Dynamic optimization of steelmaking alloying cost and precise feeding control system

Abstract

This invention provides a dynamic optimization and precise feeding control system for steelmaking alloying costs, belonging to the field of steelmaking optimization control technology. A coupled data acquisition module collects parameters and constructs partitioned coupled state data; a cost-coordination risk identification module identifies anomalies and constructs cost-coordination risk data; a feeding strategy optimization module formulates and optimizes precise feeding control strategies; and a feedback control module evaluates control nodes and generates risk adjustment strategies. This invention achieves early prediction and intervention of anomalies by constructing a risk identification and quantitative assessment system, dynamically optimizes control strategies based on real-time equipment status, and continuously corrects control effects through a feedback control mechanism. This solves the problems of traditional control methods being crude, lagging, and poorly adaptable, reducing alloy consumption while ensuring product quality, and improving the overall efficiency and intelligence level of smelting.

Description

Technical Field

[0001] This invention belongs to the field of steelmaking optimization control technology, specifically a dynamic optimization and precise feeding control system for steelmaking alloying costs. Background Technology

[0002] In the steel smelting process, precise control of composition is the core link that determines product quality and production cost. Among them, the alloy addition process is a key step in adjusting the composition of steel, which directly affects the mechanical properties, process properties and overall smelting economy of the final steel.

[0003] Existing technologies mostly employ a holistic, extensive control model based on historical experience, which struggles to accurately reflect the complex physical field distribution and state differences within the smelting vessel, making it impossible to achieve refined process control by region. The identification of abnormal operating conditions often relies on operators' on-site experience or post-event component testing, exhibiting significant time lag and hindering targeted intervention at the nascent stage of problems. Furthermore, existing control methods do not fully consider the impact of long-term equipment operation on the smelting process, leading to severe mismatches between preset control parameters and actual production conditions. This results in frequent issues such as excessive component fluctuations or over-addition of alloys, making it difficult to guarantee product quality consistency and stability, and causing unnecessary waste of precious metal raw materials. It fails to achieve an optimal balance between quality compliance and cost control, severely restricting further improvements in the automation and intelligence levels of the steel smelting process. Summary of the Invention

[0004] This invention provides a dynamic optimization and precise feeding control system for steelmaking alloying costs, in order to overcome the deficiencies in the existing technology.

[0005] This invention provides a dynamic optimization and precise feeding control system for steelmaking alloying costs, comprising:

[0006] The coupled data acquisition module is used to collect all-dimensional operating parameters of the steelmaking alloying system, divide the ladle molten pool into zones, collect the physical field parameters of each zone in real time, and correlate the physical field parameters to construct zone coupled state data.

[0007] The cost collaboration risk identification module is used to identify alloying anomaly scenarios based on partitioned coupled state data. It combines the current stage of the steelmaking alloying process and the aging state of the ladle to construct cost collaboration risk data under different alloying anomaly scenarios.

[0008] The feeding strategy optimization module is used to formulate a corresponding initial feeding control strategy based on the preset steel composition compliance constraints and the goal of minimizing the alloy cost per ton of steel. According to the cost coordination risk data, the set of parameters to be adjusted is marked in the initial feeding control strategy. Based on the sensitivity coefficient of alloy yield to each adjustment parameter, the adjustment range and priority are determined to obtain an optimized and precise feeding control strategy.

[0009] The feedback control module is used to collect the comprehensive composition balance index of the steel coupling state data of each zone after feeding control in real time. It performs a comprehensive evaluation of each control node in the optimized precision feeding control strategy, defines the control node that fails the evaluation as a risk node, and generates a risk adjustment strategy pointing to the risk node.

[0010] The dynamic optimization and precise feeding control system for steelmaking alloying costs provided by the present invention includes a coupled data acquisition module comprising a molten pool partitioning deployment unit, a multi-source data acquisition unit, and a coupled data construction unit.

[0011] The molten pool partitioning deployment unit is used to collect the ladle structure dimensions, bottom-blown argon port distribution locations, electrode insertion depth parameters, and molten steel loading volume as full-dimensional operating parameters. Based on the steel grade specifications and target alloying cycle, it ensures that the bottom-blown argon port distribution meets the requirements for uniform molten steel stirring. Based on the bottom-blown argon port distribution, the ladle molten pool is uniformly divided to determine the number of partitions. Based on the fixed distance between the molten steel flow trajectory and the center of the argon column, a ring-shaped partition centered on the argon column is formed to determine the partition boundaries.

[0012] The multi-source data acquisition unit is used to combine the number of partitions and the partition boundaries, deploy corresponding detection sensors in each partition and perform partition calibration, and collect data on the temperature field distribution of molten steel, composition concentration gradient, argon gas stirring flow field, alloy melting rate and slag basicity of each partition in real time as physical field parameters.

[0013] The coupling data construction unit is used to align the physical field parameters of the same partition in time and space based on the system timestamp, mark the correlation characteristics between different parameters, and generate partition coupling state data.

[0014] The dynamic optimization and precise feeding control system for steelmaking alloying cost provided by the present invention includes a multi-source data acquisition unit comprising a temperature field acquisition subunit, a composition field acquisition subunit, a flow field acquisition subunit, a melting rate calculation subunit, and a slag basicity detection subunit.

[0015] The temperature field acquisition subunit is used to acquire the temperature of the upper, middle and lower layers of molten steel and the edge points of the ladle wall in each zone, calculate the maximum temperature difference and temperature gradient, and form the temperature field distribution of molten steel.

[0016] The composition field acquisition subunit is used to collect the mass fraction of the target alloying elements in each zone, and at the same time record the viscosity and flow rate of molten steel to form composition concentration data. The composition concentration gradient is obtained by correcting the current temperature of molten steel.

[0017] The flow field acquisition subunit is used to collect the molten steel flow velocity value and the flow velocity difference between the argon column and the edge of the molten pool in each zone according to a preset frequency as flow field data, calculate the flow velocity gradient in the zone, analyze the fluctuation amplitude and frequency of the flow field and the time period of abrupt flow velocity change, and form an argon gas stirring flow field.

[0018] The melting rate calculation subunit is used to detect the rate of temperature change and concentration change around the alloy particles in each zone, and to calculate the alloy melting rate.

[0019] The slag alkalinity detection subunit is used to collect slag samples from the slag layers in each zone, detect the mass fraction of calcium oxide and silicon dioxide, calculate the binary alkalinity of the slag according to the slag alkalinity formula, and record the alkalinity differences in different layers to obtain slag alkalinity data.

[0020] According to the dynamic optimization and precise feeding control system for steelmaking alloying costs provided by the present invention, the cost collaborative risk identification module includes an abnormal scenario matching unit. The abnormal scenario matching unit is used to standardize the partition coupling state data, align the spatiotemporal dimensions according to the partition number and timestamp to obtain the coupling state processing data, extract the parameter features and coupling features related to alloying anomalies as alloying anomaly features, and set three-level thresholds for the extracted alloying anomaly features in combination with the composition requirements and cost control requirements of the target steel grade. Based on the three-level thresholds and alloying anomaly features, an abnormal scenario matching rule base is established, the coupling state processing data of each partition is traversed, and the corresponding alloying anomaly scenario is determined in combination with the matching rule base.

[0021] According to the dynamic optimization and precise feeding control system for steelmaking alloying costs provided by the present invention, the cost collaborative risk identification module also includes a ladle aging calculation unit. The ladle aging calculation unit is used to calculate the cumulative number of heats used by the ladle, the current thermal conductivity of the inner lining and the erosion thickness of the inner lining, and calculate the ladle aging coefficient after normalization.

[0022] According to the dynamic optimization and precise feeding control system for steelmaking alloying costs provided by this invention, the cost synergy risk identification module further includes a risk data calculation unit. This unit couples deviation characteristics of abnormal alloying scenarios, process-sensitive factors, and ladle aging coefficients to calculate a cost synergy risk index. Based on a preset numerical range, the cost synergy risk index is divided into multiple risk levels, and the risk characteristics and control requirements corresponding to each risk level are determined, generating cost synergy risk data. The process of calculating the cost synergy risk index includes:

[0023] For alloying anomaly scenarios, risk factors for anomaly scenarios are classified based on the degree of cost impact, the scope of component impact, and the duration of the anomaly.

[0024] The process sensitivity factor is calculated by combining the basic weights and relative deviation rates of parameters at the current process stage, and the cost synergy risk index is calculated by weighted summation formula.

[0025] According to the dynamic optimization and precise feeding control system for steelmaking alloying cost provided by the present invention, the feeding strategy optimization module includes an initial strategy generation unit. The initial strategy generation unit is used to decompose the target composition range of molten steel, composition hit rate constraints, and alloy cost threshold per ton of steel as control requirements into target values ​​for the feeding amount of each alloying element, and set the target weight coefficient of each element according to the performance requirements of the steel grade. Based on historical production data, thermal parameters, stirring parameters, alloy parameters, and slag parameters are selected as key process parameters. A target correlation model with the target value of the feeding amount is established. The parameter threshold range of each process is set according to the alloying process combined with the target value of the feeding amount. The initial feeding control strategy is formulated according to the linkage rules between different parameters.

[0026] The dynamic optimization and precise feeding control system for steelmaking alloying cost provided by the present invention further includes a parameter sensitivity analysis unit and a strategy adjustment verification unit in the feeding strategy optimization module.

[0027] The parameter sensitivity analysis unit is used to sort out the impact parameters corresponding to the cost coordination risk data of different alloying anomaly scenarios, determine the risk mitigation logic of impact parameter adjustment, obtain the risk parameter association rule library, and calculate the sensitivity coefficient of each impact parameter to alloy yield for the risk type of each alloying anomaly scenario.

[0028] The strategy adjustment and verification unit is used to mark the parameter categories, ranges, and priorities that need to be adjusted in the initial feeding control strategy based on the risk parameter association rule base to obtain the parameter set to be adjusted. The adjustment direction is determined according to the mitigation logic, and the adjustment range is calculated by combining different risk levels and sensitivity coefficients. For multi-parameter linkage adjustment scenarios, the order of parameter adjustment is designed to obtain the parameter adjustment scheme. The parameter adjustment scheme is substituted into the target association model for verification to determine whether the component constraints and cost constraints are satisfied at the same time. If so, an optimized and precise feeding control strategy is generated; otherwise, the adjustment range and adjustment direction are corrected.

[0029] According to the dynamic optimization and precise feeding control system for steelmaking alloying cost provided by the present invention, the feedback control module includes an equilibrium calculation unit. The equilibrium calculation unit is used to collect the steel coupling state data of each zone after feeding control, and set a differentiated collection frequency according to the state change characteristics, perform normalization processing to obtain collaborative processing data, calculate the variation coefficient of the same alloy element composition in each zone of the whole ladle and the composition fluctuation amplitude of a single zone in the continuous collection period, calculate the single parameter composition equilibrium index, and combine the influence weight of different alloy elements on the steel performance to obtain a weighted comprehensive composition equilibrium index.

[0030] The dynamic optimization and precise feeding control system for steelmaking alloying costs provided by the present invention further includes a node risk assessment unit and a risk adjustment generation unit in the feedback control module.

[0031] The node risk assessment unit is used to classify and sort all control nodes according to the optimized and precise feeding control strategy, obtain node characteristics, calculate the weight of each control node's deviation on the comprehensive composition balance index and the cost per ton of steel alloy using historical control data, evaluate the nodes based on the preset comprehensive composition balance index threshold and the preset node control deviation threshold, calculate the node risk value based on the node evaluation results, classify the node risk level according to the preset range, and select nodes that meet the preset requirements as risk nodes.

[0032] The risk adjustment generation unit is used to analyze the root causes of control failure for each risk node, and generate risk adjustment strategies based on the problem type of the root cause and the risk level of the node.

[0033] This invention provides a dynamic optimization and precise feeding control system for steelmaking alloying costs. By constructing a risk identification and quantitative assessment system, it can identify potential risks at different production stages and under different equipment conditions in advance, enabling early prediction and proactive intervention of abnormal problems, and avoiding quality losses and cost increases caused by post-event remediation. Combined with a dynamic optimization control strategy based on the real-time operating status of the equipment, it can adjust various control parameters in real time according to actual operating conditions, ensuring that the control effect always matches production needs.

[0034] By constructing a feedback control mechanism, the control effect is continuously evaluated and dynamically corrected in real time, thereby continuously improving control accuracy and process stability. This solves the core problems of traditional control methods, such as being crude, having a slow response, and being poor adaptable to operating conditions. While ensuring that product quality is stable and meets standards, it also reduces the consumption of alloy raw materials and comprehensively improves the overall production efficiency and intelligent management level of the smelting process. Attached Figure Description

[0035] The invention will now be further described with reference to the accompanying drawings.

[0036] Figure 1This is a flowchart illustrating the dynamic optimization and precise feeding control system for steelmaking alloying costs in this invention.

[0037] Figure 2 This is a schematic diagram of the dynamic optimization and precise feeding control system for steelmaking alloying costs in this invention;

[0038] Figure 3 This is a schematic diagram of the structure of the multi-source data acquisition unit in this invention. Detailed Implementation

[0039] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0040] like Figures 1 to 3 As shown, the dynamic optimization and precise feeding control system for steelmaking alloying cost provided in this embodiment of the invention includes a coupled data acquisition module, a cost collaborative risk identification module, a feeding strategy optimization module, and a feedback control module.

[0041] The coupled data acquisition module is used to collect all-dimensional operating parameters of the steelmaking alloying system, divide the ladle molten pool into zones, collect the physical field parameters of each zone in real time, and correlate the physical field parameters to construct zone coupled state data.

[0042] The coupled data acquisition module includes a melt pool partitioning deployment unit, a multi-source data acquisition unit, and a coupled data construction unit.

[0043] The molten pool partition deployment unit is used to collect ladle structure dimensions, bottom blowing argon port distribution locations, electrode insertion depth parameters, and molten steel loading volume as full-dimensional operating parameters.

[0044] The structural dimensions of the ladle include the inner diameter, height, furnace lining thickness, and bottom thickness, which are obtained through the ladle's factory technical documents and periodic laser scanning measurements.

[0045] The distribution of bottom-blown argon ports includes the number of ports, the coordinates of their centers, and the spacing between them. Common bottom-blown argon ports are 2, 4, or 6 in number and are centrally symmetrically distributed. The coordinates of their centers are determined by establishing a polar coordinate system with the center of the bottom of the ladle as the origin.

[0046] The electrode insertion depth is the distance from the electrode tip to the surface of the molten steel, which is measured by the displacement sensor of the electrode lifting mechanism, and the value ranges from 100mm to 300mm.

[0047] The amount of molten steel charged into the furnace is the actual weight of molten steel loaded into the ladle, measured by an electronic scale in the ladle, with an allowable fluctuation range of ±5% of the design capacity.

[0048] The molten pool zoning deployment unit is based on the steel grade and target alloying cycle to ensure that the distribution of bottom-blown argon ports meets the requirements for uniform steel mixing. The steel grade includes the carbon content grade and the total content of alloying elements in the molten steel. The carbon content grade is divided into low-carbon steel, medium-carbon steel and high-carbon steel, and the total content of alloying elements is divided into low-alloy steel, medium-alloy steel and high-alloy steel.

[0049] The target alloying cycle is the time required from the start of feeding to the completion of the composition, which is usually 3 to 8 minutes, depending on the steel grade, specifications and production schedule.

[0050] The number of zones in the molten pool is determined by uniformly dividing the molten pool of the ladle based on the distribution of bottom-blown argon ports. For a 120-ton ladle, the number of zones is determined to be 4 when there are 2 bottom-blown argon ports, 8 when there are 4 bottom-blown argon ports, and 12 when there are 6 bottom-blown argon ports. Each zone corresponds to the radiation range of one argon column.

[0051] The molten pool partition deployment unit is based on the fixed distance between the molten steel flow trajectory and the center of the argon column, forming a ring partition centered on the argon column to determine the partition boundary. The molten steel flow trajectory is obtained through computational fluid dynamics simulation. The simulation adopts the standard k-ε turbulence model, and the boundary conditions are set as follows: bottom blowing argon flow rate 100 L / min and molten steel temperature 1600℃.

[0052] The center of the argon column is the projection point of the bottom-blown argon port onto the surface of the molten steel. The fixed distance is the effective radius of influence of argon stirring, which is determined through water model experiments and industrial field verification, and the value ranges from 800mm to 1200mm.

[0053] For the radiation overlap area of ​​adjacent argon columns, the boundary is divided according to the principle of equivalent heat flux intensity. The heat flux intensity is obtained through numerical simulation calculation. When the heat flux intensity of the overlap area is more than 10% higher than that of the surrounding area, the boundary is shifted 100mm to 200mm to the side with lower heat flux intensity.

[0054] For the edge area near the ladle wall, the partition range should be appropriately reduced by 10% to 15% to compensate for the temperature and flow field deviations caused by heat dissipation from the furnace wall.

[0055] After the melt pool partitioning deployment unit completes the partitioning, it generates partition numbers and partition boundary coordinates, and sends them to the multi-source data acquisition unit.

[0056] The multi-source data acquisition unit is used to combine the number of partitions and the partition boundaries, deploy corresponding detection sensors in each partition and perform partition calibration, and collect data on the temperature field distribution of molten steel, composition concentration gradient, argon gas stirring flow field, alloy melting rate and slag basicity of each partition in real time as physical field parameters.

[0057] The multi-source data acquisition unit includes a temperature field acquisition subunit, a composition field acquisition subunit, a flow field acquisition subunit, a melting rate calculation subunit, and a slag alkalinity detection subunit.

[0058] The temperature field acquisition subunit is used to acquire the temperature of the upper, middle and lower layers of molten steel and the edge points of the ladle wall in each zone, calculate the maximum temperature difference and temperature gradient, and form the temperature field distribution of molten steel.

[0059] Temperature acquisition was performed using a tungsten-rhenium thermocouple, with a calibration number of WRe5 / 26, and a measurement range of 0℃ to 1800℃.

[0060] Each zone has one thermocouple deployed in the upper, middle, and lower layers. The upper layer thermocouple is located 100 mm below the molten steel surface, the middle layer thermocouple is located at 1 / 2 of the molten steel height, and the lower layer thermocouple is located 1000 mm below the molten steel surface. The thermocouples at the edge of the ladle wall are 50 mm away from the inner wall of the furnace lining.

[0061] Temperature is sampled once per second, and the collected data is transmitted to the data processing unit via industrial Ethernet.

[0062] The maximum temperature difference is the difference between the highest and lowest temperatures at all temperature collection points within the same zone at the same time.

[0063] The temperature gradient is the temperature change per meter along the height of the molten steel, and the calculation formula is:

[0064]

[0065] In the formula, The upper temperature, The temperature of the lower layer, This represents the vertical distance between the upper and lower thermocouples.

[0066] The uniformity of the temperature field is characterized by the average maximum temperature difference over 10 consecutive sampling cycles. When the average value is less than 5℃, the temperature field is considered uniform.

[0067] The temperature field acquisition subunit integrates the calculated maximum temperature difference, temperature gradient, and uniformity characteristics into steel temperature field distribution data and sends it to the coupled data construction unit.

[0068] The composition field acquisition subunit is used to collect the mass fraction of target alloying elements C, Si, Mn, and Cr in each zone, and simultaneously record the viscosity and flow rate of molten steel to form composition concentration data. The composition concentration gradient is obtained by correcting the current molten steel temperature.

[0069] The component was collected using a laser-induced breakdown spectroscopy analyzer, which was mounted on a movable robotic arm above the ladle. The detection angle was 45 degrees, the detection spot diameter was 5 mm, and the acquisition frequency was once every 2 seconds.

[0070] The viscosity of molten steel was measured using a rotational viscometer, with a measurement range of 0.001 Pa·s to 0.1 Pa·s.

[0071] The molten steel flow rate is measured using an electromagnetic velocity meter, with a measurement range of 0 m / s to 2 m / s.

[0072] Temperature correction uses a linear correction formula, and the correction coefficient is determined through offline calibration experiments. For element C, the mass fraction measurement is corrected by -0.005% for every 10°C increase in temperature. The correction formula is as follows:

[0073]

[0074] In the formula, To measure the C content, This represents the current temperature of the molten steel.

[0075] For Si, the mass fraction measurement is corrected by -0.003% for every 10°C increase in temperature. The correction formula is as follows:

[0076]

[0077] For Mn, the mass fraction measurement is corrected by -0.002% for every 10°C increase in temperature. The correction formula is as follows:

[0078]

[0079] For Cr, the mass fraction measurement is corrected by -0.001% for every 10°C increase in temperature. The correction formula is as follows:

[0080]

[0081] The composition concentration gradient is the change in composition per meter along the radial direction of the molten steel, and the calculation formula is:

[0082]

[0083] In the formula, As the partition center component, For partition edge components, The radius of the partition.

[0084] The component field acquisition subunit integrates the corrected component data and the calculated component concentration gradient into component concentration gradient data, which is then sent to the coupled data construction unit.

[0085] The flow field acquisition subunit is used to collect the molten steel flow velocity value and the flow velocity difference between the argon column and the edge of the molten pool in each zone according to a preset frequency as flow field data, calculate the flow velocity gradient in the zone, analyze the fluctuation amplitude and frequency of the flow field and the time period of abrupt flow velocity change, and form an argon gas stirring flow field.

[0086] The flow rate is sampled every 500 milliseconds, and the collected data is transmitted to the data processing unit via industrial Ethernet.

[0087] The velocity gradient is the change in velocity per meter along the radial direction of the molten steel, and the calculation formula is:

[0088]

[0089] In the formula, The velocity at the center of the argon column is... The flow velocity at the edge of the molten pool. The radius of the partition.

[0090] The fluctuation range is the difference between the maximum and minimum flow rates within 10 consecutive sampling cycles.

[0091] The fluctuation frequency is the number of flow velocity fluctuations per unit time, with a statistical period of 1 minute, and is calculated using a fast Fourier transform.

[0092] The criterion for determining a sudden change in flow velocity is that the change in flow velocity exceeds 0.5 m / s within 1 second. The start and end times of the sudden change in flow velocity are recorded as the period of the sudden change in flow velocity.

[0093] The flow field acquisition subunit integrates the calculated velocity gradient, fluctuation amplitude, fluctuation frequency, and velocity change periods into argon gas stirring flow field data, and sends it to the coupled data construction unit.

[0094] The melting rate calculation subunit is used to detect the rate of temperature change and concentration change around the alloy particles in each zone, and calculates the alloy melting rate using the Arrhenius form of melting kinetics formula.

[0095] The rate of temperature change is the change in temperature of molten steel within a 100mm radius around the alloy particles in one second. The calculation formula is:

[0096]

[0097] In the formula, for Temperature at any moment for Temperature at any moment This is the time interval, with a value of 1 second.

[0098] The rate of concentration change is the change in the mass fraction of the corresponding alloying element within 1 second, and is calculated using the following formula:

[0099]

[0100] In the formula, for The components of time, for The components of time, This is the time interval, with a value of 1 second.

[0101] The formula for calculating the melting rate of an alloy is as follows:

[0102]

[0103] In the formula, It is the alloy melting rate in g / s; It is the alloy melting frequency factor, which is determined by offline calibration experiments based on the alloy type and particle size. It is the activation energy for alloy melting; It is the universal gas constant, with a value of 8.314 J / (mol·K); It is the thermodynamic temperature of molten steel; It is the difference between the saturated concentration and the actual concentration of alloying elements in molten steel. It is the alloy density. It is the melting interface area of ​​a single alloy particle.

[0104] Alloy melting frequency factor and melting activation energy Offline calibration experiments determined that for ferrosilicon alloys, The range of values ​​is m / s to m / s, The value ranges from 40 kJ / mol to 60 kJ / mol; for ferromanganese alloys, The range of values ​​is m / s to m / s, The value ranges from 35 kJ / mol to 55 kJ / mol; for ferrochrome alloys, The range of values ​​is m / s to m / s, The value ranges from 45 kJ / mol to 65 kJ / mol.

[0105] The calibration experiment used a high-temperature tube furnace, measuring the melting time of alloy particles of different sizes within a temperature range of 1500℃ to 1650℃, and obtaining the results through nonlinear fitting. and The value of .

[0106] The melting rate calculation subunit sends the calculated alloy melting rate data to the coupled data construction unit.

[0107] The slag alkalinity detection subunit is used to collect slag samples from the slag layers in each zone, detect the mass fraction of calcium oxide and silicon dioxide, calculate the binary alkalinity of the slag according to the slag alkalinity formula, and record the alkalinity differences in different layers to obtain slag alkalinity data.

[0108] Slag samples were collected using an immersion sampler. The upper, middle and lower layers of slag were collected from each zone. The upper layer was located 50 mm below the molten slag surface, the middle layer was located at 1 / 2 of the slag height, and the lower layer was located 50 mm above the slag-molten steel interface.

[0109] The mass fractions of calcium oxide and silicon dioxide were determined by X-ray fluorescence spectroscopy.

[0110] The formula for calculating the binary basicity of slag is as follows:

[0111]

[0112] In the formula, It is the binary basicity of slag; It is the mass fraction of calcium oxide in the slag; It is the mass fraction of silica in the slag.

[0113] The slag basicity data includes the basicity value of each layer and the basicity difference between layers. The basicity difference between layers is the absolute value of the basicity of the upper layer and the basicity of the lower layer.

[0114] The slag alkalinity detection subunit sends the calculated slag alkalinity data to the coupled data construction unit.

[0115] The coupling data construction unit is used to align the physical field parameters of the same partition in time and space based on the system timestamp, mark the correlation characteristics between different parameters, and generate partition coupling state data.

[0116] Spatiotemporal alignment uses linear interpolation to interpolate data collected by different sensors to a unified time point with a time interval of 1 second.

[0117] For high-frequency data such as temperature and flow rate, the values ​​at the corresponding time points are directly taken; for low-frequency data such as composition and slag alkalinity, linear interpolation is used to calculate the values ​​at the corresponding time points.

[0118] The correlation characteristics include causal and correlational relationships between parameters. Causal relationships, such as increased temperature leading to faster component diffusion rate and increased argon flow rate leading to increased flow velocity, are determined through process mechanism analysis. Correlational relationships are calculated using the Pearson correlation coefficient.

[0119] When the absolute value of the correlation coefficient is greater than 0.7, it is marked as a strong correlation feature; when the absolute value of the correlation coefficient is between 0.3 and 0.7, it is marked as a moderate correlation feature; when the absolute value of the correlation coefficient is less than 0.3, it is marked as a weak correlation feature.

[0120] The partitioned coupling state data is stored in a three-dimensional array, with the array dimensions being partition number, timestamp, and parameter type. Each element stores the measured value of the corresponding parameter and the associated feature identifier.

[0121] The coupled data construction unit sends the generated partitioned coupled status data to the cost collaboration risk identification module.

[0122] The cost collaboration risk identification module is used to identify alloying anomaly scenarios based on partitioned coupled state data. It combines the current stage of the steelmaking alloying process and the aging state of the ladle to construct cost collaboration risk data under different alloying anomaly scenarios.

[0123] The cost collaboration risk identification module includes an abnormal scenario matching unit, a ladle aging calculation unit, and a risk data calculation unit.

[0124] The abnormal scene matching unit is used to standardize the partitioned coupling state data, align the spatiotemporal dimensions according to the partition number and timestamp to obtain the coupling state processing data, and extract the parameter features and coupling features related to alloying anomalies as alloying anomaly features.

[0125] The standardization process uses the Z-score standardization method to convert the parameter values ​​into standard values ​​with a mean of 0 and a standard deviation of 1.

[0126] Missing data is filled using linear interpolation, and the filling method is the same as the interpolation method used in spatiotemporal alignment.

[0127] Spatiotemporal alignment is first sorted by partition number from smallest to largest, and then sorted by timestamp in ascending order to ensure that the data of all partitions under the same timestamp correspond one-to-one.

[0128] The parameter characteristics include the absolute value of the parameter, the degree of deviation from the target value, and the rate of change. The formula for calculating the degree of deviation from the target value is:

[0129]

[0130] In the formula, These are measured values. This is the target value.

[0131] The formula for calculating the rate of change is:

[0132]

[0133] In the formula, This is the value at the current moment. This is the value from the previous moment. For time intervals.

[0134] Coupling characteristics include the correlation coefficient and the consistency of the changing trends between different parameters. The consistency of the changing trends is determined by comparing the signs of the changing rates of the two parameters. If the signs are the same, they are changing in the same direction; if the signs are opposite, they are changing in opposite directions.

[0135] The abnormal scenario matching unit combines the composition requirements and cost control requirements of the target steel grade and sets three threshold levels for the extracted alloying abnormal characteristics. The three threshold levels are based on the statistical data of 1,000 heats of normal production. The warning threshold is the 95th percentile of the normal data, the mild abnormal threshold is the 99th percentile of the normal data, and the severe abnormal threshold is the 99.9th percentile of the normal data.

[0136] The abnormal scene matching unit establishes a matching rule base for abnormal scenes based on the three-level threshold and alloying abnormal features. The matching rule base adopts a decision tree structure, with each node corresponding to the threshold judgment of a feature and the leaf node corresponding to an abnormal scene type.

[0137] Common abnormal scenarios include abnormal compositional segregation, abnormally low temperature, abnormal insufficient argon stirring, abnormal slag basicity, and abnormal slow alloy melting.

[0138] The abnormal scene matching unit traverses the coupling state processing data of each partition, combines the matching rule library to determine the corresponding alloying abnormal scene, and traverses once every 10 seconds, recording the partition where the abnormal scene occurs, the time of occurrence, and the deviation of features.

[0139] The abnormal scenario matching unit sends the identified alloying abnormal scenario data to the risk data calculation unit.

[0140] The ladle aging calculation unit is used to calculate the ladle aging coefficient after statistical analysis of the cumulative number of heats used, the current thermal conductivity of the lining, and the erosion thickness of the lining, and after normalization.

[0141] The cumulative number of heats used by the ladle is counted from the start of its use. The count is incremented by 1 after each steel tapping. The count is reset after the ladle is repaired. The data is stored in the ladle management system.

[0142] The thermal conductivity of the lining is currently measured using the heat flow meter method. The measurement is performed after the ladle baking is completed and the furnace lining temperature reaches 1000℃. The measurement points are distributed on the side wall and bottom of the ladle, and three points are measured in each area and the average value is taken.

[0143] The erosion thickness of the lining was measured using a laser rangefinder after slag removal from the ladle and before furnace loading. The measurement points corresponded one-to-one with the thermal conductivity measurement points. The erosion thickness was calculated as the initial lining thickness minus the current lining thickness.

[0144] The normalization process uses the minimum-maximum normalization method, mapping each parameter to the interval between 0 and 1. The calculation formula is as follows:

[0145]

[0146] In the formula, These are the original parameter values. The minimum value of the parameter. This represents the maximum value of the parameter.

[0147] For the cumulative number of heats used in steel ladle =0, Design life of steel ladle heats Regarding the thermal conductivity of the lining, Thermal conductivity of the new steel ladle lining , For the thermal conductivity of the lining of the scrapped steel ladle; for the lining corrosion thickness, =0, Initial thickness of the ladle lining .

[0148] The formula for calculating the aging coefficient of steel ladles is as follows:

[0149]

[0150] In the formula, Indicates the aging coefficient of the ladle; This indicates the cumulative number of heats the ladle has been used. Indicates the current thermal conductivity of the lining; Indicates the erosion thickness of the lining; , and The weighting coefficients are obtained by regression based on historical furnace heat balance data, and ; It refers to the design life of the ladle in heats; It is the thermal conductivity of the new steel ladle lining; It is the initial thickness of the ladle lining.

[0151] Weighting coefficient , , The ridge regression method was used to determine the model's goodness of fit, which was based on the full lifecycle data of 500 historical steel ladles. The goodness of fit of the regression model was required to be greater than or equal to 0.7.

[0152] The ladle aging calculation unit will calculate the ladle aging coefficient. Send to the risk data calculation unit.

[0153] The risk data calculation unit is used to couple the deviation characteristics of alloying anomalies, process sensitivity factors and ladle aging coefficients to calculate the cost synergy risk index. Based on the preset value range, the cost synergy risk index is divided into multiple risk levels, and the risk characteristics and control requirements corresponding to each risk level are determined to generate cost synergy risk data.

[0154] The process of calculating the cost synergy risk index includes classifying risk factors for alloying anomaly scenarios based on the degree of cost impact of the anomaly type, the scope of component impact, and the duration of the anomaly. .

[0155] The degree of cost impact is divided according to the increase in alloy cost per ton of steel: 0 to 5 yuan is low impact, corresponding to a value of 0.2; 5 to 15 yuan is medium impact, corresponding to a value of 0.5; and above 15 yuan is high impact, corresponding to a value of 0.8.

[0156] The influence range of the component is divided according to the number of abnormal partitions: 1 partition is a local influence with a value of 0.3; 2 to 3 partitions are a regional influence with a value of 0.6; and 4 or more partitions are a global influence with a value of 0.9.

[0157] The duration is divided according to the abnormal duration: less than 5 minutes is short duration, corresponding to a value of 0.2; 5 to 15 minutes is medium duration, corresponding to a value of 0.5; and more than 15 minutes is long duration, corresponding to a value of 0.8.

[0158] Abnormal scenario risk factors The weighted average of the values ​​from the three dimensions is calculated using the following formula: (The weights are 0.4, 0.3, and 0.3 respectively.)

[0159]

[0160] In the formula, To determine the degree of cost impact, Values ​​are assigned to the range of influence of the components. The value is set for the duration.

[0161] The risk data calculation unit calculates the process sensitivity factor by combining the basic weights and parameter relative deviation rates of the current process stage. Based on the aging coefficient of steel ladle Calculate the aging amplification factor.

[0162] The steelmaking alloying process is divided into three stages: initial stage, intermediate stage, and final stage. The initial stage is the first 30% of the time from the start of feeding to the completion of feeding; the intermediate stage is the 30% to 70% time from the completion of feeding; and the final stage is the 70% to the end of feeding. The basic weights of the corresponding process stages are 0.3, 0.6, and 0.1, respectively.

[0163] The relative deviation rate of a parameter is the absolute value of the difference between the measured value and the target value of the parameter, divided by the target value. The calculation formula is as follows:

[0164]

[0165] Process sensitive factors The formula is: The basic weight of each process stage multiplied by the relative deviation rate of the parameters.

[0166]

[0167] In the formula, This represents the basic weight of each process stage.

[0168] The aging amplification factor is calculated using a piecewise function, when the ladle aging coefficient... When it is less than 0.2, the aging amplification factor is 1; when When the aging amplification factor is greater than or equal to 0.2 and less than 0.6, it is 1 plus 2.5 multiplied by 1. Subtract 0.2 from the difference; when When the value is greater than or equal to 0.6, the aging amplification factor is 2.0, and the calculation formula is as follows:

[0169]

[0170] In the formula, This is an aging amplification factor. This represents the aging coefficient of the steel ladle.

[0171] The risk data calculation unit calculates the cost synergy risk index using a weighted summation formula. Weighting coefficient , , Based on historical risk incident data, the values ​​are determined to be 0.5, 0.3, and 0.2 respectively, calculated using the following formula:

[0172]

[0173] In the formula, This is a cost synergy risk index; , , These are the weighting coefficients, and ; Risk factors for abnormal scenarios; Process-sensitive factors; This represents the aging coefficient of the steel ladle.

[0174] The cost synergy risk index is divided into five levels: 0 to 1 is Level I risk, 1 to 2.5 is Level II risk, 2.5 to 4 is Level III risk, 4 to 6 is Level IV risk, and 6 and above is Level V risk.

[0175] Level I risk is characterized by localized, minor anomalies that have no significant impact on product quality and cost. The control requirement is continuous monitoring of parameter changes.

[0176] Level II risk is characterized by single-zone anomaly with a parameter deviation rate of less than 100%. The control requirement is to fine-tune the mixing and feeding parameters of the corresponding zone.

[0177] Level III risk is characterized by multiple zonal anomalies and a deviation rate of more than 100% for key parameters. The control requirements are to initiate zonal coordinated regulation and adjust the bottom blowing argon flow rate and alloy feeding speed.

[0178] Level IV risk is characterized by systemic anomalies and severe parameter deviations. The control requirements are to initiate full furnace parameter correction, reduce the feeding rate, and suspend feeding if necessary.

[0179] Level V risk is characterized by a major systemic anomaly, which may lead to serious exceedances of component limits or equipment failure. The control requirement is to immediately stop feeding and investigate equipment and process problems.

[0180] The risk data calculation unit sends the generated cost-coordination risk data to the feeding strategy optimization module.

[0181] The feeding strategy optimization module is used to formulate an initial feeding control strategy based on the preset steel composition compliance constraints and the goal of minimizing the alloy cost per ton of steel. According to the cost coordination risk data, the set of parameters to be adjusted is marked in the initial feeding control strategy. The adjustment range and priority are determined based on the sensitivity coefficient of the alloy yield to each adjustment parameter, and the optimized and precise feeding control strategy is obtained through adjustment.

[0182] The feeding strategy optimization module includes an initial strategy generation unit, a parameter sensitivity analysis unit, and a strategy adjustment and verification unit.

[0183] The initial strategy generation unit is used to decompose the target composition range of molten steel, composition hit rate constraints, and alloy cost threshold per ton of steel as control requirements into target values ​​for the amount of each alloy element to be added, and to set the target weight coefficient of each element according to the performance requirements of the steel grade.

[0184] The target composition range of molten steel is determined according to industry standards and customer requirements. Taking Q235 steel as an example, the target range of C content is 0.14% to 0.22%, Si content is 0.12% to 0.30%, and Mn content is 0.30% to 0.60%.

[0185] The composition hit rate constraint requires that the composition compliance rate of the target element be no less than 98%, and the alloy cost threshold per ton of steel is determined based on the market alloy price and production cost.

[0186] The target feed rate is calculated by multiplying the weight of the molten steel by the difference between the target composition and the initial composition, and then dividing by the historical average yield. The formula is as follows:

[0187]

[0188] In the formula, The target value for the amount of material to be added. This is the weight of the molten steel. For the target component, As the initial components, This represents the historical average yield.

[0189] The target weighting coefficients are set according to the degree of influence of elements on the performance of steel. For high-strength steel, the weighting coefficients for C and Mn are 0.3, Si is 0.2, and other elements are 0.2. For stainless steel, the weighting coefficients for Cr and Ni are 0.35, C is 0.15, and other elements are 0.15.

[0190] The initial strategy generation unit selects thermal parameters, stirring parameters, alloy parameters, and slag parameters as key process parameters based on historical production data, and establishes a target correlation model with the target value of the feed amount.

[0191] Pearson correlation analysis was used to screen key process parameters. Parameters with an absolute value of correlation coefficient greater than 0.5 with the target value of the feed amount were selected. Thermal parameters included the initial temperature and final temperature of molten steel. Stirring parameters included the bottom-blown argon flow rate and stirring time. Alloy parameters included alloy particle size and alloy addition order. Slag parameters included slag basicity and slag oxidizing properties.

[0192] The target association model adopts a BP neural network model. The number of nodes in the input layer is the number of key process parameters. There are two hidden layers with 10 nodes in each layer. The number of nodes in the output layer is the actual yield of each alloying element.

[0193] The model training uses production data from more than 2,000 historical batches, with a training set to test set ratio of 8:2. Gradient descent is used for training, and the model's yield prediction error is required to be no more than 5%.

[0194] The initial strategy generation unit sets the parameter threshold range for each process according to the alloying process and the target value of the feeding amount, and formulates the initial feeding control strategy based on the linkage rules between different parameters.

[0195] The threshold range of bottom-blown argon flow rate in the initial stage of alloying is 50 L / min to 100 L / min, in the middle stage it is 100 L / min to 150 L / min, and in the final stage it is 50 L / min to 80 L / min.

[0196] The alloy feeding rate is initially 50 kg / min to 100 kg / min, in the middle stage 100 kg / min to 150 kg / min, and in the final stage 30 kg / min to 50 kg / min.

[0197] The linkage rules include: when the bottom-blown argon flow rate increases by 20 L / min, the alloy feeding rate increases by 10 kg / min; when the slag basicity is below 2.0, the amount of lime added increases by 50 kg; when the molten steel temperature is below 1550℃, the feeding rate decreases by 20%; and when the alloy yield is below 10% of the historical average, the amount of the corresponding alloy added increases by 5%.

[0198] The initial feeding control strategy includes the feeding time, feeding amount, feeding speed and corresponding stirring parameters of each alloy element, which are stored in tabular form. Each parameter corresponds to a time interval and a set value.

[0199] The initial strategy generation unit sends the generated initial feeding control strategy to the strategy adjustment and verification unit.

[0200] The parameter sensitivity analysis unit is used to sort out the impact parameters corresponding to the cost coordination risk data of different alloying anomaly scenarios, determine the risk mitigation logic of impact parameter adjustment, obtain the risk parameter association rule library, and calculate the sensitivity coefficient of each impact parameter to alloy yield for the risk type of each alloying anomaly scenario.

[0201] The risk parameter association rule base is built based on more than 100 historical anomaly handling data. It records the mitigation effect of adjusting each parameter on risk under different anomaly scenarios. The mitigation logic is described as the correspondence between the direction of parameter adjustment and the degree of risk reduction. For example, for composition segregation anomaly, increasing the bottom blowing argon flow rate can improve the stirring intensity, reduce the composition variation coefficient, and mitigate the segregation risk; for low temperature anomaly, increasing the molten steel temperature can accelerate the alloy melting rate, improve the yield, and mitigate the risk of increased cost.

[0202] The sensitivity coefficient is calculated using the controlled variable method. With other influencing parameters fixed, the value of a single parameter is changed, and the rate of change in alloy yield is measured. The sensitivity coefficient is the ratio of the rate of change in yield to the rate of change in the parameter. The calculation formula is as follows:

[0203]

[0204] In the formula, It is the alloy yield sensitivity coefficient; It is the first The actual yield of the alloy; It is the first Measured values ​​of several influencing parameters; It is the partial derivative of the yield with respect to that parameter.

[0205] A sensitivity coefficient greater than 0.5 indicates a highly sensitive parameter, 0.2 to 0.5 indicates a moderately sensitive parameter, and less than 0.2 indicates a lowly sensitive parameter.

[0206] The parameter sensitivity analysis unit sends the calculated sensitivity coefficients and the constructed risk parameter association rule base to the strategy adjustment verification unit.

[0207] The strategy adjustment verification unit is used to mark the parameter categories, ranges and priorities that need to be adjusted in the initial feeding control strategy based on the risk parameter association rule base to obtain the parameter set to be adjusted, determine the adjustment direction according to the mitigation logic, calculate the adjustment range in combination with different risk levels and sensitivity coefficients, and design the order of parameter adjustment for multi-parameter linkage adjustment scenarios to obtain the parameter adjustment scheme.

[0208] The annotation of the parameter set to be adjusted follows the principles of risk level priority and sensitivity priority. Parameters corresponding to high risk levels are annotated first, and highly sensitive parameters under the same risk level are annotated first.

[0209] The adjustment direction is determined based on the mitigation logic. For example, if the melting rate decreases due to low temperature, the adjustment direction is to increase the temperature of the molten steel; if the cost increases due to high composition, the adjustment direction is to reduce the amount of the corresponding alloy added; if composition segregation is caused by insufficient stirring, the adjustment direction is to increase the bottom blowing argon flow rate.

[0210] The adjustment range is calculated by multiplying the base adjustment range by the risk level coefficient and the sensitivity coefficient. The base adjustment range is 5% of the parameter threshold range. The risk level coefficient is 1 for level 1, 1.5 for level 2, 2 for level 3, 2.5 for level 4, and 3 for level 5. The calculation formula is as follows:

[0211]

[0212] In the formula, To adjust the range, Based on the adjustment range, This is the risk level coefficient. This is the sensitivity coefficient.

[0213] The order of multi-parameter linkage adjustment is as follows: first adjust the stirring parameters, then adjust the feeding parameters, and finally adjust the slag parameters to ensure the coordination of parameter adjustments and avoid mutual interference between parameters.

[0214] The strategy adjustment and verification unit substitutes the parameter adjustment scheme into the target association model for verification, and determines whether the component constraints and cost constraints are satisfied at the same time. If so, it generates an optimized and precise feeding control strategy; otherwise, it corrects the adjustment range and direction.

[0215] The verification process first inputs the adjusted parameters into the target association model to predict the adjusted endpoint composition and determine whether it is within the target composition range; then, based on the adjusted feed amount and the current alloy price, it calculates the adjusted alloy cost per ton of steel and determines whether it is lower than the cost threshold.

[0216] If both constraints are satisfied, an optimized and precise feeding control strategy is generated; if not, the gradient descent method is used to correct the adjustment magnitude and direction, with no more than 3 iterations. If the constraints are still not satisfied after 3 iterations, the component constraints are prioritized, and the cost constraints are appropriately relaxed.

[0217] The optimized precision feeding control strategy includes the adjusted feeding time, feeding amount, feeding speed and corresponding stirring parameters for each alloy element, as well as adjustment instructions and expected results.

[0218] The strategy adjustment and verification unit sends the generated optimized and precise feeding control strategy to the feedback control module and the field actuator.

[0219] The feedback control module is used to collect the comprehensive composition balance index of the steel-molten steel coupling state data of each zone in real time after the feeding control. It performs a comprehensive evaluation of each control node in the optimized and precise feeding control strategy, defines the control nodes that fail the evaluation as risk nodes, and generates risk adjustment strategies that point to the risk nodes.

[0220] The feedback control module includes an equilibrium calculation unit, a node risk assessment unit, and a risk adjustment generation unit.

[0221] The balance calculation unit is used to collect the steel coupling state data of each zone after the feeding control, and set the differentiated acquisition frequency according to the state change characteristics. After normalization processing, the collaborative processing data is obtained. The coefficient of variation of the same alloy element composition in each zone of the whole ladle and the composition fluctuation amplitude of a single zone in the continuous acquisition period are calculated. The single parameter composition balance index is calculated. Combined with the influence weight of different alloy elements on the steel performance, the comprehensive composition balance index is obtained by weighted calculation.

[0222] The differentiated sampling frequency is set according to the rate of change of the parameters: temperature and composition parameters are sampled once per second, flow rate parameters are sampled once every 500 milliseconds, and slag alkalinity parameters are sampled once every 5 minutes.

[0223] The normalization process uses the minimum-maximum normalization method to map all parameter values ​​to the interval between 0 and 1.

[0224] The coefficient of variation is the component standard deviation of the same element across all partitions divided by the mean, and is calculated using the following formula:

[0225]

[0226] In the formula, For the first The coefficient of variation of each alloying element This represents the standard deviation of the element's composition. This represents the average value of the element's composition.

[0227] The component fluctuation amplitude is the difference between the maximum and minimum component values ​​in a single zone over 10 consecutive acquisition cycles, calculated using the following formula:

[0228]

[0229] In the formula, For the first The fluctuation range of the composition of each alloying element For the maximum component value, This is the minimum component value.

[0230] The formula for calculating the single-parameter component balance index is as follows:

[0231]

[0232] In the formula, It is the first Single-parameter compositional balance index of each alloying element; , The weighting coefficients are set based on the steel grade's sensitivity to segregation, and ; It is the first The coefficient of variation of the alloy element composition; It is the amplitude of the component fluctuation of the element within the continuous collection period; It is the maximum allowable fluctuation range of the element's composition.

[0233] Weighting coefficient and Based on the sensitivity of steel grades to segregation, for steel grades that are sensitive to segregation, the spatial consistency weight is... Set to 0.6, time stability weight Take 0.4; for steel grades that are not sensitive to segregation, Take 0.4, Take 0.6.

[0234] The formula for calculating the overall component balance index is as follows:

[0235]

[0236] In the formula, It is a comprehensive component balance index; It is the first The influence weight of each alloying element on the properties of steel, and ; It is the number of types of elements involved in alloying; It is the first The single-parameter compositional balance index of each alloying element.

[0237] The comprehensive component balance index ranges from 0 to 1. The larger the value, the better the component uniformity. When the index is greater than or equal to 0.85, it is judged as having uniform components.

[0238] The balance calculation unit sends the calculated comprehensive component balance index to the node risk assessment unit.

[0239] The node risk assessment unit is used to classify and sort all control nodes according to process stage, control zone, and parameter type to optimize the precise feeding control strategy. It determines the parameter value, control range, and control object of each control node to obtain node characteristics. It calculates the deviation impact weight of each control node on the comprehensive composition balance index and the alloy cost per ton of steel using historical control data. Based on the node characteristics and deviation impact weight of each control node, it evaluates the node according to the preset comprehensive composition balance index threshold and the preset node control deviation threshold to obtain the node assessment result. Based on the node assessment result, it calculates the node risk value, divides the node risk level according to the preset range, and selects the nodes that meet the preset requirements as risk nodes.

[0240] Control nodes are divided into initial nodes, intermediate nodes and final nodes according to the process stage, into 1 to 8 zone nodes according to the control zone, and into temperature nodes, flow rate nodes, feed rate nodes and pressure nodes according to parameter type.

[0241] The node characteristics include the current set value of the parameter, the target value, the upper limit of the control, the lower limit of the control, and the actuator number, which are stored in the form of structured data.

[0242] The weighting of deviation impact was calculated using multiple linear regression. Based on historical data from 500 adjustments, the changes in the comprehensive component balance index and alloy cost per ton of steel were analyzed when the parameter at each node changed by 1%. The calculation formula is as follows:

[0243]

[0244] In the formula, For nodes The overall deviation affects the weight. This represents the weight of the node's influence on the overall component balance index. This represents the weight of the impact of this node on the cost of alloy steel per ton.

[0245] The preset threshold for the comprehensive component balance index is 0.85, and the preset threshold for node control deviation is 5%, meaning that the deviation rate between the actual control value and the target value of the node does not exceed 5%.

[0246] The node evaluation uses a 100-point scoring system, with the overall component balance index accounting for 60 points, the control deviation accounting for 20 points, and the control timeliness accounting for 20 points.

[0247] A comprehensive component balance index of 0.85 or higher earns 60 points, with 10 points deducted for each 0.1 point below it; a control deviation rate of 5% or lower earns 20 points, with 2 points deducted for each 1% above it; and a control timeliness index improvement of 20% or higher within 1 minute after the adjustment of the control point earns 20 points, with 5 points deducted for each 5% below it.

[0248] A score of 80 or above is considered a passing grade, 60 to 80 is considered a potential risk grade, and below 60 is considered a risk grade.

[0249] The formula for calculating the node risk value is:

[0250]

[0251] In the formula, It is the first The node risk value of each control node; It is the first The overall evaluation score of each control node, on a 100-point scale; This is the weight of the impact of this node on cost and ingredient compliance rate; It is the risk diffusion coefficient.

[0252] Node influence weight The risk diffusion coefficient is the weighted average of the impact of the deviation of this node on the comprehensive component balance index and the cost per ton of steel. Based on the influence range of the node, local nodes are set. Set the value to 1, the value for regional nodes to 1.5, and the value for global nodes to 2.

[0253] The preset range for node risk values ​​is as follows: less than 0.2 is low risk, 0.2 to 0.4 is mild risk, 0.4 to 0.8 is moderate risk, and above 0.8 is major risk. Nodes with moderate and major risks are selected as risk nodes.

[0254] The node risk assessment unit sends the selected risk node data to the risk adjustment generation unit.

[0255] The risk adjustment generation unit is used to analyze the root causes of control failure for each risk node, and generate risk adjustment strategies based on the problem type of the root cause and the risk level of the node.

[0256] In summary, this embodiment provides a dynamic optimization and precise feeding control system for steelmaking alloying costs. By constructing a risk identification and quantitative assessment system, it can identify potential risks at different production stages and under different equipment conditions in advance, enabling early prediction and proactive intervention of abnormal problems, and avoiding quality losses and cost increases caused by post-event remediation. Combined with a dynamic optimization control strategy based on the real-time operating status of the equipment, it can adjust various control parameters in real time according to actual operating conditions, ensuring that the control effect always matches production needs.

[0257] By constructing a feedback control mechanism, the control effect is continuously evaluated and dynamically corrected in real time, thereby continuously improving control accuracy and process stability. This solves the core problems of traditional control methods, such as being crude, having a slow response, and being poor adaptable to operating conditions. While ensuring that product quality is stable and meets standards, it also reduces the consumption of alloy raw materials and comprehensively improves the overall production efficiency and intelligent management level of the smelting process.

[0258] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0259] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dynamic optimization and precise feeding control system for steelmaking alloying costs, characterized in that, include: The coupled data acquisition module is used to collect all-dimensional operating parameters of the steelmaking alloying system, divide the ladle molten pool into zones, collect the physical field parameters of each zone in real time, and correlate the physical field parameters to construct zone coupling state data. The cost collaboration risk identification module is used to identify alloying anomaly scenarios based on the partitioned coupling state data, and to construct cost collaboration risk data under different alloying anomaly scenarios by combining the current stage of the steelmaking alloying process and the aging state of the ladle. The feeding strategy optimization module is used to formulate a corresponding initial feeding control strategy based on the preset steel composition compliance constraints and the goal of minimizing the alloy cost per ton of steel. According to the cost coordination risk data, the set of parameters to be adjusted is marked in the initial feeding control strategy. The adjustment range and priority are determined based on the sensitivity coefficient of alloy yield to each adjustment parameter, so as to obtain an optimized and precise feeding control strategy. The feedback control module is used to collect the comprehensive composition balance index of the steel-molten steel coupling state data of each zone in real time after the feeding control, to comprehensively evaluate each control node in the optimized and precise feeding control strategy, to define the control node that fails the evaluation as a risk node, and to generate a risk adjustment strategy pointing to the risk node.

2. The dynamic optimization and precise feeding control system for steelmaking alloying costs according to claim 1, characterized in that, The coupled data acquisition module includes a melt pool partitioning deployment unit, a multi-source data acquisition unit, and a coupled data construction unit; The molten pool partitioning deployment unit is used to collect the ladle structure dimensions, bottom-blown argon port distribution locations, electrode insertion depth parameters, and molten steel loading amount as the full-dimensional operating parameters. Based on the steel grade specifications and target alloying cycle, it ensures that the bottom-blown argon port distribution meets the requirements for uniform molten steel stirring. Based on the bottom-blown argon port distribution, the ladle molten pool is uniformly divided to determine the number of partitions. Based on the fixed distance between the molten steel flow trajectory and the center of the argon column, a ring partition centered on the argon column is formed to determine the partition boundaries. The multi-source data acquisition unit is used to combine the number of partitions and the partition boundaries, deploy corresponding detection sensors in each partition and perform partition calibration, and collect in real time the steel temperature field distribution, composition concentration gradient, argon gas stirring flow field, alloy melting rate and slag basicity data of each partition as the physical field parameters; The coupled data construction unit is used to align the physical field parameters of the same partition in time and space based on the system timestamp, mark the correlation features between different parameters, and generate the partition coupled state data.

3. The dynamic optimization and precise feeding control system for steelmaking alloying costs according to claim 2, characterized in that, The multi-source data acquisition unit includes a temperature field acquisition subunit, a composition field acquisition subunit, a flow field acquisition subunit, a melting rate calculation subunit, and a slag alkalinity detection subunit; The temperature field acquisition subunit is used to acquire the temperature of the upper, middle and lower layers of molten steel and the edge points of the ladle wall in each zone, calculate the maximum temperature difference and temperature gradient, and form the molten steel temperature field distribution. The composition field acquisition subunit is used to acquire the mass fraction of the target alloying element in each zone, and at the same time record the viscosity and flow rate of molten steel to form composition concentration data. The composition concentration gradient is obtained by correcting the current molten steel temperature. The flow field acquisition subunit is used to acquire the molten steel flow velocity value and the flow velocity difference between the argon column and the edge of the molten pool in each zone at a preset frequency as flow field data, calculate the flow velocity gradient in the zone, analyze the fluctuation amplitude and frequency of the flow field and the time period of sudden flow velocity changes, and form the argon gas stirring flow field. The melting rate calculation subunit is used to detect the rate of temperature change and the rate of concentration change around the alloy particles in each zone, and to calculate the alloy melting rate. The slag alkalinity detection subunit is used to collect slag samples from the slag layers in each zone, detect the mass fraction of calcium oxide and silicon dioxide, calculate the binary alkalinity of the slag according to the slag alkalinity formula, and record the alkalinity differences in different layers to obtain the slag alkalinity data.

4. The dynamic optimization and precise feeding control system for steelmaking alloying costs according to claim 1, characterized in that, The cost collaboration risk identification module includes an abnormal scenario matching unit. This unit is used to standardize the partition coupling state data, align the partition numbers and timestamps to obtain coupling state processing data, extract parameter features and coupling features related to alloying anomalies as alloying anomaly features, set three-level thresholds for the extracted alloying anomaly features based on the composition requirements and cost control requirements of the target steel grade, establish an abnormal scenario matching rule base based on the three-level thresholds and the alloying anomaly features, traverse the coupling state processing data of each partition, and determine the corresponding alloying anomaly scenario based on the matching rule base.

5. The dynamic optimization and precise feeding control system for steelmaking alloying costs according to claim 4, characterized in that, The cost-coordination risk identification module also includes a ladle aging calculation unit, which is used to calculate the ladle aging coefficient after normalization based on the cumulative number of furnaces used, the current thermal conductivity of the lining, and the erosion thickness of the lining.

6. The dynamic optimization and precise feeding control system for steelmaking alloying costs according to claim 4, characterized in that, The cost collaboration risk identification module also includes a risk data calculation unit. The risk data calculation unit is used to couple the deviation characteristics of the alloying anomaly scenario, the process sensitivity factor, and the ladle aging coefficient to calculate the cost collaboration risk index. The cost collaboration risk index is divided into multiple risk levels according to a preset value range, and the risk characteristics and control requirements corresponding to each risk level are determined to generate the cost collaboration risk data. The process of calculating the cost synergy risk index includes: For the aforementioned alloying anomaly scenarios, risk factors for the anomaly scenarios are categorized based on the degree of cost impact, the scope of component impact, and the duration of the anomaly. The process sensitivity factor is calculated by combining the basic weights and relative deviation rates of parameters at the current process stage, and the cost synergy risk index is calculated by weighted summation formula.

7. The dynamic optimization and precise feeding control system for steelmaking alloying costs according to claim 1, characterized in that, The feeding strategy optimization module includes an initial strategy generation unit. This unit takes the target composition range of molten steel, the composition hit rate constraint, and the alloy cost threshold per ton of steel as control requirements, decomposes them into target values ​​for the feeding amount of each alloying element, sets target weight coefficients for each element according to the performance requirements of the steel grade, selects thermal parameters, stirring parameters, alloy parameters, and slag parameters as key process parameters based on historical production data, establishes a target correlation model with the target value of the feeding amount, sets parameter threshold ranges for each process according to the alloying process and the target value of the feeding amount, and formulates the initial feeding control strategy based on the linkage rules between different parameters.

8. The dynamic optimization and precise feeding control system for steelmaking alloying costs according to claim 7, characterized in that, The feeding strategy optimization module also includes a parameter sensitivity analysis unit and a strategy adjustment verification unit; The parameter sensitivity analysis unit is used to sort out the impact parameters corresponding to the cost coordination risk data of different alloying anomaly scenarios, determine the risk mitigation logic of impact parameter adjustment to obtain the risk parameter association rule library, and calculate the sensitivity coefficient of each impact parameter to alloy yield for each risk type of alloying anomaly scenario. The strategy adjustment verification unit is used to mark the parameter categories, ranges, and priorities that need to be adjusted in the initial feeding control strategy based on the risk parameter association rule base to obtain the parameter set to be adjusted, determine the adjustment direction according to the mitigation logic, calculate the adjustment range by combining different risk levels and the sensitivity coefficient, design the order of parameter adjustment for multi-parameter linkage adjustment scenarios to obtain the parameter adjustment scheme, substitute the parameter adjustment scheme into the target association model for verification, and determine whether the component constraints and cost constraints are satisfied at the same time. If so, the optimized and precise feeding control strategy is generated; otherwise, the adjustment range and the adjustment direction are corrected.

9. The dynamic optimization and precise feeding control system for steelmaking alloying costs according to claim 1, characterized in that, The feedback control module includes a balance calculation unit, which is used to collect steel coupling state data of each zone after feeding control, and set differentiated collection frequency according to the state change characteristics, perform normalization processing to obtain collaborative processing data, calculate the coefficient of variation of the same alloy element composition of each zone of the whole ladle and the composition fluctuation amplitude of a single zone in the continuous collection period, calculate the single parameter composition balance index, and combine the influence weight of different alloy elements on the steel performance to obtain the comprehensive composition balance index.

10. The dynamic optimization and precise feeding control system for steelmaking alloying costs according to claim 9, characterized in that, The feedback control module also includes a node risk assessment unit and a risk adjustment generation unit; The node risk assessment unit is used to classify and sort all control nodes of the optimized and precise feeding control strategy to obtain node characteristics. It calculates the deviation impact weight of each control node on the comprehensive composition balance index and the cost per ton of steel alloy using historical control data. It evaluates the nodes based on the preset comprehensive composition balance index threshold and the preset node control deviation threshold to obtain node assessment results. It calculates the node risk value based on the node assessment results, divides the node risk level according to the preset interval, and selects the nodes that meet the preset requirements as the risk nodes. The risk adjustment generation unit is used to analyze the root cause of the control failure for each risk node, and generate the risk adjustment strategy based on the problem type of the root cause and the risk level of the node.

Images