Ironworks operation assistance method, operation assistance device, display device, and operation assistance program

Abstract

Provided is an operation assistance method, which is used in ironworks having a blast furnace section, a coke section, and an energy section and executed by an information processing device, comprising: inputting data of blast furnace operation specifications influencing the energy section and data of a coke quantity influencing the energy section into a learning model; estimating a total cost by the learning model on the basis of the input data of the blast furnace operation specifications and the input data of the coke quantity; and outputting the estimated total cost. The learning model is a model trained on the basis of record data including records of the blast furnace operation specifications, the coke quantity, and the total cost.

Description

Steelworks operation support method, operation support device, display device, and operation support program

[0001] The present disclosure relates to an operation support method, an operation support device, a display device, and an operation support program for supporting operations in a steelworks.

[0002] For example, Patent Literature 1 describes a method for determining the conditions for minimizing costs by performing an optimization process using the total cost of a steelworks as an objective function and operational conditions such as the crude steel production volume, the molten iron blending ratio, and the types and amounts of raw materials used as variables in a calculation model.

[0003] Japanese Patent Application Laid-Open No. 2005-55997

[0004] In actual operation, the material heat balance and power plant efficiency of a blast furnace often deviate from the theoretical values ​​derived by the calculation model. Therefore, the calculation model needs to be tuned each time using the most recent actual results. However, it is difficult to tune the calculation model using the genetic algorithm processing, which is the optimization processing method used in Patent Document 1. As described above, in conventional techniques, optimization conditions are searched for based on theoretical values, which may result in an inability to obtain an optimal solution that is suited to actual operation. Thus, there is room for improvement in the technology for supporting operations in steelworks.

[0005] In view of the above circumstances, an object of the present disclosure is to improve the technology that supports operations in steelworks.

[0006] (1) An operation support method according to one embodiment of the present disclosure is an operation support method executed by an information processing device and used in a steelworks having a blast furnace department, a coke department, and an energy department, the operation support method comprising: inputting data on blast furnace operation specifications that have an impact on the energy department and data on coke volume that have an impact on the energy department into a learning model; estimating a total cost using the learning model based on the input data on blast furnace operation specifications and coke volume data; and outputting the estimated total cost; the learning model being a model trained based on actual data including actual data on the blast furnace operation specifications, coke volume, and total cost.

[0007] (2) An operation support method according to an embodiment of the present disclosure is the operation support method described in (1), wherein the learning model includes a blast furnace department model, a coke department model, and an energy department model, inputting data on blast furnace operation specifications that have an impact on the energy department into the blast furnace department model, outputting estimated values ​​of molten iron components and molten iron temperature, and B gas generation conditions from the blast furnace department model based on the input blast furnace operation specification data, inputting coke amount data that has an impact on the energy department into the coke department model, outputting C gas generation conditions from the coke department model based on the input coke amount data, inputting the B gas generation conditions and the C gas generation conditions into the energy department model, estimating the total cost using the energy department model based on the B gas generation conditions and the C gas generation conditions that have been input, and outputting the estimated total cost.

[0008] (3) An operation support method according to one embodiment of the present disclosure is the operation support method described in (2), wherein the steelworks further includes a steelmaking department, the learning model further includes a steelmaking department model, and the operation support method further includes: inputting estimated values ​​of molten iron components, estimated values ​​of molten iron temperature, and iron blending into the steelmaking department model to determine LD gas generation conditions for the steelmaking department; inputting the B gas generation conditions, the LD gas generation conditions, and the C gas generation conditions into the energy department model; and outputting the total cost based on the input B gas generation conditions, the LD gas generation conditions, and the C gas generation conditions using the energy department model.

[0009] (4) An operation support method according to an embodiment of the present disclosure is the operation support method according to (3), wherein the pig iron ratio is determined based on the amount of crude steel, the type of steel, and the main unit price.

[0010] (5) An operation support method according to an embodiment of the present disclosure is the operation support method described in any one of (1) to (4), wherein the blast furnace department includes a plurality of blast furnaces, the total amount of molten iron is allocated to each blast furnace as a molten iron amount based on an upper and lower limit molten iron productivity ratio for each blast furnace, and data of the blast furnace operation specifications for each blast furnace is determined based on the allocated molten iron amount.

[0011] (6) An operation support method according to an embodiment of the present disclosure is the operation support method described in (3), wherein the learning model further estimates carbon dioxide emissions from the blast furnace division, steelmaking division, coke division, and energy division based on the input blast furnace operation specifications, coke volume, and pig iron distribution data, and outputs the estimated carbon dioxide emissions.

[0012] (7) An operation support method according to an embodiment of the present disclosure is the operation support method described in any one of (1) to (6), in which the total cost includes a cost equivalent to carbon dioxide emissions.

[0013] (8) An operation support method according to an embodiment of the present disclosure is the operation support method according to any one of (1) to (7), wherein the blast furnace operation specifications that have an impact on the energy sector include reducing material specifications and air blast specifications.

[0014] (9) An operation support method according to an embodiment of the present disclosure is the operation support method described in (8), wherein the reducing material specifications include a coke ratio and a pulverized coal ratio, and the blast specifications include a blast moisture content, a blast temperature, and a blast oxygen concentration.

[0015] (10) An operation support method according to an embodiment of the present disclosure is the operation support method according to (9), in which the reducing material specifications include a composition and a carbon content related to coke.

[0016] (11) An operation support method according to one embodiment of the present disclosure is the operation support method described in (1), comprising inputting data on a predetermined range of blast furnace operation specifications into the learning model, estimating total costs for the blast furnace department, the coke department, and the energy department corresponding to each input, and outputting multiple candidate operation data based on the estimated total costs.

[0017] (12) An operation support method according to an embodiment of the present disclosure is the operation support method according to (11), in which a total cost related to the plurality of candidate operation data satisfies a predetermined standard.

[0018] (13) An operation support method according to an embodiment of the present disclosure is the operation support method according to (11) or (12), in which the plurality of candidate operation data satisfies a predetermined constraint condition.

[0019] (14) An operation support method according to one embodiment of the present disclosure is the operation support method according to (13), in which the predetermined constraint condition includes a constraint condition related to a coke amount.

[0020] (15) An operation support method according to one embodiment of the present disclosure is the operation support method according to any one of (1) to (14), in which the coke amount includes a coke production amount and a coke inventory amount in the coke department.

[0021] (16) An operation support method according to an embodiment of the present disclosure is the operation support method according to (15), in which the coke inventory amount is restored by reducing a coke rate.

[0022] (17) An operation support method according to one embodiment of the present disclosure is the operation support method according to (15) or (16), in which the coke inventory amount is restored by increasing coke production.

[0023] (18) An operation support method according to one embodiment of the present disclosure is the operation support method described in (15), wherein, in a period including a first period and a second period in which the amount of iron produced in the blast furnace department is greater than that in the first period, the coke rate is reduced in the first period and the coke production amount is increased in the second period, thereby recovering the coke inventory.

[0024] (19) An operation support method according to one embodiment of the present disclosure is the operation support method described in (15), wherein the coke rate is increased when the cost of increasing the coke rate to adjust the coke inventory amount is lower than the cost of reducing the coke production amount.

[0025] (20) An operation support method according to one embodiment of the present disclosure is the operation support method described in (15), in which, when the coke inventory cannot be restored by reducing the coke rate and increasing the coke production volume, coke is transported from another steelworks to restore the coke inventory.

[0026] (21) An operation support method according to one embodiment of the present disclosure is the operation support method described in (20), wherein the other steelworks include a first steelworks and a second steelworks, and even if the cost of transporting coke from the first steelworks is higher than the cost of transporting coke from the second steelworks, coke is transported from the first steelworks to restore the coke inventory if the cost associated with energy supply and demand within the first steelworks is lower than the cost associated with energy supply and demand within the second steelworks.

[0027] (22) An operation support device according to one embodiment of the present disclosure is an operation support device for a steelworks having a blast furnace department, a coke department, and an energy department, the operation support device including a control unit, wherein the control unit inputs data on blast furnace operation specifications that have an impact on the energy department and data on coke volume that have an impact on the energy department into a learning model for the steelworks, estimates a total cost using the learning model based on the input data on blast furnace operation specifications and coke volume data, and outputs the estimated total cost, and the learning model is a model trained based on actual data including actual data on the blast furnace operation specifications, coke volume, and total cost.

[0028] (23) A display device according to one embodiment of the present disclosure is a display device that displays the output of an operation support device equipped with a control unit in a steelworks having a blast furnace department, a coke department, and an energy department, and displays a total cost estimated by the control unit equipped in the operation support device using a learning model related to the steelworks based on data on blast furnace operation specifications that have an impact on the energy department and data on coke volume that has an impact on the energy department, and the learning model is a model trained based on actual data including actual data on blast furnace operation specifications, coke volume, and total cost.

[0029] (24) A display device according to one embodiment of the present disclosure is the display device described in (23), which filters from among a plurality of candidate operation data based on the total costs of the blast furnace department, the coke department, and the energy department estimated corresponding to each input of data of blast furnace operation parameters within a predetermined range input into the learning model, and displays the filtered candidate operation data.

[0030] (25) An operation support program according to one embodiment of the present disclosure is an operation support program for a steelworks having a blast furnace department, a coke department, and an energy department, executed by an information processing device, and causes a computer to execute the following: inputting data on blast furnace operation specifications that have an impact on the energy department and data on coke volume that have an impact on the energy department into a learning model for the steelworks; estimating a total cost using the learning model based on the input data on blast furnace operation specifications and coke volume data; and outputting the estimated total cost, wherein the learning model is a model trained based on actual data including actual data on blast furnace operation specifications, coke volume, and total cost.

[0031] According to one embodiment of the present disclosure, technology supporting operations in steel plants can be improved.

[0032] 1 is a block diagram showing a schematic configuration of an operation support device according to an embodiment of the present disclosure; FIG. 2 is a flowchart showing the operation of an operation support device according to an embodiment of the present disclosure; FIG. 3 is a diagram showing the configuration of a learning model according to an embodiment of the present disclosure; FIG. 4 is a flowchart showing the operation of an operation support device according to an embodiment of the present disclosure; FIG. 5 is a case study result for operations under six different operating conditions; FIG. 6 is a case study result for cases where preconditions are different; and FIG. 7 is a flowchart showing the operation of the operation support device when the steelmaking department model is omitted.

[0033] Hereinafter, an operation support technology for a steelworks according to an embodiment of the present disclosure will be described with reference to the drawings.

[0034] In each drawing, the same or corresponding parts are denoted by the same reference numerals. In the description of this embodiment, the description of the same or corresponding parts will be omitted or simplified as appropriate.

[0035] An overview and configuration of an operation support technology for a steelworks according to this embodiment will be described with reference to FIG. 1 . The operation support technology for a steelworks according to this embodiment is executed by an operation support device 10. The operation support device 10 is also referred to as an information processing device. This support method is used, for example, in a steelworks having a blast furnace department, a steelmaking department, a coke department, and an energy department. Hereinafter, the support technology according to this embodiment will be described as being used in a steelworks having a blast furnace department, a steelmaking department, a coke department, and an energy department.

[0036] First, an overview of this embodiment will be described, and details will be provided later. Generally, steelworks prepare production plans based on steel product demand, including the amount of steel tapped, the amount of pig iron produced, and the amount of coke produced. Such production plans are prepared for each plant with the aim of maintaining production activities without surpluses or shortages. Furthermore, the energy department, which manages overall energy, prepares supply and demand plans to ensure the supply of utilities required at each plant without shortages. Here, utilities include gas, electricity, steam, water, compressed air, and the like. When preparing operation plans, each department, including the energy department, performs cost management not only to realize the production plan but also to minimize manufacturing costs. Specifically, under conditions that satisfy the production plan, the operation plan is prepared by optimizing various operating conditions, such as the types and amounts of raw materials used, the types and amounts of utilities used, and equipment operating conditions, to minimize manufacturing costs. In steelworks, combustible gases (B gas, C gas, and LD gas) produced as by-products in blast furnaces, coke ovens, and converters are used directly or as a mixed gas (M gas) within the plant, with surplus gases being used in private power generation facilities managed by the energy department. In the private power generation facilities, the by-product gases and some purchased fuel are combusted, and the resulting flue gas is heat-exchanged with pure water to generate steam, which is then used to power a steam turbine. This process also involves extracting the necessary amount of steam for the steelworks, i.e., steam extraction. Therefore, the supply and demand management of utilities managed by the energy department, particularly gas, electricity, and steam, is deeply related not only to the conditions of energy equipment such as turbines, but also to the operating conditions managed by each plant, particularly the conditions for by-product gas generation. However, gas, electricity, and steam supply and demand plans are typically prepared based on detailed operating plans for each plant, and it is difficult to say that the operating plans adequately take into account the energy department's cost situation. Therefore, the result of aggregating the optimal costs for each department does not necessarily represent the optimal cost for the entire steelworks. Therefore, the operation support technology for steelworks according to this embodiment aims to provide a comprehensive energy simulation technology for blast furnaces, steelmaking, and coke making that enables more accurate cost management.

[0037] When operating a blast furnace, first, operating conditions are determined using a model constructed from material balance and heat balance data. However, in actual operation, due to the influence of changes in the state of the blast furnace or changes in the surrounding environment, which change every day, a discrepancy occurs between the theoretical values ​​calculated from the model and the actual data obtained from actual operation. In this embodiment, in order to suppress this discrepancy, machine learning is performed based on data from actual operation, and the model is corrected.

[0038] Specifically, in the steelworks operation support technology according to this embodiment, the operation support device 10 uses a physical model of the steelworks to perform estimation processing necessary for operation support at the steelworks. The physical model of the steelworks is also referred to as a learning model. The learning model is a learning model that takes blast furnace operation specification data as input and costs for the blast furnace department, steelmaking department, coke department, and energy department as output. The costs for the blast furnace department, steelmaking department, coke department, and energy department are also referred to as total costs. Such a learning model is trained using blast furnace operation specification data and actual data including actual total costs as training data. Here, in this embodiment, the operation support device 10 inputs blast furnace operation specification data that have an impact on the energy department into the learning model. The operation support device 10 then estimates the total cost using the learning model based on the input blast furnace operation specification data. The operation support device 10 then outputs the estimated total cost.

[0039] As described above, according to this embodiment, the operation support device 10 inputs data on blast furnace operation parameters that have an impact on the energy sector into a learning model and outputs a total cost. The learning model is trained using actual data as training data. In other words, this embodiment is provided with a feedback function based on actual data to correct any discrepancy between actual operation and the theoretical value derived from the model. This improves the accuracy of estimating total costs in a steelworks. In other words, this embodiment can provide a comprehensive blast furnace and energy simulation technology that enables more accurate cost management, thereby improving the technology supporting operations in a steelworks.

[0040] Here, data on blast furnace operation specifications that affect the energy sector include, for example, reducing material conditions and blast conditions. There are a wide variety of operating conditions that affect the overall costs of the blast furnace sector and the energy sector. For example, in a blast furnace, reducing material conditions such as the coke ratio and pulverized coal ratio, which are closely related to the B gas generation conditions, and blast conditions such as the blast moisture content, blast temperature, and blast oxygen concentration also have a significant impact on the overall cost. In this embodiment, by considering such reducing material conditions and blast conditions as variables, more accurate cost management and the creation of highly feasible operation plans are realized.

[0041] Next, each component of the operation support device 10 will be described in detail.

[0042] The operation support device 10 is any device used by a user, including, for example, a manager or person in charge of a steelworks. For example, a general-purpose electronic device or a dedicated electronic device can be adopted as the operation support device 10. As shown in FIG. 1 , the operation support device 10 includes a control unit 11, a memory unit 12, a communication unit 13, an input unit 14, and an output unit 15.

[0043] The control unit 11 includes at least one processor, at least one dedicated circuit, or a combination of these. The processor is a general-purpose processor such as a central processing unit (CPU) or a graphics processing unit (GPU), or a dedicated processor specialized for a specific process. The dedicated circuit is, for example, a field-programmable gate array (FPGA) or an application specific integrated circuit (ASIC). The control unit 11 controls each part of the operation support device 10 and executes processes related to the operation of the operation support device 10.

[0044] The memory unit 12 includes at least one semiconductor memory, at least one magnetic memory, at least one optical memory, or a combination of at least two of these. The semiconductor memory is, for example, a random access memory (RAM) or a read-only memory (ROM). The RAM is, for example, a static random access memory (SRAM) or a dynamic random access memory (DRAM). The ROM is, for example, an electrically erasable programmable read-only memory (EEPROM). The memory unit 12 functions as, for example, a main memory device, an auxiliary memory device, or a cache memory. The memory unit 12 stores data used in the operation of the operation support device 10 and data obtained by the operation of the operation support device 10. For example, the memory unit 12 stores a learning model. The learning model may be stored in an external device different from the operation support device 10. In this case, the operation support device 10 uses the learning model by accessing the external device storing the learning model via the communication unit 13.

[0045] The communication unit 13 includes at least one external communication interface. The communication interface may be either a wired or wireless communication interface. In the case of wired communication, the communication interface is, for example, a local area network (LAN) interface or a universal serial bus (USB). In the case of wireless communication, the communication interface is, for example, an interface compatible with mobile communication standards such as long term evolution (LTE), fourth generation (4G), or fifth generation (5G), or an interface compatible with short-range wireless communication such as Bluetooth (registered trademark). The communication unit 13 receives data used in the operation of the operation support device 10 and transmits data obtained by the operation of the operation support device 10.

[0046] The input unit 14 includes at least one input interface. The input interface may be, for example, a physical key, a capacitive key, or a pointing device. The input interface may also be a touch screen integrated with a display. The input interface may also be, for example, a microphone that accepts voice input or a camera that accepts gesture input. The input unit 14 accepts operations to input data used in the operation of the operation support device 10. The input unit 14 may be connected to the operation support device 10 as an external input device instead of being provided in the operation support device 10. Any connection method may be used, for example, a Universal Serial Bus (USB), a High-Definition Multimedia Interface (HDMI) (registered trademark), or Bluetooth (registered trademark).

[0047] The output unit 15 includes at least one output interface. The output interface is, for example, a display that outputs information visually or a speaker that outputs information audibly. The display is, for example, a liquid crystal display (LCD) or an organic electroluminescence (EL) display. The output unit 15 displays and outputs data obtained by the operation of the operation support device 10. The output unit 15 may be connected to the operation support device 10 as an external output device instead of being provided in the operation support device 10. Any connection method can be used, for example, USB, HDMI (registered trademark), or Bluetooth (registered trademark).

[0048] The functions of the operation support device 10 are realized by executing a program according to this embodiment on a processor corresponding to the operation support device 10. That is, the functions of the operation support device 10 are realized by software. The program causes a computer to execute the operations of the operation support device 10, thereby causing the computer to function as the operation support device 10. That is, the computer functions as the operation support device 10 by executing the operations of the operation support device 10 in accordance with the program.

[0049] In this embodiment, the program can be recorded on a computer-readable recording medium. The computer-readable recording medium includes non-transitory computer-readable media, such as a magnetic recording device, an optical disc, a magneto-optical recording medium, or a semiconductor memory. The program can be distributed, for example, by selling, transferring, or lending a portable recording medium, such as a DVD (digital versatile disc) or a CD-ROM (compact disc read only memory), on which the program is recorded. The program can also be distributed by storing the program in the storage of an external server and transmitting the program from the external server to another computer. The program can also be provided as a program product.

[0050] Some or all of the functions of the operation support device 10 may be realized by a dedicated circuit equivalent to the control unit 11. In other words, some or all of the functions of the operation support device 10 may be realized by hardware.

[0051] (Operation Example 1 of the Operation Support Device 10) Next, an operation example 1 of the operation support device 10 according to this embodiment will be described. Fig. 2 is a flowchart showing an example of an operation support method executed by the operation support device 10 according to this embodiment.

[0052] Step S100: The control unit 11 of the operation support device 10 trains the learning model based on the actual data. Specifically, the control unit 11 trains the learning model using actual data including actual blast furnace operation parameters, coke volume, iron mix, and overall cost as training data. Here, iron mix represents the ratio between the total amount of molten iron and the amount of scrap. As described above, a discrepancy may occur between the theoretical value calculated from the physical model and the actual data obtained from actual operation. In this embodiment, to suppress this discrepancy, the learning model is trained based on data from actual operation.

[0053] In this embodiment, the blast furnace operation specifications include reducing agent specifications and blast specifications. The reducing agent specifications include a coke rate and a pulverized coal rate. The reducing agent specifications may also include the composition and carbon content of the coke. The blast specifications include blast moisture, blast temperature, and blast oxygen concentration. All of these blast furnace operation specifications have a significant impact on the energy sector. In other words, these blast furnace operation specifications have a significant impact on cost management in the energy sector, i.e., costs in the energy sector. Therefore, in this embodiment, a learning model is adopted that uses blast furnace operation specifications that have an impact on the energy sector as input. By adopting such a learning model in this embodiment, the estimation accuracy of the total cost at the steelworks is improved. Hereinafter, in this embodiment, the blast furnace operation specifications will be described as including the coke rate, pulverized coal rate, blast moisture, blast temperature, and blast oxygen concentration.

[0054] The total cost includes the main costs of the blast furnace department, the main costs of the steelmaking department, the main costs of the coke department, and the main costs of the energy department. The main costs of the blast furnace department include main raw material-related costs and energy costs, and the energy costs include reducing material-related costs and utility costs. Utility costs are the costs of utilities required at each factory, etc. The main costs of the steelmaking department include main raw material-related costs, auxiliary raw material-related costs, and utility costs. The main costs of the coke department include utility costs. The main costs of the energy department include purchased fuel costs and electricity buying and selling costs. In this embodiment, the total cost is determined by the following formula (1). [Formula 1] Total cost = main cost of the blast furnace department + main cost of the steelmaking department + main cost of the coke department + main cost of the energy department = (main raw material cost + auxiliary raw material cost) + (reducing material-related cost + utility cost) + (purchased fuel cost + electricity buying and selling cost) (1)

[0055] The learning model may be, for example, a machine learning model generated based on a multilayer perceptron configured with an input layer, a hidden layer, and an output layer, but is not limited to this. The learning model may also be any model generated based on a machine learning algorithm such as a convolutional neural network (CNN), a recurrent neural network (RNN), or other deep learning. The learning model is not limited to a model generated by machine learning. For example, the learning model may be a multiple regression model or a model based on a Kalman filter. The learning model may also be a theoretical model constructed from mass balance and heat balance data. In the present embodiment, "training" includes, but is not limited to, training in a machine learning algorithm. In the present embodiment, "training" includes, for example, optimization or correction of various parameters in a multiple regression model or a theoretical model.

[0056] Step S110: The control unit 11 determines the blast furnace operation parameters, coke volume, and pig iron mix based on conditions determined in accordance with the steelworks' production plan, etc. Conditions determined in accordance with the steelworks' production plan, etc., are also referred to as preconditions. In this embodiment, the coke volume is the coke production volume in the coke department. The preconditions are set, for example, by user input. These preconditions include pig iron production volume (or pig iron productivity ratio), coke unit price, rolling operation plan, coal market conditions, and crude oil market conditions. Specifically, the control unit 11 determines the blast furnace operation parameters using, for example, a lookup table in which pig iron production volume and blast furnace operation parameters are associated. In this case, the memory unit 12 stores the lookup table. The control unit 11 also determines the blast furnace operation parameters corresponding to the preconditions by referring to the memory unit 12. The control unit 11 may determine the pig iron mix based on the crude steel volume, steel type, and major unit price. The crude steel volume, steel type, and main unit price may be included in the preconditions or may be derived from the preconditions. The crude steel volume is the volume of crude steel produced in the blast furnace. The steel type is the type of crude steel produced in the blast furnace. The main unit price includes unit prices of coal, scrap, LNG (Liquefied Natural Gas), etc. The control unit 11 may calculate an objective function within a search range set by user input and determine the iron distribution by adopting higher-level specifications including the iron distribution.

[0057] Step S120: The control unit 11 inputs the blast furnace operation specifications, coke amount, and pig iron distribution data determined in step S110 into the learning model learned in step S100.

[0058] Step S130: The control unit 11 uses a learning model to estimate the total costs of the blast furnace, steelmaking, coke, and energy departments based on the input data on blast furnace operation specifications, coke volume, and pig iron distribution. The above-mentioned preconditions are used in this estimation process. Specifically, the learning model uses the rolling operation plan, coal market conditions, and crude oil market conditions, and estimates the total costs of the blast furnace, steelmaking, coke, and energy departments based on the input data on blast furnace operation specifications, coke volume, and pig iron distribution.

[0059] Step S140: The control unit 11 outputs the total cost estimated in step S130 through the output unit 15.

[0060] Here, in the above, an example has been described in which the output of the learning model is a total cost, but this is not limited to this. The learning model may further estimate carbon dioxide emissions from the blast furnace department, steelmaking department, coke department, and energy department based on input data on blast furnace operation specifications, coke volume, and pig iron distribution. The carbon dioxide emissions include carbon dioxide emissions from reducing materials, carbon dioxide emissions from purchased fuel, and carbon dioxide emissions from purchased electricity. Furthermore, the carbon dioxide emissions from sold electricity are subtracted from the carbon dioxide emissions. In other words, the carbon dioxide emissions are determined, for example, by the following formula (2): [Formula 2] Carbon dioxide emissions = carbon dioxide emissions from reducing materials + carbon dioxide emissions from purchased fuel + carbon dioxide emissions from purchased electricity - carbon dioxide emissions from sold electricity ... (2)

[0061] When estimating the amount of carbon dioxide emissions, the actual data in step S100 described above includes the actual amount of carbon dioxide emissions. That is, in step S100, the control unit 11 trains the learning model using actual data including the blast furnace operation specifications, coke volume, pig iron mix, total cost, and carbon dioxide emissions as training data. In addition, in step S130, the control unit 11 estimates the total cost of the blast furnace department, steelmaking department, coke department, and energy department, and the amount of carbon dioxide emissions using the learning model based on the input data of the blast furnace operation specifications. Furthermore, in step S140, the control unit 11 outputs the estimated total cost and carbon dioxide emissions.

[0062] The total cost may also include the cost equivalent to the amount of carbon dioxide emissions. The cost equivalent to the amount of carbon dioxide emissions is determined by the amount of carbon dioxide emissions and the unit price of carbon dioxide. In this case, the above-mentioned preconditions include the unit price of carbon dioxide. Hereinafter, the total cost including the cost equivalent to the amount of carbon dioxide emissions will also be referred to as the total cost including carbon assessment. The total cost including carbon assessment is determined, for example, by the following formula (3). [Formula 3] Total cost including carbon assessment = Total cost + Carbon dioxide emissions × Carbon dioxide unit price (3)

[0063] Although the learning model has been described as a single model corresponding to a steelworks including a blast furnace department, a steelmaking department, a coke department, and an energy department, this is not limited to this. For example, the learning model may be a combination of four models: a model corresponding to the blast furnace department, i.e., a blast furnace department model, a model corresponding to the steelmaking department, i.e., a steelmaking department model, a model corresponding to the coke department, i.e., a coke department model, and a model corresponding to the energy department, i.e., an energy department model. Furthermore, the learning model may be a model that combines models that are further subdivided from the blast furnace department model, the steelmaking department model, the coke department model, and the energy department model.

[0064] Referring to FIG. 3 , an example of the configuration of a learning model is shown. The learning model shown in FIG. 3 includes a blast furnace department model 100, a steelmaking department model 200, a coke department model 300, and an energy department model 400. The blast furnace department model 100 includes a blast furnace operation model 110 and a hot stove model 120. The blast furnace operation model 110 and the hot stove model 120 are models based on the material and heat balance of the blast furnace and the hot stove, respectively. The control unit 11 trains the blast furnace operation model 110 and the hot stove model 120 using actual data as training data. Blast furnace operation specifications based on preconditions are input to the blast furnace operation model 110 and the hot stove model 120. Such preconditions include a tapping plan. The tapping plan includes plans such as the tapping volume. As described above, the control unit 11 determines the blast furnace operation specifications to be input to the model based on the tapping plan and the look-up table. The blast furnace operation model 110 and the hot stove model 120 output data necessary for calculating the overall cost and data on molten iron conditions, including molten iron composition and molten iron temperature, based on the input blast furnace operation specifications. The data necessary for calculating the overall cost is used as input to the energy department model 400. The data necessary for calculating the overall cost includes, for example, data on B gas, which is a combustible gas by-produced in the blast furnace. The data on B gas includes the amount of B gas generated and the heat value per unit volume of B gas. The data on molten iron conditions, including molten iron composition and molten iron temperature, is used as input to the steelmaking department model 200.

[0065] The steelmaking department model 200 includes a converter operation model 210 and an electric furnace operation model 220. The converter operation model 210 and the electric furnace operation model 220 are models based on the material and heat balances of the converter and the electric furnace, respectively. The control unit 11 trains the converter operation model 210 and the electric furnace operation model 220 using actual data as training data. Data on molten iron conditions, including the total amount of molten iron, the amount of scrap, and the molten iron composition and temperature, based on preconditions, is input to the converter operation model 210. The amount of scrap based on preconditions is input to the electric furnace operation model 220. The converter operation model 210 and the electric furnace operation model 220 output data necessary for calculating the overall cost based on the input preconditions. This data is used as input to the energy department model 400. The data necessary for calculating the overall cost includes, for example, data on LD gas, a combustible gas by-produced in the converter. The data relating to the LD gas includes the amount of LD gas generated and the calorific value per unit volume of the LD gas.

[0066] The coke department model 300 is a model based on the utility balance of the coke oven and peripheral equipment. The control unit 11 trains the coke department model 300 using actual data as training data. Data on the amount of coke produced based on preconditions is input to the coke department model 300. The coke department model 300 outputs data necessary for calculating the overall cost based on the input preconditions. This data is used as input to the energy department model 400. The data necessary for calculating the overall cost includes, for example, data related to C gas, which is a combustible gas produced as a by-product in coke ovens. The data related to C gas includes the amount of C gas produced and the heat value of C gas per unit volume.

[0067] The energy sector model 400 includes a utility balance logic 410. The control unit 11 trains the utility balance logic 410 using actual data as training data. The utility balance logic 410 outputs a total cost and a carbon dioxide emission amount based on input preconditions and data related to B gas, LD gas, and C gas. These preconditions include a rolling operation plan, coal market conditions, and crude oil market conditions. In other words, the energy sector model 400 receives as input the preconditions, the output of the blast furnace sector model 100, the output of the steelmaking sector model 200, and the output of the coke sector model 300, and outputs a total cost and a carbon dioxide emission amount.

[0068] (Operation Example 2 of the Operation Support Device 10) Next, operation example 2 of the operation support device 10 according to this embodiment will be described. FIG. 4 is a flowchart showing an example of an operation support method executed by the operation support device 10 according to this embodiment. Here, an example of an operation in which the operation support device 10 outputs multiple operation data candidates based on the overall cost estimated by the learning model will be described. The multiple operation data candidates are also referred to as candidate operation data. It is assumed that the operation support device 10 uses a learning model that has been trained by the processing in step S100 of FIG. 3.

[0069] Step S210: The control unit 11 of the operation support device 10 determines representative values ​​of the blast furnace operation parameters, coke rate, and pig iron ratio based on the preconditions. The representative values ​​of the blast furnace operation parameters, coke rate, and pig iron ratio are also referred to as base conditions. These preconditions are set, for example, by user input. The preconditions include pig iron production rate (or pig iron production ratio), coke price, rolling operation schedule, coal market conditions, and crude oil market conditions. Specifically, the control unit 11 determines the base conditions using, for example, a lookup table in which pig iron production rate and base conditions are associated. In this case, the memory unit 12 stores the lookup table. The control unit 11 also determines the base conditions corresponding to the preconditions by referring to the memory unit 12. The base conditions include the coke rate, pulverized coal rate, blast moisture, blast temperature, and blast oxygen concentration. Note that at least some of the base conditions may be set manually by the user.

[0070] Step S220: The control unit 11 inputs data on the blast furnace operation parameters, coke amount, and pig iron mixture within a predetermined range into the learning model based on the base conditions. That is, the control unit 11 inputs data on the blast furnace operation parameters, coke amount, and pig iron mixture within a predetermined range into the learning model, using the five operating factors included in the base conditions: coke rate, pulverized coal rate, blast moisture, blast temperature, and blast oxygen concentration. A predetermined range is set for each of the five operating factors. Specifically, the predetermined range for the coke rate has a lower limit of 280 kg / ton and an upper limit of 450 kg / ton. The predetermined range for the pulverized coal rate has a lower limit of 80 kg / ton and an upper limit of 260 kg / ton. The predetermined range for the blast moisture has a lower limit of 5 g / Nm ３ and the upper limit is 70 g / Nm ３ The predetermined range of the blast temperature has a lower limit of 900°C and an upper limit of 1200°C. The predetermined range of the blast oxygen concentration has a lower limit of 0% and an upper limit of 10%. These predetermined ranges are determined based on the upper and lower limit values ​​of operational performance in a steady state. The predetermined ranges may also be determined based on constraints in facility management. For example, the predetermined range of the blast temperature is determined to be, for example, the above range from the viewpoint of maintaining stable function of the hot stove.

[0071] Furthermore, the control unit 11 inputs all combinations of operating factors within a predetermined range into the learning model based on a specified mesh, i.e., an increment within the predetermined range. For example, the predetermined range for the airflow temperature is set to 900°C to 1200°C. The mesh can be specified as 50°C. When the predetermined range and mesh are specified in this manner, airflow temperatures of 900°C, 950°C, ..., 1200°C are input into the learning model. In other words, the control unit 11 performs a grid search for the five operating factors, using a predetermined range for each operating factor, for example, the settable upper and lower limit values ​​for each operating factor, as the search range. The mesh size is determined appropriately depending on the calculation load associated with the grid search.

[0072] Step S230: The control unit 11 estimates the total cost corresponding to each input input in step S220 using the learning model.

[0073] Step S240: The control unit 11 outputs multiple candidate operation data through the output unit 15 based on the total cost estimated in step S230. In other words, the control unit 11 outputs multiple candidate operation data by searching using the total cost as an objective function. The multiple candidate operation data are, for example, the top 20 pieces of operation data in descending order of total cost estimated corresponding to each input. The operation data include the coke rate, pulverized coal rate, blast moisture, blast temperature, and blast oxygen concentration. In other words, the user can easily identify the top 20 pieces of operation data with the lowest total cost. The user can also select realistically usable operation data from the operation data.

[0074] While the above describes an example in which the output of the learning model is the total cost, this is not limiting. The learning model may further estimate the total carbon dioxide emissions based on the input blast furnace operation specification data. The carbon dioxide emissions are expressed by the above formula (2). In this case, in steps S220 to S240, the control unit 11 outputs multiple candidate operation data by searching using the total cost and carbon dioxide emissions as objective functions. In this case, the candidate operation data is the top 20 operation data based on a ranking determined based on the total cost and carbon dioxide emissions estimated for each input. The ranking determined based on the total cost and carbon dioxide emissions is appropriately determined based on an evaluation value based on the total cost and carbon dioxide emissions. For example, the learning model may also estimate the total cost including carbon evaluation. The total cost including carbon evaluation is determined, for example, by the above formula (3). In this case, in steps S220 to S240, the control unit 11 outputs multiple candidate operation data by searching using the total cost including carbon evaluation as an objective function. In this case, the candidate operational data is, for example, the top 20 operational data in order of lowest total cost including carbon assessment estimated in response to each input.

[0075] As described above, according to this embodiment, the operation support device 10 inputs data on blast furnace operation parameters, coke volume, and pig iron distribution that affect the energy sector into a learning model and outputs total costs, etc. The learning model is trained based on actual data. Therefore, this embodiment can improve the accuracy of estimating total costs, etc. in a steelworks. Furthermore, according to this embodiment, the operation support device 10 inputs data on a predetermined range of blast furnace operation parameters, coke volume, and pig iron distribution, estimates total costs, etc. corresponding to each input, and outputs multiple candidate operation data. Therefore, it is possible to output multiple candidate operation data that are desirable from the perspective of total costs, etc., from among data on blast furnace operation parameters, coke volume, and pig iron distribution within a predetermined range. Thus, this embodiment can improve the technology for supporting operations in a steelworks.

[0076] There are a wide variety of operating conditions that affect the overall cost, etc. For example, in a blast furnace, reducing material conditions such as the coke ratio and pulverized coal ratio, which are closely related to the B gas generation conditions, and blast conditions such as the blast moisture content, blast temperature, and blast oxygen concentration also have a significant effect on the overall cost and carbon dioxide emissions. According to this embodiment, these reducing material conditions and blast conditions are considered as variables, allowing for more accurate cost management and the creation of a more feasible operation plan.

[0077] Figure 5 shows the results of a case study in which six different blast furnace operation data and coke rates were used under the same preconditions and pig iron mix. All six blast furnace operation parameters were thermally equivalent, combining multiple parameter changes equivalent to a coke rate of 2 kg / ton. In the case study results shown in Figure 5, the cost benefits of the blast furnace, coke, and energy departments are shown in bar graphs. The cost benefit is calculated by multiplying the cost by -1. The overall cost benefit, calculated by combining the cost benefits of the blast furnace, coke, and energy departments, is shown in a line graph. The overall cost benefit is the sum of the cost benefits of the blast furnace, coke, and energy departments. As shown in Figure 5, the cost benefits of the blast furnace, coke, and energy departments are maximized in Cases III and VI for the blast furnace department, Cases IV, V, and VI for the coke department, and Case II for the energy department. On the other hand, the overall cost benefit is maximized in Case III. As described above, the case where the cost benefit of the blast furnace department is greatest, the case where the cost benefit of the coke department is greatest, the case where the cost benefit of the energy department is greatest, and the case where the overall cost benefit is greatest do not necessarily coincide. In other words, it is necessary to determine the blast furnace specifications based on the overall cost, not just the individual costs of the blast furnace department, the coke department, and the energy department. By estimating the overall cost using the operation support device 10 according to this embodiment, it is possible to optimize the entire steelworks.

[0078] Figure 6 shows the results of a case study of trial operation under different assumptions. The six blast furnace operating parameters in Figure 6 were all thermally equivalent, combining multiple parameter changes equivalent to a coke rate of 2 kg / t. In other words, Figure 6 shows the results of a case study of trial operation under different assumptions, with the same coke volume and pig iron mix. Rolling operation plans A and B represent the cases where all major rolling mills are operated and none of the major rolling mills are operated, respectively. The normal and elevated coke prices and crude oil market prices are appropriately set within the normal and elevated price ranges, respectively. Figure 6 also highlights the cells with the highest and second-highest overall cost merits. As Figure 6 shows, the cases with the highest cost merits are different for all six assumptions. In other words, if the assumptions change, the overall costs in the blast furnace and energy sectors must be reevaluated. According to the operation support device 10 of this embodiment, the total cost can be easily estimated, so that even if the preconditions change, recalculation related to the optimization of the entire steelworks can be performed in a timely manner.

[0079] In step S240, the multiple candidate operational data are described as being the top 20 operational data based on the total cost, but this is not limited to this. The total cost of the multiple candidate operational data may be operational data that meets a predetermined criterion. For example, the predetermined criterion may be a criterion based on ranking, or may be that the total cost is lower than a predetermined threshold. The predetermined threshold may be a fixed value, or may be determined based on the average value of the total costs corresponding to the inputs found by grid search.

[0080] The plurality of candidate operational data may be operational data that satisfy predetermined constraints. In other words, constraints on the operation or facility management of each of the blast furnace division, steelmaking division, coke division, and energy division may be set, and if candidate operational data conflicts with the constraints, it may be excluded when ranking. Such constraints include tuyere temperature, blast volume, etc. For example, the lower and upper limits of the tuyere temperature may be specified as 2250°C and 2320°C, respectively. Also, for example, the lower and upper limits of the blast volume may be specified as 6500 Nm ３ / min, and 7400 Nm ３ / min. The constraints may also include a minimum amount of heavy oil used in the power generation facility. The plurality of candidate operational data may be limited to data that satisfies these constraints.

[0081] The control unit 11 may highlight and output candidate operation data that is identical or similar to the operating conditions of the performance data, among the plurality of candidate operation data output by the output unit 15. In this way, the user can more easily select operational data that can actually be used from the candidate operation data.

[0082] In the present embodiment, the learning model outputs either the total cost, the total cost and carbon dioxide emissions, or the total cost including carbon assessment. However, this is not limited to this. The learning model may also output the carbon dioxide emissions. In other words, the operation support device 10 may estimate and output only the carbon dioxide emissions without estimating the total cost. In this case, in step S130 of the above-described Operation Example 1, the control unit 11 uses the learning model to estimate the carbon dioxide emissions of the blast furnace department, steelmaking department, coke department, and energy department based on the input blast furnace operation specifications, coke volume, and pig iron distribution data. Then, in step S140, the control unit outputs the estimated carbon dioxide emissions via the output unit 15. Also, in step S230 of the above-described Operation Example 2, the control unit 11 uses the learning model to estimate the carbon dioxide emissions of the blast furnace department, steelmaking department, coke department, and energy department corresponding to each input. Then, in step S240, the control unit 11 outputs multiple candidate operation data via the output unit 15 based on the estimated carbon dioxide emissions. In other words, in this case, the control unit 11 performs a search using the carbon dioxide emission amount as an objective function, and outputs a plurality of candidate operation data.

[0083] In the present embodiment, the operation support device 10 includes an output unit 15, and the control unit 11 of the operation support device 10 outputs the results of calculations performed by the output unit 15. However, the present invention is not limited to this. For example, the technology according to the present embodiment may be implemented in a Software as a Service (SaaS) format. In this case, the operation support device 10 functions as a cloud server. The operation support device 10 outputs the calculation results on a display device separate from the operation support device 10. The display device includes a communication unit, receives the calculation results from the operation support device 10 via the communication unit, and displays information such as total costs. In other words, the display device displays the total costs of the blast furnace department, steelmaking department, coke department, and energy department estimated by the control unit 11 of the operation support device 10 using a learning model related to the steelworks, based on data on blast furnace operation parameters, coke volume, and pig iron distribution that affect the energy department. Alternatively, the display device displays carbon dioxide emissions from the blast furnace department, steelmaking department, coke department, and energy department estimated by the control unit 11 included in the operation support device 10 using a learning model related to the steelworks, based on data on blast furnace operation specifications, coke volume, and pig iron distribution that have an impact on the energy department. Any method can be used to request the calculation results from the operation support device 10. For example, the display device may include an input unit that accepts input from a user, and transmit a request for the calculation results to the operation support device 10 based on the input content accepted by the input unit.

[0084] This device is a device for outputting blast furnace operating parameters, coke volume, and pig iron ratios in order to optimize the overall costs and carbon dioxide emissions of the blast furnace, steelmaking, coke, and energy divisions combined. This embodiment is intended for blast furnace, steelmaking, and coke operators to consider and implement changes to blast furnace operating parameters, coke volume, and pig iron ratios accordingly, and is therefore different from devices that optimize the supply and demand of gas, steam, or electricity only within a specific division, such as the energy division.

[0085] The display device may output multiple candidate operation data based on the estimated total cost. That is, the operation support device 10, functioning as a cloud server, inputs a predetermined range of blast furnace operation specification data into the learning model and estimates the total costs of the blast furnace department, coke department, energy department, etc. corresponding to each input. Based on the estimated total cost, the display device may display multiple candidate operation data. The display device may filter the multiple candidate operation data based on filtering conditions and display the filtered candidate operation data. The filtering conditions may be appropriately set. In other words, the display device may filter from the multiple candidate operation data based on the total costs of the blast furnace department, coke department, energy department, etc. estimated corresponding to each input of a predetermined range of blast furnace operation specification data input into the learning model, and display the filtered candidate operation data. The filtering conditions may include, for example, the ranking of the total cost. Specifically, for example, the filtering conditions may be the top five rankings of the total cost. In this way, the display device may display only candidate operation data whose total cost is at or above a predetermined ranking.

[0086] In this embodiment, the LD gas conditions are output by the steelmaking department model and used to calculate the total cost. However, this is not limited to this. For example, in the operation support according to this embodiment, the steelmaking department model may be omitted from the learning model. That is, the operation support device 10 according to this embodiment may be used in a steelworks having a blast furnace department, a coke department, and an energy department. In this case, the operation support device 10 inputs data on blast furnace operation specifications that affect the energy department and data on coke volume that affect the energy department into the learning model, and estimates the total cost using the learning model based on the input data on blast furnace operation specifications and coke volume data. The operation support device 10 then outputs the estimated total cost.

[0087] More specifically, the operation support device 10 according to this embodiment calculates the total cost using three models: a blast furnace department model, a coke department model, and an energy department model. In this case, the learning model includes the blast furnace department model, the coke department model, and the energy department model. The operation support device 10 inputs data on blast furnace operation specifications that affect the energy department into the blast furnace department model, and outputs estimated values ​​of molten iron composition and molten iron temperature, and the B gas generation condition from the blast furnace department model based on the input blast furnace operation specification data. The operation support device 10 also inputs data on coke volume that affects the energy department into the coke department model, and outputs the C gas generation condition from the coke department model based on the input coke volume data. The operation support device 10 also inputs the B gas generation condition and the C gas generation condition into the energy department model, and estimates the total cost using the energy department model based on the input B gas generation condition and the C gas generation condition. The operation support device 10 then outputs the estimated total cost. FIG. 7 shows a flowchart illustrating an example of an operation support method executed by the operation support device 10 according to this embodiment in the case where the steelmaking department model is omitted.

[0088] Step S310: The control unit 11 of the operation support device 10 trains the learning model based on the performance data. Specifically, the control unit 11 trains the learning model using performance data including the performance of the blast furnace operation specifications, the coke volume, and the total cost as training data.

[0089] Step S320: The control unit 11 determines the blast furnace operation parameters and the coke amount based on the preconditions. The preconditions are set, for example, by user input. Specifically, the control unit 11 determines the blast furnace operation parameters using, for example, a lookup table in which the iron production rate is associated with the blast furnace operation parameters.

[0090] Step S330: The control unit 11 inputs the blast furnace operation specifications and coke amount data determined in step S320 into the learning model learned in step S310.

[0091] Step S340: The control unit 11 estimates the total costs of the blast furnace department, the coke department, and the energy department using a learning model based on the input data on blast furnace operation specifications and coke volume. The above-mentioned preconditions are used in this estimation process. Specifically, the learning model uses the rolling operation plan, coal market conditions, and crude oil market conditions, and estimates the total costs of the blast furnace department, the coke department, and the energy department based on the input data on blast furnace operation specifications and coke volume.

[0092] Step S350: The control unit 11 outputs the total cost estimated in step S340 through the output unit 15.

[0093] In particular, since the C gas condition has a large impact on the total cost, by using the C gas condition from the coke department model and omitting the steelmaking department model, the total cost can be estimated with a simpler configuration.

[0094] In this embodiment, the blast furnace department in the steelworks has been described as including one blast furnace, but this is not limited thereto. The blast furnace department may include multiple blast furnaces. In this case, the total amount of molten iron may be allocated to each blast furnace based on the upper and lower limit productivity ratios of each blast furnace. Furthermore, the blast furnace operation specifications for each blast furnace may be determined based on the allocated amount of molten iron.

[0095] The above-mentioned predetermined constraints may include conditions related to the gas permeability in the blast furnace. The gas permeability in the blast furnace section may be determined by the degree of porosity in the blast furnace. In other words, the predetermined constraints may include conditions related to the porosity in the blast furnace. A higher porosity is preferable because it reduces the pressure loss of the reducing material in the blast furnace. Here, the porosity in the blast furnace varies depending on the region within the blast furnace. The blast furnace has an upper region, a middle region, and a lower region. The upper region is a region extending downward from the top end of the blast furnace to approximately 10 m. The middle region is a region extending downward from the bottom end of the upper region to approximately 10 m. The lower region is the remaining region extending downward from the bottom end of the middle region. The upper region contains solids and gases, i.e., pig iron and coke, as well as reducing material. The porosity in the upper region is determined mainly based on the ratio of pig iron to coke, i.e., the coke ratio. The middle and lower regions contain liquid, solid, and gas, i.e., pig iron, coke, and reducing material. The void fractions of the middle and lower regions are determined based on the viscosity of the liquid pig iron, in addition to the coke ratio. Therefore, in this embodiment, the void fraction may be determined by the average value of the void fractions of the upper, middle, and lower regions. For example, the operation support device 10 may output only candidate operation data for which the void fraction is equal to or greater than a predetermined value.

[0096] The filtering conditions may include a condition related to the air permeability of the blast furnace. In other words, the filtering conditions may include a condition related to the porosity of the blast furnace. For example, the display device may output only candidate operation data for which the porosity is equal to or greater than a predetermined value.

[0097] In addition, the operation support technology according to this embodiment may take into account the inventory amount of coke produced in the coke department. In other words, the coke amount may include the coke production amount and the coke inventory amount in the coke department. Here, a lower limit and an upper limit are set for the coke inventory amount. The lower limit may be set, for example, based on the coke required to continuously operate the steelworks. On the other hand, the upper limit may be set, for example, based on the size of the coke storage area in the steelworks. In particular, if the coke inventory amount falls below the lower limit, it may become necessary to purchase coke from outside, which may have an impact on the operating costs of the steelworks. Therefore, the operation support device 10 may provide operation support based on a total cost that also takes into account the coke inventory amount.

[0098] The above-mentioned predetermined constraints may include a condition related to the amount of coke. That is, the predetermined constraints may include a condition related to the amount of coke produced and the amount of coke in the coke department. As described above, the amount of coke in stock has a lower limit and an upper limit. Since the trading price of coke can be high, by setting the condition related to the amount of coke as a predetermined constraint, it is possible to maintain an appropriate amount of coke and reduce overall costs.

[0099] The relationship between coke volume and preconditions is explained below. Coke consumption in steelworks is also affected by preconditions. Coke consumption may be included in blast furnace operation specifications. For example, when crude steel production is high, almost without exception, pig iron production and coke consumption are also high. Therefore, B gas, C gas, and LD gas generation are also high, almost proportionally. Meanwhile, the rolling process is also operating at nearly full capacity, resulting in high electricity consumption. Gas consumption in steelworks is strongly dependent on pig iron production, coke production, and rolling operation time, so gas consumption is also high in this case. On the other hand, when crude steel production is low, almost all elements are low. However, because the decline in electricity and gas consumption in the rolling process does not have the same impact as the decline in production, electricity and gas shortages are likely. Furthermore, if the decline in pig iron production is large and the decline in B gas generation is significant, gas shortages may occur in generators that are nearly exclusively B gas-fired. As a result, more external electricity may need to be purchased than necessary. For these reasons, when crude steel production is at a low level, the purchase ratio of externally purchased energy such as electricity, LNG, or heavy oil is likely to increase due to a shortage of by-product gas.

[0100] Furthermore, when at least some coke ovens are repaired, i.e., when coke ovens are shut down, this leads to a decrease in coke production and a decrease in C gas generation, regardless of the production level. This often leads to an increase in externally purchased energy due to gas shortages and a decrease in coke inventory. Furthermore, when rolling mills are shut down, it is common to plan for a certain amount of reduction in production level, but even in such a situation, the impact of electricity and gas demand being reduced to zero due to the rolling mill shutdown is significant, and often results in a surplus of by-product gas. This can lead to a decrease in externally purchased energy. In some cases, electricity may even be sold to external parties.

[0101] It is preferable to perform coke inventory management based on the relationship between the coke amount and the prerequisites. An example of coke inventory management will be described below.

[0102] (Example 1 of Coke Inventory Management) In a steelworks with a coke inventory management lower limit of 10,000 tons and an upper limit of 80,000 tons, the current inventory was 15,000 tons. A coke oven shutdown is scheduled for the following month, resulting in a 10,000 tons reduction in coke production. Therefore, the inventory must be restored to at least 20,000 tons within the current month. Therefore, the coke strength was improved for the current month only, and the coke consumption per unit of blast furnace iron production (coke rate) was reduced. For example, when iron production is 900,000 tons per month, a decrease in the coke rate of 6 kg / ton can increase the coke inventory by 5,400 tons.

[0103] (Second embodiment of coke inventory management) In the above case, the inventory amount can also be recovered by increasing the coke production amount rather than reducing the coke rate in the blast furnace, provided that the upper limit of the coke production amount has not already been reached.

[0104] (Third embodiment of coke inventory management) In the above case, the inventory amount can be restored by combining the first and second embodiments.

[0105] (Example 4 of coke inventory management) In a steel mill with a coke inventory management lower limit of 10,000 tons and an upper limit of 80,000 tons, the current inventory was 15,000 tons. The coke ovens are scheduled to be shut down in the month after next, resulting in a reduction in coke production of 20,000 tons. The main production plan for the next three months is as follows:

[0106]

[0107] If the coke rate is assumed to be 360 ​​kg / ton, and the coke yield is assumed to be 90%, the change in coke inventory at the end of the month will be: 36 x 0.9 - (88 x 360 / 1000) = +0.7,200 tons Similarly, the change in coke inventory for the next month will be ±0,000 tons, and the change in coke inventory for the month after that will be -18,000 tons. Therefore, coke inventory recovery measures worth 10,800 tons are required. As described in Examples 1 to 3, the coke inventory recovery measures are a reduction in the coke rate or an increase in coke production, or a combination of both.

[0108] According to the production plan, a rolling shutdown is scheduled for the current month. From an energy supply and demand perspective, by-product gas is more likely to be needed the following month when there is no rolling shutdown than the current month when there is a rolling shutdown, so it can be said that the following month is the most desirable period for high C gas production due to increased coke production. On the other hand, from the perspective of blast furnace permeability, it is easier to reduce the coke rate in the current month when iron production is relatively low. For these reasons, it can be said that it is desirable from an overall operational strategy perspective to reduce the coke rate in the current month and increase coke production the following month.

[0109] In this embodiment, the production plan is scheduled on a monthly basis, but may also be scheduled on various other basis such as a weekly basis or a bimonthly basis. In this embodiment, the current month in which the pig iron production rate is low may be replaced by a first period. The following month in which the pig iron production rate is high may be replaced by a second period. A coke rate reduction may be performed in the first period in which the pig iron production rate is low, and a coke rate increase may be performed in the second period in which the pig iron production rate is high.

[0110] (Example 5 of Coke Inventory Management) In a steel mill with a coke inventory management lower limit of 10,000 tons and an upper limit of 80,000 tons, the current inventory was 60,000 tons. A large-scale reduction in pig iron production is planned for the month after next, and production is expected to return to its original level within the next few months. The main production plans for the next three months are as follows:

[0111]

[0112] If the coke rate is assumed to be 360 ​​kg / ton, the amount of surplus coke in the month after next will be 36 x 0.9 - (75 x 360 / 1000) = +36,000 tons, which means that a coke inventory adjustment of at least 16,000 tons will be necessary.After comparing the total costs of a plan to reduce coke production for the current and next months by 9,000 tons / month with a plan to increase the coke rate for the current and next months by 10 kg / ton, in other words, to use the surplus 9,000 tons / month, the latter was selected.

[0113] In other words, an increase in the coke rate may be implemented if the cost of increasing the coke rate to adjust the coke inventory is less than the cost of decreasing the coke production rate, and conversely, a decrease in the coke production rate may be implemented if the cost of decreasing the coke production to adjust the coke inventory is less than the cost of increasing the coke rate.

[0114] (Example 6 of Coke Inventory Management) At Steel Works A, whose coke inventory management has a lower limit of 10,000 tons and an upper limit of 80,000 tons, the current inventory has fallen to 15,000 tons. According to the current month's production plan, even at the upper limit of coke production, there is still a coke shortage, and it is predicted that the inventory will fall to 7,000 tons by the end of the month. Due to poor conditions such as poor ventilation in the blast furnace, a reduction in the coke rate is not expected. If this continues, Steel Works B will be forced to purchase expensive coke from outside. Meanwhile, at Steel Works B, whose coke inventory management has a lower limit of 10,000 tons and an upper limit of 100,000 tons, the inventory has remained stable at around 40,000 tons, and there is still room for an increase in coke production. Therefore, it has been decided to reduce coke purchases from outside by increasing coke production at Steel Works B by 10,000 tons per month and transporting it to Steel Works A.

[0115] In other words, if the coke inventory cannot be restored by reducing the coke ratio and increasing coke production at the steelworks that is the target of operational support, coke may be transported from another steelworks to restore the coke inventory.

[0116] (Example 7 of Coke Inventory Management) Steel Works A, which has a coke inventory management lower limit of 10,000 tons and an upper limit of 80,000 tons, has currently seen its inventory fall to 15,000 tons. According to the production plan for this month, even at the upper limit of coke production, there will still be a coke shortage, and it is predicted that the inventory will fall to 7,000 tons by the end of the month. Due to poor ventilation and other conditions in the blast furnace, it is not possible to reduce the coke rate, and if this continues, the steel works will be forced to purchase expensive coke from outside.

[0117] Meanwhile, at Steelworks B, where the coke inventory management limit is 15,000 tons and the upper limit is 100,000 tons, inventory has been stable at around 40,000 tons, and there is still room for increasing coke production. Furthermore, at Steelworks C, where the coke inventory management limit is 8,000 tons and the upper limit is 60,000 tons, inventory has been stable at around 30,000 tons, and there is still room for increasing coke production. To recover coke inventory at Steelworks A, the company considered transporting coke from Steelworks B or C. Due to distance and location, transportation costs were lower for Steelworks B than for Steelworks C, but a comparison of costs related to energy supply and demand within the steelworks showed that Steelworks B was higher than Steelworks C. After comprehensively assessing these factors, the company decided to transport coke from Steelworks C to Steelworks A.

[0118] In other words, when other steel mills that can transport coke to the steel mill that is the target of operational support include a first steel mill and a second steel mill, even if the cost of transporting coke from the first steel mill is higher than the cost of transporting coke from the second steel mill, coke may be transported from the first steel mill if the cost associated with energy supply and demand within the first steel mill is lower than the cost associated with energy supply and demand within the second steel mill.

[0119] Although the present disclosure has been described based on the drawings and examples, it should be noted that those skilled in the art can make various modifications and alterations based on the present disclosure. Therefore, it should be noted that these modifications and alterations are included in the scope of the present disclosure. For example, the functions included in each means or step can be rearranged so as not to be logically inconsistent, and multiple means or steps can be combined or divided into one.

[0120] REFERENCE SIGNS LIST 10 Operation support device 11 Control unit 12 Memory unit 13 Communication unit 14 Input unit 15 Output unit 100 Blast furnace division model 110 Blast furnace operation model 120 Hot stove model 200 Steelmaking division model 300 Coke division model 400 Energy division model 410 Utility balance logic

Claims

1. An operations support method executed by an information processing device and used in a steelworks having a blast furnace department, a coke department, and an energy department, comprising: inputting data on blast furnace operation parameters that have an impact on the energy department and data on coke volume that has an impact on the energy department into a learning model; estimating a total cost based on the input data on blast furnace operation parameters and coke volume data using the learning model; and outputting the estimated total cost; and the learning model is a model trained based on actual data including actual data on blast furnace operation parameters, coke volume, and total cost.

2. The operation support method according to claim 1, wherein the learning model includes a blast furnace sector model, a coke sector model, and an energy sector model, and the learning model includes inputting data on blast furnace operation parameters that have an impact on the energy sector into the blast furnace sector model, outputting estimated values ​​of molten iron components and molten iron temperature, and B gas generation conditions from the blast furnace sector model based on the input data on blast furnace operation parameters, inputting coke volume data that has an impact on the energy sector into the coke sector model, outputting C gas generation conditions from the coke sector model based on the input data on coke volume, inputting the B gas generation conditions and the C gas generation conditions into the energy sector model, estimating the total cost by the energy sector model based on the input B gas generation conditions and C gas generation conditions, and outputting the estimated total cost.

3. The operation support method according to claim 2, wherein the steelworks further includes a steelmaking department, the learning model further includes a steelmaking department model, and the operation support method further includes: inputting estimated values ​​of molten iron components, estimated values ​​of molten iron temperature, and iron mix into the steelmaking department model to determine LD gas generation conditions for the steelmaking department; inputting the B gas generation conditions, the LD gas generation conditions, and the C gas generation conditions into the energy department model, and outputting the total cost based on the input B gas generation conditions, the LD gas generation conditions, and the C gas generation conditions by the energy department model.

4. The operation support method according to claim 3, wherein the pig iron ratio is determined based on the amount of crude steel, the type of steel, and the main unit price.

5. An operation support method according to any one of claims 1 to 4, wherein the blast furnace department includes a plurality of blast furnaces, a total amount of molten iron is allocated to each blast furnace as a molten iron amount based on an upper and lower limit iron production ratio for each blast furnace, and data of the blast furnace operation parameters for each blast furnace is determined based on the allocated molten iron amount.

6. The operation support method according to claim 3, wherein the learning model further estimates carbon dioxide emissions from the blast furnace division, the steelmaking division, the coke division and the energy division based on the input blast furnace operation specifications, coke volume and pig iron mix data, and outputs the estimated carbon dioxide emissions.

7. An operation support method according to any one of claims 1 to 6, wherein the total cost includes a cost equivalent to carbon dioxide emissions.

8. An operation support method according to any one of claims 1 to 7, wherein the blast furnace operation parameters having an impact on the energy sector include reducing material parameters and blast parameters.

9. The operation support method according to claim 8, wherein the reducing material specifications include a coke ratio and a pulverized coal ratio, and the blast specifications include a blast moisture content, a blast temperature, and a blast oxygen concentration.

10. The operation support method according to claim 9, wherein the reducing material specifications include a composition and a carbon content of coke.

11. The operation support method according to claim 1, further comprising inputting data on a predetermined range of blast furnace operation parameters into the learning model, estimating a total cost for the blast furnace department, the coke department, and the energy department corresponding to each input, and outputting a plurality of candidate operation data based on the estimated total cost.

12. The operation support method according to claim 11, wherein a total cost relating to the plurality of candidate operation data satisfies a predetermined criterion.

13. The operation support method according to claim 11 or 12, wherein the plurality of candidate operation data satisfy a predetermined constraint condition.

14. The operation support method according to claim 13, wherein the predetermined constraint conditions include a constraint condition relating to the amount of coke.

15. An operation support method according to any one of claims 1 to 14, wherein the coke amount includes a coke production amount and a coke inventory amount in the coke department.

16. The method of claim 15, further comprising recovering the coke inventory by reducing the coke rate.

17. The method for supporting operations according to claim 15 or 16, further comprising recovering the coke inventory by increasing coke production.

18. An operation support method as described in claim 15, wherein, in a period including a first period and a second period in which the amount of iron produced in the blast furnace department is greater than in the first period, the coke rate is reduced in the first period and the coke production amount is increased in the second period, thereby recovering the coke inventory.

19. The method of claim 15, further comprising increasing the coke rate when the cost of increasing the coke rate to adjust the coke inventory is less than the cost of decreasing coke production.

20. The method of claim 15, further comprising the step of transporting coke from another steelworks to replenish the coke inventory when the coke inventory cannot be replenished by reducing the coke rate and increasing the coke production.

21. The operation support method described in claim 20, wherein the other steel mills include a first steel mill and a second steel mill, and even if the cost of transporting coke from the first steel mill is higher than the cost of transporting coke from the second steel mill, coke is transported from the first steel mill to restore the coke inventory when costs associated with energy supply and demand within the first steel mill are lower than costs associated with energy supply and demand within the second steel mill.

22. An operation support device for a steelworks having a blast furnace department, a coke department, and an energy department, comprising a control unit, wherein the control unit inputs data on blast furnace operation parameters that have an impact on the energy department and data on coke volume that has an impact on the energy department into a learning model for the steelworks, estimates a total cost using the learning model based on the input data on blast furnace operation parameters and coke volume data, and outputs the estimated total cost, wherein the learning model is an operation support device that is a model trained based on actual data including actual data on blast furnace operation parameters, coke volume, and total cost.

23. A display device for displaying the output of an operation support device equipped with a control unit in a steelworks having a blast furnace department, a coke department, and an energy department, the display device displays a total cost estimated by the control unit equipped in the operation support device using a learning model related to the steelworks based on data on blast furnace operation specifications that have an impact on the energy department and data on coke volume that has an impact on the energy department, the learning model being a model trained based on actual data including actual data including blast furnace operation specifications, coke volume, and total cost.

24. A display device as described in claim 23, which filters from among a plurality of candidate operation data based on the total costs of the blast furnace department, the coke department, and the energy department estimated corresponding to each input of data of a predetermined range of blast furnace operation parameters input to the learning model, and displays the filtered candidate operation data.

25. An operations support program for a steelworks having a blast furnace department, a coke department, and an energy department, executed by an information processing device, which causes a computer to execute the following: inputting data on blast furnace operation parameters that have an impact on the energy department and data on coke volume that has an impact on the energy department into a learning model for the steelworks; estimating a total cost using the learning model based on the input data on blast furnace operation parameters and coke volume data; and outputting the estimated total cost, wherein the learning model is a model trained based on actual data including actual data on blast furnace operation parameters, coke volume, and total cost.

Images