JP3882705B2 - Converter blowing control method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a converter blowing control method for performing control to make the composition of molten steel coincide with a target composition at the end of blowing for a converter.
[0002]
[Prior art]
The converter uses pig iron supplied from the blast furnace and scrap prepared separately as a main raw material. After adding auxiliary materials such as lime to this, oxygen is blown from above, sulfur S and phosphorus P contained inside It has a function of removing impurities such as oxidization, refining a steel having a desired composition and temperature, and supplying it to the next rolling step. Then, the main raw material and auxiliary raw material are supplied to the converter, oxygen is sprayed, and one process from the time when the steel having the desired composition and temperature is output (output steel) at the final point (end point) is “charged” ".
[0003]
The amount, composition, and temperature of pig iron and scrap supplied to the converter vary from charge to charge, so in order to produce steel having a uniform composition and a uniform temperature over all charges. It is necessary to carry out optimal blowing control for each charge.
[0004]
Therefore, conventionally, in this blowing control, the carbon contained in the molten steel that is produced at the end of the blowing operation against the blowing conditions such as pig iron and scrap composition and temperature that fluctuate for each charge of the converter. In order to make the composition containing the contents of C, phosphorus P, silicon Si, manganese Mn, etc. coincide with the target composition, the total amount of blown oxygen that indicates the total amount of oxygen blown into the converter during the blowing process is calculated. It was.
[0005]
Conventionally, the calculation of the total amount of blown oxygen has been performed by a static material balance calculation model as the amount of oxygen necessary for combustion.
In addition, as a dynamic calculation model for the total amount of blown oxygen, the target carbon amount, the amount of oxygen required to reach the target temperature, and the coolant are calculated based on the measurement results of the molten steel temperature and carbon amount using the sublance during the blowing. The amount is corrected.
[0006]
However, in these calculations, variations in blowing conditions and changes over time cannot be expressed sufficiently in the model, and so far, methods for correcting theoretical equations using learning terms have been studied.
[0007]
For example, according to Japanese Patent No. 2703254, in the static model, the blown integrated oxygen amount y is expressed as a function of blowing conditions.
[0008]
y = F (x₁, X₂, X_Three, ..., x_n)
Where x₁, X₂, X_Three, ..., x_nIs composed of blowing conditions such as pig iron, scrap materials and other main raw materials and auxiliary raw materials such as lime input, composition of the main raw materials, temperature of the pig iron to be charged, etc. The value (variable) of each item of n.
[0009]
At this time, an actual model correction (learning) term Δa is added to the above equation and rewritten below.
y = F (x₁, X₂, X_Three, ..., x_n) + Δa
Here, an exponential smoothing method or the like has been studied as a learning procedure for updating the learning term Δa at the end of each blowing in each charge.
[0010]
In JP-A-6-264129, the blowing control for the converter is performed with each item constituting the blowing conditions as an input layer, and the total amount of blown-off oxygen and the temperature and composition at the end of the blowing are output layers. A method is proposed in which the weighting coefficient of the intermediate layer in the neural network is updated by using the neural network as a model and periodically using blowing performance data.
[0011]
[Problems to be solved by the invention]
However, with the above-described methods, there are still the following problems to be solved even in the converter blowing control method in which the composition at the end of the actual blowing in the newly implemented charge matches the target composition. It was.
[0012]
That is, the method of expressing the blown integrated oxygen amount y as a function of the blowing conditions and correcting the learning term Δa for each charge is basically centered on the correction based on the blowing performance performed in the immediately preceding charge. Thus, the influence of many past charges is reflected only by the exponential smoothing circuit, and the similarity of the entire blowing conditions is not reflected in the calculation of the current blown integrated oxygen amount y.
[0013]
Furthermore, in the method of modeling the blowing control for the converter with a neural network, the operation (blowing conditions) used for learning each weighting factor for the integrated amount of blown oxygen calculated using this neural network. And fluctuations in performance data) are reflected to some extent, but lack general versatility for states that are not used for learning. In addition, learning of a neural network requires a large amount of calculation, and unless it is re-learned frequently, it cannot follow changes in blowing conditions and performance data.
[0014]
The present invention has been made in view of such circumstances, and by defining the entire charge item to be newly implemented as a single vector (set), it is possible to truly A similar actual charge can be selected, and the highly accurate blown accumulated oxygen amount can be calculated from this actual charge, and the composition of the molten steel produced at the end of actual blowing can be surely matched with the target composition. It aims at providing the converter blowing control method which can aim at the load reduction of secondary refining while improving the quality of the steel products manufactured with the steel produced from a converter.
[0015]
[Means for Solving the Problems]
The present invention provides a target composition at the end of blowing of each charge by adjusting the amount of acid sent to the converter in each charge in which blowing is performed with at least pig iron and various auxiliary materials being charged into the converter. It is applied to the converter blowing control method to match.
[0016]
In order to solve the above problems, in the converter blowing control method of the present invention, first, at least the amount of pig iron input, the composition at the time of charging, and the target composition at the end of blowing are determined in the charge to be newly performed. The blowing condition consisting of multiple items is defined as a new blowing vector. Next, the blowing conditions for each charge stored in the blowing performance database that stores the blowing conditions for each charge carried out in the past and the actual blown-off integrated oxygen amount at the end of the blowing as performance data, respectively. Defined as a smelting vector.
[0017]
Further, a predetermined number of actual blowing vectors similar to the new blowing vector are selected from the plurality of actual blowing vectors, and each blowing condition and each actual blowing of the selected predetermined number of actual blowing vectors are selected. Create an approximate model for estimating the total blown-off oxygen amount of the charge to be newly implemented from the stop-off integrated oxygen amount.
[0018]
Then, using this approximate model, the estimated amount of accumulated blow-off oxygen for the charge to be newly implemented is estimated, and based on this estimated accumulated amount of blow-off oxygen, the amount of oxygen delivered to the converter for the newly implemented charge Determine.
[0019]
  The features of the converter blowing control method configured as described above will be described.
  Consider a method for efficiently predicting the total amount of blown oxygen that reaches the target composition of the molten steel that is produced after execution of this charge. This search is performed by searching the database for past charges having blowing conditions similar to those of the newly executed charge.WasWith reference to the actual accumulated blow-off accumulated oxygen amount of the past charge, the blow-off accumulated oxygen amount in the charge refining execution period to be newly performed is predicted.
[0020]
In this case, the problem is that there are a wide variety of items that make up the blowing conditions for each charge, and the number of items is very large, such as a dozen or so. There is rarely a past charge with blowing conditions. Therefore, it is necessary to select a past charge having similar blowing conditions. In this case, if only a specific item is selected and selected, the blown integrated oxygen amount that is far from the actual is estimated. Therefore, it is important how to select a past charge having similar blowing conditions.
[0021]
Therefore, in the present invention, each item such as the amount of pig iron input, composition at the time of injection, target composition at the end of blowing is not treated as an individual item, which constitutes the blowing conditions in the charge newly implemented. Thus, it is treated as a set of a plurality of items, and this set of items is defined as a new blowing vector. Similarly, a set of items constituting blowing conditions in each charge performed in the past is defined as a performance blowing vector.
[0022]
Then, a predetermined number of actual blown vectors similar to the new blown vector are selected from the many actual blown vectors in the past, and the actual blowing corresponding to the selected predetermined number of actual blown vectors is selected. An approximate model is created using the value of each item of the smelting conditions and the actual amount of accumulated blowout oxygen.
[0023]
In this way, each blowing condition is treated as a vector consisting of a set of multiple items, and the past between those having similar blowing conditions more accurately by evaluating the similarity between the vectors by the similarity between the sets. Can be selected. Therefore, an approximate model for estimating a more accurate integrated blown oxygen amount is created, and using this approximate model, a more accurate integrated blown oxygen amount for a new charge is estimated. Based on the estimated blown integrated oxygen amount, the amount of acid sent to the converter is obtained.
[0024]
Another invention relates to the converter blowing control method of the invention described above, wherein a norm of a difference vector between the new blowing vector and each actual blowing vector is calculated, and the calculated norm is a predetermined number that is small. Are a predetermined number of actual blown vectors similar to the new blown vector.
[0025]
Thus, by adopting a norm, which is a mathematically established technique, for quantitatively evaluating the degree of similarity between vectors, similar past charges can be selected more easily.
[0026]
Further, in another converter according to the above-described converter water lily control method, each norm includes a new blowing vector and a value of each item constituting each actual blowing vector, and statistics of each item stored in the blowing actual database. A new blowing normalization vector normalized by the value and each actual blowing normalization vector are calculated.
[0027]
Thus, by calculating the norm using the normalized vector obtained by normalizing each item constituting the blowing condition, the similarity between the vectors can be evaluated more accurately.
[0028]
According to another invention, in the converter blowing control method described above, the value of each item constituting each blowing condition of each of the predetermined number of actual blowing vectors and each actual blown integrated oxygen amount satisfy the approximate model. Assuming that, this approximate model is identified using a regression equation.
[0029]
Since the actual blown accumulated oxygen amount and the blowing conditions do not completely correspond one-to-one due to various factors including measurement errors, this approximate model is identified using a statistical regression equation.
[0030]
According to another invention, in the converter blowing control method described above, the approximate model is assigned according to each norm value of a difference vector between the selected predetermined number of actual blowing vectors and the new blowing vector. Based on the weighted value and the actual blown integrated oxygen amount of each corresponding charge, the blown integrated oxygen amount of the charge to be newly implemented is estimated.
[0031]
Thus, in the approximate model in the configured converter blowing control method, calculation is performed using the actual blown integrated oxygen amount of each charge of the predetermined number of selected records. Since the weighted value given according to the norm value is added to the integrated oxygen amount, it is possible to estimate a statistically more accurate blown integrated oxygen amount.
[0032]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
(First embodiment)
FIG. 1 is a block diagram showing a schematic configuration of a converter blowing control apparatus to which the converter blowing control method according to the first embodiment of the present invention is applied. This converter blowing control device is formed of a kind of information processing device such as a computer.
[0033]
In the converter blowing control device, an operation unit 1 for an operator such as a keyboard to input various information, a blowing performance database 2 for storing data of each charge performed in the past, and a calculation A display 3 that displays the total amount of blown oxygen and the amount of acid sent to the converter is incorporated. In addition, it is also possible to input various information directly to this converter blowing control apparatus from a high-order computer, without going through the operation part 1 which an operator operates.
[0034]
In the blowing performance database 2, as shown in FIG. 2, a blowing condition 4 and a performance heat margin 5 are stored for each charge number that identifies each charge that has been performed in the past. As blowing condition 4, actual values x in a plurality of items 1 to n₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_nIs remembered. For example, item 1 (value x₁) Is the amount (actual) of hot metal charged to this converter, and item 2 (value x₂) Is one composition of hot metal to be introduced into this converter (for example, carbon C content rate), item 3 (value x_Three) Is the temperature (actual) of the hot metal charged into the converter, and item 4 (value x_Four) Is the input amount (actual) of one brand (type) of the auxiliary material to be input to this converter, and item 5 (value x_Five) Is the composition of one record of hot metal produced from the converter at the end of blowing (end point), item 6 (value x₆) Is the actual temperature (end point actual temperature) of the hot metal discharged from the converter at the end of blowing (end point).
[0035]
  In addition to the carbon C content described above, the hot metal composition to be introduced into the converter has various contents such as manganese Mn, silicon Si, phosphorus P, etc., and each has a different item number (value x). And stored in the blowing performance database 2.
  Similarly, the input amount of the auxiliary raw material to be input to the converter is also stored in the blowing performance database 2 with item numbers (values x) having different input amounts of a plurality of brands. There is a content of each metal such as manganese Mn, silicon Si, phosphorus P, oxygen O and the like.Therefore, in addition to the six items shown in FIG. 2, the items include the contents of manganese Mn, silicon (silicon) Si, phosphorus P, and the like contained in the auxiliary material.
[0036]
Furthermore, the above item 5 (value x_Five) Is the composition of the molten steel actually measured at the end of blowing (for example, the content of carbon C). Besides this carbon C, each metal such as manganese Mn, silicon Si, oxygen O, etc. Are stored in the blowing performance database 2 with different item numbers (value x).
[0037]
The actual blown accumulated oxygen amount (h) 5 indicates the measured total amount of oxygen blown into the converter during the period of execution of blowing of the corresponding charge.
[0038]
In the converter blowing control apparatus shown in FIG. 1, in addition to the operation unit 1, the blowing performance database 2, and the display 3 described above, a new condition input unit 6 and a new condition formed on the application program are provided. Vector definition unit 7, performance condition vector definition unit 8, normalization vector creation unit 9, norm calculation unit 10, similar vector selection unit 11, approximate model creation unit 12, blown integrated oxygen amount calculation unit 13, oxygenation amount calculation unit 14 etc. are provided.
[0039]
Next, operation | movement of each part 6-14 of the converter blowing control apparatus comprised in this way is demonstrated using the flowchart shown in FIG.
[0040]
The actual value x in each item 1 to n in the charge newly performed by the operator via the operation unit 1 is described above.₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_nNew blowing conditions [x ’₁, X ’₂, X ’_Three, X ’_Four, X ’_Five, X ’₆, ..., x '_n] To the new condition input unit 6. In addition, it is also possible to directly input the above-described new blowing conditions from the host computer without using the operation unit 1.
[0041]
Specifically, item 1 (value x₁) Is the amount of hot metal to be added. Item 2 (value x₂) Is one composition of hot metal to be added (for example, carbon C), and item 3 (value x_Three) Is the temperature of the hot metal to be added. Item 4 (value x)_Four) Is the input amount of one brand (type) of the auxiliary raw material to be input, and item 5 (value x_Five) Is one composition (for example, carbon C) that is the target of hot metal to be produced from the converter at the end of blowing (end point), and item 6 (value x)₆) Is a target temperature (end point target temperature) of the hot metal to be discharged from the converter at the end of blow blowing (end point).
[0042]
In addition, the target composition mentioned above is a composition of the target molten steel at the time of completion | finish of blowing, In this composition, in addition to the carbon C mentioned above, content rate of each metal, such as manganese Mn, silicon Si, oxygen O, etc. Exists.
[0043]
Thus, among items 1 to n of the new blowing conditions in the charge newly implemented, item 5 (value x_Five) And item 6 (value x)₆) Is the target composition and end target temperature. In the actual charge blowing conditions stored in the blowing performance database 2 shown in FIG. 2, the target composition and end point target temperature in the new blowing conditions are replaced with the actual composition and end point actual temperature, respectively, in the same item.
[0044]
New blowing conditions [x ’₁, X ’₂, X ’_Three, X ’_Four, X ’_Five, X ’₆, ..., x '_n] Is input, the new condition input unit 6 is activated, and the input new blowing condition [x ′₁, X ’₂, X ’_Three, X ’_Four, X ’_Five, X ’₆, ..., x '_n] Is sent to the new condition vector definition unit 7 (step S1). The new condition vector definition unit 7 converts the input new blowing condition into a new blowing vectorVa is defined (S2).
[0045]
Va = (x ′₁, X ’₂, X ’_Three, X ’_Four, X ’_Five, X ’₆, ..., x '_n)
Next, the result condition vector definition unit 8 is activated, and the result blowing condition [x of each charge stored in the blowing result database 2 of FIG.₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_n] Each achievement blown vectorVb.
[0046]
Vb = (x₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_n)
Next, the normalized vector creation unit 9 is activated, and the actual values xi (i = 1, 2,..., N) of the items 1 to n stored in the blowing performance database 2 are all charged. The average value ui and standard deviation σi are calculated (S3). Next, the value xi (i = 1, 2,..., N) of each item in the actual blowing conditions in each charge is normalized. The normalized value xib of each item is
xib = (xi-ui) / σi
(S4).
[0047]
Next, each charge performance blowing vectorVA vector obtained by replacing the value xi of each item of b with the normalized value xibVDefine bo (S5
).
[0048]
Vbo = (x_1b, X_2b, X_3b, X_4b, X_5b, X_6b, ..., x_nb)
In addition, new blowing vectorVThe value x'i of each item of a is normalized using the average value ui and standard deviation σi obtained from each value of the blowing performance database 2 described above,
xia = (x'i-ui) / σi
Replace each normalized value xia with a new blowing normalization vectorVDefine ao (S6).
[0049]
Vao = (x_1a, X_2a, X_3a, X_4a, X_5a, X_6a, ..., x_na)
The norm calculation unit 10 is activated and this new blowing normalization vectorVao and each achievement blowing normalization vectorVEach deviation Δ from boVab is set (S7),
ΔVab =Vao―Vbo
Each deviation ΔVab vectorVao,VNorm | Δ as a quantitative criterion for similarity between boVab | is calculated (S8).
[0050]
| ΔVab ｜
= [(X_1a―X_1b)²+ (X_2a―X_2b)²+ ... + (x_na―X_nb)²]^1/2
In addition, weighting factor w for each item in blowing condition₁, W₂... w_nNorm calculated by multiplying || ΔVab |.
[0051]
| ΔVab ｜
= [W₁(X_1a―X_1b)²+ W₂(X_2a―X_2b)²+ ... + w_n(X_na―X_nb)²]^1/2Each norm for a new charge in all the actual charges stored in the blow blowing actual database 2 | ΔVWhen the calculation processing of ab | is completed, the pseudo vector selection unit 11 is activated and each calculated norm | ΔVab |, k (predetermined number) norm | ΔVab | is selected (S9).
[0052]
K selected norms | ΔVThe value x of each item in the performance blowing condition of the performance charge corresponding to ab |₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_nAnd the corresponding actual blow-off integrated oxygen amount h is read (S10).
[0053]
Note that the Akaike information criterion, prediction error, and cross-validation methods may be employed as methods for determining the approximate k pieces.
[0054]
Next, the approximate model creation unit 12 is activated, and as an approximate model of the blown integrated oxygen amount Hd, the value x of each item of the blowing conditions described above for the blown integrated oxygen amount Hd.₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_nIs set as a linear approximation expression represented by a linear function (S11).
Hd = c₁x₁+ C₂x₂+ C_Threex_Three+ C_Fourx_Four+ C_Fivex_Five+ ... + c_nx_n+ C₀
Where c₁, C₂, C_Three, ..., c_nIs a coefficient and c₀Is a constant.
[0055]
The total amount of blown-off oxygen, which is the total amount of oxygen to be blown into (supplied to) the converter during the blowing process for the converter, mainly consists of the components contained in the main raw materials such as pig iron and the like introduced into the converter. It is determined by each oxygen amount for burning (oxidizing) impurities. Specifically, in addition to the carbon C combustion oxygen amount, the silicon Si combustion oxygen amount, the phosphorus P combustion oxygen amount, and the manganese Mn combustion oxygen amount necessary for obtaining the target composition, the iron ore and other secondary oxygen that have been input are included. It is determined in consideration of the amount of fixed oxygen previously contained in the raw material.
[0056]
Each of the above-mentioned necessary oxygen amounts for determining the blown integrated oxygen amount is the value x of each item of the above-described blowing conditions.₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_nTherefore, the accumulated amount of blown oxygen required during the blowing period in this charge is also the value x of each item of the blowing conditions.₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_nIt can be expressed by the function
[0057]
At this time, it is considered that linear approximation is possible locally in the vicinity of the new blowing vector. Therefore, the blown integrated oxygen amount Hd is the value x of each item of the blowing condition.₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_nCan be expressed by a linear approximation formula using as a variable. Therefore, each coefficient c₁, C₂, C_Three, ..., c_nAnd constant c₀Is obtained, the above-described linear approximation formula is determined.
[0058]
Value x of each item of actual blowing conditions for each actual charge similar to newly implemented charge₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_n, And the corresponding actual blown accumulated oxygen amount h should substantially satisfy the above-mentioned linear approximation formula, so the value x of each item of the k actual blowing conditions₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_n, And the corresponding actual blown accumulated oxygen amount h is substituted into the above-described linear approximation formula, and each coefficient c is calculated by the least square method or the like.₁, C₂, C_Three, ..., c_nAnd constant c₀Is obtained (S12).
[0059]
The blown integrated oxygen amount calculation unit 13 is activated, and the value x of each item in the charge blowing condition (new blowing condition) to be newly performed is added to the linear approximation obtained in this way.₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_nIs substituted to calculate the predicted blown integrated oxygen amount Hd of the charge to be newly implemented. Then, the calculated blown integrated oxygen amount Hd is displayed and output on the display 3 (S13).
[0060]
Next, the oxygen supply amount calculation unit 14 is activated, and an oxygen supply amount indicating the amount of oxygen actually blown into the converter at each elapsed time of the blowing period is calculated based on the integrated blown oxygen amount Hd. To do. Then, the calculation result is displayed and output on the display 3 (S14).
[0061]
In the converter blowing control method configured in this way, each item of 1 to n constituting the blowing condition in the charge to be newly implemented is not treated as an individual item, but a set of a plurality of items The set of each item is treated as a new blowing vectorVa. Similarly, a set of items constituting the blowing conditions in each charge carried out in the past is used as the actual blowing vector.Vb.
[0062]
And many past achievements of this pastVNew blowing vector from bVk performance blow vectors similar to aVSelect b and select k actual blown vectorsVActual value x in each item of actual blowing conditions corresponding to b₁, X₂, X_Three, X_Four, X_Five, X₆, ..., x_nEach coefficient c of the linear approximation formula as an approximation model using the value and each actual blown accumulated oxygen amount h₁, C₂, C_Three, ..., c_nAnd constant c₀Has been identified.
[0063]
In this case, each blowing condition is a vector consisting of a set of multiple items.Va,Vb, and the similarity between vectors is represented by the norm of the vector difference | ΔVBy evaluating with ab |, past charges having similar blowing conditions can be selected more accurately. Therefore, it is possible to estimate a more accurate blown-off integrated oxygen amount Hd for a newly implemented charge.
[0064]
As a result, the composition at the end of blowing in the newly implemented charge can be accurately matched to the target composition, and the quality of steel products produced from the steel produced from this converter can be improved. The load of the next refining can be reduced.
[0065]
Further, since the neural network model in the conventional method is not adopted, the load on the computer for performing this blowing control does not increase.
[0066]
(Second Embodiment)
Hereinafter, a converter blowing control method according to the second embodiment of the present invention will be described.
The converter blowing control method of the second embodiment is different only in the approximation model for obtaining the blown integrated oxygen amount Hd in the converter blowing control method of the first embodiment described above. A method for obtaining an approximate model for obtaining the blown-off integrated oxygen amount Hd will be described.
[0067]
Each norm for the charge newly implemented in all the actual charges stored in the blowing performance database 2 | ΔVWhen the calculation processing of ab | is completed, the pseudo vector selection unit 11 is activated and each calculated norm | ΔVab |, the smaller of the k norms | ΔVUntil ab | is selected, it is the same as the converter blowing control method of the first embodiment described above.
[0068]
In the converter blowing control method of the second embodiment, each norm | Δ representing the similarity (distance) between k vectors.Vab |₁, D₂, D_Three, D_Four, ..., d_kAnd And this k distances d₁, D₂, D_Three, D_Four, ..., d_kThe maximum distance is dmax. K distances d₁, D₂, D_Three, D_Four, ..., d_kThe actual blow-off cumulative oxygen amount for each charge of the results corresponding to₁, H₂, H_Three, H_Four... h_kAnd
[0069]
Here, each actual blown accumulated oxygen amount h₁, H₂, H_Three, H_Four... h_kFor each weighting factor to multiply₁, W₂, W_Three... w_kThen, each weight coefficient w_i(I = 1,2,3, ..., k) for each distance d_i(I = 1,2,3, ..., k) and is defined below.
w_i= [1- (d_i/ Dmax)^Three]^Three
W = w₁+ W₂+ W_Three+ ... + w_k
That is, the shorter the distance (the more similar the vectors), the weighting factor w_iBecomes bigger.
[0070]
Then, an approximate model of the newly obtained charge blowing integrated oxygen amount Hd to be newly implemented is defined by the following equation.
[0071]
Hd = [w₁h₁+ W₂h₂+ W_Threeh_Three+ ... + w_kh_k] / W
When the blown integrated oxygen amount Hd is obtained, the oxygen supplied amount indicating the amount of oxygen actually blown into the converter at each elapsed time of the blowing period is calculated using the blown integrated oxygen amount Hd. Then, the calculation result is displayed and output on the display 3.
[0072]
Each w_iAs a setting method of (i = 1, 2, 3,..., K), various methods such as a normal distribution function can be considered in addition to the above formula.
[0073]
Also in the converter blowing control method of the second embodiment configured as described above, a new blowing vector of charge to be newly implementedVk actual blown vectors approximating aVSelect b and select k actual blown vectorsVSince the actual blow-off integrated oxygen amount hd of the actual charge corresponding to b is calculated, the blow-off integrated oxygen amount Hd of the charge to be newly implemented is calculated. Therefore, the converter blowing of the first embodiment described above It is possible to achieve substantially the same operational effects as the control method.
[0074]
Furthermore, in the converter blowing control method of the second embodiment, the actual heat margin h₁, H₂, H_Three, H_Four... h_k, Weighting factor w given according to norm value₁, W₂, W_Three... w_kTherefore, the statistically more accurate blown integrated oxygen amount Hd can be estimated.
[0075]
【The invention's effect】
As described above, in the converter blowing control method of the present invention, a predetermined number of actual blowing vectors that approximate the new blowing vector of the charge to be newly selected are selected, and each selected actual blowing is performed. An approximate model for estimating the blown integrated oxygen amount of the charge to be newly implemented is created using the actual blown integrated oxygen amount of the actual charge corresponding to the vector.
[0076]
Therefore, it is possible to select an actual charge that is truly similar from a large number of actual charges that have been carried out in the past, and it is possible to calculate a highly accurate integrated amount of blow-off from this actual charge and accurately target the composition at the end of actual blowing. The composition can be matched, and the quality of steel products produced from the steel produced from this converter can be improved, and the load of secondary refining can be reduced.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a schematic configuration of a converter blowing control apparatus to which a converter blowing control method according to a first embodiment of the present invention is applied.
FIG. 2 is a diagram showing storage contents of a blowing performance database formed in the converter blowing control device.
FIG. 3 is a flowchart showing the operation of the converter blowing control device.
[Explanation of symbols]
1 ... operation part
2 ... Blowing performance database
3 ... Display section
6 ... New condition input part
7 ... New condition vector definition part
8 ... Achievement condition vector definition part
9 ... Normalized vector generator
10: Norm calculation unit
11: Similar vector selection unit
12 ... Approximate model creation unit
13 ... Blow-up integrated oxygen amount calculation part
14 ... Amount of acid sent

Claims (5)

At least for each charge that is blown with pig iron and various auxiliary materials charged to the converter, the amount of acid sent to the converter is adjusted so that the composition at the end of blowing of each charge matches the target composition. In the furnace blowing control method,
In the newly implemented charge, a blowing condition consisting of a plurality of items including at least the amount of pig iron input, the composition at the time of charging and the target composition at the end of blowing is defined as a new blowing vector,
The blowing conditions for each charge stored in the blowing performance database that stores the blowing conditions for each charge carried out in the past and the actual blown accumulated oxygen amount at the end of blowing as performance data are the actual blowing vector, respectively. Define
A predetermined number of performance blowing vectors similar to the new blowing vector are selected from the plurality of performance blowing vectors,
Create an approximate model to estimate the total amount of blown oxygen for the newly implemented charge from each blowing condition and the actual amount of accumulated blown oxygen of the selected predetermined number of actual blown vectors,
Using this created approximate model, the newly implemented charge blow-off integrated oxygen amount is estimated,
A converter blowing control method, characterized in that, based on the estimated blown integrated oxygen amount, an amount of acid to be fed to the newly-converted converter is determined. A norm of a difference vector between the new blowing vector and each actual blowing vector is calculated, and a predetermined number of actual blowing vectors having a small calculated norm are set to a predetermined number similar to the new blowing vector. The converter blowing control method according to claim 1, wherein the actual blowing vector is used. Each norm is a new blowing normalization in which the value of each item constituting the new blowing vector and each actual blowing vector is normalized by the statistical value of each item stored in the blowing actual database. 3. The converter blowing control method according to claim 2, wherein the calculation is performed using a vector and each actual blowing normalization vector. Assuming that the value of each item constituting each blowing condition of the predetermined number of actual blowing vectors and each actual blowing integrated oxygen amount satisfies the approximate model, the approximate model is identified using a regression equation The converter blowing control method of any one of Claims 1-3 characterized by the above-mentioned. The approximate model includes a weighted value assigned according to each norm value of a difference vector between a predetermined number of selected actual blown vectors and a new blown vector, and the actual blowout integration for each corresponding charge. 4. The converter blowing control method according to claim 2 or 3, wherein a cumulative blown-off oxygen amount of charge to be newly implemented is estimated from the oxygen amount.

Images