JPH09256021A - Device for calculating charging quantity of auxiliary raw material into converter - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a calculating device for calculating charging quantity of auxiliary saw materials into a converter which make the operation of the converter economical by suitably restraining the charging quantity of the auxiliary raw material in the min. limit. SOLUTION: At the time of deciding the quantity of the auxiliary raw material to be charged in order to approach the molten steel component concn. in the converter to a target valve, in this device, the optimization calculation is executed by using physical model under commiseration of a material balance and an equilibrium reaction. B this method, unevenness of the component concn. at the time of stopping the blowing can be reduced. As a result, the accuracy of the component to be adjusted can unlimitedly be approached to the upper limit value, and the economical operation of charging quantity of the auxiliary raw material can be attained by restraining the changing quantity of the anxiusiliar raw material suitably as min. as possible. Further, the ull of a physical model renewed in succession at each time of measuring the slag component becomes possible, and the accuracy is not lowered even by the change is operational condition.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an auxiliary raw material input amount calculation device for a converter, and more particularly, to burned lime, roe stone, and a transfer material which are input to adjust a predetermined component concentration in molten steel in the converter. It relates to a device that calculates the optimum amount of auxiliary materials such as muddy, raw muddy, and fluorite.

[0002]

2. Description of the Related Art In the adjustment of molten steel components performed in a converter, in order to maintain quality, standard target values for sulfur (S), phosphorus (P), silicon (Si) It is set, and the components must be adjusted so that it does not exceed this. However, this does not mean that the concentration of the component to be adjusted may simply be set low. This is because the lower the concentration of the component to be adjusted, the larger the amount of auxiliary raw material to be input, which is disadvantageous in terms of cost. Therefore, in order to achieve both quality and cost, it is desirable to set the standard target value as an upper limit value within a range in which the concentration of the component to be adjusted does not exceed the upper limit value, and as close as possible to the upper limit value. In this respect, in the conventional auxiliary raw material input amount calculation device, the auxiliary raw material input amount with respect to the set concentration of the component to be adjusted is determined by a simple addition, subtraction, multiplication and division calculation based on a table value that summarizes the empirical data. It was done. However, the above-mentioned adjustment of the components to be adjusted is originally established on the basis of the material balance before and after the introduction of the auxiliary raw material into the converter, and has the characteristic of an equilibrium reaction in which the equilibrium of the component ratios is maintained. Highly accurate calculations that consider these principles have not been made in the past.

[0003]

As described above, the conventional auxiliary raw material input amount calculation device does not consider the chemical substance balance and the equilibrium reaction, so that the concentration of the blown molten steel component varies greatly from charge to charge. Has the drawback that
This did not take into consideration changes in the operating conditions due to melting damage or refilling of the refractory material in the converter, which was a factor that further reduced the accuracy of the blown molten steel component concentration. For this reason, the set concentration of the component to be adjusted has to be set lower than necessary in view of safety, and more auxiliary raw materials are required, resulting in high cost. Therefore, an object of the present invention is to provide a converter in which variations in blown-blown molten steel component concentration are small, accuracy is not deteriorated even when operating conditions change, and the amount of auxiliary raw materials can be appropriately adjusted to increase cost. It is to provide an auxiliary raw material input amount calculation device.

[0004]

[Means for Solving the Problems] To achieve the above object, the first invention is to calculate the amount of auxiliary raw material input for calculating the amount of auxiliary raw material to be input in order to bring the molten steel component concentration in the converter closer to the target value. It is configured as an auxiliary raw material input amount calculation device for a converter, which is characterized in that the input amount of each auxiliary raw material is calculated by performing an optimum calculation using a physical model considering the material balance and equilibrium reaction. . The second invention is
It is an auxiliary raw material input amount calculation device for a converter characterized by successively updating the physical model each time the slag component is measured. Furthermore, it is a device for calculating the amount of auxiliary raw material input to the converter, which updates the physical model by using the successive least squares method. Furthermore, it is a device for calculating the amount of auxiliary raw material input to the converter, which uses a neuro model trained by a neural network to update the physical model. Further, it is an auxiliary raw material input amount calculation device for a converter in which the entire physical model is constituted by a neuro model. Further, it is an auxiliary raw material input amount calculation device for a converter in which a target component in molten steel adjusted by the input of the auxiliary raw material is any one or a plurality of S, Si and P.

[0005]

In the first aspect of the present invention, the balance and equilibrium reaction are taken into consideration in the auxiliary raw material input amount calculation device for determining the amount of the auxiliary raw material to be input in order to bring the molten steel component concentration in the converter closer to the target value. The input amount of each auxiliary material is determined by performing the optimum calculation using the physical model. Therefore, it is possible to reduce the variation in the blown molten steel component concentration.
As a result, the concentration of the component to be adjusted can be made as close as possible to its upper limit value, and economical operation can be achieved with the amount of auxiliary raw material input minimized. Furthermore, in the second aspect of the invention, the physical model can be sequentially updated each time the slag component is measured, so that the accuracy does not decrease even if the operating state changes.

[0006]

Embodiments of the present invention will be described below with reference to the drawings to provide an understanding of the present invention. It should be noted that this embodiment and the examples that follow are specific examples of the implementation of the present invention and are not of the nature to limit the technical scope of the present invention. In the following embodiments, "phosphorus" is used as the component to be adjusted, but the present invention is not limited to adjustment of "phosphorus" as will be described later. For example, "silicon", "sulfur", etc. It is applicable as well. FIG. 1 is a schematic diagram showing a schematic configuration of an auxiliary raw material input amount calculation device for a converter according to an embodiment of the present invention, and FIG. 2 is a schematic diagram of a secondary raw material input amount calculation device for a converter according to an embodiment of the present invention. FIG. 4 is a diagram showing a processing flow, FIG. 3 is a diagram showing variation of each parameter for each slag component measurement by the recursive least squares method, and FIG. 4 is a schematic diagram showing a configuration of a neural network.

As shown in FIG. 1, an auxiliary raw material input amount calculation apparatus A0 for a converter according to an embodiment of the present invention comprises an online processing unit A1 according to the first invention and an offline processing unit according to the second invention. It is composed of A2. The online processing unit A1 includes an optimization calculation unit 1 and a storage device 2, and the storage device 2 stores a physical model (details will be described later) used in the optimization calculation. The off-line processing unit A2 includes a physical model calculation unit 3 and a storage device 4. A process computer 5 for controlling the operation process of the converter is connected to the online processing unit A1 and the offline processing unit A2. The process computer 5 stores the current operation data, the component data obtained from the sample, and the like, and the control of the input of the auxiliary raw material is also performed from here.

As the physical model used for the above optimization calculation, consider the following that considers the material balance in the converter and the equilibrium reaction. Here, "phosphorus" is considered as an example of the adjusted component. "Phosphorus" balance model Whm / Phm + Wsc / Psc + Wcm / Pcm + Wom / Pom + Wbf / Pbf + Wks / Pks = Wst / Pst + Wsl / Psl (1) However, Whm: hot metal weight, Wsc: scrap weight,
Wcm: weight of cold pig iron, Wom: weight of late pig iron, Wbf: amount of pre-charge slag, Wks: amount of blast furnace slag, Wst: amount of molten steel, Wsl: amount of slag, Phm: phosphorus concentration in hot metal, Ps
c: phosphorus concentration in scrap, Pcm: phosphorus concentration in cold pig iron,
Pom: Phosphorus concentration in late pig iron, Pbf: Phosphorus concentration in pre-charge slag, Pks: Phosphorus concentration in blast furnace slag, Pst: Phosphorus concentration in molten steel, Psl: Phosphorus concentration in slag "Phosphorus" distribution model

[Equation 1] However, Lp: phosphorus distribution, Ttd: blowing stop temperature, CS: basicity, Tfe: FeO concentration in slag, Ctd: blowing stop carbon concentration, MgO: MgO concentration in slag, P1 to P6: constant slag total amount model Wsl = Wcao + Wmgo + Wsio2 + Wmno + Wal2o3 + Wfeo + Wp2o5 + Wbf + Wks (3) However, Wcao: CaO weight in slag, Wmgo: MgO weight in slag, Wsio2: SiO2 weight in slag, Wmno: MnO weight in slag, WFeo: FeO3 in slag, 5A2 in Wag, WaOo in 5: Wa12o. : P2O5 weight in slag CaO model in slag Wcao = 0.96 · Wssk + 0.61 · Wkdr (4) where Wssk: burnt lime, Wkdr: MgO model in slag Wmgo 0.31 · Wkdr + const1 (5) However, const1: correction term SiO model in slag Wsio2 = Whm · Sihm + 0.7 · Wro + 1.07 · Wfesi (6) However, Sihm: Si concentration in hot metal, Wro: throwing rock stone Amount, Wfesi: FeSi input amount MnO model in slag Wmno = MnO · Wsl (7) where MnO: MnO concentration in slag FeO model in slag Wfeo = TFe · Wsl (8) A12O3 model in slag Wal2o3 = Al2O3 · Wsl (9) However, A12O3: A12O3 concentration in slag P2O5 model in slag Wp2o5 = P2O5 · Wsl (10) However, P2O5: P2O5 concentration in slag basicity model

[Equation 2] Furnace body protection constraint Wmgo = const2 · Wsl (12) where const2: constant

Equation (1), which is a phosphorus balance model, is a phosphorus balance equation at the start and end of blowing. here,
Material balance is considered. The values in the formula that are given as constants are Psc, Pcm, Pom, Wbf,
Wks and Pks, and W is obtained for each charge.
hm, Phm, Wsc, Wcm, Wom, Wst, Ps
Each value of t. Further, Wsl and Psl are obtained from equations (3) and (2), respectively. For Pbf, the value of Psl at the time of the previous blowing is used.

The equation (2) is an example showing the distribution of phosphorus between the molten steel and the slag at the time of blowing. Equilibrium reactions are considered in this equation. P1 to P6 are constants, and how to obtain this value will be described later. Psl is related to equation (1), and C
S is obtained from the equation (11). Ttd, CS, T on the right side
The values of Fe, Ctd, and MgO are values obtained by sampling, and Pst is the phosphorus concentration in the molten steel given as a target value. Formula (3) is the sum of the total amount of slag, the weight of each slag constituent, the amount of slag Wbf (constant) left by the precharge, and the amount of slag Wks (constant) contained in the hot metal carried from the blast furnace. It is represented by.

The respective slag constituent component models are (4) to (1
It is represented by the expression (0). In the expressions (4) to (10), Wss
k and Wro are the amounts of burned lime and roe stone to be finally obtained, and other values are the values obtained in relation to another formula or for each charge. Formula (12) is (4)
Is a formula given to determine Wkdr of the formula, and co
nst2 is the MgO concentration in the slag, which is set as a constant for simplifying the calculation. Light drop input Wkdr and Fe
Although the Si input amount Wfesi is a value actually determined by other constraints, it is treated as a known amount in this model.

First, the coefficients P1 to P6 of the equation (2) are calculated by using the recursive least squares method every time the slag component measurement by sampling is performed. Figure 3 shows the changes in each of the above coefficients for each slag component measurement. It can be seen that each coefficient is updated in sequence according to changes in operating conditions. Next, since the expression (1) is constant except for Wsl and Psl, the following expression is obtained. logWsl + logPsl = -C ₉ (4) to (10) and (12) are substituted into the formula (3) and transformed, the following formula is obtained.

(Equation 3) Further, the following equation is obtained from the equations (2) and (11).

(Equation 4) The following equation is obtained by summarizing the above three equations.

(Equation 5) However, C ₁ -C _9: constant In this, only the variable Wssk, Wro, other constants are known. The parameter α is introduced into this equation.
That is, the first term on the right side is α and the second and third terms are -α,
Solving for Wssk and Wro gives the following two equations.

(Equation 6)

(Equation 7) Here, substituting equations (14) and (15) with F = A · Wssk + B · Wro (16) as an evaluation function,

(Equation 8) However, D _{1 to} D ₆ : constants are obtained. Since this expression can be analytically differentiated twice, by using the Newton method as a nonlinear optimization method, α that minimizes expression (16), and further (17), (1
The input amounts of calcined lime and roe stone can be calculated from equation 8). However, depending on the nature of the objective function, the Simplex method or the flexible polyhedron method may be used instead of the Newton method.

In the above example, the phosphorus partition model is represented by the equation (2).
As a method, the method of determining coefficients by the recursive least squares method was adopted, but it is difficult to formulate it when the model has a strong nonlinearity. Therefore, in such a case, consider using a neuro model (FIG. 4) learned by a neural network as the phosphorus distribution model. in this case,
The following equation is used instead of equation (2).

[Equation 9] From equations (3) to (10) and (12), equation (18) is
Equation (19) is obtained from equation (2 ') and equation (20) is obtained from equation (1). Wsl = D ₁ Wssk + D ₂ Wro + D ₃ Wsl + D ₄ (18) log (Psl) = f (Ttd, CS, TFe, Ctd, MgO) + D ₅ (19) PslWsl-D ₆ = 0 (20) However, D _{1 to} D ₁₂ : constant Since a neuro model is included in the above equation and differentiation is difficult, the evaluation function is set here as F = (PslWsl−D ₆ ) ² + AWSsk + BWro (21), and Wssk Input amounts of calcined lime and loach are calculated by a nonlinear optimization method such as the flexible polyhedron method with Wro as a variable. At that time, as initial values, first, appropriate values of Wssk and Wro are started, and the values are changed so as to minimize the expression (21). At this time, (20)
To satisfy the phosphorus balance of the formula, that is, (21)
One criterion is whether or not the first term on the right side of the equation is sufficiently close to zero.

Further, the entire model can be a neuro model. PslWsl is learned using the neuro model as PslWsl = f (Ttd, CS, TFe, ..., Wssk, Wro, ...) (22). Using the learned neuro model and the evaluation function of Equation (21), similar to the above method, the non-linear optimization method such as the flexible polyhedron method with Wssk and Wro as variables is used to inject the burnt lime and the roe stone. The amount can be determined. The above formulas (2 ′), (2
The number of input layers and intermediate layers of the neuro model in 2) is determined according to the model properties, learning accuracy, and learning speed. Although the number of input layers is 6 and the number of intermediate layers is 7 in FIG. 4, the number of input layers may be increased if necessary, or may be reduced if redundant.

Here, the flow of processing by the present apparatus having the above characteristics will be described. First, when the apparatus is started up, the physical model calculation unit 3 of the offline processing unit A2 uses the data accumulated in the past to execute the initial method by the successive least squares method or the learning method of the neural network. A physical model of the storage device 2 of the online processing unit A1.
To be stored. At this time, in the case of the neural network method, a neuro model as shown in FIG. 2 is provided on the storage device 4 of the offline processing unit A2. When the blowing starts, the online processing unit A1 receives the operation data of the current charge from the process computer 5, performs optimization calculation using the physical model stored in the storage device 2, Determine the input amount.
The auxiliary raw material input amount thus determined is instructed from the online processing unit A1 to the process computer 5, and the auxiliary raw material is input. This online processing is repeated for each charge using the same physical model while there is no sampling. When the sample is taken, the offline processing unit A2 receives the necessary component data from the process computer 5 and updates the physical model by the method described above. When the physical model is updated, the new physical model is sent to the storage device 2 of the online processing unit A1 and replaced with the old physical model. While a new sample is taken and the physical model is not updated, online processing is always continued with the same physical model. When updating the physical model, it is not necessary to change the optimum calculation part other than the physical model, so it can be updated without stopping the online processing.

In the above example, the component to be adjusted is limited to phosphorus, but it is also possible to create a physical model that considers only sulfur, silicon, or a plurality of them.

[0017]

As described above, since the auxiliary raw material input amount calculation apparatus for a converter according to the present invention is configured as described above, it is possible to reduce the variation in the blown molten steel component concentration, so that the adjustment target is adjusted. It is possible to bring the concentration of a component close to its upper limit without limit. As a result, it is possible to economically operate the converter with the minimum amount of auxiliary materials input.

[Brief description of drawings]

FIG. 1 is a schematic diagram showing a schematic configuration of an auxiliary raw material input amount calculation device for a converter according to an embodiment of the present invention.

FIG. 2 is a diagram showing a processing flow of an auxiliary raw material input amount calculation device for a converter according to an embodiment of the present invention.

FIG. 3 is a diagram showing variation of each parameter for each slag component measurement by the recursive least squares method.

FIG. 4 is a schematic diagram showing the configuration of a neural network.

[Explanation of symbols]

 1 ... Optimization calculation unit 2 ... Storage device 3 ... Physical model calculation unit 4 ... Storage device 5 ... Process computer 6 ... Component analysis device A0 ... Auxiliary raw material input amount calculation device A1 ... Online processing unit A2 ... Offline processing unit

Claims (6)

[Claims] 1. An optimum amount of an auxiliary raw material input calculator for calculating the amount of an auxiliary raw material to be input in order to bring the molten steel component concentration in a converter close to a target value, using a physical model in consideration of material balance and equilibrium reaction. An auxiliary raw material input amount calculation device for a converter, characterized in that the input amount of each auxiliary raw material is calculated. 2. The auxiliary raw material input amount calculation device for a converter according to claim 1, wherein the physical model is sequentially updated every time the slag component is measured. 3. The auxiliary raw material input amount calculation device for a converter according to claim 1 or 2, wherein the updating of the physical model is performed by using a sequential least squares method. 4. The auxiliary raw material input amount calculation device for a converter according to claim 1, wherein the physical model is updated by using a neuro model learned by a neural network. 5. The auxiliary raw material input amount calculation device for a converter according to claim 1, wherein the entire physical model is constituted by a neuro model. 6. The auxiliary raw material input calculation device for a converter according to claim 1, wherein the target component in the molten steel adjusted by the input of the auxiliary raw material is any one or a plurality of S, Si and P.