Refining control device and refining control method

Abstract

To provide a refining control device and refining control method that can accurately calculate the optimal amount of operations for the refining process and produce molten metal with a desired component concentration and temperature with high yield. [Solution] The smelting control device according to the present invention calculates a predicted value of the state quantity of the smelting process to be calculated using the planned values ​​of the smelting process conditions and the amount of operation for the smelting process to be calculated, generates a control model that represents the relationship between the change in the state quantity of the smelting process and the amount of operation for the smelting process using similar past smelting process operation data, evaluates the difference between the predicted value and the actual value of the state quantity in past smelting processes as an error using similar past smelting process operation data, corrects the target value of the state quantity of the smelting process to be calculated based on the error evaluation result, and inputs the difference between the predicted value and the corrected target value of the state quantity as the change in the state quantity into the control model, thereby calculating the amount of operation for the smelting process to bring the state quantity of the smelting process closer to the corrected target value.

Description

[Technical Field]

[0001] The present invention relates to a refining control device and a refining control method for refining equipment in the steel industry. [Background technology]

[0002] In steel mills, the component concentration and temperature of molten iron tapped from the blast furnace are adjusted in refining facilities such as pre-treatment facilities, converters, and secondary refining facilities. Refining facilities are used to remove impurities from the molten metal and raise its temperature by adding auxiliary materials to the ladle or furnace and blowing in oxygen, etc. They play an important role in steel quality control and rationalization of manufacturing costs. In refining facilities, measurement information on the component concentration and temperature of the molten metal obtained during the refining process is limited. Therefore, the amount of auxiliary materials added and the amount of acid supplied are often determined based on predicted values ​​of state variables such as the component concentration and temperature of the molten metal, which are predicted using predictive models such as physical reaction models and statistical models. For this reason, it is important in the refining process to establish highly accurate state variable prediction and the logic for determining appropriate control variables based on that prediction. Against this backdrop, Patent Document 1 proposes a method in which a predictive model is used to calculate the rate constant of the dephosphorization reaction from the refining conditions, the calculated rate constant is used to predict the phosphorus concentration in the molten metal at the time of stopping the blowing, and the manipulated amount is changed so that the predicted value of the phosphorus concentration is less than or equal to the target value. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Patent No. 5582105 [Overview of the project] [Problems that the invention aims to solve]

[0004] However, the method described in Patent Document 1 does not take into account the prediction error of the state variables in the prediction model when calculating the manipulated amount. Therefore, in the method described in Patent Document 1, for example, in the case of refining conditions with little past experience, when the prediction accuracy of the state variables in the prediction model decreases, there may be cases where the phosphorus concentration cannot be reduced to below the target value. Against this backdrop, there has been a need for a technology that can accurately calculate the optimal manipulated amount for the refining process and produce molten metal with the desired component concentration and temperature with good yield.

[0005] The present invention was made to solve the above problems, and its objective is to provide a refining control device and a refining control method that can accurately calculate the optimal amount of operations for the refining process and produce molten metal with a desired component concentration and temperature with a high yield. [Means for solving the problem]

[0006] The smelting control device according to the present invention is a smelting control device that calculates the amount of operation for smelting in a smelting facility and controls the smelting facility according to the calculated amount of operation for smelting, and comprises: a state quantity prediction unit that calculates predicted values ​​of the state quantities of the smelting process to be calculated using the planned values ​​of the smelting process conditions and the amount of operation for smelting, and a control model generation unit that generates a control model representing the relationship between the amount of change in the state quantities of smelting and the amount of operation for smelting, using past smelting operation data where the actual values ​​of the smelting process conditions, the amount of operation for smelting, and the actual values ​​of the results of smelting are similar to the target values ​​of the smelting process conditions, the planned values ​​of the amount of operation for smelting, and the results of smelting, and The system includes: a target value correction unit that uses past refining operation data similar to the refining conditions, planned values ​​of the refining operation volume, and target values ​​of the refining results to evaluate the difference between the predicted and actual values ​​of the state quantities in past refining processes as an error, and corrects the target value of the state quantities of the refining process based on the error evaluation result; and an operation volume determination unit that inputs the difference between the predicted value of the state quantities calculated by the state quantity prediction unit and the target value corrected by the target value correction unit as a change in the state quantities to the control model, thereby calculating the amount of refining operation to bring the state quantities of the refining process closer to the target value corrected by the target value correction unit.

[0007] The control model generation unit and the target value correction unit may extract similar operational performance data from the same population of operational performance data.

[0008] The refining control method according to the present invention is a refining control method that calculates the amount of refining operations performed in a refining facility and controls the refining facility according to the calculated amount of refining operations, and comprises: a state quantity prediction step that calculates a predicted value of the state quantity of the refining process to be calculated using the planned values ​​of the refining conditions and the amount of refining operations to be calculated; a control model generation step that generates a control model representing the relationship between the change in the state quantity of the refining process and the amount of refining operations, using past refining operation data where the actual values ​​of the refining conditions, the amount of refining operations, and the actual values ​​of the refining results are similar to the planned values ​​of the refining conditions, the amount of refining operations, and the target values ​​of the refining results to be calculated; and the actual values ​​of the refining conditions, the amount of refining operations, and The system includes: a target value correction step in which the actual value of the refining process result is evaluated as an error based on the difference between the predicted value and the actual value of the state quantity in past refining processes, using past refining process operation data similar to the refining process conditions to be calculated, the planned value of the refining process operation amount, and the target value of the refining process result, and corrects the target value of the state quantity of the refining process to be calculated based on the error evaluation result; and an operation amount determination step in which the difference between the predicted value of the state quantity calculated in the state quantity prediction step and the target value corrected in the target value correction step is input to the control model as the change in the state quantity, thereby calculating the amount of refining process operation required to bring the state quantity of the refining process closer to the target value corrected in the target value correction step. [Effects of the Invention]

[0009] According to the refining control device and refining control method of the present invention, the optimal amount of operation for the refining process can be calculated with high accuracy, and molten metal having the desired component concentration and temperature can be produced with a high yield. [Brief explanation of the drawing]

[0010] [Figure 1] Figure 1 is a schematic diagram showing the configuration of a smelting control device, which is one embodiment of the present invention. [Figure 2] Figure 2 is a flowchart showing the flow of a refining control process, which is one embodiment of the present invention. [Modes for carrying out the invention]

[0011] Hereinafter, referring to the drawings, the configuration and operation of a refining control device according to an embodiment of the present invention will be described in detail.

[0012] 〔Configuration〕 First, referring to FIG. 1, the configuration of a refining control device according to an embodiment of the present invention will be described.

[0013] FIG. 1 is a schematic diagram showing the configuration of a refining control device according to an embodiment of the present invention. As shown in FIG. 1, a refining control device 1 according to an embodiment of the present invention is a device for controlling the component concentration and temperature of molten metal 101 being processed in a refining facility 2 in the steel industry and the component concentration of slag 103. In this embodiment, the refining facility 2 includes a converter 100, a lance 102, and a duct 104. A lance 102 is disposed above the molten metal 101 in the converter 100. High-pressure oxygen is ejected from the tip of the lance 102 toward the lower molten metal 101. Impurities in the molten metal 101 are oxidized by this high-pressure oxygen and taken into the slag 103 (refining process).

[0014] A duct 104 for exhaust gas guiding and fuming is installed above the converter 100. Stirring gas is blown into the molten metal 101 in the converter 100 through a vent hole 105 formed at the bottom of the converter 100. The stirring gas is an inert gas such as Ar. The stirring gas blown into the molten metal 101 stirs the molten metal 101 and promotes the reaction between the high-pressure oxygen and the molten metal 101. A flow meter 106 measures the flow rate of the stirring gas blown into the molten metal 101.

[0015] Auxiliary raw materials are charged into the converter 100 from above the furnace. The auxiliary raw materials include heating materials containing C, Si, etc., cooling materials containing FeO, fluxing agents containing Ca, Mn, Mg, etc. Before and immediately after the start of the refining process, the component concentration and temperature of the molten metal 101 are measured. Also, the component concentration and temperature of the molten metal 101 are measured once or multiple times during the refining process, and based on the measured component concentration and temperature, the supply amount (oxygen supply amount) and rate (oxygen supply rate) of high-pressure oxygen, the flow rate of the stirring gas (stirring gas flow rate), the input amount of the auxiliary raw materials, and other operation amounts are determined. The operation amounts may include the height of the lance 102, the input amount and chronological change of the input timing of each auxiliary raw material brand.

[0016] The refining control system to which the refining control device 1 is applied includes a control terminal 10, a refining control device 1, and a display device 20 as main components.

[0017] The control terminal 10 is composed of an information processing device such as a personal computer or a workstation. The control terminal 10 controls the above-described operation amounts so that the component concentration and temperature of the molten metal 101 are within a desired range, and collects the actual value of the operation amount, the measured value of the component concentration and temperature of the molten metal 101, and information regarding the state of the refining facility 2.

[0018] The refining control device 1 is composed of an information processing device such as a personal computer or a workstation. The refining control device 1 includes an input device 11, a past performance database (past performance DB) 12, and an arithmetic processing unit 13.

[0019] The input device 11 is an input interface into which various measured values and actual values regarding the refining facility 2 are input. Examples of the input device 11 include a keyboard, a mouse, a pointing device, a data receiving device, and a graphical user interface (GUI). The input device 11 receives actual performance data, parameter setting values, etc. from the outside, writes them into the past performance DB 12 of the information, and transmits them to the arithmetic processing unit 13. Refining information is input to the input device 11 from the control terminal 10.

[0020] Here, the refining information includes information on the refining process conditions, the planned and actual values ​​of the refining process operations, and the results of the refining process. The refining process conditions include information on the actual values ​​of the component concentration and temperature of the molten metal 101 before refining, the component concentration and target values ​​of the molten metal 101, and the state of the refining equipment 2, including the number of times the refining equipment 2 has been used. The refining process conditions may also include information on the results of the refining process performed immediately before, the mixing ratio of molten metal 101 and cold iron, information on the pretreatment process of the molten metal 101, measurement results including the shape of the refining equipment 2, and information on the actual values ​​of the component concentration and temperature of the molten metal 101 and the component concentration of the slag 103 after refining in the refining process performed immediately before at the refining equipment 2 under prediction. Information on the results of the refining process includes information on the target (or planned) and actual values ​​of the component concentration and temperature of the molten metal 101 and the component concentration of the slag 103 after refining.

[0021] The Past Performance DB12 is a memory device that stores information on the model equation of the refining reaction in the refining equipment 2, actual values ​​of the refining process, and calculation results of the refining control device 1. The Past Performance DB12 may also store the parameters of the model equation as information on the model equation of the refining reaction. In addition, the Past Performance DB12 stores various information input to the input device 11 and predicted values ​​of the state quantities of the refining process calculated by the calculation processing unit 13. Examples of model equations for the refining reaction include equations describing the equilibrium reaction of components contained in the metal and slag, equations describing the reaction rate of components in the molten metal 101 or slag 103, equations describing the mass balance and heat balance in the system, empirical rules expressing the relationship between the refining process conditions, the manipulated amount of the refining process, and the state quantities of the refining process, and statistical models or machine learning models that identify the relationship between the refining process conditions, the manipulated amount of the refining process, and the state quantities of the refining process from past performance.

[0022] The arithmetic processing unit 13 is an arithmetic processing unit such as a CPU. By executing a computer program, the arithmetic processing unit 13 functions as a state variable prediction unit 14, a control model generation unit 15, a target value correction unit 16, and a manipulated variable determination unit 17. The functions of the state variable prediction unit 14, the control model generation unit 15, the target value correction unit 16, and the manipulated variable determination unit 17 will be described later. Note that the state variable prediction unit 14, the control model generation unit 15, the target value correction unit 16, and the manipulated variable determination unit 17 may be configured using dedicated arithmetic devices or arithmetic circuits.

[0023] The display device 20 is composed of a known display output device such as a CRT, and displays and outputs various information according to the control signals from the refining control device 1.

[0024] The refining control device 1, having this configuration, accurately calculates the optimal amount of operations for the refining process by executing the refining control process described below, and produces molten metal 101 with the desired component concentration and temperature with a high yield. The operation of the refining control device 1 when executing the refining control process will be explained below with reference to the flowchart shown in Figure 2.

[0025] [Smelting control process] Figure 2 is a flowchart showing the flow of a refining control process according to one embodiment of the present invention. The flowchart shown in Figure 2 starts when the operator instructs the execution of the refining control process by operating the input device 11, and the refining control process proceeds to step S1. However, the refining control process may also be started when information regarding the refining process conditions and the amount of refining process input is received by the input device 11.

[0026] In step S1, the input device 11 acquires information regarding the refining process conditions to be calculated, the planned values ​​of the refining process operations, and the target value (or planned value) of the refining process result. With this, step S1 is completed, and the refining control process proceeds to step S2.

[0027] In step S2, the state quantity prediction unit 14 calculates predicted values ​​of the state quantities for the refining process by inputting the planned values ​​of the refining process conditions and the operating quantities of the refining process, which were obtained in step S1, into the refining reaction model equation stored in the past performance DB 12. Here, the state quantities for which prediction values ​​of the refining process are calculated are the control quantities for the refining process and include the C concentration of molten metal 101, the P concentration of molten metal 101, the Si concentration of molten metal 101, the S concentration of molten metal 101, the temperature of molten metal 101, the iron oxide concentration of slag 103, the basicity of slag 103 (CaO concentration / SiO2 concentration), the P oxide concentration of slag 103, etc. With this, the process of step S2 is completed, and the refining control process proceeds to step S3.

[0028] In step S3, the control model generation unit 15 uses the information acquired in step S1 and past refining operation data stored in the past performance DB 12 to generate a control model that represents the relationship between the change in the refining state quantities and the refining operation quantities, based on the relationship between the actual values ​​of the refining state quantities and the actual values ​​of the refining operation quantities. Here, the actual values ​​of the state quantities are the actual values ​​of the state quantities collected by the control terminal 10.

[0029] Specifically, the control model generation unit 15 first extracts past refining operation data from the past performance DB 12 where the refining conditions, actual values ​​of the refining operation volume, and actual values ​​of the refining result are similar to the information to be calculated obtained in step S1. Whether or not the past refining operation data is similar to the information obtained in step S1 is determined by whether or not the difference in numerical values ​​is below a predetermined threshold, or by the similarity index W described later. n It can be determined based on this.

[0030] Similarity W n When calculating the required points x(≡[x1,x2,…,x M ] T ) is the operational data of past refining processes stored in the past performance DB12 x nFor the distance L from the required point x, it is calculated using the following mathematical formula (1). n Here, M represents the unique identification number assigned to each variable constituting the input variable space, and n represents the unique identification number assigned to each operation performance data. Also, in the mathematical formula (1), the parameter λ m is a weighting coefficient for scaling input variables measured in different scales such as chemical components and temperature.

[0031]

Number

[0032] Then, the control model generation unit 15 calculates the similarity W of the point at the distance L from the required point x for the operation performance data x of the past refining process stored in the past performance DB 12 using the following mathematical formula (2). n Here, in the mathematical formula (2), the parameter σ n is the standard deviation of the distance L represented by the mathematical formula (1) with respect to the operation performance data, and the parameter p is an adjustment parameter. n Here, in the mathematical formula (2), the parameter σ L is the standard deviation of the distance L represented by the mathematical formula (1) with respect to the operation performance data, and the parameter p is an adjustment parameter. n Here, in the mathematical formula (2), the parameter σ

[0033]

Number

[0034] Note that the control model generation unit 15 may calculate the similarity W as the product of the similarity with respect to the information obtained in the process of step S1 and the temporal similarity as shown in the following mathematical formula (3). The parameter λ in the mathematical formula (3) is a forgetting factor and has a value greater than 0 and less than 1. By introducing this forgetting factor, the similarity of the new operation performance data x n increases, and the similarity of the old operation performance data x n decreases. n increases, and the similarity of the old operation performance data x

[0035]

Number

[0036] Furthermore, similarity can be evaluated not only by the Euclidean distance shown in equation (1), but also by methods for evaluating the distance between k-dimensional vectors, including the Cityblock distance, Minkowski distance, Mahalanobis distance, and cosine similarity. Here, high similarity is equivalent to a short distance between the calculated k-dimensional vectors.

[0037] In step S1, the control model generation unit 15 extracts past refining operation data similar to the information to be calculated from the past performance DB 12. Next, using the extracted operation data, the control model generation unit 15 generates a control model that represents the relationship between the change in the refining state variables and the refining operation, based on the relationship between the actual values ​​of the refining state variables and the actual values ​​of the refining operation. Specifically, the control model generation unit 15 prepares a model equation in advance that represents the relationship between the change in the refining state variables and the refining operation. Then, the control model generation unit 15 fits the parameters of the model equation to the relationship between the change in the refining state variables and the refining operation in the extracted operation data.

[0038] Furthermore, if there are multiple state variables, it is advisable to generate a control model for each state variable. Also, the manipulated variables may be one or more items that have a significant impact on changes in the state variables. When dealing with multiple manipulated variables, an independent control model may be generated for each variable, or a control model representing the relationship between multiple manipulated variables and the changes in the state variables may be generated.

[0039] Furthermore, when generating the control model, the operational data may be selected from operational data of the same smelting facility 2 as the calculation target, or from operational data of the same smelting process as the calculation target, or from operational data of which the brand and composition of the auxiliary raw materials do not differ significantly from those of the calculation target. In addition, the number of operational data is preferably at least 100, more preferably 500 or more, and even more preferably 1000 or more. The number of operational data may also be determined by setting a retrospective period.

[0040] Furthermore, the control model generation unit 15 may generate a control model using a statistical model or machine learning model that identifies the relationship between the change in the state variables of the refining process and the operation amount of the refining process from similar past refining process operation performance data. For example, the control model generation unit 15 may use N operation performance data (input variable operation performance data x) stored in the past performance DB 12. n ) and the similarity W between it and the required point x n Using this, a local control model may be generated that emphasizes past operational performance data similar to the required point x. For example, the control model generation unit 15 may generate a control model represented by the following formula (4).

[0041]

number

[0042] The model parameter θ, expressed by the following equation (5) which constitutes equation (4), is the same as the similarity W, expressed by the following equations (6) to (9). n This can be calculated by solving an optimization problem that minimizes the value of the evaluation function J, which is the sum of squared errors between the actual and predicted values ​​of the changes in the state variables of the refining process, weighted by y. In equation (7), the parameter y n(where n=1,2,…,N) are the values ​​of the output variables corresponding to the nth operational performance data, and in equation (8), the parameter diag(s) represents a diagonal matrix with the elements of s as the principal diagonal elements. By calculating the model parameter θ that minimizes the weighted sum of squares between the predicted and measured values ​​of the changes in the state variables of the refining process, a local control model can be generated that better fits operational performance data with high similarity, i.e., data close to the required point x.

[0043]

number

[0044]

number

[0045]

number

[0046]

number

[0047]

number

[0048] Furthermore, when solving the optimization problem, the following constraints may be imposed. Specifically, as constraints, the range of the partial regression coefficient φ of the input variable in the model parameters represented by equation (10) may be restricted by equations (11) to (13). Here, the lower and upper limits represented by equations (12) and (13) shall be based on physical foresight information between the input and output variables. By adding constraints regarding the foresight information obtained from the physical model, it is possible to better fit actual data close to the required point and generate a local prediction model with partial regression coefficients that match the physical characteristics of the target to be predicted. With this, the processing in step S3 is completed, and the refining control process proceeds to the processing in step S4.

[0049]

number

[0050]

number

[0051]

number

[0052]

number

[0053] In step S4, the target value correction unit 16 corrects the target value (or planned value) of the state quantities of the refining process using the past refining process operation performance data stored in the past performance DB 12 and the predicted values ​​of the state quantities of the refining process calculated by the state quantity prediction unit 14 in the past. Specifically, first, the target value correction unit 16 evaluates the difference between the predicted value and the actual value of the state quantities of the refining process in the past operation performance data, based on the past refining process operation performance data extracted in step S3, as an error. In this embodiment, the error is evaluated based on the operation performance data extracted from the same population as the population used in step S3, but the error may be evaluated based on the operation performance data extracted from a different population than the population used in step S3. Furthermore, the error may be evaluated based on statistics such as the mean, standard deviation, median, maximum, or minimum value of the error, assuming that the errors in similar operational performance data follow a normal distribution. Alternatively, it may be evaluated based on the proportion of operational performance data where the error exceeds a predetermined upper or lower threshold value.

[0054] The target value correction unit 16 then corrects the target value (or planned value) of the state quantities in the refining process based on the error evaluation result. The corrected value of the target value may be calculated based on the error statistics. For example, assuming that the error follows a normal distribution, the target value correction unit 16 may raise or lower the target value by a constant multiple of the standard deviation. Alternatively, the upper and lower thresholds, i.e., the correction amount of the target value, may be calculated on the condition that the proportion of actual results that are greater than or less than the target value when the target value is corrected by a predetermined upper and lower threshold of the error is less than or equal to a desired value. For example, when correcting the target value to a higher value, it is advisable to determine the correction amount such that the proportion of actual results that fall below the original target value due to the resulting variability of the state quantities resulting from residual errors is greater than or equal to a desired value. The target value to be corrected is the target value of the state quantities in the refining process. If there are multiple state quantities, the target value correction calculation is performed for each state quantity.

[0055] When correcting the target value, the operational performance data may be selected from the same refining facility 2 as the calculation target, or from the same refining method as the calculation target, or from the same refining method as the calculation target.

[0056] In step S5, the manipulated amount determination unit 17 uses the control model generated in step S3 to determine the manipulated amount of the refining process to bring the state quantities of the refining process closer to the target values ​​corrected in step S4. Specifically, the manipulated amount determination unit 17 calculates the manipulated amount of the refining process by inputting the difference between the predicted value of the state quantity calculated in step S2 and the target value corrected in step S4 as the change in the state quantity into the control model.

[0057] Furthermore, if there are multiple state variables in the refining process, the manipulated variable determination unit 17 determines a manipulated variable for each state variable to bring it closer to the corrected target value. Also, if multiple manipulated variables are applicable to a single state variable, and the control model generation unit 15 generates independent control models for each, the manipulated variable determination unit 17 may determine the manipulated variables for each. In addition, if the control model generation unit 15 generates a control model that represents the relationship between multiple input variables and the change in state variables, the manipulated variable determination unit 17 may simultaneously determine a combination of manipulated variables that realize the desired change in state variables, taking into account pre-set constraints and evaluation functions.

[0058] The manipulated amount determination unit 17 then inputs information regarding the calculated manipulated amount to the control terminal 10. The manipulated amount determination unit 17 may also output information regarding the processing result of the refining control device 1 to the display device 20. Thereafter, the control terminal 10 controls the refining equipment 2 based on the information regarding the manipulated amount input from the manipulated amount determination unit 17. In addition, the actual value of the refining process result obtained after the refining process is input to the input device 11 and stored in the past performance DB 12. With this, the processing of step S5 is completed, and the series of refining control processes is finished.

[0059] As is clear from the above explanation, in the refining control process, which is one embodiment of the present invention, the refining control device 1 generates a model representing the relationship between the change in state variables and the manipulated variable using similar past operational performance data, and further corrects the target value of the state variables based on the error of the state variable prediction model in similar past operational performance data. As a result, optimal refining control is performed according to the refining conditions, and molten metal with the desired component concentration and temperature can be produced with good yield.

[0060] Although embodiments applying the invention made by the present inventors have been described above, the present invention is not limited by the descriptions and drawings that constitute part of the disclosure of the present invention in this embodiment. For example, although this embodiment relates to a refining control device and a refining control method for a converter, similar effects can be achieved in secondary refining equipment and pre-treatment equipment as long as the process determines the amount of manipulation based on the prediction of the internal state after processing. Thus, all other embodiments, examples, and operational techniques made by those skilled in the art based on this embodiment are included in the scope of the present invention. [Explanation of symbols]

[0061] 1. Refining control device 2. Refining equipment 10 Control terminals 11 Input devices 12. Past Performance Database (Past Performance DB) 13. Arithmetic Processing Unit 14 State Quantity Prediction Unit 15 Control Model Generation Unit 16. Target Value Correction Unit 17 Manipulated amount determination section 20 Display device 100 Converters 101 Molten metal 102 Lance 103 Slag 104 Duct 105 Ventilation holes 106 Flow meter

Claims

1. A refining control device that calculates the amount of refining operation in a refining facility and controls the refining facility according to the calculated amount of refining operation, A state quantity prediction unit calculates predicted values ​​of the state quantities of the refining process to be calculated using the planned values ​​of the refining process conditions and the refining process operation amounts to be calculated, A control model generation unit generates a control model that represents the relationship between the change in the state quantities of the refining process and the amount of refining process operations, using past refining process operation data where the refining process conditions, actual values ​​of the refining process operations, and actual values ​​of the refining process results are similar to the planned values ​​of the refining process conditions, the refining process operations, and the target values ​​of the refining process results being calculated. A target value correction unit evaluates the difference between the predicted and actual values ​​of state quantities in past refining processes as an error, using past refining process operation data where the refining process conditions, actual values ​​of the refining process operation volume, and actual values ​​of the refining process results are similar to the refining process conditions, planned values ​​of the refining process operation volume, and target values ​​of the refining process results being calculated, and corrects the target values ​​of the state quantities of the refining process being calculated based on the error evaluation results. The control model is input to the control model the difference between the predicted value of the state quantity calculated by the state quantity prediction unit and the target value corrected by the target value correction unit, thereby calculating the amount of operation for the refining process to bring the state quantity of the refining process closer to the target value corrected by the target value correction unit, and the operation amount determination unit calculates the amount of operation for the refining process. A refining control device equipped with the following:

2. The smelting control device according to claim 1, wherein the control model generation unit and the target value correction unit extract similar operational performance data from a population of identical operational performance data.

3. A refining control method that calculates the amount of refining operation in a refining facility and controls the refining facility according to the calculated amount of refining operation, A state quantity prediction step that calculates predicted values ​​of the state quantities of the refining process to be calculated using the planned values ​​of the refining process conditions and the refining process operation amounts to be calculated, A control model generation step generates a control model that represents the relationship between the change in the state quantities of the refining process and the amount of refining process operations, using past refining process operation data where the refining process conditions, actual values ​​of the refining process operations, and actual values ​​of the refining process results are similar to the planned values ​​of the refining process conditions, the refining process operations, and the target values ​​of the refining process results being calculated. A target value correction step is performed by using past refining operation data where the refining conditions, actual values ​​of the refining operation volume, and actual values ​​of the refining results are similar to the planned values ​​of the refining conditions, actual values ​​of the refining operation volume, and target values ​​of the refining results being calculated, evaluating the difference between the predicted and actual values ​​of the state quantities in past refining processes as an error, and correcting the target values ​​of the state quantities of the refining process being calculated based on the result of the error evaluation. An operation amount determination step calculates an operation amount for the refining process to bring the state quantities of the refining process closer to the target values ​​corrected in the target value correction step by inputting the difference between the predicted values ​​of the state quantities calculated in the state quantity prediction step and the target values ​​corrected in the target value correction step into the control model as the change in state quantities. A refining control method, including the following.

Images