 | RH endpoint temperature prediction method based on particle swarm optimization case reasoning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1298458C (en) | 2007-02-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1298458C (en) | Method for real-time estimating temperature of liquid steel in RH fining furnace | |
| CN109447346B (en) | Converter oxygen consumption prediction method based on gray prediction and neural network combined model | |
| Jiang et al. | Molecular dynamics simulation on the effect of MgO/Al 2 O 3 ratio on structure and properties of blast furnace slag under different basicity conditions | |
| Wang et al. | Prediction model of end-point manganese content for BOF steelmaking process | |
| CN113192568B (en) | Method and system for forecasting desulfurization end point of refining furnace | |
| Feng et al. | Endpoint temperature prediction of molten steel in RH using improved case-based reasoning | |
| JP5527180B2 (en) | Converter blowing method and converter blowing system | |
| CN111020118B (en) | RH endpoint temperature prediction method based on particle swarm optimization case reasoning | |
| Jenab et al. | The use of ANN to predict the hot deformation behavior of AA7075 at low strain rates | |
| Trzaska | Calculation of the steel hardness after continuous cooling | |
| CN110261223A (en) | Detection method, device, medium and the equipment of vermicular cast iron tensile strength | |
| CN116469481A (en) | LF refined molten steel composition forecasting method based on XGBoost algorithm | |
| Yan et al. | New insight in predicting martensite start temperature in steels | |
| Zheng et al. | Method to predict alloy yield based on multiple raw material conditions and a PSO-LSTM network | |
| CN1603424A (en) | Bessemerizing control method based on intelligent compound dynamic model with sublance converter | |
| CN104775006A (en) | Furnace gas analysis model-based decarburization control method of vacuum oxygen decarburization refining | |
| CN110119595B (en) | Design method of die-casting aluminum alloy material | |
| CN100507018C (en) | Method for determining oxygen blowing amount and cold material feeding amount in RH refining process | |
| Xu et al. | Prediction of static globularization of Ti-17 alloy with starting lamellar microstructure during heat treatment | |
| Trzaska et al. | Computer programme for prediction steel parameters after heat treatment | |
| CN111933221A (en) | Method for predicting dynamic recrystallization fraction of Nb microalloyed steel | |
| Gumienny et al. | Predicting the microstructure of compacted graphite iron using a fuzzy knowledge-based system | |
| Liujie et al. | Artificial neural network prediction of heat-treatment hardness and abrasive wear resistance of High-Vanadium High-Speed Steel (HVHSS) | |
| Huang et al. | Prediction of alloy yield based on optimized BP neural network | |
| Wang et al. | A real-time temperature prediction method based on CNN-LSTM with MODE in steelmaking process |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CX01 | Expiry of patent term | ||
| CX01 | Expiry of patent term |
Granted publication date: 20070207 |