Sintered ore return control device, sintered ore return control method, and sintered ore production method

Abstract

A return mine control device (10) is used in a sintering manufacturing facility and performs control relating to return mine, and is provided with: an acquisition unit (12) that acquires data relating to operating conditions of the sintering manufacturing facility as input data; a return hopper level prediction unit (13) that predicts a return hopper level after a predetermined time on the basis of the acquired input data; an operation amount calculation unit (14) that calculates an operation amount of an operation variable selected from the operation condition and the raw material loading condition on the basis of the predicted return hopper level and a predetermined target range of the return hopper level; and an output unit (15) that outputs the calculated operation amount to the sintering manufacturing equipment or presents the operation amount as a guide operation amount.

Description

Technical Field

[0001] This disclosure relates to a return ore control device, a return ore control method, and a method for manufacturing sinter. Background Technology

[0002] In the ironmaking industry, the quality of iron ore has declined due to years of mining. Therefore, the use of fine iron ore with a high powder content, which has been beneficiated in the mining area, has increased, making the sintering process, in which the fine iron ore is solidified and sintered before being charged into the blast furnace, more important.

[0003] Figure 1 This diagram illustrates the general outline of the sintering process. Sinter is used as the iron source for the blast furnace, and is thus manufactured by solidifying small-particle iron ore using the combustion heat of the sintering material. The raw material for sinter is stored in a hopper and cut out from it. If powdered iron ore is directly charged into the sintering machine, the combustion reaction is inhibited due to poor ventilation. Therefore, in the granulation process, the iron ore is mixed with quicklime, coke, and other raw materials in a granulation mixer with water to produce granules with a larger particle size than the original raw material. The granules are then charged into the sintering machine, ignited in a furnace, and the combustion reaction is gradually advanced from top to bottom in layers by air suction from below. After sintering, the sinter is discharged from the sintering machine, pulverized by a crusher, and then sent to a cooler. After being cooled by air in the cooler's pipes, it is screened, with the larger-particle sinter being considered the qualified product. Figure 1 The "products" are fed into the blast furnace. Small-sized sinter (for example, less than 4 mm in diameter) is returned to the return ore hopper as return ore, then cut out from the return ore hopper and fed back into the sintering machine.

[0004] Here, the return rate is defined as the proportion of returned ore in the sinter after firing. As a method to improve yield by reducing the return rate, promoting combustion is effective in preventing the sinter from passing through the sintering machine in an unfired state. For example, increasing the proportion of the binder material (as a heat source), increasing the proportion of quicklime (which acts as a binder during granulation) to improve aeration, and increasing the proportion of coke powder in the upper layer to facilitate ignition in the ignition furnace are all effective methods. Additionally, reducing the tray speed is effective in ensuring sufficient firing time. On the other hand, reducing the return rate is a trade-off with productivity; increasing the proportions of binder material, quicklime, and coke powder increases production costs, while reducing the tray speed decreases production output. Therefore, it is necessary to set appropriate target values ​​based on operating conditions to control the return rate.

[0005] The return rate is measured in the process after screening using a sieve. Therefore, it takes time from the start of firing to the measurement of the return rate, for example, about 2 hours. Additionally, there is another time until the return ore is fed into the return ore hopper (for example, about 0.5 hours). Therefore, even if the return rate increases, if it is detected, for example, 2 hours later, corrective action will be delayed by 2 hours. Similarly, if the return rate decreases, corrective action will also be delayed.

[0006] In order to detect and control changes in the return rate in the early stages, it is necessary to predict the future return rate and take action in advance. For example, Patent Document 1 discloses a method that predicts the amount of return ore based on the content of calcium ferrite, slag content, pore diameter distribution index and porosity of the minerals that make up the sinter.

[0007] Patent Document 1: Japanese Patent Application Publication No. 7-11349

[0008] Here, the method in Patent Document 1 requires sampling tests for physical property identification, making it impossible to continuously predict the return rate. Therefore, it is impossible to detect changes in the return rate in the early stages.

[0009] Furthermore, the return rate may fluctuate in a relatively short cycle of about one hour due to changes in raw material composition or calcination conditions. Therefore, if an operation to reduce the return rate is performed, such as increasing the quicklime blending ratio when the return rate temporarily increases, it is possible to excessively increase the quicklime blending ratio, leading to increased production costs. Conversely, if an operation to increase the return rate is performed, such as decreasing the quicklime blending ratio when the return rate temporarily decreases, it is possible to excessively decrease the quicklime blending ratio, resulting in an increased return rate and decreased productivity. Additionally, frequently performing operations to correct for fluctuations in the return rate within a one-hour cycle may cause furnace conditions to become unstable. Hereinafter, such operational actions corresponding to temporary increases or decreases in the return rate—that is, excessively increasing or excessively decreasing the quicklime blending ratio—are also referred to as over-operations. Summary of the Invention

[0010] The purpose of this disclosure, made in view of this situation, is to provide a return ore control device, a return ore control method, and a method for manufacturing sinter that can reduce excessive action to suppress the increase in production costs.

[0011] The inventors conducted in-depth research on methods for solving the aforementioned problems and discovered that controlling the return ore hopper level instead of directly controlling the return ore rate is effective in preventing over-action. The amount of return ore corresponding to changes in the return ore rate is accumulated and reflected in the return ore hopper level, thus allowing indirect management of the return ore rate by controlling the return ore hopper level within a target range. Furthermore, changes in the amount of return ore are evaluated using cumulative values, thereby reducing the impact of temporary increases or decreases in the return ore rate.

[0012] (1) One embodiment of the present disclosure of the return ore control device is used in a sintering manufacturing equipment to perform control related to return ore, wherein it comprises:

[0013] The acquisition unit acquires data related to the operating conditions of the aforementioned sintering manufacturing equipment as input data.

[0014] The return ore hopper level prediction unit predicts the return ore hopper level after a predetermined time based on the above-mentioned input data.

[0015] The operation quantity calculation unit calculates the operation quantity of the operation variable selected from the above-mentioned operating conditions and raw material loading conditions, based on the predicted return ore hopper level and the predetermined target range of the above-mentioned return ore hopper level; and

[0016] The output unit outputs the calculated operation quantity to the sintering manufacturing equipment or provides a prompt as a guide operation quantity.

[0017] (2) As one embodiment of this disclosure, based on (1),

[0018] The above input data includes the measured value of the return ore hopper level and the return ore cutting volume, as well as the raw material type, raw material moisture content, water spray flow rate in the granulation mixer, the proportion of coagulant, the proportion of quicklime, the proportion of iron ore, the proportion of return ore, and the exhaust NO. x At least one of the following: concentration, exhaust CO concentration, exhaust CO2 concentration, exhaust O2 concentration, exhaust temperature, cooler air supply pressure, tray speed, layer thickness, and production volume.

[0019] (3) As one embodiment of this disclosure, based on (1) or (2),

[0020] The aforementioned return ore hopper level prediction unit uses a prediction model to predict the level of the return ore hopper.

[0021] The aforementioned prediction model was generated using machine learning with learning data, which is the data corresponding to the input data extracted from the actual data of the sintering process, taking into account the lag time, and the learning data corresponding to the return ore hopper level extracted from the actual data.

[0022] (4) One embodiment of the return ore control method of this disclosure is used in a sintering manufacturing equipment to perform control related to return ore, wherein it comprises:

[0023] The acquisition step involves acquiring data related to the operating conditions of the aforementioned sintering manufacturing equipment as input data.

[0024] The return ore hopper level prediction step predicts the return ore hopper level after a predetermined time based on the obtained input data.

[0025] The operation quantity calculation step involves calculating the operation quantity of the selected operation variables from the aforementioned operating conditions and raw material loading conditions, based on the predicted return ore hopper level and the predetermined target range of the aforementioned return ore hopper level; and

[0026] The output step involves outputting the calculated operation quantities to the sintering manufacturing equipment or providing prompts as guiding operation quantities.

[0027] (5) In one embodiment of the present disclosure, the method for manufacturing sintered ore uses the above-mentioned operating quantity output to the above-mentioned sintering manufacturing equipment to manufacture sintered ore by means of the return ore control method in (4).

[0028] According to this disclosure, a return ore control device, a return ore control method, and a method for manufacturing sinter can be provided that can reduce excessive action to suppress the increase in production costs. Attached Figure Description

[0029] Figure 1 This is a diagram that shows an overview of the sintering process.

[0030] Figure 2 It is a graph representing the error of the predicted values ​​of the predictive model corresponding to the number of explanatory variables.

[0031] Figure 3 This is a diagram illustrating an example configuration of a ore return control device according to one embodiment of this disclosure.

[0032] Figure 4 This is a flowchart illustrating a method for controlling ore return according to one embodiment of this disclosure.

[0033] Figure 5 This is a graph showing the results of Example 1.

[0034] Figure 6 This is a graph showing the results of Example 2. Detailed Implementation

[0035] Hereinafter, a return ore control device, a return ore control method, and a method for manufacturing sinter according to one embodiment of the present disclosure will be described with reference to the accompanying drawings. The return ore control method according to this embodiment, as a summary, uses a machine learning model to predict the return ore hopper level in real time. The machine learning model is a learned model generated through machine learning. Furthermore, based on the predicted values, it is possible to calculate and output the operating quantity in a manner that brings the return ore hopper level within a target range, thereby controlling the return ore hopper level within an appropriate range or implementing appropriate operational actions.

[0036] In the return ore control method described in this embodiment, a predictive model is used to predict the return ore hopper level. For example... Figure 1 As shown, the level of the return ore hopper is the amount of return ore that is returned and fed into the return ore hopper. For example, it can be expressed as a percentage (%) relative to the maximum amount of return ore hopper, as in this embodiment. The level of the return ore hopper is not limited to a percentage; for example, it can also be expressed as the height from the bottom of the return ore hopper (height level), or as the volume of return ore, etc.

[0037] The estimation of the return hopper level is performed in each control cycle by predicting the increase or decrease in the return hopper level at a predetermined future time based on the current level. The increase or decrease in the return hopper level can be based on the predicted value Y of the return ore production from the current time to the predetermined future time. p The amount of ore cut-out u from the hopper is predicted by the following equation (1).

[0038] H p (t+k)=H(t)+Y p (t+k)-u(t+k) … (1)

[0039] Here, H(t) is the measured value of the return ore hopper level at time t. If time t is the current time, then H(t) corresponds to the current return ore hopper level. p (t+k) is the predicted value of the return ore hopper level at a future time t+k, which is a predetermined time k starting from time t. Y p(t+k) is the predicted value of the return ore production from time t to a future time t+k after a predetermined time k. u(t+k) is the return ore cut-out amount from the return ore hopper from time t to a future time t+k after a predetermined time k. Here, the return ore cut-out amount from the return ore hopper, i.e., u(t+k), can be calculated as the actual value of the return ore cut-out amount at time t. The return ore cut-out amount is not frequently revised. Therefore, even if it is processed as the actual value to maintain the status quo, the prediction accuracy will generally not have a major problem. In addition, even if the return ore cut-out amount is revised, the revision is only reflected in the next control cycle after the revision, so it will not have a major impact on the prediction accuracy.

[0040] The predicted value of the amount of ore produced from time t to a future time t+k, i.e., Y p The calculation of (t+k) can use a predictive model. Here, Y is sometimes not shown below. p (t+k) is used, but only the term "predicted value of return ore production" is used. In addition, time k is determined based on the equipment structure or the raw material delivery time. For example, time k can be determined by considering the time from the start of firing to the measurement of the return ore rate (e.g., 2 hours) or the time from when the return ore is put into the return ore hopper (e.g., 0.5 hours).

[0041] The predictive model is input with data relating to the operating conditions of the equipment used to manufacture sinter (sintering equipment), including sintering machines and return ore hoppers. This input data is used as the explanatory variable, and the amount of return ore produced from time t to time t+k is used as the target variable. Based on the predicted return ore production, the return ore hopper level at the predicted time, and the amount of return ore cut off from the hopper, the predicted value of the return ore hopper level is calculated. Furthermore, the sintering process executed by the sintering equipment can be controlled based on the predicted return ore hopper level. For example, the sintering equipment can automatically or manually operate to keep the return ore hopper level within a predetermined target range during the sintering process.

[0042] The input data for the predictive model of return ore production includes raw material type [id], raw material moisture [%], water flow rate in the granulation mixer [ton / hr], proportion of coagulant [%], proportion of quicklime [%], proportion of iron ore [%], proportion of return ore [%], and exhaust NO. xAt least one of the following: concentration [%), exhaust CO concentration [%), exhaust CO2 concentration [%), exhaust O2 concentration [%), exhaust temperature [°C], cooler air pressure [kPa], tray speed [m / min], layer thickness [mm], and production rate [ton / hr]. For example, the input data can also include all of these characteristics and can be combined with other operating factors. Here, raw material type and raw material moisture content refer to the type of iron ore and the moisture content of the raw material, respectively. Raw material type is classified into multiple groups based on its composition, and a label is attached according to which group it belongs to, and is considered as weight data corresponding to each label. Additionally, raw material moisture content is expressed as a weight ratio of the moisture content in each raw material. Furthermore, as mentioned above, iron ore is mixed with water in a granulation mixer along with other raw materials such as quicklime and coke, but the water spray flow rate is the flow rate of the mixed water when it is sprayed. The proportions of the coagulant, quicklime, iron ore, and return ore are the proportions of the coagulant, quicklime, iron ore, and return ore sintered ore, respectively. Exhaust NO x Concentration, exhaust CO concentration, exhaust CO2 concentration, exhaust O2 concentration, and exhaust temperature are the concentrations and temperatures of the various components in the exhaust gas produced during firing. Cooler air supply pressure is the pressure of the air supplied to the cooler during the cooling of the sinter. Pallet speed is the conveying speed of the pallet carrying the raw material during firing. Layer thickness is the thickness of the layer of raw material on the pallet. Furthermore, production volume is the amount of sinter produced. Specifically, production volume can be determined based on the weight and conveying speed measured by a Merrick conveyor belt scale or similar device installed on the discharge side of the sintering machine.

[0043] Here, among the aforementioned characteristic quantities, the cooler supply air pressure and exhaust NO are... X Concentration, return ore blending ratio, and production volume are particularly important. The input data should ideally include at least the cooler supply air pressure and exhaust NO₂. X Concentration, return ore ratio, and production volume. A higher return ore ratio results in finer sinter particle size, worse aeration, and a decrease in air suction pressure (increased negative pressure) when cooling the sinter in the cooler. Therefore, a higher cooler air pressure leads to a higher return ore hopper level. Here, the sensor for measuring the cooler air pressure only needs to be able to measure pressure; for example, a strain gauge, metal gauge, semiconductor gauge, or semiconductor diaphragm pressure gauge can be used. Additionally, under insufficient heat conditions, the CO partial pressure decreases while the exhaust NO... x The concentration increases. Therefore, the exhaust NO... x As the concentration increases, the return rate increases, leading to a higher level of the return ore hopper. Conversely, as production increases, the return ore increases, further increasing the level of the return ore hopper. Furthermore, as the return ore blending ratio increases, the cut-out rate of the return ore hopper increases, thus lowering the level of the return ore hopper.

[0044] The predictive model is not limited to a specific model as long as it is constructed to obtain the target variable (return ore production) from the explanatory variables. The predictive model can be, for example, a physical model or a machine learning model. In this embodiment, the predictive model is generated through machine learning. As a machine learning method, linear regression, neural networks, decision trees, GBDT (Gradient Boosting Decision Tree), random forests, etc., can be used, without particular limitation. Here, the predictive model can be a model that directly predicts the return ore hopper level. In this case, in addition to the explanatory variables of the predictive model for return ore production, the measured value of the return ore hopper level and the return ore cut-out amount can be used as necessary input data to construct the learning model. Alternatively, a model can be constructed using hierarchical Bayesian inference methods, which infers the future return ore hopper level based on the measured value of the return ore hopper level, using the explanatory variables of the predictive model for return ore cut-out amount and return ore production. Before predicting the return ore hopper level, for example, the predictive model can be generated using historical data from the sintering process (past measurements and past setpoints, etc.).

[0045] Figure 2 This is an example of evaluating predictive models generated using neural network methods. The number (types) of explanatory variables in the machine learning were increased, and the error between the predicted and actual return ore production values ​​based on each machine learning model was measured. For example, the prediction error of a model based solely on cooler air pressure as the explanatory variable was 7.3%. Here, the error was evaluated using Mean Absolute Error (MAE). In contrast, models based on cooler air pressure and exhaust NO₂ as explanatory variables... x The prediction error of the model for concentration, production volume, and return ore blending ratio was 3.2%. This shows a trend where the error decreases as the number of explanatory variables increases. Here, Figure 2 This is an example illustrating the order in which explanatory variables are added. For instance, the ratio of returned ore can be added after the cooler air pressure as an explanatory variable. It can be seen that... Figure 2 Regardless of the order in which the variables are added, the more the number of variables is increased, the lower the error tends to be.

[0046] Furthermore, various operating factors (operating variables related to the sintering process) can be used as input data for the predictive model. However, depending on the operating variable, the time it takes for the return ore hopper level to be affected—that is, the time it takes for the amount of return ore fed into the hopper to have an impact—varies. For example, data related to raw materials, such as the quicklime mixing ratio, is located upstream in the sintering process and therefore can affect the return ore hopper level after 3.5 hours. Data related to calcination, such as exhaust NO₂… xThe concentration is in the middle of the sintering process, and therefore can affect the return ore hopper level after 2.5 hours. For example, data related to the cooler, i.e., the cooler air pressure, is in the downstream of the sintering process, and therefore can affect the return ore hopper level after 1.5 hours. In addition, the time until the return ore production is affected can also vary depending on the tray speed. Therefore, in order to make accurate predictions, the input data of the prediction model is preferably obtained with a lag time as a reference for determining the time when the return ore hopper level can be determined. Here, the time when the return ore hopper level can be determined is, for example, when the return ore is transported and fed into the return ore hopper after being screened by a sieve to determine whether the particle size is below a reference after being cooled by the cooler. The operational variables of the sintered ore that affect this specific time are obtained with the lag time mentioned above. Similarly, the prediction model is preferably generated by machine learning using data corresponding to the explanatory variables (input data) extracted from the actual data of the sintering process, taking into account the lag time, and learning data corresponding to the return ore hopper level extracted from the actual data.

[0047] The process of predicting the return ore hopper level is performed using input data including the measured value of the return ore hopper level and the return ore cutting amount, as described above, and a prediction model. To avoid action lag for correction, the return ore hopper level is predicted after a predetermined time (e.g., 2.5 hours in an example). Furthermore, the operation amount of the sintering process operation variable is calculated every other predetermined control period (e.g., every hour) to ensure that the predicted value of the return ore hopper level falls within the target range. Actions to reduce the return ore rate are implemented when the return ore hopper level is above the upper limit of the target range, actions to increase the return ore rate are implemented when it is below the lower limit of the target range, and actions are reserved in other cases, thereby preventing over-action. In the calculation of the operation amount, a machine learning model with the operation amount as the explanatory variable can be constructed, and the operation amount that makes the predicted value of the return ore hopper level consistent with the target value (e.g., the midpoint of the target range) is calculated through inverse analysis.

[0048] As an action to reduce the level of the return ore hopper, promoting combustion is effective in preventing the sinter from passing through the sintering machine in an unburned state. For example, increasing the proportion of the coagulant material, which serves as a heat source, is considered. Additionally, increasing the proportion of quicklime, which acts as a binder during granulation, to improve aeration, increasing the proportion of coke powder in the upper layer to facilitate ignition in the ignition furnace, and ensuring sufficient firing time by reducing the tray speed are also effective actions to increase the level of the return ore hopper. Reducing the proportion of coagulant material, reducing the proportion of quicklime, reducing the proportion of coke powder in the upper layer, and increasing the tray speed are effective actions to increase the level of the return ore hopper. Here, as mentioned above, the reduction of the return ore hopper level is in a trade-off with productivity; increasing the proportions of coagulant material, quicklime, and coke powder increases production costs, while reducing the tray speed reduces production output. Therefore, the target range for the return ore hopper level is preferably determined from the viewpoint of production cost and production output, and it is preferable to keep the return ore hopper level within the target range. Here, to suppress over-operation (excessive operating volume), the corrective action can determine the range of operating volume for each operating variable and operate within that range.

[0049] Here, the calculated values ​​of the operational variables can be output in a manner that the process computer managing the sintering process can reflect. The output of these operational values ​​includes output as operational guidance for the operator running the sintering machine. That is, operational values ​​can be output to reflect the appropriate values ​​of the operational variables in the sintering process based on the operator's judgment. The information as operational guidance output includes at least the calculated values ​​of the operational variables and can be displayed on a monitor visible to the operator.

[0050] Figure 3 This diagram illustrates the structure of the ore return control device 10 according to this embodiment. The ore return control device 10 is used in sintering manufacturing equipment to perform ore return-related controls, including the ore return control method described above. Figure 3 As shown, the return ore control device 10 includes a storage unit 11, an acquisition unit 12, a return ore hopper level prediction unit 13, an operation quantity calculation unit 14, and an output unit 15. The return ore control device 10 acquires data related to the operating conditions of the sintering manufacturing equipment, i.e., the aforementioned input data, from the operation data server 60. In addition to the measured value of the return ore hopper level and the amount of return ore cut from the hopper, the input data also includes the cooler air supply pressure and exhaust NO₂. XThe actual values ​​of characteristic quantities such as concentration, return ore blending ratio, and production volume are displayed. These actual values ​​include measured values ​​and set values ​​for the operating variables. Furthermore, the return ore control device 10 can obtain the target range of the return ore hopper level from the operation data server 60. The operation data server 60 can communicate with the return ore control device 10 via a network, for example, a computer that manages the production of sinter. The network is, for example, the Internet. The return ore control device 10 performs the aforementioned processing, namely, using a predictive model to predict the return ore hopper level and determining the operating quantities of operating variables such as cooler air pressure in order to maintain the future return ore hopper level within the target range. In this embodiment, the return ore control device 10 also has the function of outputting the operating quantities of the operating variables to the sintering equipment via the output unit 15, or providing a prompt as a guide operating quantity. When the output unit 15 prompts a guide operating quantity, the return ore control device 10 functions as an operation guide device. The display unit 30 displays the guide operating quantity output from the return ore control device 10 (operation guide device). The ore return control device 10 can be configured as a computer (e.g., a process computer that manages the operation of the sintering machine or a sintering operation guidance server) that is different from the operation data server 60. The display unit 30 can be a display device such as a liquid crystal display (LCD) or an organic electroluminescence panel (OLED). Alternatively, the display unit 30 can be implemented using a display of a terminal device such as a smartphone or tablet. The terminal device can communicate with the ore return control device 10 via a network. A sintering operation guidance system can be configured by a sintering operation guidance server having the functions of the ore return control device 10 and a terminal device having the functions of the display unit 30. The sintering operation guidance server and the terminal device can be located in the same place (e.g., within the same factory) or physically separated. The sintering operation guidance system can also include the operation data server 60.

[0051] Here, the prediction model can be generated and stored in the storage unit 11 by the ore return control device 10, or it can be generated and stored in the storage unit 11 by another computer. When the ore return control device 10 generates the prediction model, a model generation unit may also be provided, which generates the prediction model and stores it in the storage unit 11 using the method described above.

[0052] The following describes the constituent elements of the return ore control device 10. The storage unit 11 stores the prediction model. Additionally, the storage unit 11 stores programs and data related to the level control of the return ore hopper. The storage unit 11 can store acquired input data. The storage unit 11 can store the target range for the level of the return ore hopper. The storage unit 11 can store various information obtained through processing for the level control of the return ore hopper. The storage unit 11 can be configured as any storage device, including semiconductor storage devices, optical storage devices, and magnetic storage devices. Semiconductor storage devices may include, for example, semiconductor memories. The storage unit 11 can include various types of storage devices.

[0053] The acquisition unit 12 acquires input data. Preferably, the acquisition unit 12 considers a characteristic quantity for acquiring input data based on a lag time that allows the timing of determining the level of the return ore hopper.

[0054] The return ore hopper level prediction unit 13 predicts the return ore hopper level after a predetermined time based on the acquired input data. A prediction model is used in the prediction of the return ore hopper level.

[0055] The operation quantity calculation unit 14 calculates the operation quantity of the operation variables selected from the operating conditions and raw material loading conditions based on the predicted return ore hopper level and the predetermined target range of the return ore hopper level. The raw material loading conditions include, for example, the mixing ratio of binders, quicklime, iron ore, or return ore.

[0056] The output unit 15 outputs the calculated operation quantity to the sintering manufacturing equipment, or provides it as a guide operation quantity to the display unit 30.

[0057] When the operation variable is output from the output unit 15 to the sintering manufacturing equipment, the sintering manufacturing equipment can automatically update the operation variable to the output operation value to manufacture sinter. That is, the return ore control method according to this embodiment can be executed as part of the manufacturing method for manufacturing sinter. In addition, the operator can change the operating conditions of the sintering machine based on the guide operation value displayed on the display unit 30. If the return ore hopper level is higher than a preset target range in the future, the operator can reduce the return ore rate to lower the return ore hopper level. In addition, if the return ore hopper level is lower than the target range, the operator can increase the return ore rate to increase the return ore hopper level. This sintering machine operation guidance can be executed as part of the manufacturing method for manufacturing sinter.

[0058] The ore return control device 10, as described above, can be implemented by a computer, for example. The computer includes, for example, a memory, a hard disk drive (storage device), and a CPU (processing unit). The program can be stored on the hard disk drive and read from the hard disk drive into the memory when executed by the CPU. Additionally, data generated during processing is stored in the memory and, if necessary, on the hard disk drive (HDD). The storage unit 11 can be implemented by a storage device, for example. The acquisition unit 12, the ore return hopper level prediction unit 13, the operation quantity calculation unit 14, and the output unit 15 can be implemented by a CPU that reads and executes the program, for example.

[0059] Figure 4 This is a flowchart illustrating the ore return control method according to this embodiment. The ore return control device 10 can be based on... Figure 4 The flowchart shown calculates the operand of the operand variable and outputs it as the guide operand. Figure 4 The ore return control method shown is also a running guidance method, which can be implemented as part of the sinter manufacturing process.

[0060] The acquisition unit 12 acquires input data (step S1, acquisition step). The return ore hopper level prediction unit 13 uses the input data and the prediction model to predict the return ore hopper level after a specified time (step S2, return ore hopper level prediction step). The operation quantity calculation unit 14 calculates the operation quantity of the operation variable to ensure that the predicted value of the return ore hopper level is within the target range (step S3, operation quantity calculation step). The output unit 15 outputs the calculated operation quantity of the operation variable (step S4, output step).

[0061] (Example 1)

[0062] The following describes a specific example (implementation) of determining the optimal operating variable using the control method described above. In Example 1, on a sintering production line, a neural network-based machine learning model was used to predict the amount of return ore produced after 2.5 hours, and the return ore hopper level was predicted based on the measured value of the return ore hopper level and the return ore cut-out amount. The input data included raw material type, raw material moisture content, water spray flow rate in the granulation mixer, proportion of flocculating materials, proportion of quicklime, proportion of iron ore, proportion of return ore, and exhaust NO. x Concentration, exhaust CO concentration, exhaust CO2 concentration, exhaust O2 concentration, exhaust temperature, cooler air supply pressure, tray speed, layer thickness, and production rate. The target range for the return ore hopper level is set at 40%–60%. When the return ore hopper level is below 40%, quicklime reduction is implemented. Conversely, when the return ore hopper level is above 60%, quicklime increase is implemented. The quicklime processing volume is fixed at 0.1% per cycle, and operations are conducted hourly. Figure 5As shown, in existing control (direct control of return ore rate) operation, the average quicklime utilization rate (quicklime mixing ratio) is 1.5% over 100 hours of operation. By applying the method of this embodiment, the average quicklime utilization rate is 1.3% over 100 hours of operation. In this embodiment, the quicklime mixing ratio will not become excessively high. That is, excessive action will not occur, and the quicklime mixing ratio can be suppressed to curb the increase in production costs.

[0063] (Example 2)

[0064] In addition, in Example 2, a neural network-based machine learning model was used on the sintering production line to predict the amount of return ore generated after 2.5 hours. The return ore hopper level was also predicted based on measured values ​​and the amount of return ore cut out. Input data included raw material type, raw material moisture content, water flow rate in the granulation mixer, proportion of bridging materials, proportion of quicklime, proportion of iron ore, proportion of return ore, and exhaust NO₂. x Concentration, exhaust CO concentration, exhaust CO2 concentration, exhaust O2 concentration, exhaust temperature, cooler air supply pressure, tray speed, layer thickness, and production rate. The target range for the return ore hopper level was set at 40%–60%. When the return ore hopper level was below 40%, quicklime was reduced. Conversely, when the return ore hopper level was above 60%, quicklime was increased. The amount of quicklime operated was determined through reverse analysis based on the calculation results of a machine learning model. Reverse analysis is not limited to a specific method, but in this embodiment, the input data of the machine learning model was changed to perform multiple calculations, and the optimal amount of quicklime operated was explored using a bisection method. Operations (actions) were performed every hour. Figure 6 As shown, in existing control (direct control of return ore rate) operation, the average quicklime utilization rate (quicklime mixing ratio) is 1.5% over 100 hours of operation. By applying the method of this embodiment, the average quicklime utilization rate is 1.2% over 100 hours of operation. In this embodiment, the quicklime mixing ratio will not become excessively high. That is, excessive action will not occur, and the quicklime mixing ratio can be suppressed to curb the increase in production costs.

[0065] As described above, according to the above embodiments, the return ore control device 10, the return ore control method, and the sinter manufacturing method involved in this embodiment can reduce excessive movement and suppress the increase in production costs by controlling the level of the return ore hopper.

[0066] While the embodiments described herein are based on the accompanying drawings and examples, it should be noted that those skilled in the art can readily make various modifications or alterations based on this disclosure. Therefore, it should be understood that such modifications or alterations are included within the scope of this disclosure. For example, the functions included in each structural component or step can be reconfigured in a logically consistent manner, and multiple structural components or steps can be combined into one or divided. The embodiments described herein can also be implemented as a program executed by a processor of a device or a storage medium containing a program. These should also be understood to be included within the scope of this disclosure.

[0067] Figure 3 The structure of the ore return control device 10 shown is an example. The ore return control device 10 may not include... Figure 3 All of the constituent elements shown. Additionally, the return ore control device 10 may include... Figure 3 Other components not shown. For example, the ore return control device 10 may also include a display unit 30.

[0068] Furthermore, in the above embodiments, it was explained that the prediction model can be either a "prediction model for return ore production (first prediction model)" or a "model that directly predicts the level of the return ore hopper (second prediction model)." From the viewpoint of obtaining the predicted value of the return ore hopper level, the two models can be considered to have no substantial difference. However, the second prediction model, as described above, is constructed by further using the measured value of the return ore hopper level and the return ore cutting-out amount as input data, enabling the return ore hopper level to exhibit dynamic characteristics including lag actions. For example, in cases where there are temporary return ore buffers along the actual return ore conveying path, constructing a model that includes lag actions can be expected to further improve prediction accuracy.

[0069] Explanation of reference numerals in the attached figures

[0070] 10…Return ore control device; 11…Storage unit; 12…Acquisition unit; 13…Return ore hopper level prediction unit; 14…Operation quantity calculation unit; 15…Output unit; 30…Display unit; 60…Running data server.

Claims

1. A return ore control device, used in sintering manufacturing equipment, for controlling return ore-related processes, characterized in that, have: The acquisition unit acquires data related to the operating conditions of the sintering manufacturing equipment as input data. The return ore hopper level prediction unit predicts the return ore hopper level after a predetermined time based on the acquired input data. The operation quantity calculation unit calculates the operation quantity of the operation variable selected from the operating conditions and the raw material loading conditions based on the predicted level of the return ore hopper and the predetermined target range of the level of the return ore hopper. as well as The output unit outputs the calculated operation quantity to the sintering manufacturing equipment or provides a prompt as a guiding operation quantity.

2. The ore return control device according to claim 1, characterized in that, The input data includes the measured value of the return ore hopper level and the return ore cutting volume, as well as the raw material type, raw material moisture content, water spray flow rate in the granulation mixer, the proportion of coagulant, the proportion of quicklime, the proportion of iron ore, the proportion of return ore, and the exhaust NO. x At least one of the following: concentration, exhaust CO concentration, exhaust CO2 concentration, exhaust O2 concentration, exhaust temperature, cooler air supply pressure, tray speed, layer thickness, and production volume.

3. The ore return control device according to claim 1 or 2, characterized in that, The return ore hopper level prediction unit uses a prediction model to predict the level of the return ore hopper. The prediction model is generated by machine learning using learning data, which is learning data corresponding to the input data extracted from the actual data of the sintering process, taking into account lag time, and the return ore hopper level extracted from the actual data.

4. A method for controlling ore return, used in sintering manufacturing equipment, for controlling ore return, characterized in that, have: The acquisition step involves acquiring data related to the operating conditions of the sintering manufacturing equipment as input data. The return ore hopper level prediction step predicts the return ore hopper level after a predetermined time based on the acquired input data. The operation quantity calculation step calculates the operation quantity of the operation variable selected from the operating conditions and raw material loading conditions based on the predicted level of the return ore hopper and the predetermined target range of the level of the return ore hopper. as well as The output step involves outputting the calculated operation quantity to the sintering manufacturing equipment or providing a prompt as a guiding operation quantity.

5. A method for manufacturing sintered ore, characterized in that, The return ore control method of claim 4 uses the operating quantity output to the sintering manufacturing equipment to manufacture sinter.

Images