Converter oxygen consumption prediction method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of metallurgical industry, and discloses a converter oxygen consumption prediction method and device, electronic equipment and a storage medium, the method comprises the steps that target feature data and predicted feeding data of a target converter are obtained, the target feature data are determined according to preset influence factors, and the predicted feeding data are determined according to the preset influence factors; the predicted feeding data is obtained through prediction according to a preset feeding prediction model, oxygen consumption prediction is conducted on the target characteristic data and the predicted feeding data through a preset oxygen consumption prediction model, target oxygen consumption is obtained, and oxygen supply of the target converter is controlled according to the target oxygen consumption, the oxygen consumption prediction model is obtained by training a training data sample with an oxygen consumption label; by means of the method, the accuracy, stability and process change adaptability of converter oxygen consumption prediction are improved.

Description

Technical Field

[0001] The present application relates to the technical field of metallurgical industry, and in particular to a method, device, electronic equipment and storage medium for predicting converter oxygen consumption. Background Art

[0002] In the steel production industry, converter steelmaking is a core process, and its quality control and cost efficiency are directly related to the market competitiveness of steel products and the economic benefits of enterprises. Oxygen blowing in converter steelmaking, as a key technology in the process, has a decisive impact on the quality control of molten steel and production costs.

[0003] Oxygen blowing is mainly done by injecting oxygen into molten iron to induce a violent oxidation reaction. In this process, oxygen reacts with elements such as carbon, silicon, manganese, sulfur and phosphorus in the molten iron, which can not only effectively remove harmful impurities such as sulfur and phosphorus, improve the purity and quality of steel, but also use the chemical heat released by the oxidation reaction to provide the high-temperature heat required for the steelmaking process. This oxidation reaction is crucial for accurately controlling the terminal carbon content and temperature of molten steel, which is directly related to the final performance of the steel.

[0004] However, the complexity of oxygen blowing operation lies in the fact that the supply of oxygen needs to be strictly controlled. Excessive oxygen supply will lead to excessive oxidation of alloy elements in molten iron, thereby increasing the cost of alloy compensation. At the same time, excessively high temperature may also accelerate the wear of the converter lining and shorten the service life of the equipment. On the contrary, if the oxygen supply is insufficient, the carbon content in the molten steel will exceed the standard, seriously affecting the mechanical properties and quality of the steel, and even requiring an increase in the number of re-blowing, thereby reducing production efficiency and increasing production costs. Obviously, there is an urgent need for a new converter oxygen consumption prediction method to solve at least one of the above problems.

[0005] It should be noted that the above content only provides background technical information related to the present application and does not necessarily constitute prior art. Summary of the invention

[0006] In view of the above-mentioned shortcomings of the prior art, the present application provides a method, device, electronic equipment and storage medium for predicting converter oxygen consumption to improve the accuracy, stability and adaptability of converter oxygen consumption prediction.

[0007] Other features and advantages of the present application will become apparent from the following detailed description, or may be learned in part by the practice of the present application.

[0008] According to one aspect of an embodiment of the present application, a method for predicting oxygen consumption of a converter is provided, comprising: obtaining target characteristic data and predicted charging data of a target converter, wherein the target characteristic data is determined according to preset influencing factors, and the predicted charging data is predicted according to a preset charging prediction model; performing oxygen consumption prediction on the target characteristic data and the predicted charging data by using a preset oxygen consumption prediction model to obtain a target oxygen consumption, so as to control the oxygen supply of the target converter according to the target oxygen consumption, wherein the oxygen consumption prediction model is trained based on training data samples with oxygen consumption labels.

[0009] In one embodiment of the present application, based on the aforementioned scheme, before obtaining the target characteristic data of the target converter, the method also includes: collecting production data corresponding to the target converter according to the preset influencing factors to obtain initial characteristic data, wherein the initial characteristic data include molten iron weight, scrap steel weight, target element content in molten iron, heat generated when the target element is oxidized, and molten iron temperature; constructing initial characteristic variables based on the initial characteristic data, wherein the initial characteristic variables include chemical heat entering the furnace, physical heat entering the furnace, proportion of physical heat entering the furnace, and molten iron ratio, wherein the chemical heat entering the furnace is calculated based on the molten iron weight, the scrap steel weight, the target element content in the molten iron, and the heat generated when the target element is oxidized, the physical heat entering the furnace is calculated based on the molten iron weight, the scrap steel weight, and the molten iron temperature, the proportion of physical heat entering the furnace is calculated based on the chemical heat entering the furnace and the physical heat entering the furnace, and the molten iron ratio is calculated based on the molten iron weight and the scrap steel weight; feature screening is performed on the initial characteristic data and the initial characteristic variables to obtain the target characteristic data.

[0010] In one embodiment of the present application, based on the above scheme, the expression of the chemical heat entering the furnace includes: Q chem =W iron ×(β Si ×ω[Si] ％ +β Mn ×ω[Mn] ％ +β C ×ω[C] ％ +β P ×ω[P] ％ +β S ×ω[S] ％ ) / [100×(W iron +W scrap )]; the representation of the physical heat entering the furnace includes: Q physical =W iron ×[(T e -T r )×C iron_below +Liron +(T iron -T e )×C iron_above ] / (W iron +W scrap ); the representation of the proportion of physical heat entering the furnace includes: Q ratio =Q physical / (Q physical +Q chem ); The expression of the molten iron ratio includes: Iron ratio =W iron / (W iron +W scrap ), where Q chem is the chemical heat of entering the furnace, W iron is the weight of the molten iron, W scrap is the weight of the scrap steel, ω[Si] ％ is the silicon content of molten iron, ω[Mn] ％ is the manganese content of molten iron, ω[C] ％ is the carbon content of molten iron, ω[P] ％ is the phosphorus content of molten iron, ω[S] ％ is the sulfur content of molten iron, β Si Silicon oxidation reaction exothermic heat, β Mn The heat released by the oxidation reaction of manganese, β C The heat released by the carbon oxidation reaction, β P The heat released by the phosphorus oxidation reaction, β S is the heat released by the sulfur oxidation reaction, Q physical is the physical heat of the furnace, T e is the eutectic temperature of iron, T iron is the molten iron temperature, T r is the ambient temperature, L iron is the latent heat of fusion of pig iron, C iron_below is the specific heat capacity of iron below the eutectic point, C iron_above is the specific heat capacity of iron above the eutectic point, Q ratio is the physical heat ratio of the furnace, Iron r3tio is the iron-water ratio.

[0011] In one embodiment of the present application, based on the aforementioned scheme, feature screening is performed on the initial feature data and the initial feature variables to obtain the target feature data, including: performing data preprocessing on the initial feature data and the initial feature variables, wherein the data preprocessing includes null value processing, outlier processing and data normalization; removing redundant features from the initial feature data and the initial feature variables according to a preset feature correlation algorithm to obtain initial features; performing feature analysis on the initial features through a lightweight gradient boosting algorithm, and performing feature screening from the initial features according to the analysis results to obtain the target feature data.

[0012] In one embodiment of the present application, based on the aforementioned scheme, the oxygen consumption prediction model is trained according to training data samples with oxygen consumption labels, including: obtaining sample training data and sample oxygen consumption corresponding to the sample training data, wherein the sample training data includes sample characteristic data and sample feeding data; labeling the sample training data according to the sample oxygen consumption to generate training data samples with oxygen consumption labels; training a preset oxygen consumption prediction model based on the training data samples with oxygen consumption labels to obtain the oxygen consumption prediction model, wherein the hyperparameter combination of the preset oxygen consumption prediction model is determined by a preset optimization algorithm.

[0013] In one embodiment of the present application, based on the aforementioned scheme, a preset oxygen consumption prediction model is trained based on the training data samples with oxygen consumption labels, and after obtaining the oxygen consumption prediction model, the method also includes: obtaining sample test data and test oxygen consumption corresponding to the sample test data, wherein the sample test data includes test feature data and test feeding data; inputting the sample test data into the oxygen consumption prediction model to obtain predicted oxygen consumption, and calculating a predicted residual based on the predicted oxygen consumption and the test oxygen consumption; generating a sample data set based on the sample test data and the predicted residual, and obtaining a residual correction prediction model through training with the sample data set; predicting the residual of the sample test data through the residual correction prediction model to obtain a residual correction value, and correcting the predicted oxygen consumption based on the residual correction value, so as to optimize the oxygen consumption prediction model according to the corrected predicted oxygen consumption.

[0014] In one embodiment of the present application, based on the aforementioned scheme, after correcting the predicted oxygen consumption according to the residual correction value, the method further includes: calculating the mean square error of the oxygen consumption and the determination coefficient of the test set according to the corrected predicted oxygen consumption and the test oxygen consumption; inputting the sample training data into the oxygen consumption prediction model to obtain a reference oxygen consumption, and calculating the determination coefficient of the training set according to the reference oxygen consumption and the sample oxygen consumption; and performing performance evaluation on the oxygen consumption prediction model according to at least one of the mean square error of the oxygen consumption, the determination coefficient of the test set, and the determination coefficient of the training set.

[0015] According to one aspect of an embodiment of the present application, a converter oxygen consumption prediction device is provided, comprising: an acquisition module, used to acquire target characteristic data and predicted charging data of a target converter, wherein the target characteristic data is determined according to preset influencing factors, and the predicted charging data is predicted according to a preset charging prediction model; a prediction module, used to predict the oxygen consumption of the target characteristic data and the predicted charging data through a preset oxygen consumption prediction model to obtain a target oxygen consumption, so as to control the oxygen supply of the target converter according to the target oxygen consumption, wherein the oxygen consumption prediction model is trained based on training data samples with oxygen consumption labels.

[0016] According to one aspect of an embodiment of the present application, an electronic device is provided, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device implements a method for predicting converter oxygen consumption as described in any one of the above embodiments.

[0017] The present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor of a computer, the computer is enabled to execute the method for predicting converter oxygen consumption as described in any one of the above embodiments.

[0018] Beneficial effects of the present application: The present application obtains target characteristic data and predicted charging data of a target converter, wherein the target characteristic data is determined according to preset influencing factors, and the preset influencing factors include influencing factors of multiple dimensions such as conditions of raw materials entering the furnace, conditions of oxygen lance used in blowing, conditions of converter use, conditions of process charging, and blowing targets, and initial variables constructed based on converter process principles, process mode changes, and operator habits. The application can capture the key features that affect the oxygen consumption of the converter in all directions, and ensure the high accuracy of the obtained prediction results and enhance the stability of the model by improving the multi-dimensional input of the model, thereby breaking through the limitations of traditional models and significantly optimizing the accuracy of oxygen consumption prediction. The predicted charging data is predicted according to a preset charging prediction model, and is obtained through a preset The oxygen consumption prediction model predicts the oxygen consumption of the target characteristic data and the predicted charging data to obtain the target oxygen consumption, so as to control the oxygen supply of the target converter according to the target oxygen consumption. The oxygen consumption prediction model is trained according to the training data samples with oxygen consumption labels. By predicting the oxygen consumption of the target converter through the above oxygen consumption prediction model, the number of oxygen supplementary blowing can be effectively reduced, and the production cycle can be shortened, thereby improving the overall production efficiency. It can also effectively reduce the oxidation loss of alloy elements caused by excessive oxygen supply and reduce the alloy compensation cost. It can also accurately predict the oxygen consumption according to the set target carbon content and temperature requirements of the molten steel, ensuring high-quality endpoint control. Further, it helps to reduce oxide and gas emissions, which is beneficial to environmental protection.

[0019] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings herein are incorporated into the specification and constitute a part of the specification, showing embodiments consistent with the present application, and together with the specification, are used to explain the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0021] Figure 1 is a schematic flow chart of a method for predicting converter oxygen consumption shown in an exemplary embodiment of the present application;

[0022] Figure 2 It is a schematic diagram of a process of predicting oxygen consumption of a converter according to an exemplary embodiment of the present application;

[0023] Figure 3 is a flow chart of an integrated machine learning method for predicting converter oxygen consumption shown in an exemplary embodiment of the present application;

[0024] Figure 4 is a schematic diagram of the comparison results of the actual oxygen consumption and the predicted oxygen consumption of the converter oxygen consumption prediction method shown in an exemplary embodiment of the present application;

[0025] Figure 5 is a flow chart of a method for predicting converter oxygen consumption shown in another exemplary embodiment of the present application;

[0026] Figure 6 is a block diagram of a converter oxygen consumption prediction device shown in an exemplary embodiment of the present application;

[0027] Figure 7 A schematic diagram of the structure of a computer system suitable for implementing an electronic device of an embodiment of the present application is shown. DETAILED DESCRIPTION

[0028] The following will describe the implementation methods of the present application with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for illustrating the present application, not for limiting the scope of protection of the present application.

[0029] It should be noted that the illustrations provided in the following embodiments are only used to illustrate the basic concept of the present application in a schematic manner, and therefore the illustrations only show components related to the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0030] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application. However, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present application difficult to understand.

[0031] First of all, it should be explained that the interquartile range method is a method used in statistics to measure the degree of data dispersion. Its core lies in calculating the interquartile range (IQR). The interquartile range refers to the difference between the upper quartile (Q3) and the lower quartile (Q1), that is, IQR = Q3-Q1. Quartiles are a set of data arranged from small to large and divided into four equal parts, each containing 25% of the data. Among them, Q1 is the value of the 25th percentile, and Q3 is the value of the 75th percentile.

[0032] Min-Max Normalization is a commonly used data preprocessing technique that aims to scale the original data to a specified range (usually [0,1]) for subsequent data analysis or machine learning model training. Before training a machine learning model, normalizing the data helps speed up the convergence of the model and improve the performance of the model. Normalizing data of different dimensions to the same range facilitates comparison and analysis.

[0033] In the model selection criteria, R 2 R and MSE are two commonly used indicators for evaluating model performance, which represent the coefficient of determination and mean squared error respectively. 2 Also known as the coefficient of determination, it is a statistical indicator used to evaluate the goodness of fit of a regression model. It indicates the proportion of the variability of the dependent variable that can be explained by the model, that is, how well the model fits the data. MSE is the abbreviation of mean square error, which is a commonly used indicator to measure the difference between the model's predicted value and the actual observed value. It is obtained by calculating the average of the squares of the differences between the predicted value and the actual observed value.

[0034] The Shapiro-Wilk test is a method for testing normality, which is used to determine whether a data set follows a normal distribution.

[0035] In machine learning and statistics, mutual information regression is a method for feature selection and feature correlation quantification. Mutual information is a measure of the dependency between two random variables. In regression problems, mutual information regression can be used to determine the strength of the relationship between each input feature and the output variable, thereby helping to select the most relevant features. It is based on the concept of information entropy, which measures the relationship between input features and output variables by calculating the difference between their joint distribution and their respective marginal distributions.

[0036] In the context of machine learning, hyperparameters are parameters whose values ​​are set before starting the learning process, rather than parameter data obtained through training. Typically, hyperparameters need to be optimized to select a set of optimal hyperparameters for the learning machine to improve learning performance and results.

[0037] Spearman correlation analysis is a nonparametric method widely used in statistics to measure the strength and direction of the monotonic relationship between two variables. Spearman correlation analysis is based on the concept of sorting. It first sorts the observed values ​​of the two variables separately (i.e., rank transformation), and then calculates the correlation between the ranks of the two variables after sorting. This method does not depend on the specific distribution of the data, but only considers the order relationship between the data.

[0038] The LightGBM algorithm is an efficient machine learning algorithm based on the gradient boosting framework, which is used to solve classification and regression problems. It uses a histogram-based decision tree algorithm, a leaf node splitting strategy, and support for parallel learning. It has the advantages of fast training speed, high memory efficiency, support for category features, and high precision, and performs well when processing large-scale data sets.

[0039] The KNN (K-Nearest Neighbors) algorithm is an instance-based learning method that performs classification or regression by measuring the distance between different feature values. Specifically, it finds the K nearest neighbors of a sample in the feature space and predicts the category or value of the sample based on the categories (for classification problems) or values ​​(for regression problems) of these neighbors.

[0040] Ensemble machine learning algorithms, also known as ensemble learning, are a strategy that combines the outputs of multiple machine learning models (also known as base learners or weak learners) to produce a more powerful and stable prediction model. The core idea is to use the diversity of multiple models to improve the overall performance and reduce the bias and variance that may exist in a single model. In ensemble learning, the base learners can be models of the same type (such as decision trees) or different types of models (such as decision trees, neural networks, and support vector machines). These base learners are usually diversified in some way (such as random initialization, different training data subsets, or feature subsets).

[0041] Figure 1 is a schematic flow chart of a method for predicting converter oxygen consumption according to an exemplary embodiment of the present application, and the method for predicting converter oxygen consumption can be executed by a computing and processing device. Figure 1 As shown, the converter oxygen consumption prediction method includes at least steps S110 to S120, which are described in detail as follows:

[0042] In step S110, target characteristic data and predicted charging data of a target converter are acquired.

[0043] The target characteristic data is determined according to preset influencing factors, and the predicted feeding data is predicted according to a preset feeding prediction model.

[0044] In one embodiment of the present application, the process before obtaining the target characteristic data of the target converter also includes the following steps: collecting production data corresponding to the target converter according to preset influencing factors to obtain initial characteristic data, wherein the initial characteristic data include molten iron weight, scrap steel weight, target element content in molten iron, heat generated when the target element is oxidized, and molten iron temperature; constructing initial characteristic variables based on the initial characteristic data, wherein the initial characteristic variables include chemical heat entering the furnace, physical heat entering the furnace, proportion of physical heat entering the furnace, and molten iron ratio, the chemical heat entering the furnace is calculated based on the molten iron weight, scrap steel weight, target element content in molten iron, and heat generated when the target element is oxidized, the physical heat entering the furnace is calculated based on the molten iron weight, scrap steel weight, and molten iron temperature, the proportion of physical heat entering the furnace is calculated based on the chemical heat entering the furnace and the physical heat entering the furnace, and the molten iron ratio is calculated based on the molten iron weight and the scrap steel weight; performing feature screening on the initial characteristic data and the initial characteristic variables to obtain target characteristic data.

[0045] In this embodiment, the key factors affecting the oxygen consumption of the converter are preliminarily screened by comprehensively considering multiple dimensions such as the conditions of the raw materials entering the furnace, the oxygen lance used in blowing, the use of the converter, the process charging conditions and the blowing target, so as to consider all factors that may affect the oxygen consumption of the converter as much as possible. Exemplarily, the conditions of the raw materials entering the furnace include the weight of molten iron, the temperature of molten iron, the weight of scrap steel and the carbon, silicon, manganese, phosphorus and sulfur content in the molten iron, the oxygen lance used in blowing covers the oxygen lance model, the throat diameter of the lance head and the number of times the lance head is used, the use of the converter includes the furnace age and the empty furnace time, the process charging conditions involve the types and weights of materials added during the blowing process such as lime blocks, limestone, raw dolomite, light-burned magnesium balls, coal blocks, coke, granular steel, etc., and the blowing target includes the target carbon content and the target molten steel temperature.

[0046] In one embodiment of the present application, based on the converter process principle, process mode changes and operator habits, the following new characteristic variables, namely initial characteristic variables, are creatively constructed, thereby strengthening the complex nonlinear relationship related to the converter oxygen consumption prediction. Among them, the expression of the chemical heat entering the furnace includes formula (1), the expression of the physical heat entering the furnace includes formula (2), the expression of the proportion of the physical heat entering the furnace includes formula (3), and the expression of the molten iron ratio includes formula (4):

[0047] Q chem =W iron ×(β Si ×ω[Si] ％ +β Mb ×ω[Mn] ％ +β C ×ω[C] ％ +β P ×ω[P] ％ +β S ×

[0048] ω[S] ％ ) / [100×(W iron +W scrap )] Formula (1)

[0049] Q physical =W iron ×[(T e -T r )×C iron_below +L iron +(T iron -T e )×

[0050] C iron_anove ] / (W iron +W scrap ) Formula (2)

[0051] Q ratio =Q physical / (Q physical +Q chem ) Formula (3)

[0052] Iron ratio =W iron / (W iron +W scrap ) Formula (4)

[0053] Among them, Q chem is the chemical heat of entering the furnace, W iron is the weight of molten iron, W scrap is the weight of scrap steel, ω[Si] ％ is the silicon content of molten iron, ω[Mn] ％ is the manganese content of molten iron, ω[C] ％ is the carbon content of molten iron, ω[P] ％ is the phosphorus content of molten iron, ω[S] ％ is the sulfur content of molten iron, β Si Silicon oxidation reaction exothermic heat, β Mn The heat released by the oxidation reaction of manganese, β C The heat released by the carbon oxidation reaction, β P The heat released by the phosphorus oxidation reaction, β S is the heat released by the sulfur oxidation reaction, Q physical is the physical heat of entering the furnace, T e is the eutectic temperature of iron, T iron is the molten iron temperature, T r is the ambient temperature, L iron is the latent heat of fusion of pig iron, C iron_below is the specific heat capacity of iron below the eutectic point, C iron_above is the specific heat capacity of iron above the eutectic point, Q ratio is the proportion of physical heat entering the furnace, Ironratio is the iron-water ratio.

[0054] In this embodiment, the heat generated by silicon oxidation is 29177, the heat generated by manganese oxidation is 12860, the heat generated by carbon oxidation is 6270, the heat generated by phosphorus oxidation is 11150, and the heat generated by sulfur oxidation is 5100. Substituting into formula (1) to obtain the furnace chemical heat Q chem =W iron ×(29177×ω[Si] ％ +12860×ω[Mn] ％ +6270×ω[C] ％ +11150×ω[P] ％ +5100×ω[S] ％ ) / [100×(W irn1 +W scrap )], where W iron , W scrap Represent the weight of molten iron and scrap steel respectively; ω[Si] ％ ,ω[Mn] ％ ,ω[C] ％ ,ω[P] ％ ,ω[S] ％ are the contents of the elements in the molten iron respectively; taking the eutectic temperature of iron as 1093°C, the molten iron temperature and the ambient temperature as 25°C, the specific heat capacity of iron below the eutectic point as 0.745, the specific heat capacity of iron above the eutectic point as 0.837, and the latent heat of fusion of pig iron as 217.56 kJ / kg as an example, substituting into formula (2) we can obtain the physical heat Q physical =W iron ×[(1093-25)×0.745+217.56+(T iron -1093)×0.837] / (W iron +W scrap ), where W iron , W scrap Represent the weight of molten iron and scrap steel respectively; the proportion of physical heat entering the furnace is Q ratio =Q physical / (Q physical +Q chem ), molten iron ratio ratio =W iron / (W iron +W scrap ). It can be understood that, in this embodiment, the values ​​of parameters such as the heat generated when the corresponding elements are oxidized, the eutectic temperature of iron, the molten iron temperature, the ambient temperature, the specific heat capacity of iron below the eutectic point, the specific heat capacity of iron above the eutectic point, and the latent heat of fusion of pig iron are all examples, and the present application does not limit this.

[0055] In this embodiment, in order to adapt to different conditions of raw materials entering the furnace and their impact on on-site process operations, this application uses a feature segmentation algorithm to classify the conditions of raw materials entering the furnace. Preferably, this embodiment uses a binning method based on a decision tree to classify the carbon, manganese, silicon, and phosphorus contents of molten iron, as well as the iron-to-metal ratio, molten iron temperature, and the total amount of charge.

[0056] In this embodiment, considering the habit and continuity of on-site operation, the oxygen consumption, oxygen supply time and oxygen flow rate of the previous furnace are introduced. Preferably, this embodiment incorporates the average of the oxygen supply time of the first two furnaces, the average of the oxygen consumption of the first two furnaces, and the average of the oxygen flow rate of the first two furnaces into the characteristic variables.

[0057] In one embodiment of the present application, a charging prediction model is established in advance based on the standard blowing information of the target converter. According to the standard raw material conditions and standard production operation process requirements of the target converter, a charging prediction model is established, wherein the standard converter raw material conditions include the standard percentage of each element, the standard temperature of the molten iron entering the furnace and the standard loading amount, and the standard production operation requirements include no splashing during the production process, ensuring production safety, improving production efficiency, and ensuring product quality. Based on the actual blowing information of the target converter, the charging situation of the charging prediction model is corrected. It should be noted that the establishment method and establishment process of the above-mentioned charging prediction model are only examples, and the charging prediction model can be established by other establishment methods or establishment processes according to actual needs, and the present application does not limit this. The charging prediction model is a prediction model for the amount of each slag added. The charging prediction model will calculate the predicted charging data of the target converter according to the blowing target and process charging situation of the target converter, such as the required amount of lime, dolomite, magnesium balls, cold materials, combustion aids, etc.

[0058] In one embodiment of the present application, data preprocessing is performed on initial feature data and initial feature variables, wherein the data preprocessing includes null value processing, outlier processing and data normalization; redundant features are eliminated from the initial feature data and initial feature variables according to a preset feature correlation algorithm to obtain initial features; feature analysis is performed on the initial features using a lightweight gradient boosting algorithm, and feature screening is performed from the initial features according to the analysis results to obtain target feature data.

[0059] In this embodiment, the process of null value processing includes filling and deleting null value data. If the material added weight data is null, it is assumed that the material is not put in and filled with 0. For the remaining variables, the furnace data containing null values ​​are removed to ensure data integrity. The process of outlier processing includes: using the interquartile range (IQR) method to process extreme outliers of all continuous numerical variables. According to the quartiles of the variables (Q1, Q2, Q3), any value beyond the specified range of [Q3-α1×IQR, Q1+α2×IQR] will be considered abnormal and removed, where IQR=Q3-Q1, α1 and α2 are thresholds determined based on the distribution characteristics of each variable, that is, the values ​​of α1 and α2 are determined according to the degree of deviation. For example, α1 and α2 are substituted into the range of each variable distribution to calculate the values ​​of α1 and α2. For example, the data distribution of the variable scrap steel weight is severely left-skewed, so the value of α1 is set to 3.2 and the value of α2 is set to 1.5; the data distribution of the variable molten iron weight is slightly right-skewed, so the value of α1 is set to 1.5 and the value of α2 is set to 2.0. It is understandable that the above values ​​of α1 and α2 are only for illustration, and the specific values ​​are determined according to the distribution characteristics of each variable, and this application does not limit this. The process of outlier processing also includes the exclusion of abnormal furnaces, excluding the furnace data in which unconventional operations occur during the production process, and any furnace that is interrupted during the blowing process or requires supplementary blowing operations will be considered as abnormal data and deleted, thereby improving the overall quality of the training set and the accuracy of model prediction. The process of data normalization includes performing a normality test on numerical variables and adopting a suitable data normalization method according to the test results. For example, a Shapiro-Wilk normality test is performed on all numerical variables. Then, considering that all variables do not follow a normal distribution, a minimum-maximum normalization method is used to preprocess the data. It should be noted that the use of the minimum-maximum normalization method to perform data normalization processing on the data in this embodiment is only an example. In actual applications, a suitable data normalization method is used to perform data normalization processing on the data according to the test results obtained by performing a normality test on the numerical variables.

[0060] In this embodiment, the initial feature data and initial feature variables are feature screened in the following ways to obtain the target feature data: 1) Feature correlation screening: Combine multiple feature correlation algorithms to preliminarily eliminate redundant features to capture the diverse dependencies between variables and ensure that features with significant correlations with the target variable are selected. In order to simplify the model structure and enhance the generalization ability, this embodiment uses two methods, mutual information regression and Spearman correlation method between variables, to preliminarily eliminate redundant features. 2) Model-driven feature optimization: This embodiment further identifies and excludes feature variables with lower influence based on the feature importance analysis (Feature Importance Score) output by the LightGBM model, thereby reducing the model complexity and preventing overfitting. In this way, in addition to improving the prediction efficiency of the model, it also ensures that the model only relies on features that make significant contributions to the prediction target. In the end, a total of 38 feature variables were selected to participate in model training.

[0061] In step S120, oxygen consumption is predicted for the target characteristic data and the predicted charging data by using a preset oxygen consumption prediction model to obtain a target oxygen consumption, so as to control the oxygen supply of the target converter according to the target oxygen consumption.

[0062] Among them, the oxygen consumption prediction model is trained based on training data samples with oxygen consumption labels.

[0063] In one embodiment of the present application, referring to Figure 2 , Figure 2 It is a flow chart of oxygen consumption prediction of a converter oxygen consumption prediction method shown in an exemplary embodiment of the present application. If the input data, i.e., the target characteristic data and the predicted charging data, meet the model calculation conditions, the model calculation conditions include that the target characteristic data and the predicted charging data are not null values, and meet the aforementioned range of [Q3-α1×IQR, Q1+α2×IQR], the model will make the first oxygen consumption prediction at the beginning of the main blowing. The initial prediction value is calculated based on the real-time characteristic data collected at the beginning of the main blowing and the charging demand estimated by the charging model, aiming to provide accurate oxygen consumption guidance for the operator. In view of the fact that irregular events such as slag overflow, splashing or back drying may force the operator to add more than the pre-calculated amount of charging during the converter blowing process, the oxygen consumption prediction model of the present application provides a dynamic prediction mechanism. After a preset interval time (e.g., five minutes) from the start of blowing, as long as the charging operation is performed, a model prediction is performed based on the current actual cumulative charging amount to continuously correct the oxygen consumption prediction value. This dynamic calculation method not only enhances the accuracy of oxygen consumption prediction, but also significantly improves the model's ability to respond to the ever-changing environment of the factory production site.

[0064] In one embodiment of the present application, sample training data and sample oxygen consumption corresponding to the sample training data are obtained, wherein the sample training data includes sample characteristic data and sample feeding data; the sample training data are labeled according to the sample oxygen consumption to generate training data samples with oxygen consumption labels; a preset oxygen consumption prediction model is trained based on the training data samples with oxygen consumption labels to obtain the oxygen consumption prediction model, wherein a hyperparameter combination of the preset oxygen consumption prediction model is determined by a preset optimization algorithm.

[0065] In this embodiment, historical data is obtained, and the historical data includes historical feature data and historical feeding data. The historical data is randomly divided into a training set and a test set, wherein the training set accounts for 85% of the total data volume, and the test set accounts for 15% of the total data volume. The training set includes sample training data and sample oxygen consumption corresponding to the sample training data, and the test set includes sample test data and test oxygen consumption corresponding to the sample test data. Data preprocessing is performed on the acquired historical data, wherein the data preprocessing includes null value processing, outlier processing and data normalization. The specific processing process has been specifically described in the aforementioned embodiment and will not be repeated here.

[0066] In this embodiment, the sample training data is labeled according to the sample oxygen consumption to generate training data samples with oxygen consumption labels. It can be understood that each sample training data has its corresponding sample oxygen consumption, and each sample training data is labeled according to the sample oxygen consumption corresponding to the sample training data to generate training data samples with oxygen consumption labels.

[0067] In this embodiment, the preset oxygen consumption prediction model is obtained by training the lightweight gradient boosting algorithm, and the preset optimization algorithm is the Bayesian optimization algorithm as an example. A hyperparameter optimization algorithm is used to select the hyperparameter combination that can reduce the objective function to the greatest extent. In this embodiment, the Bayesian optimization algorithm is used as the main learner-LightGBM regressor to find the optimal parameter combination. First, a multidimensional hyperparameter search space is defined as the range for the Bayesian optimization algorithm to find the optimal parameter combination. In this search space, Bayesian optimization uses the Gaussian process model to establish a mapping relationship between hyperparameters and objective function values ​​using the negative mean square error as the objective function. Through continuous iterative steps, the Gaussian process model is continuously optimized to obtain a more accurate estimate of the probability distribution of the objective function, and then determine the hyperparameter combination that can reduce the objective function to the greatest extent. The above-mentioned hyperparameters include but are not limited to the number of leaf nodes, the maximum depth of the tree, the minimum number of samples of leaf nodes, the proportion of data used for training at each iteration, and the proportion of features used for training at each iteration. Determining the hyperparameter combination of the preset oxygen consumption prediction model through a preset optimization algorithm can improve the accuracy of model prediction, help avoid overfitting or underfitting problems in the model, and improve the stability of the oxygen consumption prediction model.

[0068] In one embodiment of the present application, sample test data and test oxygen consumption corresponding to the sample test data are obtained, wherein the sample test data includes test characteristic data and test feeding data; the sample test data are input into an oxygen consumption prediction model to obtain predicted oxygen consumption, and a prediction residual is calculated based on the predicted oxygen consumption and the test oxygen consumption; a sample data set is generated based on the sample test data and the prediction residual, and a residual correction prediction model is obtained by training the sample data set; the residual of the sample test data is predicted by the residual correction prediction model to obtain a residual correction value, and the predicted oxygen consumption is corrected based on the residual correction value, so as to optimize the oxygen consumption prediction model based on the corrected predicted oxygen consumption.

[0069] In this embodiment, the sample test data and the test oxygen consumption corresponding to the sample test data can be understood as each sample test data has its corresponding test oxygen consumption, so that each sample test data is tested according to the test oxygen consumption corresponding to the sample test data. The residual correction prediction model can be obtained by training with a nearest neighbor algorithm and a sample data set. It can be understood that the nearest neighbor algorithm is a preferred algorithm. In practical applications, the residual correction prediction model can also be obtained by training with other algorithms and sample data sets. This application does not limit this, nor should it bring any limitations to the functions and scope of use of the embodiments of this application.

[0070] In this embodiment, the LightGBM algorithm is used as the main learner to generate preliminary prediction results, and then the KNN algorithm is introduced as an auxiliary learner to precisely correct the prediction residuals that the main learner cannot fully explain. The above-mentioned integrated processing not only effectively utilizes the fast and efficient data processing capabilities of the LightGBM algorithm, but also can use the sensitivity of the KNN algorithm in capturing subtle data changes to improve the adaptability and prediction accuracy of the model for diversified data scenarios. This application adopts an integrated machine learning algorithm to improve the accuracy, flexibility, stability and computational efficiency of the model by integrating the unique advantages of two different algorithms. First, an algorithm is used as the main learner to generate preliminary prediction results. Then the residual generated by it is used as input and applied to an auxiliary learner based on another algorithm to further correct the prediction error. The unique advantages of multiple algorithms are effectively integrated, and the accuracy, flexibility, stability and computational efficiency of the model are significantly improved. In addition, through meticulous feature selection and hyperparameter optimization, combined with systematic evaluation measures, the model performance fluctuations caused by randomness are effectively reduced, ensuring the high stability and accuracy of the model in the actual production environment. Figure 3 is a flow chart of an integrated machine learning method for predicting converter oxygen consumption according to an exemplary embodiment of the present application, referring to Figure 3As shown, the integrated machine learning method mainly includes three steps. The first step is to train and optimize the main learner, including four steps: feature correlation screening, model-driven feature optimization, finding the optimal hyperparameter combination and eliminating model randomness. The second step is to calculate the residual generated by the calculation result of the main learner. The third step is to use the auxiliary learner to fit the residual. Each specific step is described in detail in other embodiments and will not be repeated here.

[0071] In one embodiment of the present application, the mean square error of oxygen consumption and the coefficient of determination of the test set are calculated based on the corrected predicted oxygen consumption and the test oxygen consumption; the sample training data is input into the oxygen consumption prediction model to obtain the reference oxygen consumption, and the coefficient of determination of the training set is calculated based on the reference oxygen consumption and the sample oxygen consumption; the performance of the oxygen consumption prediction model is evaluated based on at least one of the mean square error of oxygen consumption, the coefficient of determination of the test set and the coefficient of determination of the training set.

[0072] In this embodiment, the randomness in the model training and evaluation process is fully considered, and the model randomness is eliminated by the following methods, such as data splitting, regularization selection, Bayesian optimization process and cross-validation. In order to enhance the robustness of model parameter selection and reduce excessive sensitivity to data segmentation methods, a systematic method for comparing model performance under different random states is proposed. Model selection is based on the following criteria: (1) The determination coefficient R of the test set is preferentially selected. 2 The one with the highest score; (2) the coefficient of determination R in the test set 2 If the scores are the same, the one with lower mean square error (MSE) is preferred; (3) If the coefficient of determination R of the test set is 2 If the mean square error (MSE) is the same as the training set’s determination coefficient (R), then 2 Score.

[0073] It can be seen from the above embodiments that the present application can adjust the performance of the LightGBM algorithm to the best. When the model training is completed, the selected hyperparameters and random states are applied to the model to generate predictions and calculate residuals. These residuals are then refitted as inputs to the auxiliary learner KNN to fine-tune the final prediction results. Through these optimization strategies, the present invention significantly improves the practicality and accuracy of the converter oxygen consumption prediction model in an actual production environment. Table 1 compares the prediction results corresponding to the experimental data of the present application with the prediction results of the comparative invention. Through the comparison, it can be clearly seen that the present application has achieved a significant improvement in the accuracy of the converter oxygen consumption prediction. In addition, Figure 4 As shown, Figure 4 It is a schematic diagram of the comparison results of the actual oxygen consumption and the predicted oxygen consumption of the converter oxygen consumption prediction method shown in an exemplary embodiment of the present application. The oxygen consumption prediction model proposed in the present invention has an excellent degree of fitting and can accurately capture the change characteristics of most data.

[0074] Table 1

[0075]

[0076] It should be noted that this embodiment is only for illustration, and the data in Table 1 (including error percentage, furnace times, proportion, etc.) can be increased, decreased or changed according to actual needs, and its values ​​can also be modified according to actual needs, and this application does not limit this.

[0077] Figure 5 is a flow chart of a method for predicting converter oxygen consumption shown in another exemplary embodiment of the present application, referring to Figure 5 As shown, the present application mainly includes four steps, namely, feature initial screening and construction, data preprocessing, model training and optimization, and model dynamic prediction. The specific process has been described in detail in the above embodiments and will not be repeated here.

[0078] The present application is not limited to the single consideration of the converter raw material feeding conditions as the model input feature, but integrates the target feature data corresponding to multiple related factors such as process principle related variables, the specific usage status of the converter and the oxygen gun, the actual charging process, the operating habits of the operators and the production mode, aiming to capture the key features affecting the converter oxygen consumption in all directions, and to ensure the high accuracy of the obtained prediction results and enhance the stability of the model by improving the multi-dimensional input of the model, thereby breaking through the limitations of the traditional model and significantly optimizing the oxygen consumption prediction; the present application adopts an integrated learning method, which effectively improves the accuracy, flexibility, stability and computational efficiency of the model by integrating the unique advantages of a variety of different algorithms; the present application also proposes A dynamic prediction mechanism is provided, which enables the oxygen consumption prediction model to continuously adjust the oxygen consumption prediction value according to the changes in the blowing process, so as to enhance the accuracy of the prediction result and the adaptability of the model to the changes in the production process; the oxygen consumption prediction model in the present application provides an important reference for the oxygen distribution in each stage of the blowing process and the timing of the gun pressing in the later stage of blowing, which can effectively reduce the number of oxygen supplementary blowing and shorten the production cycle, thereby improving the overall production efficiency, effectively reducing the oxidation loss of alloy elements caused by excessive oxygen supply, reducing the alloy compensation cost, and being able to accurately predict the oxygen consumption according to the set target carbon content and temperature requirements of the molten steel, thereby ensuring high-quality endpoint control, and helping to reduce oxide and gas emissions, so as to benefit environmental protection.

[0079] Figure 6 is a block diagram of a converter oxygen consumption prediction device shown in an exemplary embodiment of the present application. The device can be applied to Figure 1 The implementation environment shown in the figure is specifically configured in the computer device 102. The device can also be applied to other exemplary implementation environments and specifically configured in other devices. This embodiment does not limit the implementation environment to which the device is applied.

[0080] like Figure 6 As shown, the exemplary converter oxygen consumption prediction device includes: an acquisition module 610 and a prediction module 620. The acquisition module 610 is used to acquire target characteristic data and predicted charging data of a target converter, wherein the target characteristic data is determined according to preset influencing factors, and the predicted charging data is predicted according to a preset charging prediction model; the prediction module 620 is used to predict oxygen consumption of the target characteristic data and the predicted charging data through a preset oxygen consumption prediction model to obtain a target oxygen consumption, so as to control the oxygen supply of the target converter according to the target oxygen consumption, wherein the oxygen consumption prediction model is trained according to training data samples with oxygen consumption labels.

[0081] It should be noted that the converter oxygen consumption prediction device provided in the above embodiment and the converter oxygen consumption prediction method provided in the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs the operation has been described in detail in the method embodiment, and will not be repeated here. In actual application, the converter oxygen consumption prediction device provided in the above embodiment can allocate the above functions to different functional modules as needed, that is, divide the internal structure of the device into different functional modules to complete all or part of the functions described above, and this is not limited here.

[0082] An embodiment of the present application also provides an electronic device, comprising: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by one or more processors, the electronic device implements the converter oxygen consumption prediction method provided in the above-mentioned embodiments.

[0083] Figure 7 The structure diagram of the computer system suitable for implementing the electronic device of the embodiment of the present application is shown. It should be noted that: Figure 7 The computer system 700 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0084] like Figure 7As shown, the computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 702 or the program loaded from the storage part 708 to the random access memory (RAM) 703, such as executing the methods provided in the above-mentioned various embodiments. In the RAM 703, various programs and data required for system operation are also stored. The CPU 701, the ROM 702 and the RAM 703 are connected to each other through the bus 704. The input / output (I / O) interface 705 is also connected to the bus 704.

[0085] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, a mouse, etc.; an output section 707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 710 as needed so that a computer program read therefrom is installed into the storage section 708 as needed.

[0086] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication section 709, and / or installed from a removable medium 711. When the computer program is executed by a central processing unit (CPU) 701, various functions defined in the system of the present application are executed.

[0087] It should be noted that the computer-readable medium shown in the embodiment of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a 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. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a computer-readable computer program is carried. This propagated data signal can take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. A computer program contained on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0088] The flowchart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operations of the system, method and computer program product according to various embodiments of the present application. Wherein, each box in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0089] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. The names of these units do not, in some cases, constitute limitations on the units themselves.

[0090] Another aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor of a computer, causes the computer to execute the converter oxygen consumption prediction method provided in the above-mentioned embodiments. The computer-readable storage medium may be included in the electronic device described in the above-mentioned embodiments, or may exist independently without being assembled into the electronic device.

[0091] It should be noted that, although several modules or units of the equipment for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into being embodied by multiple modules or units.

[0092] Another aspect of the present application also provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the converter oxygen consumption prediction method provided in each of the above embodiments.

[0093] Through the description of the above implementation methods, it is easy for those skilled in the art to understand that the example implementation methods described here can be implemented by software or by combining software with necessary hardware. Therefore, the technical solution according to the implementation methods of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a touch terminal, or a network device, etc.) to execute the method according to the implementation methods of the present application.

[0094] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the embodiments disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application.

[0095] The above embodiments are merely illustrative of the principles and effects of the present application, and are not intended to limit the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.

Claims

1. A method for predicting converter oxygen consumption, characterized in that: include: Obtaining target characteristic data and predicted charging data of a target converter, wherein the target characteristic data is determined according to preset influencing factors, and the predicted charging data is predicted according to a preset charging prediction model; The target characteristic data and the predicted charging data are subjected to oxygen consumption prediction using a preset oxygen consumption prediction model to obtain a target oxygen consumption, so as to control the oxygen supply of the target converter according to the target oxygen consumption, wherein the oxygen consumption prediction model is trained based on training data samples with oxygen consumption labels.

2. The method for predicting converter oxygen consumption according to claim 1, characterized in that: Before acquiring the target characteristic data of the target converter, the method further includes: According to the preset influencing factors, the production data corresponding to the target converter is collected to obtain initial characteristic data, wherein the initial characteristic data includes the weight of molten iron, the weight of scrap steel, the content of the target element in the molten iron, the heat generated when the target element is oxidized, and the temperature of the molten iron; Constructing initial characteristic variables based on the initial characteristic data, wherein the initial characteristic variables include chemical heat entering the furnace, physical heat entering the furnace, proportion of physical heat entering the furnace and molten iron ratio, the chemical heat entering the furnace is calculated according to the weight of the molten iron, the weight of the scrap steel, the content of the target element in the molten iron and the heat generated when the target element is oxidized, the physical heat entering the furnace is calculated according to the weight of the molten iron, the weight of the scrap steel and the molten iron temperature, the proportion of physical heat entering the furnace is calculated according to the chemical heat entering the furnace and the physical heat entering the furnace, and the molten iron ratio is calculated according to the weight of the molten iron and the weight of the scrap steel; Feature screening is performed on the initial feature data and the initial feature variables to obtain the target feature data.

3. The method for predicting converter oxygen consumption according to claim 2, characterized in that: The expression of the chemical heat entering the furnace includes: Q chem =W iron ×(β Si ×ω[Si] ％ +β Mn ×ω[Mn] ％ +β C ×ω[C] ％ +β P ×ω[P] ％ +β S ×ω[S] ％ ) / [100×(W iron +W scrap )]; The physical heat entering the furnace is expressed in the following ways: Q physical =W iron ×[(T e -T r )×C iron_below +L iron +(T iron -T e )×C iron_above ] / (W iron +W scrap ); The representation of the proportion of physical heat entering the furnace includes: Q ratio =Q physical / (Q physical +Q chem ); The expression of the iron-water ratio includes: ratio =W iron / (W iron +W scrap ); Among them, Q chem is the chemical heat of entering the furnace, W iron is the weight of the molten iron, W scrap is the weight of the scrap steel, ω[Si] ％ is the silicon content of molten iron, ω[Mn] ％ is the manganese content of molten iron, ω[C] ％ is the carbon content of molten iron, ω[P] ％ is the phosphorus content of molten iron, ω[S] ％ is the sulfur content of molten iron, β Si Silicon oxidation reaction exothermic heat, β Mn The heat released by the oxidation reaction of manganese, β C The heat released by the carbon oxidation reaction, β P The heat released by the phosphorus oxidation reaction, β S is the heat released by the sulfur oxidation reaction, Q physical is the physical heat of entering the furnace, T e is the eutectic temperature of iron, T iron is the molten iron temperature, T r is the ambient temperature, L iron is the latent heat of fusion of pig iron, C iron_below is the specific heat capacity of iron below the eutectic point, C iron_above is the specific heat capacity of iron above the eutectic point, Q ratio is the physical heat ratio of the furnace, Iron ratio is the iron-water ratio.

4. The method for predicting converter oxygen consumption according to claim 2, characterized in that: Performing feature screening on the initial feature data and the initial feature variables to obtain the target feature data includes: Performing data preprocessing on the initial feature data and the initial feature variables, wherein the data preprocessing includes null value processing, outlier processing and data normalization; Eliminating redundant features from the initial feature data and the initial feature variables according to a preset feature correlation algorithm to obtain initial features; The initial features are subjected to feature analysis by using a lightweight gradient boosting algorithm, and feature screening is performed from the initial features according to the analysis results to obtain the target feature data.

5. The method for predicting converter oxygen consumption according to claim 1, characterized in that: The oxygen consumption prediction model is trained based on training data samples with oxygen consumption labels, including: Acquire sample training data and sample oxygen consumption corresponding to the sample training data, wherein the sample training data includes sample characteristic data and sample feeding data; Labeling the sample training data according to the sample oxygen consumption to generate training data samples with oxygen consumption labels; The preset oxygen consumption prediction model is trained based on the training data samples with oxygen consumption labels to obtain the oxygen consumption prediction model, wherein a hyperparameter combination of the preset oxygen consumption prediction model is determined by a preset optimization algorithm.

6. The method for predicting converter oxygen consumption according to claim 5, characterized in that: After training a preset oxygen consumption prediction model based on the training data samples with oxygen consumption labels to obtain the oxygen consumption prediction model, the method further includes: Acquire sample test data and test oxygen consumption corresponding to the sample test data, wherein the sample test data includes test characteristic data and test feed data; Inputting the sample test data into the oxygen consumption prediction model to obtain predicted oxygen consumption, and calculating the prediction residual according to the predicted oxygen consumption and the test oxygen consumption; Generate a sample data set according to the sample test data and the prediction residual, and obtain a residual correction prediction model through training with the sample data set; The residual of the sample test data is predicted by the residual correction prediction model to obtain a residual correction value, and the predicted oxygen consumption is corrected according to the residual correction value, so as to optimize the oxygen consumption prediction model according to the corrected predicted oxygen consumption.

7. The method for predicting converter oxygen consumption according to claim 6, characterized in that: After correcting the predicted oxygen consumption according to the residual correction value, the method further includes: Calculate the mean square error of oxygen consumption and the coefficient of determination of the test set according to the corrected predicted oxygen consumption and the tested oxygen consumption; Inputting the sample training data into the oxygen consumption prediction model to obtain a reference oxygen consumption, and calculating a determination coefficient of a training set based on the reference oxygen consumption and the sample oxygen consumption; The performance of the oxygen consumption prediction model is evaluated according to at least one of the mean square error of the oxygen consumption, the coefficient of determination of the test set, and the coefficient of determination of the training set.

8. A converter oxygen consumption prediction device, characterized in that: include: An acquisition module, used to acquire target characteristic data and predicted charging data of a target converter, wherein the target characteristic data is determined according to preset influencing factors, and the predicted charging data is predicted according to a preset charging prediction model; A prediction module is used to predict the oxygen consumption of the target characteristic data and the predicted charging data through a preset oxygen consumption prediction model to obtain a target oxygen consumption, so as to control the oxygen supply of the target converter according to the target oxygen consumption, wherein the oxygen consumption prediction model is trained based on training data samples with oxygen consumption labels.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; A storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, enables the electronic device to implement the converter oxygen consumption prediction method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor of a computer, the computer is caused to execute the method for predicting the converter oxygen consumption according to any one of claims 1 to 7.

Images