Method, system and equipment for judging comprehensive metallurgical performance of steelmaking lime and medium

Abstract

The invention relates to the technical field of steel smelting, in particular to a comprehensive metallurgical performance judgment method, system and equipment for steelmaking lime and a medium, and the method comprises the steps that furnace information of multiple historical furnaces is obtained, the furnace information comprises independent variable parameters and dependent variable parameters, and a historical database is established; dividing a secondary database according to a preset interval; stepwise regression analysis and symmetric neural network operation are performed on each secondary database, first and second optimal combination information is extracted, and third optimal combination information is generated through a recurrent neural network; constructing a metallurgical performance rating model based on all the third optimal combination information; and inputting the heat information of the to-be-evaluated heat into the lime metallurgical performance rating model to obtain the metallurgical performance grade of the lime used by the heat, supplementing the heat information into the historical database, and updating the lime metallurgical performance rating model. According to the method, the judgment comprehensiveness, precision, real-time performance and long-term adaptability can be improved.

Description

Background Art

[0002] Lime for steelmaking is a key slag-making material in the converter steelmaking process. Its performance directly affects the efficiency of dephosphorization and desulfurization and the fluidity of the slag, and thus determines the quality of molten steel and production costs. During the converter steelmaking process, the chemical composition (such as the content of effective CaO and SiO2) and physical and chemical indexes (such as activity and particle size) of lime for steelmaking act together and affect the end-point parameters of smelting (such as the basicity and temperature of the final slag) through kinetic processes such as dissolution and mass transfer. With the upgrading of the steel industry's demand for quality control and cost reduction and efficiency improvement, establishing an accurate evaluation system for the metallurgical properties of lime has become the core link to improve the stability of the steelmaking process.

[0003] In the prior art, the comprehensive metallurgical property evaluation of lime mainly relies on single-index detection (such as titration of the content of effective CaO) or empirical formula calculation of the lime addition amount, and combines statistical methods to perform linear regression analysis on historical smelting data. For example, the theoretical amount of lime is calculated by the material balance formula or a fixed empirical model, and then the parameters are adjusted according to the actual smelting results. Such methods can initially correlate the lime quality with the smelting effect and provide basic support for process optimization.

[0004] However, the existing methods for evaluating the comprehensive metallurgical properties of lime have the following problems: single indexes or static formulas cannot reflect the dynamic coupling relationship among the chemical composition, physical properties of lime and smelting conditions, resulting in one-sided evaluation; traditional regression models are difficult to capture the multi-parameter non-linear interaction and cannot adapt to the different differential smelting scenarios in different preset intervals; empirical formulas rely on artificially set weights and cannot dynamically optimize the lime addition amount according to real-time smelting data; historical data is updated laggingly, the model lacks self-learning ability, and long-term application is prone to deviate from the actual working conditions. Summary of the Invention

[0005] Aiming at the technical problems that the existing methods for evaluating the comprehensive metallurgical properties of lime rely on single indexes, static formulas and linear models, cannot analyze the multi-parameter dynamic coupling relationship and non-linear smelting interaction of lime, and lack self-learning ability, resulting in one-sided evaluation, low model accuracy, large deviation of lime addition amount and poor long-term adaptability, the present application provides a method, system, device and medium for determining the comprehensive metallurgical properties of lime for steelmaking. Through multi-dimensional database division, fusion regression and neural network modeling, cyclic network dynamic optimization and self-learning update, it realizes the accurate mapping of all elements of lime performance, the analysis of multi-parameter non-linear association, the adaptive regulation of lime addition amount and the continuous evolution of the model, and improves the comprehensiveness, accuracy, real-time performance and long-term adaptability of the determination.

[0006] [[ID=#ID=17]]In the first aspect, the present application provides a method for determining the comprehensive metallurgical properties of lime for steelmaking, including the following steps: S1. Obtain the furnace information of multiple historical furnace batches, including independent variable parameters and dependent variable parameters, and establish a historical database; The independent variable parameters include lime parameters and converter charging parameters, and the dependent variable parameter is the tapping end parameter; S2. Divide all the heat information in the historical database according to the preset intervals of the independent variable parameters to form a secondary database for this preset interval; S3. Conduct stepwise regression analysis and symmetric neural network operations on each secondary database respectively to generate the first optimal combination information and the second optimal combination information of the independent variable parameters and the corresponding dependent variable parameters under each preset interval; Conduct a recurrent neural network operation on the first optimal combination information and the second optimal combination information to generate the third optimal combination information of the corresponding dependent variable under each preset interval; S4. Construct a lime metallurgical property rating model based on all the third optimal combination information. The input of the lime metallurgical property rating model is the heat information, and the output is the metallurgical property grade of the lime used in this heat; Among them, the lime metallurgical property rating model obtains the metallurgical property grade of the lime used in this heat by calculating the deviation degree from the third optimal combination information; S5. Input the heat information of the heat to be evaluated into the lime metallurgical property rating model to obtain the metallurgical property grade of the lime used in this heat, and supplement the heat information into the historical database to update the lime metallurgical property rating model.

[0007] Furthermore, it should be noted that in step S1, the lime parameters include the chemical composition, physical and chemical indexes and addition amount of lime. Among them, the chemical composition includes at least the content of effective CaO and SiO2, and the physical and chemical indexes include at least particle size, bulk density and activity; The converter charging parameters include the composition of hot metal, tapping temperature and scrap addition amount. The hot metal composition includes at least the contents of C, Si, Mn, P and S; The tapping end parameters include the temperature at the tapping end, the carbon content of molten steel and the final slag composition. The final slag composition includes at least the contents of CaO, SiO2, FeO and MgO.

[0008] Furthermore, it should be noted that the determination of lime activity adopts the hydrochloric acid titration method, and the activity value is defined as: the volume of hydrochloric acid consumed after reacting 50 g of lime with 4 mol / L hydrochloric acid for 10 minutes at 40 ± 1 °C.

[0009] Furthermore, it should be noted that in step S2, the preset intervals include: dividing the lime chemical composition into intervals of 80%-85%, 85%-90%, 90%-95%, ≥95% according to the content of effective CaO; Dividing the lime activity into intervals of 260-280 mL, 280-300 mL, 300-320 mL; Divide the molten iron temperature into intervals of 1360 - 1370 °C, 1370 - 1380 °C, and 1380 - 1390 °C; Divide the final slag basicity into intervals of 2.5 - 3.0, 3.0 - 3.5, 3.5 - 4.0, and ≥4.0 according to the target range; All the preset intervals are divided according to the principle of left - closed and right - open, and the maximum interval of each parameter includes the endpoint value of the upper limit.

[0010] Furthermore, it should be noted that in step S3, the step - by - step regression analysis of each secondary database includes: S301. Take the lime parameters and converter charging parameters of the heat number information in the secondary database as independent variables, and take the smelting end - point parameters as the dependent variable; S302. Establish a zero model that only contains a constant term and does not contain independent variables as the initial regression model; S303. Introduce the independent variables into the initial regression model one by one. After introducing each independent variable, conduct a significance test. If the significance level of this independent variable is higher than the set value, retain this independent variable; otherwise, remove this independent variable; S304. After introducing a new independent variable each time, re - test the significance of the existing independent variables for those whose significance level is lower than the set value due to the introduction of the new variable; S305. Repeat S303 - S304 until no new independent variable can significantly improve the model fitting effect and all independent variables in the model remain significant, obtaining an optimized regression model; S306. Match the optimal parameter combinations of all independent variables in the optimized regression model with the corresponding smelting end - point parameters to form the first best combination information. The optimal parameter combination represents the optimal mathematical relationship between the lime parameters and converter charging parameters and the smelting end - point within the current preset interval.

[0011] Furthermore, it should be noted that in step S3, the symmetric neural network operation for each secondary database includes: Take the lime parameters and converter charging parameters of the heat number information in the secondary database as input features, and take the smelting end - point parameters as output labels; Construct a three - layer neural network with a symmetric structure, where the number of neurons in the input layer and the output layer respectively matches the input and output dimensions, and the number of neurons in the hidden layer is automatically adjusted according to the input and output dimensions; Calculate the prediction result through forward propagation, and use the backpropagation algorithm to adjust the network weights, and iterate repeatedly until the prediction error converges; Select the parameter combination with the smallest prediction error and the corresponding smelting end - point parameters to form the second best combination information.

[0012] Further, it should be noted that in step S3, the recurrent neural network operation on the first optimal combination information and the second optimal combination information includes: Performing time series alignment and normalization processing on the first optimal combination information, the second optimal combination information, and the corresponding smelting end-point parameters in each preset interval, and constructing them into a combined information training data set; Using the first optimal combination information and the second optimal combination information as input features, and the corresponding smelting end-point parameters as output labels to construct a recurrent neural network model with bidirectional LSTM units; Training the recurrent neural network model using the combined information training data set to obtain a pre-trained parameter optimization model; Extracting the weight matrix of the pre-trained parameter optimization model as the third optimal combination information.

[0013] Further, it should be noted that in step S4, the construction of the lime metallurgical performance rating model includes: Summarizing the third optimal combination information corresponding to each preset interval to form the optimal combination information set for this preset interval; Determining the preset interval to which the independent variable parameters of the furnace to be evaluated belong, and extracting the optimal combination information set of this preset interval as the candidate reference group; Using a dynamic feature matching algorithm to select the third optimal combination information with the highest similarity to the furnace information of the furnace to be evaluated from the candidate reference group as the target reference combination; Calculating the deviation degree between the furnace information of the furnace to be evaluated and the target reference combination; Specifying the mapping relationship between the deviation degree and the lime metallurgical performance grade, and outputting the evaluation result of the lime metallurgical performance grade of the furnace to be evaluated.

[0014] Further, it should be noted that the weighted Euclidean distance algorithm is used to calculate the deviation degree between the furnace information of the furnace to be evaluated and the target reference combination :

[0015] In the formula, is the total number of evaluation parameters, and the evaluation parameters are independent variable parameters or dependent variable parameters in the furnace information; is the weight coefficient of the i-th evaluation parameter; is the actual value of the i-th evaluation parameter of the furnace to be evaluated; is the value of the i-th evaluation parameter in the target reference combination.

[0016] Further, it should be noted that the evaluation parameters include: The effective CaO content of lime, and the weight coefficient is 0.3; Lime activity, with a weight coefficient of 0.25; Final tapping temperature of converter steelmaking, with a weight coefficient of 0.2; Final slag basicity of converter steelmaking, with a weight coefficient of 0.15; Phosphorus removal rate of converter steelmaking, with a weight coefficient of 0.1.

[0017] It should be further noted that in step S4, the lime metallurgical property grade is divided into grades 1 - 20.

[0018] It should be further noted that in step S5, the steps for determining the furnace information of the furnace to be evaluated include: Set the converter charging parameters and ideal final tapping parameters required for the furnace to be evaluated; Use the converter charging parameters and ideal final tapping parameters required for the furnace to be evaluated to perform material balance calculation and empirical formula calculation respectively, to obtain the balanced lime addition amount and the empirical lime addition amount; Perform recurrent neural network calculation on the balanced lime addition amount and the empirical lime addition amount to obtain the actual lime addition amount of this furnace; Use the actual lime addition amount and the required converter charging parameters to carry out the steelmaking of this furnace, obtain the final tapping parameters after the steelmaking of this furnace, and jointly form the furnace information of the furnace to be evaluated with the required converter charging parameters, the actual lime addition amount, and the actual chemical composition and physical and chemical indexes of the lime.

[0019] It should be further noted that the material balance calculation formula for calculating the balanced lime addition amount is:

[0020] The empirical formula for calculating the empirical lime addition amount is:

[0021] In the formula, and the unit of the empirical lime addition amount is kg; is the mass percentage of Si in the hot metal of this furnace, with the unit of %; is the mass of hot metal, with the unit of kg; is the number of moles of SiO2 generated by the oxidation of Si; is the mass percentage of effective CaO in the lime used in this furnace, with the unit of %; is the required final slag basicity of this furnace, ; is the mass percentage of P in the hot metal of this furnace, with the unit of %; is the mass percentage of effective SiO2 in the lime used for this heat, with the unit of %.

[0022] Furthermore, it should be noted that the recurrent neural network calculation for the lime balance addition amount and the lime empirical addition amount includes: According to the converter charging parameters and the smelting end point parameters in the historical database, material balance calculations and empirical formula calculations are respectively carried out to obtain the lime balance addition amount and the lime empirical addition amount for each historical heat, and a training data set is constructed; Taking the lime balance addition amount and the lime empirical addition amount as input features and the actual lime addition amount as the output label, a recurrent neural network model containing LSTM units is constructed, and the training data set is used for training to obtain a pre-trained lime addition amount prediction model; The lime balance addition amount and the lime empirical addition amount of the heat to be evaluated are input into the pre-trained lime addition amount prediction model, and the actual lime addition amount is output.

[0023] Furthermore, it should be noted that the model update in step S5 includes: after supplementing the heat information of the heat to be evaluated into the historical database, re-performing the regression analysis and neural network operation in steps S2 - S4 to optimize the parameters of the lime metallurgical performance rating model.

[0024] In the second aspect, the present application provides a comprehensive metallurgical performance determination system for lime used in steelmaking, which is used to implement the above comprehensive metallurgical performance determination method, including: A data acquisition and storage module, which is used to obtain the heat information of multiple historical heats and store it in the historical database; A data preprocessing module, which is used to divide all the heat information in the historical database according to the preset intervals of the independent variable parameters, and summarize all the heat information that meets each preset interval to form a secondary database for this preset interval; A data analysis and modeling module, which is used to perform stepwise regression analysis and symmetric neural network operation on each secondary database respectively to generate the first best combination information and the second best combination information of the independent variable parameters and the corresponding dependent variable parameters under each preset interval; Performing recurrent neural network operation on the first best combination information and the second best combination information to generate the third best combination information of the corresponding dependent variable under each preset interval; Constructing a lime metallurgical performance rating model based on all the third best combination information; A model update module, which is used to obtain the heat information of the heat to be evaluated, input it into the lime metallurgical performance rating model to obtain the metallurgical performance grade of this batch of lime, and supplement the heat information of the heat to be evaluated into the historical database to update the lime metallurgical performance rating model.

[0025] In a third aspect, the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. The processor is configured to implement the steps of the above-described comprehensive metallurgical property determination method for lime used in steelmaking when executing the computer program.

[0026] In a fourth aspect, the present application provides a storage medium having a computer program stored thereon. The computer program, when executed by a processor, implements the steps of the above-described comprehensive metallurgical property determination method for lime used in steelmaking.

[0027] As can be seen from the above technical solutions, the present application has the following advantages: 1. By obtaining the independent variable parameters and dependent variable parameters of historical furnace batches, establishing a multi-dimensional historical database and dividing preset intervals, the present application solves the problem of one-sided evaluation of single indicators, realizes accurate data mapping of lime properties and smelting conditions in multiple dimensions and multiple intervals, provides a full-factor analysis basis for comprehensive performance determination, and improves the comprehensiveness of lime comprehensive metallurgical property determination.

[0028] 2. Based on the secondary database, the present application performs stepwise regression and symmetric neural network operations respectively to generate the best combination information in different intervals. By integrating the significance test of the regression model and the non-linear fitting advantage of the neural network, the present application solves the technical bottleneck that traditional linear models cannot analyze the dynamic coupling relationship of multi-parameters, and significantly improves the modeling accuracy of the correlation between lime properties and smelting end points.

[0029] 3. By inputting the new furnace batch data into the rating model to calculate the deviation degree and update the historical database, and continuously iterating the best combination information through the model self-learning mechanism, the present application solves the problem of reduced adaptability caused by data aging of static models, ensures that the rating model evolves dynamically with the production process, and maintains the determination accuracy and process guiding value in the long term.

[0030] 4. By dynamically calculating the material balance amount and empirical addition amount through a recurrent neural network and training the weight parameters in combination with historical data, the present application solves the limitation of relying on manual setting of weights for fixed formulas, realizes real-time adaptive optimization of lime addition amount, and reduces the influence of human experience deviation on the smelting process. Description of the Drawings

[0031] To more clearly illustrate the technical solutions of the present application, the drawings required for description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0032] Figure 1 It is a flowchart of the comprehensive metallurgical property determination method for lime used in steelmaking in an embodiment of the present application.

[0033] Figure 2 It is a schematic block diagram of a comprehensive metallurgical property determination system for lime used in steelmaking in an embodiment of the present application.

[0034] Figure 3 It is a schematic diagram of the hardware structure of an electronic device in an embodiment of the present application. Detailed implementation manners

[0035] In order to make the application objectives, features, and advantages of the present application more obvious and understandable, the technical solutions protected by the present application will be clearly and completely described below by using specific embodiments and the accompanying drawings. Obviously, the embodiments described below are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in this patent, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this patent.

[0036] The comprehensive metallurgical property determination method for lime used in steelmaking involved in the present application mainly aims at the field of iron and steel smelting technology. By obtaining the independent variable parameters and dependent variable parameters of historical furnace batches, establishing a multi-dimensional historical database and dividing preset intervals, it solves the problem of one-sided evaluation of single indicators, realizes the accurate data mapping of lime properties and smelting conditions in multiple dimensions and multiple intervals, provides a full-element analysis basis for comprehensive property determination, and improves the comprehensiveness of lime comprehensive metallurgical property determination; based on the secondary database, stepwise regression and symmetric neural network operations are respectively performed to generate the best combination information in different intervals. By integrating the significance test of the regression model and the non-linear fitting advantages of the neural network, it solves the technical bottleneck that traditional linear models cannot analyze the dynamic coupling relationship of multi-parameters, and significantly improves the modeling accuracy of the correlation between lime properties and smelting end points; through the recurrent neural network, the material balance and empirical addition amount are dynamically calculated, and the weight parameters are trained in combination with historical data, which solves the limitation of the fixed formula depending on manual setting of weights, realizes the real-time adaptive optimization of lime addition amount, and reduces the influence of human experience deviation on the smelting process; the new furnace batch data is input into the rating model to calculate the deviation degree and update the historical database. Through the model self-learning mechanism, the best combination information is continuously iterated, which solves the problem of reduced adaptability caused by data aging of the static model, ensures that the rating model evolves dynamically with the production process, and maintains the determination accuracy and process guiding value for a long time.

[0037] The comprehensive metallurgical property determination method for lime used in steelmaking involved in the present application mainly aims at the technical problems existing in the existing lime comprehensive metallurgical property evaluation method, such as relying on single indicators, static formulas and linear models, being unable to analyze the dynamic coupling relationship of multi-parameters of lime and non-linear smelting interaction, and lacking self-learning ability, resulting in one-sided evaluation, low model accuracy, large deviation in lime addition amount, and poor long-term adaptability.

[0038] The comprehensive metallurgical property determination method of lime for steelmaking involved in this application will be described in detail below. For illustration rather than limitation, specific details such as specific system structures and technologies are presented to thoroughly understand the embodiments of this application. However, those skilled in the art should clearly understand that this application can also be implemented in other embodiments without these specific details.

[0039] In the comprehensive metallurgical property determination method of lime for steelmaking involved in this application, the term "including" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations. The terms "including", "comprising", "having" and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0040] For the convenience of clearly describing the technical solutions of this application, terms such as "first" and "second" are used to distinguish the same items or similar items with basically the same functions and roles. Those skilled in the art can understand that the terms "first", "second", etc. do not limit the quantity and execution order, and the terms "first", "second", etc. do not necessarily limit to being different.

[0041] The statements such as "an embodiment" or "some embodiments" described in this application mean that the specific features, structures, or characteristics described in the embodiment are included in one or more embodiments of this application. Thus, the statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments" and the like that appear in different places in this application do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways.

[0042] Next, the technical solutions in the embodiments of this application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application.

[0043] The comprehensive metallurgical property determination method of lime for steelmaking provided by the embodiments of this application is executed by a computer device. Correspondingly, the comprehensive metallurgical property determination system of lime for steelmaking runs in the computer device.

[0044] The following are some noun explanations in this solution to facilitate a better understanding of this solution: Effective CaO: Effective CaO refers to calcium oxide (CaO) that can actually participate in reactions and play an effective role in metallurgical reactions during the metallurgical process. In metallurgical slag, CaO may form complex compounds with other substances, or due to factors such as the presence of some impurities in the slag, not all CaO can immediately participate in the desired reactions. Only those CaO that can play a role under specific conditions, such as participating in reactions like dephosphorization and desulfurization, and regulating the properties of the slag, are called effective CaO. The content of effective CaO is of great significance for evaluating the reaction ability of the slag and the metallurgical effect. It directly affects the efficiency of the metallurgical process and the quality of the products.

[0045] Final slag basicity: Final slag basicity refers to the ratio of the content of basic oxides to acidic oxides in the slag at the end of the metallurgical process. It is usually expressed by the mass ratio of calcium oxide (CaO) to silicon dioxide (SiO2), that is, final slag basicity = (%CaO) / (%SiO2) (mass fraction). Final slag basicity is an important indicator for measuring the properties of the slag. It has a significant impact on the melting point, viscosity, fluidity of the slag, and the progress of metallurgical reactions. Appropriate final slag basicity can promote the separation of slag and metal, improve the metal recovery rate, and at the same time is conducive to controlling the removal of impurity elements, such as the progress of reactions like desulfurization and dephosphorization, and plays a key role in ensuring the quality of metallurgical products and reducing production costs. Under different metallurgical processes and raw material conditions, the required final slag basicity is also different.

[0046] Figure 1 It is a flowchart of a method for determining the comprehensive metallurgical performance of lime for steelmaking in an embodiment of the present application. Among them, Figure 1 The execution subject can be a system for determining the comprehensive metallurgical performance of lime for steelmaking. According to different requirements, the order of the steps in this flowchart can be changed, and some can be omitted.

[0047] As Figure 1 shown, the method for determining the comprehensive metallurgical performance of lime for steelmaking includes: Step S1, obtaining the furnace information of multiple historical furnace campaigns, including independent variable parameters and dependent variable parameters, and establishing a historical database; The independent variable parameters include lime parameters and converter charging parameters, and the dependent variable parameter is the smelting end point parameter.

[0048] By collecting the full-process data of lime parameters, converter charging parameters, and smelting end point parameters in historical furnace campaigns through the system, a multi-dimensional metallurgical process database is constructed, providing a full-element data basis covering chemical composition, physical indicators, process operation parameters, and smelting results for subsequent model training, effectively solving the problem of insufficient model interpretability caused by the single data dimension in traditional methods, and creating necessary conditions for accurately analyzing the influence mechanism of lime performance on the smelting end point.

[0049] In some specific embodiments, the lime parameters include the chemical composition, physical and chemical indexes, and addition amount of lime. The chemical composition includes at least the content of effective CaO and SiO2, and the physical and chemical indexes include at least particle size, bulk density, and activity. The parameters of the converter's charged materials include the composition of hot metal, the charged temperature, and the scrap addition amount. The composition of hot metal includes at least the contents of C, Si, Mn, P, and S. The parameters at the end of smelting include the temperature at the end of smelting, the carbon content of molten steel, and the composition of final slag. The composition of final slag includes at least the contents of CaO, SiO2, FeO, and MgO.

[0050] By clearly defining that the lime parameters include core chemical components such as effective CaO and SiO2, and key physical and chemical indexes such as particle size and activity, standardizing the parameters of the converter's charged materials to cover basic data such as the composition and temperature of hot metal, and defining the parameters at the end of smelting to include temperature, the carbon content of molten steel, and the composition of final slag, a complete metallurgical process parameter system is constructed, ensuring the comprehensiveness of data collection and the scientificity of index selection, providing a standardized data input specification for the subsequent establishment of a high-precision model, and effectively avoiding model deviation caused by parameter missing or ambiguous definition.

[0051] In some specific embodiments, the activity of lime is measured by the hydrochloric acid titration method, and the activity value is defined as: the volume of hydrochloric acid consumed after reacting 50 g of lime with 4 mol / L hydrochloric acid for 10 minutes at 40 ± 1 °C.

[0052] By standardizing the test conditions and calculation method of lime activity, using the reaction volume of 4 mol / L hydrochloric acid for 10 minutes under the constant temperature condition of 40 ± 1 °C as a quantitative index, the activity measurement standard is unified, the data discreteness problem caused by different detection methods is eliminated, the comparability and consistency of activity data among different batches and different detection devices are ensured, and a reliable quantitative basis is provided for accurately evaluating the reaction performance of lime.

[0053] Step S2: Divide all the heat information in the historical database according to the preset intervals of the independent variable parameters, and summarize all the heat information that meets each preset interval to form a secondary database for that preset interval.

[0054] Using the preset interval division method to decompose the full amount of data into sub-datasets with similar process characteristics, effectively reducing the interference of data noise through parameter interval processing, enhancing the parameter sensitivity of the model under specific process conditions, laying a foundation for the subsequent establishment of an interval-optimized model, not only retaining the comprehensiveness of big data analysis but also meeting the refined modeling requirements, and significantly improving the ability of subsequent algorithms to capture local feature laws.

[0055] In some specific embodiments, the preset intervals include: classifying the lime chemical components into intervals of 80%-85%, 85%-90%, 90%-95%, and ≥95% according to the effective CaO content; classifying the lime activity into intervals of 260-280 mL, 280-300 mL, and 300-320 mL; classifying the hot metal temperature into intervals of 1360-1370 °C, 1370-1380 °C, and 1380-1390 °C; classifying the final slag basicity into intervals of 2.5-3.0, 3.0-3.5, 3.5-4.0, and ≥4.0 according to the target range; The preset intervals are all divided according to the principle of left-closed and right-open, and the maximum interval of each parameter includes the endpoint value of the upper limit.

[0056] By scientifically dividing the parameter intervals of the effective CaO content, activity value, hot metal temperature, and final slag basicity, and clarifying the boundary conditions according to the left-closed and right-open principle, a parameter partition system with clear process meanings is established, which not only meets the actual control accuracy requirements of the metallurgical process, but also ensures the sufficiency of the samples in each sub-dataset, effectively balancing the relationship between the model subdivision requirements and the data statistical significance, and providing reasonable process logic support for interval-based modeling.

[0057] Step S3: Perform stepwise regression analysis and symmetric neural network operation on each secondary database respectively to generate the first best combination information and the second best combination information of the independent variable parameters and the corresponding dependent variable parameters under each preset interval; Perform recurrent neural network operation on the first best combination information and the second best combination information to generate the third best combination information of the corresponding dependent variable under each preset interval.

[0058] By stepwise regression to screen the key influencing factors, combined with the symmetric neural network to mine the non-linear relationship characteristics, and finally using the recurrent neural network for time series feature fusion, a three-layer progressive modeling system is formed, which not only ensures the scientific nature of parameter selection but also fully extracts the complex interaction information, realizes the global optimization of the metallurgical process parameter combination, and provides a multi-dimensional best parameter combination benchmark for establishing a high-precision rating model.

[0059] In some specific embodiments, performing stepwise regression analysis on each secondary database includes: S301. Take the lime parameters and the converter heat input parameters in the heat information in the secondary database as independent variables, and the smelting end point parameters as the dependent variable; S302. Establish a zero model that only contains a constant term and does not contain independent variables as the initial regression model; S303. Introduce the independent variables into the initial regression model one by one. After introducing each independent variable, perform a significance test. If the significance level of the independent variable is higher than the set value, retain the independent variable; otherwise, remove it. S304. After introducing a new independent variable each time, recheck the significance of the existing independent variables for those whose significance level is lower than the set value due to the introduction of the new variable. S305. Repeat S303 - S304 until no new independent variable can significantly improve the model fitting effect and all independent variables in the model remain significant, obtaining an optimized regression model. S306. Match the optimal parameter combinations of all independent variables in the optimized regression model with the corresponding smelting end - point parameters to form the first best combination information. The optimal parameter combination represents the optimal mathematical relationship between the lime parameters and the converter furnace - charging parameters on the smelting end - point within the current preset interval.

[0060] Adopt the step - by - step regression analysis method to dynamically screen key influencing factors, realize automatic parameter optimization through the introduction of a significance test mechanism, establish an optimized regression model with statistical significance, which not only avoids the limitations of manual experience selection but also effectively prevents overfitting, ensuring the reliability and interpretability of the mathematical relationship within each parameter interval, and providing a verified feature combination basis for subsequent neural network modeling.

[0061] In some specific embodiments, performing symmetric neural network operations on each secondary database includes: Take the lime parameters and converter furnace - charging parameters of the heat information in the secondary database as input features, and the smelting end - point parameters as output labels. Construct a three - layer neural network with a symmetric structure, where the number of neurons in the input layer and the output layer respectively matches the input and output dimensions, and the number of neurons in the hidden layer is automatically adjusted according to the input and output dimensions. Calculate the prediction result through forward propagation, use the backpropagation algorithm to adjust the network weights, and iterate repeatedly until the prediction error converges. Select the parameter combination with the minimum prediction error and the corresponding smelting end - point parameters to form the second best combination information.

[0062] By constructing a symmetric neural network structure with matching input and output dimensions, combining forward propagation and backpropagation algorithms to realize in - depth mining of non - linear relationships, automatically adjusting the number of neurons in the hidden layer to ensure the adaptability of the model complexity to data features, effectively capturing complex parameter interaction effects that are difficult to represent by traditional linear models, and providing more generalizable intelligent decision - making support for metallurgical process optimization.

[0063] In some specific embodiments, performing recurrent neural network operations on the first best combination information and the second best combination information includes: Align the first optimal combination information, the second optimal combination information of each preset interval and the corresponding smelting end point parameters in time series and perform normalization processing to construct a combined information training data set; Use the first optimal combination information and the second optimal combination information as input features, and the corresponding smelting end point parameters as output labels to construct a recurrent neural network model with bidirectional LSTM units; Use the combined information training data set to train the recurrent neural network model to obtain a pre-trained parameter optimization model; Extract the weight matrix of the pre-trained parameter optimization model as the third optimal combination information.

[0064] Construct a recurrent neural network using bidirectional LSTM units. Through deep feature extraction of the time series-aligned combined information, realize the dynamic fusion and parameter optimization of the output results of different models, effectively integrate the advantages of statistical models and neural network models, improve the ability to capture the time series correlation features of process parameters, and form a more robust parameter optimization combination benchmark.

[0065] Step S4, construct a lime metallurgical performance rating model based on all the third optimal combination information. The input of the lime metallurgical performance rating model is the heat information, and the output is the metallurgical performance grade of the lime used in this heat; Among them, the lime metallurgical performance rating model obtains the metallurgical performance grade of the lime used in this heat by calculating the deviation degree from the third optimal combination information.

[0066] Based on the dynamic feature matching algorithm, realize the intelligent alignment of real-time heat data and the optimal combination, calculate the deviation degree and establish a grading mapping mechanism, and construct a dynamically adjustable rating model, which can effectively solve the problem that traditional static rating standards are difficult to adapt to process fluctuations, realize the real-time and accurate evaluation of lime performance, and provide immediate feedback guidance for process optimization.

[0067] In some specific embodiments, the construction of the lime metallurgical performance rating model includes: Summarize the third optimal combination information corresponding to each preset interval to form a set of optimal combination information for this preset interval; Determine the preset interval to which the independent variable parameters of the heat to be evaluated belong, and extract the set of optimal combination information for this preset interval as the candidate reference group; Use the dynamic feature matching algorithm to select the third optimal combination information with the highest similarity to the heat information of the heat to be evaluated from the candidate reference group as the target reference combination; Calculate the deviation degree between the heat information of the heat to be evaluated and the target reference combination; Specify the mapping relationship between the deviation degree and the lime metallurgical performance grade, and output the evaluation result of the lime metallurgical performance grade of the heat to be evaluated.

[0068] Implement intelligent benchmarking of real-time data with the historical best combination through a dynamic feature matching algorithm, construct an adaptive rating mechanism by combining weight matrix extraction and similarity calculation, enable the evaluation model to automatically select the optimal reference benchmark according to specific process conditions, significantly improve the pertinence and accuracy of the rating results, and provide personalized quality assessment guidance for the production site.

[0069] In some specific embodiments, the weighted Euclidean distance algorithm is used to calculate the deviation degree between the furnace information of the furnace to be evaluated and the target benchmark combination. :

[0070] In the formula, is the total number of evaluation parameters, and the evaluation parameters are independent variable parameters or dependent variable parameters in the furnace information; is the weight coefficient of the i-th evaluation parameter; is the actual value of the i-th evaluation parameter of the furnace to be evaluated; is the value of the i-th evaluation parameter in the target benchmark combination.

[0071] Construct a deviation degree calculation model using the weighted Euclidean distance algorithm, reflect the importance differences of different process indicators by introducing parameter weight coefficients, combine normalization processing to eliminate the influence of dimensions, establish a scientific and reasonable comprehensive evaluation index system, make the deviation degree calculation results more in line with the actual process control requirements, and provide a quantitative basis for accurate grading.

[0072] In some specific embodiments, the evaluation parameters include: The effective CaO content of lime, with a weight coefficient of 0.3; The activity of lime, with a weight coefficient of 0.25; The tapping temperature at the end of converter steelmaking, with a weight coefficient of 0.2; The basicity of the final slag in converter steelmaking, with a weight coefficient of 0.15; The dephosphorization rate in converter steelmaking, with a weight coefficient of 0.1.

[0073] By setting the weight distribution system of the effective CaO content, activity, tapping temperature, basicity of the final slag and dephosphorization rate, accurately reflect the actual influence degree of each parameter on metallurgical properties, construct a weighted evaluation model that conforms to the characteristics of the steelmaking process, ensure that the rating results are highly consistent with the actual production requirements, and improve the engineering practicability of the evaluation index system.

[0074] In some specific embodiments, the metallurgical performance grade of lime is divided into levels 1-20.

[0075] The metallurgical performance grade is subdivided into 20 grades, and a refined grading evaluation standard is established, which can not only clearly distinguish lime products of different quality levels, but also avoid the fluctuation of the rating results caused by excessive subdivision, achieving a balance between evaluation accuracy and operation practicability, and providing a clear improvement direction guidance with a clear gradient for production quality control.

[0076] Step S5: Input the furnace information of the furnace to be evaluated into the lime metallurgical performance rating model to obtain the metallurgical performance grade of the lime used in this furnace, and supplement the furnace information into the historical database to update the lime metallurgical performance rating model.

[0077] A data closed-loop update mechanism is established. By continuously supplementing training samples with new production data and dynamically optimizing model parameters, the rating model is enabled to have the ability of continuous learning and evolution, effectively overcoming the problem of performance attenuation of traditional models caused by process condition changes, realizing the collaborative optimization development of the evaluation system and production practice, and ensuring the effectiveness and accuracy of long-term application.

[0078] In some specific embodiments, the steps of determining the furnace information of the furnace to be evaluated include: Set the converter charging parameters and ideal smelting end-point parameters required for the furnace to be evaluated; Use the converter charging parameters and ideal smelting end-point parameters required for the furnace to be evaluated to perform material balance calculation and empirical formula calculation respectively to obtain the lime balance addition amount and the lime empirical addition amount; Perform recurrent neural network calculation on the lime balance addition amount and the lime empirical addition amount to obtain the actual lime addition amount of this furnace; Use the actual lime addition amount and the required converter charging parameters to smelt this furnace, obtain the smelting end-point parameters after smelting this furnace, and jointly form the furnace information of the furnace to be evaluated with the required converter charging parameters, the actual lime addition amount, and the actual chemical composition and physical and chemical indexes of the lime.

[0079] Through the dual verification mechanism of material balance calculation and empirical formula calculation, combined with recurrent neural network for dynamic prediction of the actual addition amount, a lime feeding control model integrating theoretical calculation and data-driven is constructed, which not only retains the physical interpretability of metallurgical principles but also enhances the adaptability under complex working conditions, significantly improving the control accuracy of lime addition amount and process stability.

[0080] In some specific embodiments, the material balance calculation formula for calculating the lime balance addition amount is:

[0081] The empirical formula for calculating the lime empirical addition amount is:

[0082] In the formula, The unit of the empirical addition amount of lime is kg; is the mass percentage of Si in the hot metal of this heat, and the unit is %; is the mass of the hot metal, and the unit is kg; is the number of moles of SiO2 generated by the oxidation of silicon; is the mass percentage of effective CaO in the lime used in this heat, and the unit is %; is the final slag basicity required for this heat, ; is the mass percentage of P in the hot metal of this heat, and the unit is %; is the mass percentage of effective SiO2 in the lime used in this heat, and the unit is %.

[0083] Through the dual calculation mechanism of the material balance equation and the empirical formula, the lime addition amount is verified from two dimensions of theoretical derivation and practical accumulation respectively, and a computationally interpretable framework is constructed to ensure that the prediction results not only conform to the basic principles of metallurgy but also take into account the actual production experience, providing a reliable initial value benchmark for neural network prediction and effectively improving the convergence speed and accuracy of the final prediction model.

[0084] In some specific embodiments, the recurrent neural network calculation of the balanced addition amount of lime and the empirical addition amount of lime includes: Carry out material balance calculation and empirical formula calculation respectively according to the converter charging parameters and smelting end point parameters in the historical database, obtain the balanced addition amount of lime and the empirical addition amount of lime for each historical heat, and construct a training data set; Construct a recurrent neural network model with LSTM units using the balanced addition amount of lime and the empirical addition amount of lime as input features and the actual addition amount of lime as the output label, and use the training data set for training to obtain a pre-trained prediction model for the addition amount of lime; Input the balanced addition amount of lime and the empirical addition amount of lime of the heat to be evaluated into the pre-trained prediction model for the addition amount of lime, and output the actual addition amount of lime.

[0085] An LSTM recurrent neural network is used to establish an addition amount prediction model. By capturing the time series dependence relationship and dynamic change law in the historical data, the nonlinear time-varying characteristics of the production process parameters are effectively processed, the intelligent dynamic adjustment of the lime addition amount is realized, the control accuracy under complex working conditions is significantly improved, and the risk of subjective error in manual experience judgment is reduced.

[0086] In some specific embodiments, the model update includes: after supplementing the furnace information of the furnace to be evaluated into the historical database, re-performing the regression analysis and neural network operations in steps S2 - S4 to optimize the parameters of the lime metallurgical performance rating model.

[0087] Establish a model dynamic update mechanism. By continuously incorporating new production data to re-train the model parameters, enable the rating system to have the ability of online learning, effectively adapt to dynamic production environments such as changes in the characteristics of raw and auxiliary materials and updates of process equipment, ensure the long-term effectiveness and prediction accuracy of the evaluation model, and achieve the self-evolution and continuous optimization of the quality control system.

[0088] In a specific embodiment, the comprehensive metallurgical performance determination method for lime used in steelmaking includes: Step S1, obtain the furnace information of multiple historical furnaces, including independent variable parameters and dependent variable parameters, and establish a historical database; The independent variable parameters include lime parameters and converter charging parameters, and the dependent variable parameter is the smelting end point parameter; The lime parameters include the chemical composition, physical and chemical indexes, and addition amount of lime. The chemical composition includes at least the content of effective CaO and SiO2, and the physical and chemical indexes include at least particle size, bulk density, and activity; The converter charging parameters include the composition, charging temperature of hot metal, and scrap addition amount. The hot metal composition includes at least the contents of C, Si, Mn, P, and S; The smelting end point parameters include the temperature at the smelting end point, the carbon content of molten steel, and the final slag composition. The final slag composition includes at least the contents of CaO, SiO2, FeO, and MgO; The determination of lime activity adopts the hydrochloric acid titration method, and the activity value is defined as: the volume of hydrochloric acid consumed after reacting with 50 g of lime for 10 minutes at 40 ± 1 °C using 4 mol / L hydrochloric acid; Step S2, divide all the furnace information in the historical database according to the preset intervals of the independent variable parameters, and summarize all the furnace information that meets each preset interval to form a secondary database for that preset interval; The preset intervals include: dividing the lime chemical composition into intervals of 80% - 85%, 85% - 90%, 90% - 95%, and ≥95% according to the content of effective CaO; Dividing the lime activity into intervals of 260 - 280 mL, 280 - 300 mL, and 300 - 320 mL; Dividing the hot metal temperature into intervals of 1360 - 1370 °C, 1370 - 1380 °C, and 1380 - 1390 °C; Dividing the final slag basicity into intervals of 2.5 - 3.0, 3.0 - 3.5, 3.5 - 4.0, and ≥4.0 according to the target range; The preset intervals are all divided according to the principle of left-closed and right-open, and the maximum interval of each parameter includes the endpoint value of the upper limit; Step S3: Perform stepwise regression analysis and symmetric neural network operation on each secondary database respectively to generate the first optimal combination information and the second optimal combination information of the independent variable parameters and the corresponding dependent variable parameters under each preset interval; Performing stepwise regression analysis on each secondary database includes: S301. Take the lime parameters and converter heat charge parameters in the heat information of the secondary database as independent variables, and the smelting end point parameters as dependent variables; S302. Establish a zero model that only contains a constant term and does not contain independent variables as the initial regression model; S303. Introduce independent variables into the initial regression model one by one. After introducing each independent variable, perform a significance test. If the significance level of this independent variable is higher than the set value, retain this independent variable; otherwise, remove this independent variable; S304. After introducing a new independent variable each time, recheck the significance of the existing independent variables for those independent variables whose significance level is lower than the set value due to the introduction of the new variable; S305. Repeat S303 - S304 until no new independent variable can significantly improve the model fitting effect and all independent variables in the model remain significant, and an optimized regression model is obtained; S306. Match the optimal parameter combination of all independent variables in the optimized regression model with the corresponding smelting end point parameters to form the first optimal combination information. The optimal parameter combination represents the optimal mathematical relationship between the lime parameters and the converter heat charge parameters on the smelting end point within the current preset interval; Performing symmetric neural network operation on each secondary database includes: Take the lime parameters and converter heat charge parameters in the heat information of the secondary database as input features, and the smelting end point parameters as output labels; Construct a three-layer neural network with a symmetric structure, where the number of neurons in the input layer and the output layer respectively matches the input and output dimensions, and the number of neurons in the hidden layer is automatically adjusted according to the input and output dimensions; Calculate the prediction result through forward propagation, and use the backpropagation algorithm to adjust the network weights, and iterate repeatedly until the prediction error converges; Select the parameter combination with the smallest prediction error and the corresponding smelting end point parameters to form the second optimal combination information; Perform recurrent neural network operation on the first optimal combination information and the second optimal combination information to generate the third optimal combination information of the corresponding dependent variable under each preset interval, including: Align the first and second best combination information of each preset interval with the corresponding smelting end-point parameters in time series and perform normalization processing to construct a combined information training data set; Use the first and second best combination information as input features and the corresponding smelting end-point parameters as output labels to construct a recurrent neural network model with bidirectional LSTM units; Use the combined information training data set to train the recurrent neural network model to obtain a pre-trained parameter optimization model; Extract the weight matrix of the pre-trained parameter optimization model as the third best combination information; Step S4: Based on all the third best combination information, construct a lime metallurgical performance rating model. The input of the lime metallurgical performance rating model is the heat information, and the output is the metallurgical performance grade of the lime used in this heat; The construction of the lime metallurgical performance rating model includes: Summarize the third best combination information corresponding to each preset interval to form the best combination information set for this preset interval; Determine the preset interval to which the independent variable parameters of the heat to be evaluated belong, and extract the best combination information set of this preset interval as the candidate reference group; Adopt a dynamic feature matching algorithm to select the third best combination information with the highest similarity to the heat information of the heat to be evaluated from the candidate reference group as the target reference combination; Adopt a weighted Euclidean distance algorithm to calculate the deviation degree between the heat information of the heat to be evaluated and the target reference combination :

[0089] In the formula, is the total number of evaluation parameters, and the evaluation parameters are independent variable parameters or dependent variable parameters in the heat information; is the weight coefficient of the i-th evaluation parameter; is the actual value of the i-th evaluation parameter of the heat to be evaluated; is the value of the i-th evaluation parameter in the target reference combination.

[0090] It should be further noted that the evaluation parameters include: The effective CaO content of lime, and the weight coefficient is 0.3; The lime activity, and the weight coefficient is 0.25; The converter smelting end-point temperature, and the weight coefficient is 0.2; The converter smelting final slag basicity, and the weight coefficient is 0.15; The converter smelting dephosphorization rate, and the weight coefficient is 0.1; Define the mapping relationship between the deviation degree and the lime metallurgical performance grade, and output the evaluation result of the lime metallurgical performance grade of the furnace to be evaluated. The lime metallurgical performance grade is divided into levels 1-20; Step S5: Input the furnace information of the furnace to be evaluated into the lime metallurgical performance rating model to obtain the metallurgical performance grade of the lime used in this furnace, supplement the furnace information into the historical database, and update the lime metallurgical performance rating model; The steps to determine the furnace information of the furnace to be evaluated include: Set the converter charging parameters and ideal smelting end-point parameters required for the furnace to be evaluated; Use the converter charging parameters and ideal smelting end-point parameters required for the furnace to be evaluated to perform material balance calculations and empirical formula calculations respectively to obtain the lime balance addition amount and the lime empirical addition amount; Perform recurrent neural network calculations on the lime balance addition amount and the lime empirical addition amount to obtain the actual lime addition amount for this furnace; Use the actual lime addition amount and the required converter charging parameters to smelt this furnace, obtain the smelting end-point parameters after smelting this furnace, and jointly form the furnace information of the furnace to be evaluated with the required converter charging parameters, the actual lime addition amount, and the actual chemical composition and physical and chemical indexes of the lime; The material balance calculation formula for calculating the lime balance addition amount is:

[0091] The empirical formula for calculating the lime empirical addition amount is:

[0092] In the formula, The units of and the lime empirical addition amount are kg; is the mass percentage of Si in the hot metal of this furnace, and the unit is %; is the hot metal mass, and the unit is kg; is the number of moles of SiO2 generated by the oxidation of Si; is the mass percentage of effective CaO in the lime used in this furnace, and the unit is %; is the final slag basicity required for this furnace, ; is the mass percentage of P in the hot metal of this furnace, and the unit is %; is the mass percentage of effective SiO2 in the lime used in this furnace, and the unit is %; Performing recurrent neural network calculations on the lime balance addition amount and the lime empirical addition amount includes: Based on the converter charging parameters and tapping end parameters in the historical database, material balance calculations and empirical formula calculations are respectively carried out to obtain the lime balance addition amounts and lime empirical addition amounts for each historical heat, and a training data set is constructed. Using the lime balance addition amount and lime empirical addition amount as input features and the actual lime addition amount as the output label, a recurrent neural network model with LSTM units is constructed and trained using the training data set to obtain a pre-trained lime addition amount prediction model. Input the lime balance addition amount and lime empirical addition amount of the heat to be evaluated into the pre-trained lime addition amount prediction model to output the actual lime addition amount. Updating the lime metallurgical property rating model includes: after supplementing the heat information of the heat to be evaluated into the historical database, re-performing the regression analysis and neural network operations in steps S2 - S4 to optimize the parameters of the lime metallurgical property rating model.

[0093] The following is an embodiment of the comprehensive metallurgical property determination system for lime used in steelmaking provided by the present disclosure. This comprehensive metallurgical property determination system and the comprehensive metallurgical property determination method for lime used in steelmaking in the above embodiments belong to the same inventive concept. Details not described in detail in the embodiment of the comprehensive metallurgical property determination system for lime used in steelmaking can refer to the embodiments of the comprehensive metallurgical property determination method for lime used in steelmaking.

[0094] Now, the mobile terminals implementing various embodiments of the present application will be described with reference to the accompanying drawings. In the following description, suffixes such as "module", "component", or "unit" used to represent elements are only for the convenience of explaining the embodiments of the present application, and they have no specific meaning in themselves. Therefore, "module" and "component" can be used interchangeably.

[0095] Figure 2 is a schematic block diagram of the comprehensive metallurgical property determination system for lime used in steelmaking in this embodiment. As Figure 2 shown, the comprehensive metallurgical property determination system for lime used in steelmaking includes: A data acquisition and storage module, configured to obtain the heat information of multiple historical heats and store it in the historical database; A data preprocessing module, configured to divide all the heat information in the historical database according to the preset intervals of the independent variable parameters, and summarize all the heat information that conforms to each preset interval to form a secondary database for that preset interval; A data analysis and modeling module, configured to perform stepwise regression analysis and symmetric neural network operations on each secondary database respectively to generate the first best combination information and the second best combination information of the independent variable parameters and the corresponding dependent variable parameters under each preset interval; Perform recurrent neural network operations on the first optimal combination information and the second optimal combination information to generate the third optimal combination information of the corresponding dependent variable under each preset interval; Construct a lime metallurgical property rating model based on all the third optimal combination information; A model update module is used to obtain the furnace information of the furnace to be evaluated, input it into the lime metallurgical property rating model to obtain the metallurgical property grade of the lime in this batch, and supplement the furnace information of the furnace to be evaluated into the historical database to update the lime metallurgical property rating model.

[0096] The comprehensive metallurgical property determination system of this embodiment is used to implement the comprehensive metallurgical property determination method of lime for steelmaking. The steps include: S1. Obtain the furnace information of multiple historical furnaces, including independent variable parameters and dependent variable parameters, and establish a historical database; The independent variable parameters include lime parameters and converter charging parameters, and the dependent variable parameter is the smelting end point parameter; S2. Divide all the furnace information in the historical database according to the preset intervals of the independent variable parameters, and summarize all the furnace information that meets each preset interval to form a secondary database for this preset interval; S3. Perform stepwise regression analysis and symmetric neural network operations on each secondary database respectively to generate the first optimal combination information and the second optimal combination information of the independent variable parameters and the corresponding dependent variable parameters under each preset interval; Perform recurrent neural network operations on the first optimal combination information and the second optimal combination information to generate the third optimal combination information of the corresponding dependent variable under each preset interval; S4. Construct a lime metallurgical property rating model based on all the third optimal combination information. The input of the lime metallurgical property rating model is furnace information, and the output is the metallurgical property grade of the lime used in this furnace; S5. Input the furnace information of the furnace to be evaluated into the lime metallurgical property rating model to obtain the metallurgical property grade of the lime used in this furnace, and supplement the furnace information into the historical database to update the lime metallurgical property rating model.

[0097] This application also provides an electronic device for implementing each embodiment of this application. The electronic device includes a memory, a processor, and a computer program stored on the memory and executable on the processor.

[0098] Those skilled in the art can understand that the structure of the electronic device involved in the embodiments of this application does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements.

[0099] Figure 3Schematic diagram of the hardware structure of an electronic device for implementing various embodiments of the present application.

[0100] The electronic device includes, but is not limited to, components such as a processor and a memory. Those skilled in the art can understand that the structure of the electronic device involved in the embodiments of the present application does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0101] In the embodiments of the present application, the electronic device includes, but is not limited to, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device may also represent various forms of mobile devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of the present application described and / or claimed herein.

[0102] In the embodiments of the present application, the processor may be implemented by using at least one of an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a processor, a controller, a microcontroller, a microprocessor, and an electronic unit designed to execute the functions described herein. In some cases, such an implementation may be implemented in the controller. For a software implementation, an implementation of a process or function may be implemented with a separate software module that allows execution of at least one function or operation. The software code may be implemented by a software application (or program) written in any suitable programming language. The software code may be stored in the memory and executed by the controller.

[0103] In addition, the electronic device includes some functional modules not shown herein and will not be elaborated further.

[0104] Those skilled in the art to which the present application pertains can understand that various aspects of the electronic device provided by the present application can be implemented as a method, a system, or a program product. Therefore, various aspects of the present disclosure can be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation combining hardware and software aspects, which can be collectively referred to herein as "circuit", "module", or "system".

[0105] The present application also provides a storage medium in which a program product capable of implementing the comprehensive metallurgical property determination method for lime used in steelmaking is stored. In some possible implementation manners, each aspect of the present disclosure can also be implemented in the form of a program product, which includes program code. When the program product runs on a terminal device, the program code is used to cause the terminal device to execute the steps according to various exemplary embodiments of the present disclosure described in the above "Exemplary Method" section of this specification.

[0106] The storage medium may adopt any combination of one or more readable media. The readable media may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0107] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for determining the comprehensive metallurgical properties of lime for steelmaking, characterized in that, Including: S1. Obtain the heat information of multiple historical heats, including independent variable parameters and dependent variable parameters, and establish a historical database; The independent variable parameters include lime parameters and converter charging parameters, and the dependent variable parameter is the smelting end point parameter; S2. Divide all the heat information in the historical database according to the preset intervals of the independent variable parameters to form a secondary database for each preset interval; S3. Conduct stepwise regression analysis and symmetric neural network operations on each secondary database respectively to generate the first best combination information and the second best combination information of the independent variable parameters and the corresponding dependent variable parameters under each preset interval; Conduct a recurrent neural network operation on the first best combination information and the second best combination information to generate the third best combination information of the corresponding dependent variable under each preset interval; S4. Construct a lime metallurgical property rating model based on all the third best combination information. The input of the lime metallurgical property rating model is the heat information, and the output is the metallurgical property grade of the lime used in this heat; S5. Input the heat information of the heat to be evaluated into the lime metallurgical property rating model to obtain the metallurgical property grade of the lime used in this heat, and supplement the heat information into the historical database to update the lime metallurgical property rating model.

2. The comprehensive metallurgical property determination method according to claim 1, characterized in that In step S1, the lime parameters include the chemical composition, physical and chemical indexes and addition amount of lime. The chemical composition includes at least the content of effective CaO and SiO2, and the physical and chemical indexes include at least particle size, bulk density and activity; The converter charging parameters include the composition of hot metal, tapping temperature and scrap addition amount. The hot metal composition includes at least the contents of C, Si, Mn, P and S; The smelting end point parameters include the temperature at the smelting end point, molten steel carbon content and final slag composition. The final slag composition includes at least the contents of CaO, SiO2, FeO and MgO.

3. The comprehensive metallurgical property determination method according to claim 2, wherein In step S2, the preset intervals include: dividing the lime chemical composition into intervals of 80%-85%, 85%-90%, 90%-95% and ≥95% according to the content of effective CaO; Dividing the lime activity into intervals of 260-280 mL, 280-300 mL and 300-320 mL; Dividing the hot metal temperature into intervals of 1360-1370 °C, 1370-1380 °C and 1380-1390 °C; Dividing the final slag basicity into intervals of 2.5-3.0, 3.0-3.5, 3.5-4.0 and ≥4.0 according to the target range; The preset intervals are all divided according to the principle of left-closed and right-open, and the maximum interval of each parameter includes the endpoint value of the upper limit.

4. The comprehensive metallurgical property determination method according to claim 1, characterized in that, In step S3, the recurrent neural network operation on the first best combination information and the second best combination information includes: Align the time series and normalize the first best combination information, the second best combination information of each preset interval and the corresponding smelting end point parameters to construct a combined information training data set; Construct a recurrent neural network model with bidirectional LSTM units using the first best combination information and the second best combination information as input features and the corresponding smelting end point parameters as output labels; Use the combined information training data set to train the recurrent neural network model to obtain a pre-trained parameter optimization model; Extract the weight matrix of the pre-trained parameter optimization model as the third best combination information.

5. The comprehensive metallurgical property determination method according to claim 1, characterized in that In step S4, the construction of the lime metallurgical performance rating model includes: Summarize the third best combination information corresponding to each preset interval to form the best combination information set for this preset interval; Determine the preset interval to which the independent variable parameters of the furnace to be evaluated belong, and extract the best combination information set of this preset interval as the candidate reference group; Use the dynamic feature matching algorithm to select the third best combination information with the highest similarity to the furnace information of the furnace to be evaluated from the candidate reference group as the target reference combination; Calculate the deviation degree between the furnace information of the furnace to be evaluated and the target reference combination; Specify the mapping relationship between the deviation degree and the lime metallurgical performance grade, and output the evaluation result of the lime metallurgical performance grade of the furnace to be evaluated; Among them, the weighted Euclidean distance algorithm is used to calculate the deviation degree of the furnace information of the furnace to be evaluated from the target reference combination : Wherein, is the total number of evaluation parameters, and the evaluation parameters are independent variable parameters or dependent variable parameters in the heat information; is the weight coefficient of the i-th evaluation parameter; is the actual value of the i-th evaluation parameter of the furnace to be evaluated; is the value of the i-th evaluation parameter in the target reference portfolio.

6. The comprehensive metallurgical property determination method according to claim 2, wherein In step S5, the steps for determining the furnace information of the furnace to be evaluated include: Set the converter charging parameters and ideal smelting end-point parameters required for the furnace to be evaluated; Use the converter charging parameters and ideal smelting end-point parameters required for the furnace to be evaluated to perform material balance calculations and empirical formula calculations respectively to obtain the lime balance addition amount and the lime empirical addition amount; Perform recurrent neural network calculations on the lime balance addition amount and the lime empirical addition amount to obtain the actual lime addition amount for this furnace; Use the actual lime addition amount and the required converter charging parameters to smelt this furnace, obtain the smelting end-point parameters after the smelting of this furnace, and jointly form the furnace information of the furnace to be evaluated with the required converter charging parameters, the actual lime addition amount, and the actual chemical composition and physical and chemical indexes of the lime.

7. The comprehensive metallurgical property determination method according to claim 6, characterized in that The material balance calculation formula for calculating the lime balance addition amount is: The empirical formula for calculating the lime empirical addition amount is: In the formula, and the unit of the empirical addition amount of lime is kg; is the mass percentage of Si in the hot metal of this heat, with the unit of %; is the hot metal quality, with the unit of kg; is the number of moles of SiO2 formed by the oxidation of silicon; is the mass percentage of effective CaO in the lime used for this heat, with the unit of %; is the basicity of the final slag required for this heat, ; is the mass percentage of P in the hot metal of this heat, with the unit of %; It is the mass percentage of effective SiO2 in the lime used for this heat, with the unit of %.

8. An integrated metallurgical property determination system for lime used in steelmaking, characterized in that, For implementing the comprehensive metallurgical performance determination method according to any one of claims 1-7, including: A data acquisition and storage module, configured to obtain the furnace information of multiple historical furnaces and store it in the historical database; A data preprocessing module, configured to divide all the furnace information in the historical database according to the preset intervals of the independent variable parameters, and summarize all the furnace information that meets each preset interval to form the secondary database of this preset interval; A data analysis and modeling module, configured to perform stepwise regression analysis and symmetric neural network operations on each secondary database respectively, and generate the first best combination information and the second best combination information of the independent variable parameters and the corresponding dependent variable parameters under each preset interval; Perform recurrent neural network operations on the first best combination information and the second best combination information to generate the third best combination information of the corresponding dependent variable under each preset interval; Construct a lime metallurgical performance rating model based on all the third best combination information; A model update module, configured to obtain the furnace information of the furnace to be evaluated, input it into the lime metallurgical performance rating model to obtain the metallurgical performance grade of this batch of lime, and supplement the furnace information of the furnace to be evaluated into the historical database to update the lime metallurgical performance rating model.

9. An electronic device, characterized in that, Including a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor is used to implement the steps of the comprehensive metallurgical performance determination method according to any one of claims 1-7 when executing the computer program.

10. A storage medium, characterized in that, A computer program is stored on a storage medium, and when the computer program is executed by a processor, the steps of the comprehensive metallurgical property determination method described in any one of claims 1-7 are implemented.

Images