A method and system for predicting the viscosity of metallurgical slag

Abstract

An embodiment of the present application provides a method and system for predicting the viscosity of metallurgical slag. After setting constraint conditions, the method can obtain an initial sample set through random sampling. Then, a preset viscosity calculation model is used to calculate viscosity values to generate a set of sample viscosity values. Next, a feature sample set including sub-interval endpoint values is constructed based on the partial dependence graph generated from the set of sample viscosity values, and a set of theoretical values and a set of true values are obtained based on the feature sample set, and an error set is calculated to correct the set of sample viscosity values using the error set to generate a corrected sample set. Thus, a slag viscosity prediction model is constructed based on the corrected sample set, and the slag viscosity value is predicted using the slag viscosity prediction model. By combining the theoretically corrected viscosity data with machine learning modeling, the method captures the non-linear relationship between slag viscosity, compound composition, and temperature, and can improve the prediction accuracy.

Description

Technical Field

[0001] The present application relates to the field of metallurgical process control, and particularly to a method and system for predicting the viscosity of metallurgical slag. Background Art

[0002] The viscosity of metallurgical slag is a key physical parameter in the smelting process, which can affect the fluidity of the slag, the metal reduction efficiency, and the in-furnace reaction process. In the smelting process, it is necessary to set the smelting process parameters according to the viscosity of the metallurgical slag adaptively. However, under high-temperature conditions, there are problems such as poor composition uniformity, high time consumption, and feedback lag in the measurement of slag viscosity, which limit the application of the viscosity of metallurgical slag in the production process. Therefore, it is necessary to predict the viscosity of metallurgical slag.

[0003] In order to predict the viscosity of metallurgical slag, a viscosity prediction model can be constructed based on metallurgical slag samples. The viscosity prediction model can include empirical models, structural models, thermodynamic models, and machine learning-based models. Based on the constructed viscosity prediction model, the viscosity of metallurgical slag can be predicted, so as to set the metallurgical process parameters in advance according to the prediction results.

[0004] For example, a viscosity prediction model can be constructed based on the Arrhenius equation. That is, by selecting the slag composition range, designing a quadratic regression orthogonal experiment for the slag composition range, then preparing the slag of the above composition using analytical pure reagents, and measuring the viscosity of the slag using the rotating cylinder method to obtain the viscosity value. Thus, according to the slag composition and the viscosity value, a viscosity parameter equation is established based on the Arrhenius equation to construct a slag viscosity prediction model. However, when using the viscosity prediction model to predict the viscosity, it is easy to generate errors when exceeding the data range, resulting in low accuracy and poor real-time performance of the slag viscosity prediction method. Summary of the Invention

[0005] In view of this, the embodiments of the present application provide a method and system for predicting the viscosity of metallurgical slag to solve the problems of low accuracy and poor real-time performance of the slag viscosity prediction method.

[0006] According to one aspect of the present application, a method for predicting the viscosity of metallurgical slag is provided. The method includes:

[0007] Obtaining an initial sample set by randomly sampling within a preset constraint condition, where the constraint condition includes the mass fraction interval and temperature interval of the slag system compounds of the metallurgical slag to be predicted; the initial sample set includes multiple parameter points; the parameter points are combinations of mass fraction values and temperature values obtained by random sampling;

[0008] Calculating the viscosity values corresponding to the parameter points in the initial sample set using a preset viscosity calculation model to generate a sample viscosity value set;

[0009] Generate a partial dependence plot based on the set of sample viscosity values, and construct a set of characteristic samples based on the endpoint values of the partial dependence plot, where the partial dependence plot is used to characterize the response of the predictor variable to the viscosity value, and the predictor variable is the slag system compound of the metallurgical slag to be predicted;

[0010] Obtain a set of theoretical values, a set of true values, and a set of errors, where the set of theoretical values includes the viscosity values obtained by performing calculations on the set of characteristic samples using the viscosity calculation model; the set of true values includes the viscosity values obtained by detecting the set of characteristic samples through experiments; the set of errors includes the error values obtained by performing a difference calculation on the set of theoretical values and the set of true values;

[0011] Use the initial sample set and the set of errors to correct the set of sample viscosity values to generate a corrected sample set;

[0012] Construct a slag viscosity prediction model based on the corrected sample set, and use the slag viscosity prediction model to predict the slag viscosity value.

[0013] Optionally, calculating the viscosity values corresponding to the parameter points in the initial sample set using a preset viscosity calculation model to generate a set of sample viscosity values, including:

[0014] Call a preset viscosity calculation model, where the viscosity calculation model is one of an empirical model, a structural model, a molecular dynamics model, and a thermodynamic application model;

[0015] Input the parameter points in the initial sample set into the viscosity calculation model one by one to obtain the viscosity values output by the viscosity calculation model;

[0016] Add the viscosity values to the initial sample set to generate the set of sample viscosity values. In the set of sample viscosity values, the viscosity values have an associated relationship with the combined mass fraction values and temperature values corresponding to the parameter points.

[0017] Optionally, generating a partial dependence plot according to the set of sample viscosity values, including:

[0018] Apply multiple machine learning algorithms to the set of sample viscosity values for modeling to obtain multiple first alternative models;

[0019] Obtain the quantifiable indicators of the first alternative models, where the quantifiable indicators include one or a combination of root mean square error, mean square error, coefficient of determination, mean absolute percentage error, and mean absolute error;

[0020] Select the optimal model from the multiple alternative models according to the quantifiable indicators;

[0021] Generate a partial dependence plot of all predictor variables' response to the viscosity value through the optimal model.

[0022] Optionally, construct a feature sample set based on the endpoint values of the partial dependence plot, including:

[0023] Based on a preset interval partitioning algorithm, divide the partial dependence plot into multiple sub-intervals. The preset interval partitioning algorithm includes one or a combination of the equal distribution method, response distribution method, and random distribution method;

[0024] Record the endpoint values of each sub-interval where the predictor variable is located;

[0025] Combine the endpoint values that meet the constraint conditions to construct the feature sample set.

[0026] Optionally, based on a preset interval partitioning algorithm, divide the partial dependence plot into multiple sub-intervals, including:

[0027] Obtain the curve function and the domain of the curve function in the partial dependence plot;

[0028] Extract the monotonic intervals of the curve function in the domain;

[0029] Within the monotonic intervals, evenly divide the domain into a first number of sub-intervals and calculate the length of each sub-interval;

[0030] Calculate the split points according to the sub-interval lengths and use the split points to divide the monotonic intervals into a first number of sub-intervals.

[0031] Optionally, based on a preset interval partitioning algorithm, divide the partial dependence plot into multiple sub-intervals, including:

[0032] Obtain the curve function, the domain, and the range of the curve function in the partial dependence plot;

[0033] Extract the monotonic intervals of the curve function in the domain;

[0034] Within the monotonic intervals, evenly divide the range into a second number of sub-intervals and calculate the length of each sub-interval;

[0035] Calculate the split points according to the sub-interval lengths;

[0036] Calculate the predictor variable values corresponding to the split points and divide the monotonic intervals into a second number of sub-intervals according to the predictor variable values.

[0037] Optionally, based on a preset interval partitioning algorithm, divide the partial dependence plot into multiple sub-intervals, including:

[0038] Obtain the curve function and the domain of the curve function in the partial dependency graph;

[0039] Extract the monotonic intervals of the curve function in the domain;

[0040] In the monotonic interval, determine the number of sub-intervals, and randomly generate a third number of splitting points, where the third number is equal to the number of sub-intervals minus 1;

[0041] Divide the monotonic interval into a third number of sub-intervals according to the splitting points.

[0042] Optionally, use the initial sample set and the error set to correct the sample viscosity value set to generate a corrected sample set, including:

[0043] Extract parameter points from the initial sample set;

[0044] Search for target points in the characteristic sample set based on the parameter points, where the target point is the end point value closest to the parameter point;

[0045] Based on the error set, record the error value corresponding to the target point;

[0046] Calculate the arithmetic mean of multiple error values corresponding to the same target point, and establish a slag sample viscosity value error set according to the arithmetic mean;

[0047] Add the slag sample viscosity value error set to the viscosity values in the sample viscosity value set to obtain a corrected sample set.

[0048] Optionally, construct a slag viscosity prediction model based on the corrected sample set, including:

[0049] Divide the corrected sample set into a training set and a validation set;

[0050] Apply multiple machine learning algorithms to the training set in the corrected sample set to perform modeling to obtain multiple second alternative models;

[0051] Use the validation set in the corrected sample set to calculate the quantifiable indicators of the second alternative model, where the quantifiable indicators include the root mean square error;

[0052] Extract the slag viscosity prediction model based on the quantifiable indicators, where the slag viscosity prediction model is the model with the smallest root mean square error among multiple second alternative models.

[0053] According to another aspect of the present application, there is provided a metallurgical slag viscosity prediction system, and the system includes:

[0054] Sampling module, configured to obtain an initial sample set by randomly sampling within preset constraint conditions, where the constraint conditions include the mass fraction range and temperature range of the slag system compounds of the metallurgical slag to be predicted; the initial sample set includes multiple parameter points; the parameter points are combinations of mass fraction values and temperature values obtained by random sampling.

[0055] Theoretical viscosity calculation module, configured to calculate the viscosity values corresponding to the parameter points in the initial sample set using a preset viscosity calculation model to generate a sample viscosity value set.

[0056] Feature extraction module, configured to generate a partial dependence plot based on the sample viscosity value set and construct a feature sample set based on the endpoint values of the partial dependence plot, where the partial dependence plot is used to characterize the response of the prediction variable to the viscosity value, and the prediction variable is the slag system compound of the metallurgical slag to be predicted.

[0057] Error calculation module, configured to obtain a theoretical value set, a true value set, and an error set, where the theoretical value set includes the viscosity values obtained by performing calculations on the feature sample set using the viscosity calculation model; the true value set includes the viscosity values obtained by performing tests on the feature sample set; the error set includes the error values obtained by performing difference calculations on the theoretical value set and the true value set.

[0058] Correction module, configured to correct the sample viscosity value set using the initial sample set and the error set to generate a corrected sample set.

[0059] Viscosity prediction module, configured to construct a slag viscosity prediction model based on the corrected sample set and use the slag viscosity prediction model to predict the slag viscosity value.

[0060] According to another aspect of the present application, there is provided a computer device, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, where when the processor executes the program, the above-mentioned metallurgical slag viscosity prediction method is implemented.

[0061] According to still another aspect of the present application, there is provided a storage medium, on which a computer program is stored, and when the program is executed by a processor, the above-mentioned metallurgical slag viscosity prediction method is implemented.

[0062] With the above technical solution, the embodiments of the present application provide a method and system for predicting the viscosity of metallurgical slag. After setting constraint conditions, the method can obtain an initial sample set through random sampling. Then, a preset viscosity calculation model is used to calculate viscosity values to generate a set of sample viscosity values. Next, a characteristic sample set containing the endpoint values of subintervals is constructed based on the partial dependence graph generated from the set of sample viscosity values, and a set of theoretical values and a set of true values are obtained based on the characteristic sample set, and an error set is calculated to correct the set of sample viscosity values using the error set to generate a corrected sample set. Thus, a slag viscosity prediction model is constructed based on the corrected sample set, and the slag viscosity value is predicted using the slag viscosity prediction model. By combining the theoretically corrected viscosity data with machine learning modeling, the method captures the non-linear relationship between slag viscosity, compound composition, and temperature, and can improve the prediction accuracy. Moreover, by randomly sampling to generate a sample parameter set and optimizing the modeling process, redundant calculations and experimental complexity are reduced. Combined with the selection of the optimal model, the prediction efficiency can be improved. The method can be compatible with various metallurgical slag systems and their multi-dimensional changes in chemical composition and temperature. By introducing experimental correction and multi-level optimization, it is applicable to viscosity prediction under complex production conditions, providing reliable technical support for metallurgical process optimization and control. It can also flexibly select various machine learning models and evaluation indicators, enabling the method to be adjusted according to different scenario requirements, and having good generalization ability and application potential.

[0063] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented in accordance with the content of the specification. And in order to make the above and other purposes, features, and advantages of the present application more obvious and understandable, the specific embodiments of the present application are specifically described below. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The drawings described herein are used to provide a further understanding of the present application, and constitute a part of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation of the present application. In the drawings:

[0065] Figure 1 It is a schematic flow chart of a method for predicting the viscosity of metallurgical slag provided by an embodiment of the present application;

[0066] Figure 2 It is a schematic flow chart of the construction of the viscosity prediction model and the viscosity prediction provided by an embodiment of the present application;

[0067] Figure 3 It is a schematic diagram of the partial dependence graph of CaO provided by an embodiment of the present application;

[0068] Figure 4 It is a schematic diagram of the partial dependence graph of SiO 2 provided by an embodiment of the present application;

[0069] Figure 5 Schematic diagram of the partial dependency graph of MgO provided by the embodiment of the present application;

[0070] Figure 6 For Al provided by the embodiment of the present application 2 O 3 Schematic diagram of the partial dependency graph;

[0071] Figure 7 Schematic diagram of the partial dependency graph of temperature provided by the embodiment of the present application;

[0072] Figure 8 Schematic diagram of the structure of a metallurgical slag viscosity prediction system provided by the embodiment of the present application. Specific embodiments

[0073] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.

[0074] In the embodiment of the present application, the slag is a by-product generated during the metal smelting process. The slag is formed by the chemical reaction of impurities and fluxes in the metal ore under high-temperature conditions. As a molten mixture, the slag can include various compounds. Taking the steelmaking process as an example, the slag can include oxides in the metallurgical raw materials or oxides generated during the metallurgical process, such as CaO, FeO, MnO, MgO, Al 2 O 3 , SiO 2 , P 2 O 5 , Fe 2 O 3 etc. The slag may contain a small amount of other types of compounds, such as fluorides (CaF 2 ), chlorides (NaCl), sulfides (CaS, MnS), sulfates, etc.

[0075] As a mixture with fluidity, the slag has a certain viscosity. The viscosity of the metallurgical slag is a key physical parameter during the smelting process, which can affect the fluidity of the slag, the metal reduction efficiency, and the in-furnace reaction process.

[0076] The viscosity of the slag can be affected by various factors, including the chemical composition, temperature, and structure of the slag. For example, in copper smelting slag, the viscosity of the slag containing solid phase can be calculated by combining the Roscoe equation with the multi-component and multi-phase equilibrium calculation and pure liquid phase slag viscosity calculation functions of FactSage software. The viscosity of copper smelting slag is related to the content of iron oxides, that is, when the content of iron oxides increases, the viscosity of copper smelting slag will decrease. And when SiO2 When the mass fraction is less than 25% or higher than 40%, a large amount of solid-phase substances will appear in the system, causing the viscosity of the molten slag to increase sharply.

[0077] In the metal smelting process, the smelting process parameters can be set according to the viscosity adaptability of the metallurgical molten slag. However, under high-temperature conditions, there are problems such as poor composition uniformity, high time consumption, and feedback lag in the measurement of the molten slag viscosity, which limit the application of the molten slag viscosity in the production process. Therefore, it is necessary to predict the viscosity of the metallurgical molten slag. In some embodiments, the viscosity of the metallurgical molten slag can be predicted based on empirical models, structural models, thermodynamic models, and machine learning models. However, when using the viscosity prediction model for viscosity prediction, errors are likely to occur when exceeding the data range, resulting in low accuracy and poor real-time performance of the molten slag viscosity prediction method.

[0078] To solve the problems of low accuracy and poor real-time performance of the molten slag viscosity prediction method, in this embodiment, a method for predicting the viscosity of metallurgical molten slag is provided, as Figure 1 、 Figure 2 shown. The method includes:

[0079] S100. Obtain an initial sample set by randomly sampling within preset constraint conditions.

[0080] When constructing the viscosity prediction model, the constraint conditions can be specified first. The constraint conditions can include the types of slag system compounds, mass fraction intervals, and temperature intervals of the metallurgical molten slag to be predicted. Among them, the types of compounds in the metallurgical molten slag system can include but are not limited to: SiO 2 、Al 2 O 3 、CaO, MgO, MnO, FeO, Fe 2 O 3 、NiO, PbO, ZnO, Li 2 O, Na 2 O, B 2 O 3 、TiO 2 and Ti 2 O 3 .

[0081] The mass fraction intervals of the slag system compounds in the constraint conditions can be set separately for each compound according to the specific chemical composition of the metallurgical molten slag to be predicted. And the mass fractions of all compounds need to satisfy that the sum of the mass fractions of all compounds is 100%. The temperature interval in the constraint conditions can be set according to the melting temperature range of the metallurgical molten slag to be predicted. For different metallurgical molten slags, different temperature intervals can be set.

[0082] In some embodiments, the preset constraint conditions may further include the ratio of basic oxides to acidic oxides, including binary basicity, ternary basicity, quaternary basicity, etc. The ratio of basic oxides to acidic oxides can reflect the resistance of the material to acidity and its chemical properties. Among them, binary basicity is the mass ratio of the main basic oxides (such as calcium oxide CaO) to the main acidic oxides (such as silicon oxide SiO 2 in the iron-making process. Binary basicity can be used to determine the fluidity, viscosity of the slag and its interaction with molten metal. Ternary basicity is based on binary basicity and also considers magnesium oxide (MgO), that is, the ratio of the basic components of calcium oxide CaO and magnesium oxide (MgO) to silicon oxide. Compared with binary basicity, ternary basicity can more comprehensively describe the basic strength and reaction characteristics of the slag system. Quaternary basicity refers to the ratio of the total mass of calcium oxide (CaO) and magnesium oxide (MgO) in the slag to the mass percentage content of silicon dioxide (SiO 2 and aluminum oxide (Al 2 O 3 ). Quaternary basicity can be applicable to the case where the content of magnesium oxide (MgO) in the metallurgical slag to be predicted is relatively high.

[0083] In addition to the ratios of basic oxides to acidic oxides such as binary basicity, ternary basicity, and quaternary basicity, in the preset constraint conditions, the ratio range between specific several compounds can also be defined according to specific analysis needs, such as the mass fraction ratio of calcium oxide (CaO) and magnesium oxide (MgO), etc.

[0084] When setting the types of compounds in the metallurgical slag system, one or more of the above compounds can be included. In some embodiments, different constraint conditions can be set according to the specific application fields and component contents of the metallurgical slag.

[0085] For example, according to the specific application scenario of the metallurgical slag, the preset constraint conditions include the types of compounds, the mass fraction intervals of compounds, the temperature interval, and the binary basicity range. That is, it can be specified that the types of compounds in the metallurgical slag system to be predicted are: CaO, SiO 2 , MgO, Al 2 O 3 . The mass fraction intervals of each compound are: w (CaO): 20% - 60%, w (SiO 2 ): 20% - 60%, w (MgO): 5% - 20%, w (Al 2 O 3 ) : 5% - 20%. The temperature interval of the slag system is: 1400 - 1700 °C. Other constraint conditions are: binary basicity w(CaO) / w (SiO 2 ) ranges from 0.9 to 1.8.

[0086] After setting the constraint conditions, random sampling can be performed within the value range set by the constraint conditions to obtain an initial sample set. That is, within the compound composition, compound mass range, and temperature range specified in the constraint conditions, a number of mass fraction values and temperature values of all specified types of compounds are randomly sampled. The sampled mass fraction values and temperature values are combined to form multiple parameter points, and thus an initial sample set A is constructed based on the multiple parameter points. Among them, the initial sample set A is used to characterize the slag sample parameters. Therefore, the initial sample set includes multiple parameter points, and the parameter points are combinations of mass fraction values and temperature values obtained by random sampling.

[0087] In some embodiments, when performing random sampling within the value range set by the constraint conditions, a specific random sampling method can be used. The random sampling method includes, but is not limited to: simple random sampling method, Dirichlet distribution sampling method, low-discrepancy sequence method, etc. By any one of the above random sampling methods, an initial sample set containing at least a point set including all compound mass fractions and the corresponding slag system temperature can be obtained, that is:

[0088] ;

[0089] Among them, A represents the initial sample set, including multiple slag sample parameters; w i represents the i th compound mass fraction, T represents the corresponding slag system temperature.

[0090] For example, within the compound composition and compound mass fraction range specified in the above constraint conditions, 3000 mass fraction values and temperature values of all types of compounds are randomly sampled using the simple random sampling method to construct the initial sample set A, as shown in Table 1:

[0091] Table 1, initial sample set A;

[0092]

[0093] In some embodiments, during the sampling process using the random sampling method, it is also possible to determine whether the sampled parameter points exceed the compound composition, compound mass fraction range, and temperature range specified by the constraint conditions. If the result obtained by random sampling exceeds the specified range in the constraint conditions, this set of data can be ignored and the next set of sampling can be performed until the number of the initial sample set A reaches the preset sample size.

[0094] S200. Calculate the viscosity values corresponding to the parameter points in the initial sample set using a preset viscosity calculation model to generate a sample viscosity value set.

[0095] After setting the constraint conditions and randomly sampling within the preset constraint conditions to obtain the initial sample set, a preset viscosity calculation model can be called, and the preset viscosity calculation model is used to calculate the viscosity values corresponding to the parameter points in the initial sample set.

[0096] Among them, the preset viscosity calculation model is a model or tool for performing the coupled viscosity calculation of compound composition and temperature. In some embodiments, the preset viscosity calculation model is one of an empirical model, a structural model, a molecular dynamics model, and a thermodynamic application model. For example, the preset viscosity calculation model is a composition-temperature coupled viscosity calculation model constructed based on the National Physical Laboratory (NPL). Another example is that the viscosity calculation model is an empirical model used in the metallurgical field to predict the viscosity values of different component slag systems, such as the Riboud model and the Urbain model. The Riboud model is based on the Weymann-Frenkel viscosity formula and the Urbain model, combined with the viscosity data of blast furnace slag, to obtain the relationship between the slag composition and the viscosity value. The Urbain model is applicable to describe the relationship between the viscosity and temperature of silicate melts, supercooled liquids, and glasses. Still another example is that the viscosity calculation model is a prediction tool constructed based on thermodynamic software such as FactSage.

[0097] After calculating the viscosity values of each parameter point in the initial sample set A through the preset viscosity calculation model, the viscosity values can be combined with the initial sample set A to obtain a sample viscosity value set B. The sample viscosity value set may include the combination of each compound mass fraction value and the corresponding temperature value, and the viscosity value calculated through the viscosity calculation model under the combination of the compound mass fraction value and the corresponding temperature value. That is, the sample viscosity value set B is the calculated viscosity value corresponding to all the parameter points of the initial sample set A, that is:

[0098] ;

[0099] Among them, B represents the sample viscosity value set; A represents the initial sample set; η represents the calculated viscosity value.

[0100] Therefore, in some embodiments, when using a preset viscosity calculation model to calculate the viscosity values corresponding to the parameter points in the initial sample set to generate a sample viscosity value set, the preset viscosity calculation model can be called first, and then the parameter points in the initial sample set are input into the viscosity calculation model one by one to obtain the viscosity values output by the viscosity calculation model. Then, the viscosity values are added to the initial sample set to generate the sample viscosity value set. It can be seen that in the sample viscosity value set, there is an association relationship between the viscosity value and the combination of the mass fraction value and the temperature value corresponding to the parameter point. In the sample viscosity value set, the association relationship can be represented by setting the parameter point and the predicted viscosity value in the same row or the same column.

[0101] For example, the NPL method is used to calculate the viscosity values of all parameter points in the initial sample set A, thereby establishing a sample viscosity value set B, as shown in Table 2:

[0102] Table 2, sample viscosity value set B;

[0103]

[0104] Among them, in the process of calculating the viscosity value based on the NPL method, the viscosity value can be calculated according to the following formula:

[0105] ;

[0106] Among them, μ represents the viscosity of the system, that is, a measure of the internal resistance of the fluid; exp() represents the exponential function, that is, the power of the base e of the natural logarithm; A represents a constant related to the exponential function and serves as the coefficient of the exponential function; B represents a constant related to the system; T represents the thermodynamic temperature.

[0107] Among them, the coefficient A of the exponential function can be calculated according to the following formula:

[0108] ;

[0109] The constant B related to the system can be calculated according to the following formula:

[0110] ;

[0111] In the formula, Λ is the corrected optical basicity and can be calculated according to the following formula:

[0112] ;

[0113] Among them, Xi is the mole fraction after correction for the i i-th component; n i is the number of oxygen atoms of the i i-th component; Λ i is the optical basicity of the i i-th component. Exemplarily, the values of optical basicity are as follows: CaO is 1; SiO 2 is 0.48; MgO is 0.78; Al 2 O 3 is 0.6.

[0114] It should be noted that in order to accurately obtain the correlation relationship between the compound composition, temperature value, and viscosity value, when calculating the viscosity value through the composition-temperature coupled viscosity calculation model or tool, the selected calculation model should ensure that the viscosity can be calculated for all parameter points in the constructed initial sample set A, and only 1 model or tool can be selected for all viscosity calculations within one calculation cycle.

[0115] S300. Generate a partial dependence graph according to the sample viscosity value set, and construct a characteristic sample set based on the endpoint values of the partial dependence graph.

[0116] After obtaining the sample viscosity value set B, regression analysis can be performed according to the sample viscosity value set B to determine the response relationship function of the predictor variable to the predicted value. The response relationship function can be represented in the form of a partial dependence graph. For the prediction process of metallurgical slag viscosity, the predictor variable is the mass fraction of the slag system compound of the molten slag, and the predicted value is the viscosity value calculated by the calculation model. Therefore, the partial dependence graph can include an independent variable coordinate and a dependent variable coordinate. The independent variable coordinate is the mass fraction value or temperature value of a certain slag system compound, and the dependent variable coordinate is the viscosity value.

[0117] In order to determine the response relationship function of the predictor variable to the predicted value, in some embodiments, multiple machine learning methods can be applied for modeling, and the optimal model can be selected according to the quantifiable evaluation index, so as to generate the partial dependence graph of all predictor variables' responses to the predicted value by the selected optimal model.

[0118] That is, in some embodiments, when generating the partial dependence graph according to the sample viscosity value set, multiple machine learning algorithms can be first applied to the sample viscosity value set for modeling to obtain multiple first alternative models. Then, the quantifiable indexes of the first alternative models are obtained, and the optimal model is selected from the multiple alternative models according to the quantifiable indexes, and the partial dependence graph of all predictor variables' responses to the viscosity value is generated by the optimal model.

[0119] Among them, the multiple machine learning algorithms may include, but are not limited to, multiple models such as Support Vector Machine (SVM) models, Gaussian process regression models, efficient linear models, kernel models, ensemble models, neural network models, tree models, linear regression models, stepwise linear regression models, and sub-models of each model with different hyperparameters; the quantifiable evaluation metrics may include, but are not limited to, one evaluation metric among Root Mean Square Error (RMSE), Mean Squared Error (MSE), coefficient of determination ( R 2 ), Mean Absolute Percentage Error (MAPE), and Mean Absolute Error (MAE).

[0120] For example, for the set B of sample viscosity values shown in Table 2, multiple machine learning methods can be applied for modeling. The multiple machine learning methods include: a total of 28 models such as SVM models, Gaussian process regression models, efficient linear models, kernel models, ensemble models, neural network models, tree models, linear regression models, stepwise linear regression models, and sub-models of each model with different hyperparameters. The 5-fold cross-validation method is used for machine learning, so as to select the optimal model according to the RMSE results calculated from the validation set, and the partial dependence plot of all predictor variables on the predicted response is generated by the optimal model. The RMSE results of the validation sets of each model are shown in Table 3:

[0121] Table 3, RMSE results of the model validation sets;

[0122]

[0123] According to the principle of the minimum RMSE, the 8th model is selected as the optimal model, and the partial dependence plots of the mass fraction values of CaO, SiO 2 , MgO, Al 2 O 3 and the temperature value on the predicted viscosity value are obtained, as shown in Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 .

[0124] After generating the partial dependence plot, based on the preset interval partitioning algorithm, the partial dependence plot can be divided into multiple sub-intervals. Among them, the preset interval partitioning algorithm includes one or a combination of the equal distribution method, the response distribution method, and the random distribution method.

[0125] In some embodiments, when dividing a partial dependence graph into multiple subintervals using the uniform distribution method, the curve function and the domain of the curve function in the partial dependence graph may be obtained first, and then the monotonic intervals of the curve function in the domain may be extracted. Within the monotonic intervals, the domain may be evenly divided into a first number of subintervals, and the length of each subinterval may be calculated. Then, the splitting points may be calculated based on the lengths of the subintervals, and the monotonic intervals may be divided into a first number of subintervals using the splitting points.

[0126] For example, for each partial dependence graph curve function f i ( x ), and its domain D i = a i , b i . First, it may be determined that f i ( x ) in D i all the monotonic intervals, that is, I i,m = c i,m , d i,m ( m = 1, 2, ……, M i ). In each monotonic interval I i,m inside, its domain is evenly divided into k i,m subintervals, then the length of each subinterval is:

[0127] ;

[0128] wherein, c i,m is the left endpoint of the monotonic interval, d i,m is the right endpoint of the monotonic interval; 𝑘 i,m is the number of subinterval divisions.

[0129] Then, based on the length of each subinterval and the number of subinterval divisions, the splitting points may be calculated, that is:

[0130] ;

[0131] wherein, x i,m,j represents the i th compound's j th splitting point;c i,m is the left endpoint of the monotonic interval; h i,m is the length of the sub-interval.

[0132] Thus, according to the calculated splitting points, I i,m is split into 𝑘 i,m sub-intervals, that is, each sub-interval can be expressed as: c i,m , x i,m,1 , x i,m,1 , x i,m,2 ,…, x i,m,ki,m-1 , d i,m ; Then record the set of predicted variable values at the endpoints of all sub-intervals, that is, obtain the following set of endpoint values:

[0133] ;

[0134] Among them, c i,m is the left endpoint of the monotonic interval; x i,m,j represents the i th endpoint of the j th compound; d i,m is the right endpoint of the monotonic interval; 𝑘 i,m is the number of sub-interval divisions, that is, based on the uniform distribution method, the curve function in the partial dependence graph can be evenly divided according to the domain. In the monotonic interval of the domain, 𝑘 i,m-1 endpoint values can be recorded for each compound.

[0135] In some embodiments, when using the response distribution method to divide the partial dependence graph into multiple sub-intervals, after obtaining the curve function in the partial dependence graph and the domain and range of the curve function, the monotonic intervals of the curve function in the domain can be extracted. And in the monotonic interval, the range is evenly divided into a second number of sub-intervals, and the length of each sub-interval is calculated. Then, according to the length of the sub-interval, the splitting points are calculated, and the predicted variable values corresponding to the splitting points are calculated, so as to divide the monotonic interval into a second number of sub-intervals according to the predicted variable values.

[0136] For example, for each partial dependence graph curve function f i ( x ) and its domain Di = a i , b i , it is possible to first determine f i ( x ) in D i all the monotonic intervals on, that is I i,m = c i,m , d i,m ( m = 1, 2,..., M i ). In each monotonic interval I i,m inside, obtain the value range corresponding to the curve function, that is:

[0137] ;

[0138] Among them, f i ( I i,m ) represents the value range of the curve function in the monotonic interval I i,m inside, y i,m min represents in the monotonic interval I i,m inside f i ( I i,m ) the minimum value of; y i,m max represents in the monotonic interval I i,m inside f i ( I i,m ) the maximum value of. By evenly dividing the value range corresponding to the curve function into 𝑘 i,m sub - intervals, the length of each sub - interval can be determined as:

[0139] ;

[0140] In the formula, Δ y i,m represents the length of each sub - interval, 𝑘 i,m represents the number of sub - interval divisions. According to the length Δ y i,m of each sub - interval, the splitting points can be calculated, that is:

[0141] ;

[0142] wherein, y i,m,j represents the i -th j breaking point of the y i,m min -th compound; I i,m represents the minimum value of f i within the monotonic interval I i,m ), i.e., the lower endpoint of the value range; Δ y i,m represents the length of each sub-interval.

[0143] And calculate the x value corresponding to the breaking point:

[0144] ;

[0145] Thus, according to the sub-interval division method provided in the above embodiment, based on the x value corresponding to the breaking point, I i,m is divided into k i,m sub-intervals, and collect the set of predictor variable values at the endpoints of all sub-intervals to obtain the set of endpoint values.

[0146] In some embodiments, when using the random distribution method to divide the partial dependence graph into multiple sub-intervals, after obtaining the curve function and the domain of the curve function in the partial dependence graph, and extracting the monotonic intervals of the curve function in the domain, the number of sub-intervals can also be determined within the monotonic interval, and the third number of breaking points is randomly generated. Wherein, the third number is equal to the number of sub-intervals minus 1. Then, according to the breaking points, the monotonic interval is divided into the third number of sub-intervals.

[0147] For example, for each curve function f i ( x ) of the partial dependence graph and its domain D i = a i , b i , it is possible to determine the curve function f i ( x ) in the domain Di All the monotonic intervals, i.e., I i,m = c i,m , d i,m ( m = 1, 2, ……, M i ). In each monotonic interval I i,m Determine the number of sub - intervals 𝑘 i,m , and then, based on random distribution methods such as uniform distribution and normal distribution, randomly generate 𝑘 i,m −1 splitting points in this interval, i.e.:

[0148] ;

[0149] wherein, x i,m,j represents the i th splitting point of the j th compound. The relationship < c i,m < x i,m,1 < x i,m,2 <……< x i,m,ki,m-1 < d i,m can be satisfied between the splitting points. Then, according to the randomly generated splitting points, I i,m is divided into k i,m sub - intervals, and the set of 𝑥 values of the endpoints of all sub - intervals is collected.

[0150] Through the sub - interval division method provided in the above embodiments, after dividing the partial dependence graph into multiple sub - intervals, the endpoint values of each sub - interval where the prediction variables are located can be recorded, and the endpoint values that meet the constraint conditions can be combined to construct the characteristic sample set.

[0151] That is, after dividing the sub - intervals, the values of each prediction variable at the endpoints of each interval can be recorded, and within the compound composition mass fraction interval and temperature interval specified in the preset constraint conditions, the endpoint values of all prediction variables are combined to construct the characteristic sample set C. The characteristic sample set C is used to characterize the characteristic slag sample parameters of the metallurgical slag to be predicted.

[0152] The characteristic sample set C is a combination of each prediction variable at the endpoints of the divided intervals and is a point set that meets the range of the compound composition mass fraction interval specified in the preset constraint conditions, i.e.:

[0153] ;

[0154] Among them, C represents the characteristic sample set; x represents an element in the characteristic sample set; w i represents the compound mass fraction value; T represents the temperature value; then each w i ( i = 1, 2,..., n ) and T are divided into several interval endpoints, denoted as { a i1 , a i2 ,..., a ikj} and { b 1 , b 2 ,..., b kT}.

[0155] For example, after obtaining the partial dependence diagrams of the contents of CaO, SiO 2 , MgO, Al 2 O 3 and temperature on the predicted viscosity value, each partial dependence diagram is divided into 4 intervals according to the response distribution method, the values of each prediction variable at the interval endpoints are recorded, and the endpoint values of all prediction variables are combined within the specified compound composition mass fraction interval and temperature interval that meet the preset constraint conditions to construct the characteristic sample set C, as shown in Table 4:

[0156] Table 4, Characteristic sample set C;

[0157]

[0158] According to the characteristic sample set C shown in Table 4, the endpoint values of each prediction variable in each sub-interval can be determined as shown in Table 5:

[0159] Table 5, Endpoint values of prediction variables in each sub-interval;

[0160]

[0161] S400. Obtain the theoretical value set, the true value set, and the error set.

[0162] After constructing the characteristic sample set, theoretical value calculation and analysis experiments can be respectively performed on the constructed characteristic sample set to obtain a theoretical value set and a true value set. Among them, the theoretical value set includes the viscosity values obtained by performing calculations on the characteristic sample set using the viscosity calculation model.

[0163] To calculate the theoretical values, the same viscosity calculation model as in step S200 can be used, and the theoretical values can be calculated using this viscosity calculation model according to the data points in the characteristic sample set C. That is, in some embodiments, a preset viscosity calculation model can be called first. Among them, the viscosity calculation model is one of an empirical model, a structural model, a molecular dynamics model, and a thermodynamic application model. Then, the data points in the characteristic sample set C are input into the viscosity calculation model one by one to obtain the theoretical viscosity values output by the viscosity calculation model, and then the theoretical viscosity values are added to the characteristic sample set C to generate the theoretical value set D.

[0164] Similarly, in the theoretical value set D, the theoretical viscosity value has an associated relationship with the combination of the mass fraction endpoint value and the temperature endpoint value corresponding to the data point. It can be seen that the theoretical value set D is a point set containing the calculated viscosity values corresponding to all the data points in the characteristic sample set C, that is:

[0165] ;

[0166] Among them, D represents the theoretical value set; C represents the characteristic sample set; η represents the calculated theoretical viscosity value.

[0167] For example, based on the composition-temperature coupled viscosity calculation model or tool, the viscosity values of all points in the viscosity value calculation set C are calculated to establish the theoretical value set D, as shown in Table 6:

[0168] Table 6, theoretical value set D;

[0169]

[0170] The true value set includes the viscosity values obtained by performing tests on the characteristic sample set. The viscosity values of all data points in the characteristic sample set C can be obtained using experimental methods, and a true value set E is established. The true value set E is used to characterize the true viscosity value of the characteristic slag sample.

[0171] Among them, the experimental methods include the capillary method, the oscillating container method, the rotation method, the oscillating plate method, etc. The true value set E is a point set containing the experimental viscosity values corresponding to all the data points in the characteristic sample set C, that is:

[0172] ;

[0173] Among them, E represents the set of true values; C represents the set of characteristic samples; represents the test viscosity value.

[0174] For example, the true viscosity values of all data points in the characteristic sample set C are obtained using the rotation method, and the set of true values E as shown in Table 7 is established:

[0175] Table 7, set of true values E;

[0176]

[0177] After obtaining the set of theoretical values and the set of true values according to the method provided in the above embodiments, the error can also be calculated based on the set of theoretical values and the set of true values to obtain an error set. That is, the error set includes the error values obtained by performing a difference calculation on the set of theoretical values and the set of true values.

[0178] An error set F of the viscosity value of the characteristic slag sample can be established according to the set of theoretical values D and the set of true values E using a mathematical method. The mathematical method can be to subtract the test viscosity value in the set of true values E from the calculated viscosity value in the set of theoretical values D correspondingly, or subtract the calculated viscosity value in the set of theoretical values D from the test viscosity value in the set of true values E correspondingly. The error set F obtained through calculation is used to characterize the error of the viscosity value of the slag sample. Therefore, the error set F is a point set containing the differences of the viscosity values obtained by mathematical processing corresponding to all data points of the characteristic sample set C, that is:

[0179] ;

[0180] Among them, F represents the error set; C represents the set of characteristic samples; Δ η represents the difference of the viscosity value obtained by mathematical processing.

[0181] For example, subtract the test viscosity value in the set of true values E from the calculated viscosity value in the set of theoretical values D correspondingly to establish an error set F of the viscosity value of the characteristic slag sample, as shown in Table 8:

[0182] Table 8, error set F;

[0183]

[0184] S500. Use the initial sample set and the error set to correct the sample viscosity value set to generate a corrected sample set.

[0185] After obtaining the error set F, the sample viscosity value set B can be corrected based on the error set F and the initial sample set A, so as to correct the prediction result of the viscosity value in the sample viscosity value set B and obtain a corrected sample set.

[0186] In some embodiments, to generate a corrected sample set, when using the initial sample set and the error set to correct the sample viscosity value set, parameter points can be extracted from the initial sample set. And a target point is searched for in the characteristic sample set based on the parameter points. Wherein, the target point is the end point value closest to the parameter point. And based on the error set, the error value corresponding to the target point is recorded. Then, the arithmetic mean of multiple error values corresponding to the same target point is calculated, and a slag sample viscosity value error set is established according to the arithmetic mean. Thus, the slag sample viscosity value error set is added to the viscosity values in the sample viscosity value set to obtain a corrected sample set.

[0187] After obtaining the error set F, all the parameter points in the initial sample set A can be sequentially searched for the point closest to this parameter point in the characteristic sample set C k points, and the error value of this k point corresponding in the error set F is recorded. The k arithmetic mean of the error values of each parameter point in the initial sample set A is taken to establish a slag sample viscosity value error set G. Then, the sample viscosity value set B and the slag sample viscosity value error set G are corrected correspondingly to establish a corrected sample set H.

[0188] Wherein, the method for searching for the point closest to the k points can be the k-nearest neighbor algorithm (k-NN), and k the value of the parameter should not be less than the number of compound types. Therefore, the slag sample viscosity value error set G is the set of points of the viscosity value errors corresponding to all the parameter points of the initial sample set A, that is:

[0189] ;

[0190] Wherein, G is the slag sample viscosity value error set; A is the initial sample set; Δ η represents the error of the viscosity value of this point.

[0191] Then, according to the corresponding correction method, the sample viscosity value set B is corrected to obtain a corrected sample set H. The sample correction set H contains the set of points of the corrected viscosity values of all the data points in the sample viscosity value set B, that is:

[0192] ;

[0193] Wherein, H represents the sample correction set; A represents the initial sample set; η ' represents the corrected viscosity value of this point.

[0194] For example, for all points in the initial sample set A, search for the 6 points closest to each point in the feature sample set C in sequence, and record the error values corresponding to these 6 points in the error set F. Take the arithmetic mean of the 6 error values of each point in the initial sample set A, and add the arithmetic mean to the viscosity value of the corresponding point in the sample viscosity value set B to obtain the corrected sample set H of the slag, as shown in Table 9:

[0195] Table 9, corrected sample set H;

[0196]

[0197] S600. Construct a slag viscosity prediction model based on the corrected sample set, and use the slag viscosity prediction model to predict the slag viscosity value.

[0198] After obtaining the corrected sample set, a slag viscosity prediction model can be constructed based on the corrected sample set. That is, apply machine learning methods to the corrected sample set H for modeling.

[0199] In some embodiments, multiple machine learning methods can be applied to the corrected sample set H for modeling, and the optimal model can be selected according to quantifiable evaluation indicators to obtain the optimal slag viscosity prediction model within the mass fraction range of compound components and the temperature range under specified constraint conditions. For this purpose, the corrected sample set can be divided into a training set and a validation set, and then multiple machine learning algorithms are applied to the training set in the corrected sample set for modeling to obtain multiple second alternative models. Calculate the quantifiable indicators of the second alternative models using the validation set in the corrected sample set, and extract the slag viscosity prediction model based on the quantifiable indicators. Among them, the quantifiable indicators can include the root mean square error, and the corresponding slag viscosity prediction model is the model with the smallest root mean square error among the multiple second alternative models.

[0200] For example, when applying multiple machine learning methods to the corrected sample set H for modeling, the optimal model is selected according to the RMSE results of the validation set, and this optimal model is the optimal slag viscosity prediction model under the specified constraint conditions. Among them, the multiple machine learning methods include: SVM model, Gaussian process regression model, efficient linear model, kernel model, ensemble model, neural network model, tree model, linear regression model, stepwise linear regression model, and sub-models with different hyperparameters for each model, a total of 28 models, and a 5-fold cross-validation method is used for machine learning. The RMSE results of the validation sets of each second alternative model are shown in Table 10:

[0201] Table 10, RMSE results of the validation sets of the second alternative models;

[0202]

[0203] It can be seen that based on the principle of minimizing RMSE, the 16th model can be selected as the optimal model, that is, this model is the slag viscosity prediction model.

[0204] It should be noted that since when determining the slag viscosity prediction model, it is also necessary to evaluate the model based on various machine learning methods and quantifiable evaluation indicators, the quantifiable evaluation indicators used when determining the slag viscosity prediction model can be the same as those of various machine learning methods and quantifiable evaluation indicators when determining the response relationship function of the prediction variable to the predicted value. However, other machine learning methods and quantifiable evaluation indicators can also be selected when necessary.

[0205] After constructing the slag viscosity prediction model, the slag viscosity prediction model can be used to predict the slag viscosity value. In some embodiments, in order to predict the slag viscosity value, input data can be obtained first, where the input data includes mass fraction values and temperature values of various types of compounds within the compound type, mass fraction interval, and temperature interval corresponding to the preset constraints. Then, the input data is input into the slag viscosity prediction model to obtain the viscosity value output by the slag viscosity prediction model.

[0206] By applying the technical solution of this embodiment, the method can combine the theoretically corrected viscosity data with machine learning modeling to capture the non-linear relationship between slag viscosity and compound composition and temperature, and can improve the prediction accuracy. Moreover, by randomly sampling to generate a sample parameter set and optimizing the modeling process, redundant calculations and experimental complexity can be reduced. Combining with the optimal model screening, the prediction efficiency can be improved. The method can be compatible with various metallurgical slag systems and multi-dimensional changes in their chemical compositions and temperatures. By introducing experimental correction and multi-level optimization, it is applicable to viscosity prediction under complex production conditions and provides reliable technical support for metallurgical process optimization and control. It can also flexibly select various machine learning models and evaluation indicators, enabling the method to be adjusted according to different scenario requirements, and having good generalization ability and application potential.

[0207] Furthermore, as a specific implementation of the metallurgical slag viscosity prediction method in the above embodiment, an embodiment of the present application provides a metallurgical slag viscosity prediction system, as Figure 8 shown, the system includes:

[0208] A sampling module, configured to obtain an initial sample set by randomly sampling within preset constraints, where the constraints include the mass fraction interval and temperature interval of the slag system compounds of the metallurgical slag to be predicted; the initial sample set includes multiple parameter points; the parameter points are combinations of mass fraction values and temperature values obtained by random sampling;

[0209] A theoretical viscosity calculation module for calculating the viscosity values corresponding to the parameter points in the initial sample set using a preset viscosity calculation model to generate a sample viscosity value set;

[0210] A feature extraction module for generating a partial dependence graph based on the sample viscosity value set and constructing a feature sample set based on the endpoint values of the partial dependence graph, where the partial dependence graph is used to characterize the response of the predictor variable to the viscosity value, and the predictor variable is the slag system compound of the metallurgical slag to be predicted;

[0211] An error calculation module for obtaining a theoretical value set, a true value set, and an error set, where the theoretical value set includes the viscosity values obtained by performing calculations on the feature sample set using the viscosity calculation model; the true value set includes the viscosity values obtained by detecting the feature sample set through experiments; the error set includes the error values obtained by performing difference calculations on the theoretical value set and the true value set;

[0212] A correction module for correcting the sample viscosity value set using the initial sample set and the error set to generate a corrected sample set;

[0213] A viscosity prediction module for constructing a slag viscosity prediction model based on the corrected sample set and predicting the slag viscosity value using the slag viscosity prediction model.

[0214] By applying the technical solution of this embodiment, a metallurgical slag viscosity prediction system provided in the above embodiment can, after setting constraint conditions, obtain an initial sample set through random sampling. Then, use a preset viscosity calculation model to calculate viscosity values to generate a sample viscosity value set. Then, construct a feature sample set containing sub-interval endpoint values based on the partial dependence graph generated from the sample viscosity value set, and obtain a theoretical value set and a true value set based on the feature sample set, and calculate an error set to correct the sample viscosity value set using the error set to generate a corrected sample set. Thus, construct a slag viscosity prediction model based on the corrected sample set and predict the slag viscosity value using the slag viscosity prediction model. The system captures the non-linear relationship between slag viscosity and compound composition and temperature by combining experimentally corrected theoretical viscosity data with machine learning modeling, which can improve the prediction accuracy. And, by randomly sampling to generate a sample parameter set and optimizing the modeling process, reducing redundant calculations and experimental complexity, and combining with optimal model screening, the prediction efficiency can be improved.

[0215] It should be noted that for other corresponding descriptions of each functional unit involved in a metallurgical slag viscosity prediction system provided in an embodiment of the present application, reference can be made to the corresponding descriptions in the metallurgical slag viscosity prediction method provided in the above embodiment, which will not be elaborated here.

[0216] The embodiments of the present application further provide a computer device, which may specifically be a personal computer, a server, a network device, etc. The computer device includes a bus, a processor, a memory, and a communication interface, and may further include an input / output interface and a display device. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store location information. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the steps in the method embodiments are implemented.

[0217] Those skilled in the art can understand that the structure of the above computer device is only a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components, or combine certain components, or have different component arrangements.

[0218] In one embodiment, a computer-readable storage medium is further provided. The computer-readable storage medium may be non-volatile or volatile, and has a computer program stored thereon. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0219] In one embodiment, a computer program product is further provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0220] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data that have been authorized by the user or fully authorized by all parties.

[0221] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it may include the processes of the method embodiments as described above.

[0222] Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application may include at least one of non-volatile and volatile memories. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc.

[0223] Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0224] The databases involved in the embodiments provided in this application may include at least one of relational databases and non-relational databases. Non-relational databases may include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application may be general-purpose processors, graphics processors, digital signal processors, programmable logics, data processing logics based on quantum computing, etc., without limitation.

[0225] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0226] The above-described embodiments only represent several implementation manners of this application. Their descriptions are relatively specific and detailed, but they should not be construed as limiting the patent scope of this application. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several modifications and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of this application should be subject to the appended claims.

Claims

1. A method for predicting the viscosity of metallurgical slag, characterized in that: The method comprises: An initial sample set is obtained by random sampling within preset constraints, wherein the constraints include a mass fraction interval and a temperature interval of the slag-based compound of the metallurgical slag to be predicted; the initial sample set includes a plurality of parameter points; the parameter points are a combination of mass fraction values ​​and temperature values ​​obtained by random sampling; Calculating the viscosity values ​​corresponding to the parameter points in the initial sample set using a preset viscosity calculation model to generate a sample viscosity value set; Generating a partial dependence graph according to the sample viscosity value set, and constructing a characteristic sample set based on the endpoint values ​​of the partial dependence graph, wherein the partial dependence graph is used to characterize the response of the prediction variable to the viscosity value, wherein the prediction variable is the mass fraction of the slag-based compound and the slag-based temperature of the metallurgical slag to be predicted; Acquire a theoretical value set, a true value set, and an error set, wherein the theoretical value set includes viscosity values ​​obtained by calculating the characteristic sample set using the viscosity calculation model; the true value set includes viscosity values ​​obtained by testing the characteristic sample set through experiments; and the error set includes error values ​​obtained by performing difference calculation on the theoretical value set and the true value set; Correcting the set of sample viscosity values ​​using the initial sample set and the error set to generate a corrected sample set; A slag viscosity prediction model is constructed based on the corrected sample set, and the slag viscosity value is predicted using the slag viscosity prediction model.

2. The method according to claim 1, characterized in that Calculating the viscosity values ​​corresponding to the parameter points in the initial sample set using a preset viscosity calculation model to generate a sample viscosity value set, including: Calling a preset viscosity calculation model, wherein the viscosity calculation model is one of an empirical model, a structural model, a molecular dynamics model, and a thermodynamic application model; Inputting the parameter points in the initial sample set into the viscosity calculation model one by one to obtain the viscosity value output by the viscosity calculation model; The viscosity value is added to the initial sample set to generate the sample viscosity value set, in which the viscosity value has an associated relationship with the combination of the mass fraction value and the temperature value corresponding to the parameter point.

3. The method according to claim 1, characterized in that Generating a partial dependence graph according to the set of sample viscosity values ​​includes: Applying a plurality of machine learning algorithms to the sample viscosity value set to perform modeling to obtain a plurality of first candidate models; Obtaining quantifiable indicators of the first candidate model, wherein the quantifiable indicators include one or more combinations of root mean square error, mean square error, determination coefficient, mean absolute percentage error, and mean absolute error; Selecting the optimal model from the plurality of candidate models according to the quantifiable indicators; A partial dependence plot of all predictor variables on viscosity values ​​was generated using the optimal model.

4. The method according to claim 3, characterized in that Constructing a feature sample set based on the endpoint values ​​of the partial dependence graph includes: Based on a preset interval partitioning algorithm, the partial dependence graph is divided into a plurality of sub-intervals, wherein the preset interval partitioning algorithm includes one or more combinations of an average distribution method, a response distribution method, and a random distribution method; Record the endpoint value of each sub-interval where the predictor variable is located; The endpoint values ​​satisfying the constraint conditions are combined to construct the feature sample set.

5. The method according to claim 4, characterized in that Based on a preset interval partitioning algorithm, the partial dependency graph is divided into a plurality of sub-intervals, including: Obtaining a curve function in the partial dependence graph and a definition domain of the curve function; Extracting the monotonic interval of the curve function in the definition domain; In the monotonic interval, the domain is evenly divided into a first number of subintervals, and the length of each subinterval is calculated; A segmentation point is calculated according to the subinterval length, and the monotonic interval is divided into a first number of subintervals using the segmentation point.

6. The method according to claim 4, characterized in that Based on a preset interval partitioning algorithm, the partial dependency graph is divided into a plurality of sub-intervals, including: Obtain the curve function in the partial dependence graph and the domain and range of the curve function; Extracting the monotonic interval of the curve function in the definition domain; In the monotonic interval, the value range is evenly divided into a second number of sub-intervals, and the length of each sub-interval is calculated; Calculate the split point according to the sub-interval length; The predicted variable value corresponding to the segmentation point is calculated, and the monotonic interval is divided into a second number of sub-intervals according to the predicted variable value.

7. The method according to claim 4, characterized in that Based on a preset interval partitioning algorithm, the partial dependency graph is divided into a plurality of sub-intervals, including: Obtaining a curve function in the partial dependence graph and a definition domain of the curve function; Extracting the monotonic interval of the curve function in the definition domain; In the monotonic interval, determining the number of subintervals, and randomly generating a third number of segmentation points, where the third number is equal to the number of subintervals minus 1; The monotonic interval is divided into a third number of sub-intervals according to the segmentation points.

8. The method according to claim 1, characterized in that Correcting the sample viscosity value set using the initial sample set and the error set to generate a corrected sample set includes: Extracting parameter points from the initial sample set; Searching for a target point in the feature sample set based on the parameter point, the target point being an endpoint value closest to the parameter point; Based on the error set, recording the error value corresponding to the target point; Calculating the arithmetic mean of multiple error values ​​corresponding to the same target point, and establishing a slag sample viscosity value error set according to the arithmetic mean; The slag sample viscosity value error set is added to the viscosity values ​​in the sample viscosity value set to obtain a corrected sample set.

9. The method according to claim 1, characterized in that: A slag viscosity prediction model is constructed based on the corrected sample set, including: Dividing the modified sample set into a training set and a validation set; Applying a plurality of machine learning algorithms to the training sets in the revised sample set to perform modeling to obtain a plurality of second candidate models; Calculate a quantifiable index of the second candidate model using a validation set in the revised sample set, wherein the quantifiable index includes a root mean square error; The slag viscosity prediction model is extracted based on the quantifiable index, and the slag viscosity prediction model is a model with the smallest root mean square error among multiple second alternative models.

10. A metallurgical slag viscosity prediction system, characterized in that: The system comprises: A sampling module, used for obtaining an initial sample set by random sampling within preset constraints, wherein the constraints include a mass fraction interval and a temperature interval of the slag-based compound of the metallurgical slag to be predicted; the initial sample set includes a plurality of parameter points; the parameter points are a combination of a mass fraction value and a temperature value obtained by random sampling; A theoretical viscosity calculation module, used to calculate the viscosity values ​​corresponding to the parameter points in the initial sample set using a preset viscosity calculation model to generate a sample viscosity value set; A feature extraction module, used for generating a partial dependence graph according to the sample viscosity value set, and constructing a feature sample set based on the endpoint values ​​of the partial dependence graph, wherein the partial dependence graph is used for characterizing the response of a prediction variable to a viscosity value, wherein the prediction variable is a slag-based compound of the metallurgical slag to be predicted; an error calculation module, for obtaining a theoretical value set, a real value set and an error set, wherein the theoretical value set includes viscosity values ​​obtained by calculating the characteristic sample set using the viscosity calculation model; the real value set includes viscosity values ​​obtained by testing the characteristic sample set through experiments; and the error set includes error values ​​obtained by performing difference calculation on the theoretical value set and the real value set; A correction module, configured to correct the sample viscosity value set using the initial sample set and the error set to generate a corrected sample set; The viscosity prediction module is used to construct a slag viscosity prediction model based on the corrected sample set, and use the slag viscosity prediction model to predict the slag viscosity value.

Images