Intelligent steelmaking process based on dynamic raw material ratio

Abstract

The invention relates to the technical field of metallurgy, and discloses an intelligent steelmaking process based on dynamic raw material proportioning, the core of the process is that a functional pellet with a core-shell-film multilayer structure is adopted, functional components sequentially acting in a molten pool are packaged in different layers of the functional pellet respectively, and the functional components are sequentially separated from the molten pool. The stable isotope tracers are in one-to-one correspondence with the functional components and can be released in sequence. A mass spectrometer sensing system is used for detecting specific time sequence signals formed by sequential release of the tracer agents on line, and the system can confirm the delivery and step-by-step reaction process of materials and conduct on-line self-correction on kinetic parameters in a metallurgical process model. According to the method, a closed-loop control system integrating precise feeding, process confirmation and model adaptive capacity is constructed, and adaptive control over the steelmaking process is achieved. The method can improve the control precision of molten steel components and temperature at the smelting end point, shorten the smelting period, and improve the overall production efficiency and the resource utilization rate.

Description

Technical Field

[0001] This invention relates to the field of metallurgical technology, specifically to an intelligent steelmaking process based on dynamic raw material ratio. Background Technology

[0002] Converter steelmaking is a core production process in the modern steel industry. Its basic task is to blow oxygen into a converter containing molten iron at high temperatures, and add lime and dolomite as auxiliary materials. At high temperatures, impurities such as carbon, silicon, manganese, and phosphorus in the molten iron are oxidized and removed as slag or gas, thereby obtaining molten steel with specific composition and temperature requirements. The precision of this process control directly determines the quality of the final product, production costs, and the efficiency of resource and energy utilization.

[0003] To improve the control level of the steelmaking process, modern metallurgical practice has widely applied process control models based on metallurgical principles. These models pre-calculate the total oxygen blowing volume and auxiliary material addition regime based on the initial information of the molten iron composition, temperature, weight, and scrap steel ratio. Simultaneously, to correct and guide the model during the blowing process, various online monitoring technologies are integrated, such as a secondary lance system for intermittent temperature measurement and sampling, and a mass spectrometer or infrared analyzer for continuous analysis of the furnace mouth flue gas composition.

[0004] However, existing steelmaking process control methods still have inherent technical limitations. Their control logic relies on static or semi-static models based on historical data and theoretical derivations. This open-loop feedforward control mode makes it difficult to effectively cope with the common fluctuations in raw material composition and complex changes in furnace conditions in steelmaking production. Although process monitoring methods have been introduced, their feedback mechanisms exhibit significant lag and indirectness. For example, secondary lance measurements are low-frequency and cannot capture the reaction process within the furnace; while indirect measurement methods such as flue gas analysis can provide continuous information, they reflect the macroscopic comprehensive result of multiple reactions superimposed, making it difficult to attribute to a specific metallurgical reaction or a specific material intervention.

[0005] When auxiliary materials are added to the furnace, existing technologies cannot directly confirm in real time whether these materials have successfully reached the intended action area (such as the slag-gold interface), nor can they know their actual melting, dissolution, or reaction rates in the high-temperature molten pool. The control system can only make lagging inferences based on macroscopic changes in process parameters after material addition. This ambiguity and delay in feedback makes it difficult to accurately quantify the effect of dynamic intervention. Therefore, when there is a deviation between the initial model and the actual operating conditions of the current furnace, this deviation will accumulate as the blowing process continues. The control system lacks an effective mechanism for online self-correction and optimization of the model within a single smelting cycle.

[0006] The inadequate acquisition of process information and the lack of model adaptive capability ultimately make it difficult to maintain a consistently high hit rate for the smelting endpoint (i.e., the target composition and temperature of the molten steel). In actual production, "over-blowing" or "under-blowing" phenomena caused by inaccurate control occur frequently, necessitating remedial measures such as supplementary blowing, adding molten iron, cooling, or increasing carbon content. This not only prolongs the smelting cycle and increases material and energy consumption but also adversely affects the stability of the production rhythm. Therefore, how to achieve real-time and accurate perception of the reaction process inside the converter and establish an online adaptive closed-loop control model based on this is a pressing technical challenge in the current steelmaking technology field. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides an intelligent steelmaking process based on dynamic raw material proportioning. This solves the problem that current steelmaking processes primarily rely on static models based on historical data for feedforward control, or on intermittent measurement methods for delayed feedback. These processes lack real-time perception of instantaneous changes within the furnace during blowing, immediate confirmation of the effectiveness of dynamic intervention measures, and the ability to adaptively correct the control model online. These technological limitations result in insufficient accuracy of process control when faced with fluctuations in raw material composition, making it difficult to stably achieve the final smelting target.

[0008] To achieve the above objectives, this invention provides the following technical solution: an intelligent steelmaking process based on dynamic raw material proportioning. This process establishes a closed-loop control system that includes real-time feedback, effect confirmation, and model self-correction to achieve adaptive control of the steelmaking process. The process includes the following steps:

[0009] First, before smelting begins, a mathematical model of the metallurgical process for the current smelting furnace is established based on the obtained information on the composition, temperature, and weight of the initial raw materials, including molten iron and scrap steel. One or more target process paths are then planned using this model. The target process path defines the target curves of the changes in key process parameters (such as molten pool temperature, carbon content, and phosphorus content) over time during the smelting process.

[0010] Secondly, after the converter blowing process is started, the actual process parameters are continuously or frequently monitored in real time using sensors. The monitored actual process parameters are continuously compared with the pre-planned target process path to identify and quantify the deviation between the two.

[0011] Furthermore, when the identified deviation exceeds a preset control threshold, the control system decides to feed a specific type and quantity of functionalized microspheres into the molten pool. These functionalized microspheres have a specially designed internal structure and composition, containing at least two functional components that sequentially undergo physical or chemical reactions in the high-temperature environment of the molten pool. Additionally, they are internally encapsulated with at least two stable isotopic tracers that correspond one-to-one with the stated function and component and are sequentially released into the molten pool.

[0012] Subsequently, a sensing system installed in the converter flue gas system is used to detect in real time the tracer timing signal formed by the successive release of the at least two stable isotopic tracers. The characteristics of this timing signal, such as the order in which different tracer signal peaks appear and the time interval between them, provide direct evidence for determining whether the fed microparticles have successfully reached the action area, and the time point and process at which the different functional components inside them begin to take effect.

[0013] Finally, and most importantly, the control system corrects the kinetic parameters in the mathematical model of the metallurgical process online based on the received tracer timing signal and the changes in actual process parameters monitored after the functionalized microspheres are fed. The corrected model more accurately reflects the actual metallurgical reaction characteristics of the current furnace and is immediately applied to subsequent deviation judgments and feeding decisions for that furnace, thus forming a closed-loop adaptive control process that continuously improves itself within a single smelting cycle.

[0014] Preferably, the functionalized microspheres have a core-shell-membrane multilayer structure. By designing the physicochemical properties (such as melting point and reactivity) of each layer material, the structure decomposes or reacts in a predetermined order when the microspheres enter the molten pool, thereby controlling the release of different functional components and tracers encapsulated therein in a predetermined time sequence.

[0015] Preferably, the core-shell-membrane multilayer structure may include:

[0016] An outer layer (film) composed of a fast-melting flux encapsulates the first tracer, which is used to rapidly release the micro-pellet after it is dropped into the molten pool, generating an initial signal indicating "successful delivery".

[0017] A middle layer (shell) composed of the main functional chemicals encapsulates a second tracer. The decomposition or reaction rate of this layer is lower than that of the outer layer. The release of the tracer is used to indicate the initiation or occurrence of the main chemical reaction.

[0018] The inner layer (core) is composed of slow-release or conditional reaction components, which encapsulates a third tracer for use in the later stages of smelting or under specific conditions (such as when the slag basicity reaches a certain value). The release of the tracer indicates the start of the later-stage control reaction.

[0019] Preferably, the stable isotope tracer is selected from stable isotopes of inert gases, such as non-radioactive isotopes of neon, argon, krypton, and xenon. These tracers are chemically stable, do not participate in metallurgical reactions, and are present in extremely low concentrations in the background gas, making them easy to detect. The sensing system can employ a mass spectrometer or laser gas analyzer with a time resolution of 1 second or less to ensure accurate capture of the sequentially released, transient tracer signal peaks.

[0020] Preferably, the analysis of the tracer timing signal includes calculating the time interval between different tracer signal peaks. This time interval, as a direct measurement, is used to quantitatively evaluate the actual melting kinetics or the rate of the stepwise reaction of the functionalized microparticles within the molten pool.

[0021] Preferably, the steps of the online calibration of the metallurgical process model are as follows: the system compares the actual rate of change of process parameters monitored after feeding functionalized micro pellets with the rate of change predicted by the model based on the current state before feeding, and combines the reaction process information calculated from the tracer time series signal to calculate the correction coefficient through a preset algorithm (such as a proportional-integral algorithm or a rule-based correction algorithm), and uses the coefficient to adjust the reaction efficiency, heat exchange coefficient or kinetic constant parameters in the model.

[0022] Preferably, the process prepares various types of functionalized microspheres, such as type A microspheres mainly used for efficient dephosphorization, type B microspheres mainly used for precise temperature control, or type C microspheres mainly used for improving slag fluidity. Different types of microspheres are designed to have their own unique and identifiable tracer combinations or timing signal characteristics.

[0023] Preferably, the functionalized microparticles are fed through a solid material injection system driven by high-pressure gas (such as nitrogen), which has high-precision weighing and flow control capabilities to ensure that decision commands are executed accurately.

[0024] This invention provides an intelligent steelmaking process based on dynamic raw material proportioning. It has the following beneficial effects:

[0025] 1. This invention provides a direct means of confirming whether chemical materials dynamically added to the converter molten pool have been successfully delivered and when they begin to act by encapsulating stable isotope tracers in functionalized microspheres and detecting their signals online. This method transforms the indirect inference-based control of traditional steelmaking into confirmatory control with direct physical signal feedback, solving the problem of blind control caused by the lack of effective feedback methods in traditional processes.

[0026] 2. This invention, by designing a core-shell-membrane multilayer physical structure for microparticles and encapsulating different tracers at different levels, enables the acquisition and decoding of tracer signals containing timing information. Analyzing the time intervals between the appearance of different tracer peaks in this signal allows the process system to perceive the melting kinetics and stepwise reaction process of the microparticles in high-temperature molten slag online, providing a novel and direct measurement dimension for quantitatively evaluating the microscopic reaction environment within the furnace.

[0027] 3. This invention establishes an online self-calibration mechanism for metallurgical process models based on tracer signal feedback and actual response of process parameters. This mechanism enables the control model to no longer rely solely on historical data and initial settings, but to self-correct and evolve within a single smelting cycle based on the unique operating conditions of the current furnace, significantly improving the model's adaptability to uncertainties in raw material composition fluctuations and its predictive accuracy.

[0028] 4. This invention combines precise material feeding, real-time confirmation of the step-by-step reaction process, and the online adaptive capability of the model to construct a complete, fast, and high-frequency closed-loop control system. This system can continuously guide the actual process path to the preset ideal path, avoiding the accumulation of deviations, thereby improving the control accuracy of the molten steel composition and temperature at the smelting endpoint and the stability of the process. Detailed Implementation

[0029] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] To better understand the present invention, the above content will be described in detail below with reference to specific embodiments.

[0031] This invention provides an intelligent steelmaking process based on dynamic raw material ratio.

[0032] Example 1

[0033] 1. Initial conditions and model establishment: Smelting is carried out in a converter with a nominal capacity of 260 tons.

[0034] The initial raw material conditions for this furnace are as follows:

[0035] Iron loading volume: 230 tons

[0036] Molten iron temperature: 1355℃

[0037] Iron composition (wt%): [C]: 4.35, [Si]: 0.45, [Mn]: 0.18, [P]: 0.125, [S]: 0.028.

[0038] Scrap steel loading: 35 tons.

[0039] The initial conditions are input into the central control system, and the metallurgical process model in the system automatically generates the target process path for this smelting. This path includes the target temperature curve of the molten pool, the target decarburization curve, and the target dephosphorization curve that change over time.

[0040] 2. Blowing Process and Deviation Detection: Oxygen blowing begins according to the initial procedure. At t=6 minutes into the blowing process, the system monitors the real-time changes in CO and CO2 concentrations in the flue gas and calculates the actual dephosphorization rate of the molten pool based on material balance calculations. After comparison with the target process path, the system determines that the current actual dephosphorization rate is 18% lower than the target value, and the deviation exceeds the preset threshold.

[0041] 3. Closed-loop intervention and time-series signal decoding: The system decides to execute a high-efficiency dephosphorization intervention, instructing the TEFPs injection system to feed 4.5 kg of type A (high-efficiency dephosphorization) functionalized microspheres into the furnace. After the feeding instruction is issued, the flue gas analysis system (mass spectrometer) enters a high-sensitivity capture mode and detects the following signals in sequence:

[0042] The tracer was detected at t = 6 minutes and 3 seconds. 22 The concentration pulse peak of Ne (T1);

[0043] The tracer was detected at t = 6 minutes and 11 seconds. 38 Ar concentration pulse peak (T2);

[0044] The tracer was detected at t = 6 minutes and 35 seconds. 82 The concentration pulse peak of Kr (T3).

[0045] The system records the above time points and calculates the timing signal intervals: Δt1 = T2 - T1 = 8 seconds, Δt2 = T3 - T2 = 24 seconds.

[0046] 4. Online Model Self-Calibration: The online self-calibration module receives the decoded time-series signal and, in conjunction with the actual changes in the dephosphorization rate after feeding, performs the following operations:

[0047] Based on the parameter Δt1 = 8 seconds (which reflects the actual melting time of the outer membrane of type A micro-pellets in the current slag), the system adjusts the "slag melting kinetic coefficient" parameter in the model downward.

[0048] Based on the comparison between the actual improvement in dephosphorization efficiency after feeding and the model prediction, the system corrected the "A-type micro-pellet reaction efficiency" parameter in the model to 0.85.

[0049] All the revised parameters were immediately applied to the subsequent process control calculations for this smelting.

[0050] 5. Endpoint Control: During the subsequent blowing process, the system performed multiple adaptive adjustments based on different deviations. At t = 16 minutes and 30 seconds, the model determined that all key indicators had entered the target window, and the system issued a stop blowing command. After the blowing stopped, the measured state of the molten steel was: temperature 1655℃, composition (wt%): [C]: 0.052, [P]: 0.011.

[0051] Example 2

[0052] This embodiment provides an adaptive steelmaking process that achieves precise dual control of temperature and composition at the end of the smelting process.

[0053] 1. Initial Conditions and Model Establishment: Smelting will be carried out in the same 260-ton converter. The initial raw material conditions for this batch are as follows:

[0054] Iron loading volume: 225 tons

[0055] Molten iron temperature: 1340℃

[0056] Iron composition (wt%): [C]: 4.50, [Si]: 0.35, [Mn]: 0.22, [P]: 0.095, [S]: 0.030.

[0057] Scrap steel loading: 40 tons.

[0058] Based on the above data, the system generates an initial metallurgical process model and a target process path.

[0059] 2. Blowing Process and Multiple Deviation Detection: At t=12 minutes during blowing, the critical period for endpoint control begins. The system simultaneously identifies two deviations using data from an infrared thermometer and a flue gas analyzer:

[0060] Deviation 1: The actual temperature of the molten pool is 1680℃, which is 25℃ higher than the target temperature.

[0061] Deviation 2: The actual decarbonization rate is 12% lower than the target rate.

[0062] 3. Composite Closed-Loop Intervention and Decoding: The system decides to execute a composite intervention. First, the injection system is instructed to feed 8.0 kg of type B (precision temperature control) microspheres into the furnace to reduce the molten pool temperature. Ten seconds after the type B microspheres are fed, 2.0 kg of type A microspheres are then instructed to be fed to moderately enhance the decarburization and dephosphorization reactions. The flue gas analysis system monitors and distinguishes between the two sets of signals in parallel:

[0063] Detected the corresponding type B microparticles 86 Kr(T1') and 131 The Xe(T2') tracer signal was decoded to reveal its melting and reaction process.

[0064] Subsequently, the corresponding type A microparticles were detected. 22 Ne(T1”) and 38 The Ar(T2”) tracer signal was also decoded to determine its action process.

[0065] 4. Online self-calibration of multiple model parameters: The online self-calibration module receives two sets of independent feedback information and performs parallel model parameter corrections.

[0066] Based on the feedback signal from the B-type micro-pellets and the actual cooling rate, the relevant parameters of "heat balance and heat loss" in the metallurgical model are corrected.

[0067] Based on the feedback signal from type A micro pellets and the actual changes in decarburization rate, the "carbon-oxygen reaction kinetic coefficient at the end of blowing" in the model is corrected.

[0068] 5. Endpoint Control: Through this combined control, the molten pool temperature and composition were brought back to the target path. At t = 15 minutes and 45 seconds, the system determined that the endpoint conditions were met and issued a stop blowing command. After the blowing stopped, the measured state of the molten steel was: temperature 1660℃, composition (wt%): [C]: 0.041, [P]: 0.014.

[0069] Comparative Example 1

[0070] Compared to Example 1, the difference lies in that the process control in this comparative example relies entirely on the calculation results based on a static model before blowing, and all required auxiliary materials (lime and dolomite) are added to the converter all at once during the initial blowing stage. No dynamic material feeding intervention is performed throughout the entire blowing process. All other initial raw material conditions, equipment, and target endpoints are the same as in Example 1.

[0071] Comparative Example 2

[0072] Compared to Example 1, the difference lies in that, although dynamic material feeding is also performed during the blowing process in this comparative example, the material fed is ordinary lime powder without encapsulated tracers. Therefore, this process cannot obtain confirmation signals regarding the success of material feeding and the reaction progress, nor can it perform online self-calibration of the metallurgical process model based on this feedback. All other initial raw material conditions, equipment, and target endpoints are the same as in Example 1.

[0073] Comparative Example 3

[0074] The difference compared to Example 1 is that the functionalized microspheres fed in this comparative example only encapsulate one stable isotope tracer ( 22 Ne). The tracer is only used to confirm that the pellets have been successfully added to the molten pool, but it cannot provide information on the progress of subsequent step-by-step reactions (i.e., it cannot obtain T2 and T3 signals). Therefore, the control system cannot evaluate the melting kinetics of the pellets, nor can it provide richer quantitative information beyond "successful addition" for model self-calibration. All other initial raw material conditions, equipment, and target endpoints are the same as in Example 1.

[0075] Test Example 1

[0076] 1. Test Description

[0077] To verify the technical effect of the process method described in this invention, repeatable industrial tests were conducted on the same converter with a nominal capacity of 260 tons, respectively, according to the process methods described in Example 1 and Comparative Examples 1, 2, and 3.

[0078] The experimental steps are as follows:

[0079] For each heat of steel, raw materials with initial raw material conditions (amount of molten iron, temperature, composition, and amount of scrap steel) similar to those recorded in Example 1 were used.

[0080] The process control logic and material addition method corresponding to Example 1, Comparative Example 1, Comparative Example 2, and Comparative Example 3 are activated respectively.

[0081] For each process method, 20 heats of smelting tests were conducted consecutively.

[0082] During the experiment, the central control system and laboratory information system automatically recorded the following data for each heat of steel:

[0083] The actual temperature of the molten steel at the end of the blowing process.

[0084] The actual carbon and phosphorus content of the molten steel at the end of the blowing process.

[0085] The total smelting cycle from start of blowing to tapping steel.

[0086] The total weight of lime and dolomite auxiliary materials consumed in producing one ton of qualified molten steel.

[0087] For Examples 1, 2, and 3, the metallurgical model's predicted values ​​for the endpoint temperature and endpoint carbon content were recorded one minute before the blowing process was stopped.

[0088] 2. Definition and collection of performance evaluation indicators

[0089] Endpoint hit rate (%): Furnaces where the endpoint temperature is within ±10℃ of the target value, the endpoint carbon content is within ±0.01% of the target value, and the endpoint phosphorus content is lower than the upper limit of the target value are counted as hits. The percentage of hit furnaces out of the total number of furnaces is calculated.

[0090] Average smelting cycle (min): The average time from start of blowing to tapping of steel for 20 heats under each process method.

[0091] Auxiliary material consumption per ton of steel (kg / t): The average total weight of auxiliary materials consumed per ton of qualified molten steel produced in 20 heats under various process methods.

[0092] Model prediction mean absolute error: Calculate the mean absolute error between the model's predicted value 1 minute before the end of blowing and the actual detected value at the endpoint for each process method (excluding Comparative Example 1), and record it as temperature error (°C) and carbon content error (wt%).

[0093] 3. Experimental Data

[0094] The experimental data from 20 heats were statistically averaged, and the results are shown in Table 1.

[0095] Table 1. Comparison of performance evaluation data for different process methods

[0096]

[0097]

[0098] 3. Results Analysis

[0099] Experimental data show that, compared to the comparative examples, the process method described in Example 1 of this invention exhibits significant differences in endpoint hit rate, smelting cycle, auxiliary material consumption, and model prediction accuracy. The underlying technical mechanism lies in the fact that this invention provides a novel process paradigm that deeply couples process verification with model self-calibration. By detecting specific time-series signals formed by multiple stable isotopic tracers encapsulated within functionalized microspheres, the system can not only confirm successful material delivery but also obtain direct temporal information on the step-by-step reaction process of the material within the molten pool. This time-series information provides the system with a quantitative assessment method for the microscopic reaction kinetics within the furnace, which is not available in the processes of Comparative Examples 2 and 3. Therefore, the mean absolute error of the model prediction in Example 1 is significantly lower than other comparable schemes.

[0100] Based on the feedback containing reaction process information obtained above, the process described in this invention can correct key kinetic parameters in its metallurgical process model online and in real time. A model capable of self-correcting according to actual operating conditions within a single smelting cycle provides a more accurate description of the process, thus enabling more precise control decisions. While Comparative Examples 2 and 3 also involve dynamic intervention, their control models cannot evolve online due to a lack of effective feedback or only a single delivery confirmation feedback, limiting the accuracy of control in the face of raw material fluctuations. This model-level difference is directly reflected in the higher endpoint hit rate and shorter average smelting cycle of Example 1.

[0101] Ultimately, precise process control guided by a high-precision model makes each dynamic intervention more targeted and effective, avoiding material waste and subsequent remedial operations caused by excessive or insufficient feeding. This control method stably maintains the entire smelting process near the optimal path, reducing unnecessary resource and energy consumption. The significantly reduced auxiliary material consumption per ton of steel in Example 1 of the experimental data objectively reflects the role of this technical solution in improving resource utilization efficiency.

[0102] Test Example 2

[0103] 1. Test Description

[0104] To verify the technical effectiveness of the process method described in this invention in synchronously and comprehensively controlling temperature and composition at the end of smelting, repeatable industrial tests were conducted on the same converter with a nominal capacity of 260 tons, respectively, according to the process methods of Example 2 and the newly added Comparative Example 4.

[0105] The experimental steps are as follows:

[0106] Definition of Comparative Example 4: Compared with Example 2, the difference is that although dynamic intervention was also carried out at the end of the smelting process in this comparative example, the materials fed were ordinary iron ore (for cooling) without encapsulated tracers and carburizing agents. This process cannot obtain confirmation signals of the material action process, nor can it perform online self-calibration of the model.

[0107] For each heat of steel, raw materials with initial raw material conditions similar to those recorded in Example 2 were used, and the steel entered the final control stage with similar high temperature and high carbon conditions at the end.

[0108] The process control logic and material addition method corresponding to Example 2 and Comparative Example 4 are enabled respectively.

[0109] For each process method, 20 heats of smelting tests were conducted consecutively.

[0110] During the experiment, the central control system and laboratory information system automatically recorded the following data for each heat of steel:

[0111] The actual temperature and actual carbon content of the molten steel at the end of the blowing process.

[0112] The total smelting cycle from start of blowing to tapping steel.

[0113] The total weight of control materials (coolant, carburizer) consumed at the end of the smelting process to produce one ton of qualified molten steel.

[0114] For Example 2, the metallurgical model's predicted values ​​for the endpoint temperature and endpoint carbon content were recorded one minute before the blowing process was stopped.

[0115] 2. Definition and collection of performance evaluation indicators

[0116] Double-hit rate (%): Heats where the endpoint temperature is within ±10℃ of the target value and the endpoint carbon content is within ±0.005% of the target value are counted as double hits. The percentage of double-hit heats out of the total number of heats is calculated.

[0117] Average smelting cycle (min): The average time from start of blowing to tapping of steel for 20 heats under each process method.

[0118] End-of-life material consumption per ton of steel (kg / t): The average total weight of materials consumed in the end-of-life control stage for 20 heats under each process method.

[0119] Model prediction mean absolute error: The mean absolute error between the model's predicted value 1 minute before the end of blowing and the actual detected value at the endpoint under the process of Example 2 was calculated and recorded as temperature error (°C) and carbon content error (wt%), respectively.

[0120] 3. Experimental Data

[0121] The data from 20 heats were statistically averaged, and the results are shown in Table 2.

[0122] Table 2. Comparison of performance evaluation data of different process methods under dual control at the end stage

[0123]

[0124] 4. Results Analysis

[0125] Experimental data show that in the final stage of smelting, where simultaneous combined control of temperature and composition is required, the control results using the process method described in Example 2 are significantly different from those in Comparative Example 4. The technical mechanism lies in the fact that this invention sets specific and unique tracer combinations (such as type B temperature-controlled microspheres and type A dephosphorization / decarbonization-aiding microspheres) for microspheres with different functions (e.g., type B temperature-controlled microspheres and type A dephosphorization / decarbonization-aiding microspheres). 86 Kr / 131 Xe and 22 Ne / 38 Ar) enables the control system to identify and track the effects of two or more interventions in parallel and independently. In Comparative Example 4, after the simultaneous addition of coolant and carbon raiser, the thermal and chemical effects are coupled and indistinguishable at the system level. However, the process of this invention can effectively decouple these two effects by decoding different timing signals.

[0126] Based on independent feedback from multiple intervention measures, the online self-calibration module of this invention can more accurately and purposefully correct different parameters of the metallurgical model. For example, the system can directionally correct the thermal balance coefficient in the model based on the feedback signal of type B micro-pellets and the actual cooling rate; simultaneously, it can directionally correct the carbon-oxygen reaction kinetic constant based on the feedback signal of type A micro-pellets and the actual decarburization rate change. This multi-parameter parallel self-calibration enables the model to more accurately describe the multivariate coupling relationships under complex conditions at the end of the smelting process, which is impossible to achieve with the static model or single feedback model in Comparative Example 4.

[0127] An adaptive model capable of accurately describing multivariate relationships can calculate more optimized composite intervention schemes, thereby avoiding the "whack-a-mole" control oscillations caused by single control methods or ambiguous feedback information. The higher endpoint double-hit rate and shorter average smelting cycle in Example 2, as shown in the experimental data, directly demonstrate this precise control capability. Achieving the endpoint target through one or a few efficient composite interventions reduces the need for repeated adjustments, significantly lowering the unit consumption of materials for final control and improving the overall efficiency of the production process.

[0128] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their likenesses.

Claims

1. A smart steelmaking process based on dynamic raw material proportioning, characterized in that, Includes the following steps: (a) Based on the initial raw material information, establish a metallurgical process model for the current furnace and plan the target process path; (b) During the blowing process, the actual process parameters are monitored in real time and compared with the target process path to determine the deviation; (c) When the deviation exceeds a preset threshold, a preset number of functionalized microspheres are fed into the molten pool; wherein the functionalized microspheres contain at least two functional components that will act sequentially in the molten pool, and are encapsulated with at least two stable isotope tracers that correspond to the functional components and will be released sequentially in the molten pool. (d) Real-time detection of the tracer timing signal formed by the sequential release of the at least two tracers using a sensing system; (e) Based on the tracer timing signal and the changes in actual process parameters after feeding the functionalized microparticles, the kinetic parameters in the metallurgical process model are corrected online for use in the subsequent process control of the current furnace.

2. The process according to claim 1, characterized in that, The functionalized microparticles have a core-shell-membrane multilayer structure, with different functional components and their corresponding tracers encapsulated in different layers. The tracers are released sequentially by the different melting or reaction rates of the materials in each layer in the molten pool.

3. The process according to claim 2, characterized in that, The core-shell-membrane multilayer structure includes: Outer layer (film): Rapid-melting flux layer, encapsulating the first tracer, used to release the delivery confirmation signal; Middle layer (shell): The main functional reaction layer, which encapsulates the second tracer and is used to release signals indicating the initiation or progress of the main reaction; Inner layer (core): a sustained-release or conditional reaction layer that encapsulates a third tracer to release the initiation signal for the later reaction.

4. The process according to claim 1, characterized in that, The stable isotope tracer is selected from stable isotopes of inert gases.

5. The process according to claim 1, characterized in that, The sensing system is a mass spectrometer or laser gas analyzer with a time resolution of less than or equal to 1 second, used for online detection of the stable isotope tracer in converter flue gas.

6. The process according to claim 1, characterized in that, The tracer timing signal includes: the order in which different tracer pulse peaks appear and the time interval between each pulse peak.

7. The process according to claim 6, characterized in that, The time interval between the pulse peaks of the different tracers is used to quantitatively evaluate the melting kinetics or reaction process of the functionalized microparticles in the molten pool.

8. The process according to claim 1, characterized in that, The online correction of the kinetic parameters in the metallurgical process model in step (e) specifically involves comparing the actual rate of change of the process parameters after feeding the functionalized microparticles with the predicted rate of change of the model, and correcting the reaction efficiency coefficient or kinetic constant in the model based on the comparison result and the tracer timing signal.

9. The process according to claim 1, characterized in that, The functionalized microspheres include at least one or more of the following types: type A microspheres for efficient dephosphorization, type B microspheres for precise temperature control, or type C microspheres for slag modification, and different types of microspheres have their own exclusive tracer combinations or time-series signal characteristics.

10. The process according to claim 1, characterized in that, The functionalized microparticles are fed into the molten pool through a jetting system driven by high-pressure gas and capable of precise flow control.

Images