A method and system for predicting steelmaking alloy yield based on alloy pulverization

Abstract

The present invention relates to the technical field of iron and steel metallurgy, and discloses a steelmaking alloy yield prediction method and system based on alloy pulverization, the method comprising obtaining alloy composition data, steelmaking process parameters and alloy particle size distribution data; calculating the binary alloy pulverization rate according to the alloy composition data and the alloy particle size distribution data in combination with a binary alloy pulverization rate formula; performing alloy pulverization degree analysis according to the binary alloy pulverization rate to obtain an alloy pulverization degree factor; inputting the alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters into a pre-trained alloy yield prediction model, and outputting a predicted alloy yield. The method can improve the accuracy of alloy yield prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of output prediction, and in particular to a method and system for predicting steelmaking alloy yield based on alloy pulverization. Background Art

[0002] At present, in order to improve the performance of steel products, especially the material strength, it is necessary to add various alloying elements and mineral materials, such as silicon, manganese, chromium, nickel, titanium or niobium, or various refining agents, deoxidizers or alloy additives, during the smelting process. In the process of converter steelmaking, the amount of alloy used is accurately calculated based on the alloy composition, steel type and the yield of alloying elements. However, in actual steelmaking operations, the yield of alloying elements is a parameter that is difficult to accurately determine in advance. This parameter directly affects the weight of the added alloy material, and then affects the quality of the molten steel after the alloying operation. Despite this, most steel companies currently rely on long-term accumulated actual production data to estimate the alloy yield, but this method does not fully consider the impact of changes in process conditions between different furnaces on the alloy yield, resulting in low accuracy of alloy dosage. In recent years, with the rapid development of computer technology and information technology, traditional steelmaking processes have the conditions for building intelligent control systems. In the alloying link of converter steelmaking, operators hope to obtain more accurate alloying element yield data for the current furnace in order to reasonably calculate the amount of alloy added. Doing so can not only reduce the waste of alloy raw materials and excessive composition caused by inaccurate estimates, but also further improve the quality of molten steel. However, the mechanism of alloy pulverization is quite complex, especially when multiple elements interact with each other, and its influence on alloy yield is still under research. In actual production, engineers can only rely on rough estimates of the degree of alloy element pulverization and empirical models to grasp the alloy yield under different ratios, so as to adjust the amount of alloy added and ensure the accuracy of alloy composition.

[0003] In one prior art, a large amount of production data accumulated during long-term operation is used to establish a prediction model for alloy element yield. Before building such a model, the historical data must first be preprocessed, including screening and cleaning, to remove invalid or erroneous information. Since the historical data contains a large number of process parameters, not all factors will affect the yield of alloy elements. Therefore, correlation analysis should be used to select key parameters that have a significant impact on the yield for the construction of the prediction model.

[0004] However, the construction of the prediction model in the prior art only considers the influence of production parameters on the alloy yield, and does not consider the influence of each doping element in the multi-component alloy, resulting in low accuracy in the prediction of the steelmaking alloy yield. Summary of the invention

[0005] The present invention provides a steelmaking alloy yield prediction method and system based on alloy pulverization to achieve accurate prediction of the steelmaking alloy yield.

[0006] In a first aspect, in order to solve the above technical problems, the present invention provides a steelmaking alloy yield prediction method based on alloy pulverization, comprising:

[0007] Obtain alloy composition data, steelmaking process parameters and alloy particle size distribution data;

[0008] According to the alloy composition data and the alloy particle size distribution data, combined with the binary alloy pulverization rate formula, the binary alloy pulverization rate is calculated;

[0009] Performing alloy pulverization degree analysis according to the binary alloy pulverization rate to obtain an alloy pulverization degree factor;

[0010] The alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters are input into a pre-trained alloy yield prediction model, and the predicted alloy yield is output.

[0011] In an optional embodiment, the binary alloy pulverization rate is obtained by calculating the alloy composition data and the alloy particle size distribution data in combination with a binary alloy pulverization rate formula, including:

[0012] The powdering rate of binary alloy is calculated by the following formula:

[0013]

[0014] in, is the powdering rate of binary alloy, is the humidity factor, is the alloy element particle size, is the temperature, is an empirical constant, is a natural constant, is the activation energy, is the gas constant.

[0015] In an optional embodiment, the alloy pulverization degree analysis is performed according to the binary alloy pulverization rate to obtain the alloy pulverization degree factor, including:

[0016] The alloy powdering degree factor is calculated by the following formula:

[0017]

[0018] in, is the alloy pulverization degree factor, is the total number of alloying elements, fori The weight of alloying elements, for i The powderization rate of binary alloys of alloying elements, i The alloy element number.

[0019] In an optional embodiment, the training process of the alloy yield prediction model includes:

[0020] Based on the historical alloy pulverization degree factor, historical alloy composition data, historical steelmaking process parameters and historical alloy yield rate, an alloy yield rate prediction model is constructed, the model is trained, and the training is determined to be complete after the loss function of the model meets the conditions, thereby obtaining a trained model;

[0021] The alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters are input into the trained model to obtain a predicted alloy yield.

[0022] In an optional embodiment, the acquisition of the alloy particle size distribution data includes:

[0023] Obtaining particle settling velocity, particle density, dispersion medium viscosity and dispersion medium density;

[0024] The alloying element particle size is calculated by the following formula:

[0025]

[0026] in, is the alloy element particle size, is the particle settling velocity, is the dispersion medium viscosity, is the particle density, is the density of the dispersion medium, is the acceleration due to gravity;

[0027] Arranging the alloy element particle sizes from large to small to obtain a particle size screening sequence;

[0028] Get the number of alloy element particles;

[0029] The weight percentage is calculated by dividing the number of particles of each alloy element by the total number of particles;

[0030] The alloy particle size distribution data includes the weight percentage and the particle size screening sequence.

[0031] In an optional embodiment, the alloy composition data includes alloy element types and alloy element mass ratios;

[0032] The steelmaking process parameters include steel type, heating temperature and smelting system.

[0033] In an optional embodiment, the calculation of the humidity factor includes:

[0034] Get relative humidity;

[0035] The humidity factor is calculated using the following formula:

[0036]

[0037] in, is the humidity factor, is the base rate constant, is the relative humidity, It is the humidity sensitivity index.

[0038] In a second aspect, the present invention provides a steelmaking alloy yield prediction system based on alloy pulverization, comprising:

[0039] Data acquisition module, used to obtain alloy composition data, steelmaking process parameters and alloy particle size distribution data;

[0040] A binary pulverization module, for calculating the binary alloy pulverization rate according to the alloy composition data and the alloy particle size distribution data in combination with a binary alloy pulverization rate formula;

[0041] A multi-element pulverization module, used to analyze the alloy pulverization degree according to the binary alloy pulverization rate to obtain an alloy pulverization degree factor;

[0042] The result prediction module is used to input the alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters into a pre-trained alloy yield prediction model, and output a predicted alloy yield.

[0043] In a third aspect, the present invention further provides an electronic device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the steelmaking alloy yield prediction method based on alloy pulverization as described above is implemented.

[0044] In a fourth aspect, the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned steelmaking alloy yield prediction methods based on alloy pulverization.

[0045] Compared with the prior art, the present invention has the following beneficial effects: The present invention proposes a steelmaking alloy yield prediction method and system based on alloy pulverization, which aims to improve the accuracy of alloy yield prediction by comprehensively considering alloy composition data, steelmaking process parameters and alloy particle size distribution data. First, the alloy composition, specific operating parameters in the steelmaking process and alloy particle size distribution are obtained. Then, the alloy pulverization degree is calculated using the binary alloy pulverization rate formula. This step helps to more accurately understand the behavior of the alloy in the steelmaking process, especially how the pulverization phenomenon affects the quality of the final product. Then, based on the binary alloy pulverization rate, the alloy pulverization degree factor is calculated by weighted summation, which comprehensively reflects the pulverization degree under the combined action of multiple alloy elements. This process not only takes into account the characteristics of individual alloy elements, but also their proportions in the alloy mixture, so as to more accurately reflect the actual situation. Finally, the alloy pulverization degree factor, alloy composition data and steelmaking process parameters are used as inputs, and a more accurate alloy yield prediction value can be obtained through a pre-trained prediction model. The beneficial effect of this method is that it not only considers the two traditional factors of alloy composition and steelmaking process, but also introduces two new variables, alloy particle size distribution and degree of pulverization, thereby improving the accuracy and reliability of the prediction model. Through in-depth analysis of the degree of alloy pulverization, the amount of alloy added in the steelmaking process can be better controlled to improve production efficiency. In addition, using a pre-trained model for prediction can provide more accurate prediction results compared to traditional empirical rules or simple statistical methods.

[0046] Furthermore, this method also provides a detailed description of the acquisition of alloy particle size distribution data, including the calculation of alloy element particle size through parameters such as particle settling velocity, particle density, dispersion medium viscosity and density, and the particle size screening sequence and weight percentage are obtained accordingly. In addition, the calculation of humidity factor is also an important part of this method. The value of humidity factor is adjusted by relative humidity to more accurately simulate the humidity effect in the actual steelmaking environment.

[0047] In conclusion, the present invention makes predictions based on the alloy pulverization rate, thereby improving the accuracy of the predicted value of the steelmaking alloy yield. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a schematic flow chart of a method for predicting alloy yield of steelmaking based on alloy pulverization provided by the first embodiment of the present invention;

[0049] Figure 2 It is a schematic diagram of the structure of a steelmaking alloy yield prediction system based on alloy pulverization provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] At present, in order to improve the performance of steel products, especially the material strength, it is necessary to add various alloying elements and mineral materials, such as silicon, manganese, chromium, nickel, titanium, niobium, etc., or various refining agents, deoxidizers, alloy additives, etc., during the smelting process. In the process of converter steelmaking, the amount of alloy used is accurately calculated based on the alloy composition, steel type and the yield of alloying elements. However, in actual steelmaking operations, the yield of alloying elements is a parameter that is difficult to accurately determine in advance. This parameter directly affects the weight of the added alloy material, and then affects the quality of the molten steel after the alloying operation. Despite this, most steel companies currently rely on long-term accumulated actual production data to estimate the alloy yield, but this method does not fully consider the impact of changes in process conditions between different furnaces on the alloy yield, resulting in low accuracy of alloy dosage. In recent years, with the rapid development of computer technology and information technology, traditional steelmaking processes have the conditions for building intelligent control systems. In the alloying link of converter steelmaking, operators hope to obtain more accurate alloying element yield data for the current furnace in order to reasonably calculate the amount of alloy added. Doing so can not only reduce the waste of alloy raw materials and excessive composition caused by inaccurate estimates, but also further improve the quality of molten steel. However, the mechanism of alloy pulverization is quite complex, especially when multiple elements interact with each other, and its influence on alloy yield is still under research. In actual production, engineers can only rely on rough estimates of the degree of alloy element pulverization and empirical models to grasp the alloy yield under different ratios, so as to adjust the amount of alloy added and ensure the accuracy of alloy composition.

[0052] In one prior art, a large amount of production data accumulated during long-term operation is used to establish a prediction model for alloy element yield. Before building such a model, the historical data must first be preprocessed, including screening and cleaning, to remove invalid or erroneous information. Since the historical data contains a large number of process parameters, not all factors will affect the yield of alloy elements. Therefore, correlation analysis should be used to select key parameters that have a significant impact on the yield for the construction of the prediction model.

[0053] However, the construction of the prediction model in the prior art only considers the influence of production parameters on the alloy yield, and does not consider the influence of each doping element in the multi-component alloy, resulting in low accuracy in the prediction of the steelmaking alloy yield.

[0054] To solve the above problems, refer to Figure 1 The first embodiment of the present invention provides a steelmaking alloy yield prediction method based on alloy pulverization, comprising the following steps:

[0055] S11, obtaining alloy composition data, steelmaking process parameters and alloy particle size distribution data;

[0056] S12, calculating the binary alloy pulverization rate according to the alloy composition data and the alloy particle size distribution data in combination with a binary alloy pulverization rate formula;

[0057] S13, analyzing the alloy pulverization degree according to the binary alloy pulverization rate to obtain an alloy pulverization degree factor;

[0058] S14, inputting the alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters into a pre-trained alloy yield prediction model, and outputting a predicted alloy yield.

[0059] In step S11, alloy composition data, steelmaking process parameters and alloy particle size distribution data are obtained.

[0060] It is worth noting that the alloy particle size distribution data includes the weight percentage and the particle size screening sequence. The alloy composition data includes the type of alloy elements and the mass ratio of alloy elements; the steelmaking process parameters include steel type, heating temperature and smelting system. The alloy composition data refers to the types and mass ratios of various elements constituting the alloy. Spectroscopic analysis methods (such as X-ray fluorescence spectrometer XRF, atomic absorption spectrometer AAS, etc.) are used in industry to determine the alloy composition. The steelmaking process parameters mainly include steel type, heating temperature and smelting system, etc. Different steel types require different smelting conditions, while heating temperature and smelting system can reduce energy consumption, improve productivity and ensure the quality of steel. These parameters are set according to the production plan, and the actual data are monitored and adjusted by a computer control system. For example, the heating temperature can be monitored in real time by a temperature sensor such as a thermocouple. The alloy particle size distribution data reflects the size of the alloy element particles and their distribution. The particle size distribution has an important influence on the processing properties (such as fluidity, compaction density, etc.) and final performance of the alloy.

[0061] In one embodiment, the acquisition of the alloy particle size distribution data includes: acquiring particle settling velocity, particle density, dispersion medium viscosity and dispersion medium density;

[0062] The alloying element particle size is calculated by the following formula:

[0063]

[0064] in, is the alloy element particle size, is the particle settling velocity, is the dispersion medium viscosity, is the particle density, is the density of the dispersion medium, is the acceleration due to gravity;

[0065] The alloy element particle sizes are sorted from large to small to obtain a particle size screening sequence; the number of alloy element particles is obtained; and the weight percentage is calculated by dividing the number of each alloy element particle by the total number of particles.

[0066] It is worth mentioning that the particle settling velocity is obtained through experimental measurement. The viscosity of the dispersion medium is measured using a viscometer. The particle density is determined by a density bottle or a specific gravity bottle. The density of the dispersion medium can also be measured by a density bottle. The acceleration of gravity is a known constant. The particle size distribution of the alloy particles is directly measured by equipment such as a laser scattering particle size analyzer. After obtaining the above data, the particle sizes of the alloy elements are sorted from large to small to form a particle size sieve sequence. Calculate the proportion of the number of particles in each particle size range to the total number of particles, that is, the weight percentage. The particle size distribution data will be presented in the form of weight percentage and particle size sieve sequence for subsequent analysis.

[0067] In step S12, the binary alloy pulverization rate is calculated based on the alloy composition data and the alloy particle size distribution data in combination with the binary alloy pulverization rate formula to obtain the binary alloy pulverization rate.

[0068] In one embodiment, the binary alloy pulverization rate is calculated by the following formula:

[0069]

[0070] in, is the powdering rate of binary alloy, is the humidity factor, is the alloy element particle size, is the temperature, is an empirical constant, is a natural constant, is the activation energy, is the gas constant.

[0071] It is worth noting that the pulverization rate refers to the degree to which an alloy decomposes into fine particles under specific conditions, and is an important indicator for evaluating the stability and service life of an alloy. A binary alloy refers to an alloy doped with one element. The humidity factor reflects the effect of environmental humidity on alloy pulverization. In a high humidity environment, the alloy is more susceptible to oxidation and corrosion, which increases the pulverization rate. The smaller the particle size of the alloying element, the larger the surface area, and the higher the pulverization tendency. Increased temperature accelerates the chemical reaction rate, thereby increasing the pulverization rate. The empirical constant is taken as 1.14, which is obtained by experimental fitting. The natural constant is taken as 2.71828. Activation energy is the minimum energy required for a chemical reaction to occur. The higher the activation energy, the more difficult the reaction is to proceed and the lower the pulverization rate is. The gas constant is taken as 8.314 J / (mol·K).

[0072] In one embodiment, the calculation of the humidity factor includes:

[0073] Get the relative humidity; calculate the humidity factor using the following formula:

[0074]

[0075] in, is the humidity factor, is the base rate constant, is the relative humidity, It is the humidity sensitivity index.

[0076] It is worth mentioning that the humidity factor reflects the effect of ambient humidity on the powdering rate of the alloy. The benchmark rate constant is the rate constant under standard conditions (relative humidity is 100%). Relative humidity indicates the ratio of the actual content of water vapor in the air to the saturated water vapor content at the same temperature. The humidity sensitivity index reflects the strength of the effect of humidity on the powdering rate. The benchmark rate constant is determined by conducting a powdering experiment under standard conditions and recording the powdering rate to determine the specific value. The humidity sensitivity index is determined by conducting a series of powdering experiments under different relative humidities, recording the change of powdering rate with relative humidity, and then using regression analysis.

[0077] In another embodiment, the humidity factor monitors the ambient humidity in real time using a humidity sensor (such as a capacitive humidity sensor). The particle size of the alloy element is measured by a laser scattering particle size analyzer, a screening method, or a microscope to measure the particle size distribution of the alloy particles. The temperature is monitored in real time using a temperature sensor such as a thermocouple or an infrared thermometer. The empirical constant is determined by historical experimental data through regression analysis. The activation energy is determined by performing a pulverization experiment at different temperatures, recording the change of the pulverization rate with temperature, and then fitting the data using the Arrhenius equation to solve the activation energy. The gas constant directly uses the standard value of 8.314 J / (mol·K). For example, the above-mentioned values ​​are obtained: the ambient humidity is measured to be 60% using a humidity sensor, and the value is 1.2 after being substituted into the humidity factor formula. The average particle size measured using a laser scattering particle size analyzer is 100 μm. The temperature measured using a thermocouple is 500 K. The activation energy obtained by fitting the experimental data using the Arrhenius equation is 100 kJ / mol. Substitute the formula to calculate the pulverization rate.

[0078] In step S13, the alloy pulverization degree is analyzed according to the binary alloy pulverization rate to obtain an alloy pulverization degree factor.

[0079] In one embodiment, the alloy pulverization degree factor is calculated by the following formula:

[0080]

[0081] in, is the alloy pulverization degree factor, is the total number of alloying elements, for i The weight of alloying elements, for i The powderization rate of binary alloys of alloying elements, i Is the alloy element number.

[0082] It is worth noting that the alloy pulverization factor indicates the pulverization degree of the entire alloy under specific conditions. The total number of alloy elements is determined by spectral analysis to determine the type and quantity of alloy elements. The weight of the alloy element ratio is the mass ratio. For example, a ternary alloy contains three elements A, B and C, and their mass ratios are 0.3, 0.4, and 0.3 respectively. The pulverization rates of binary alloys are 0.001, 0.002, and 0.0015 respectively, and the total pulverization rate of multi-element alloys is 0.00145.

[0083] In step S14, the alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters are input into a pre-trained alloy yield prediction model, and the predicted alloy yield is output.

[0084] In one embodiment, an alloy yield prediction model is constructed based on a historical alloy pulverization degree factor, historical alloy composition data, historical steelmaking process parameters and historical alloy yield, the model is trained, and the training is determined to be completed after the loss function of the detection model meets the conditions, thereby obtaining a trained model; the alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters are input into the trained model to obtain a predicted alloy yield.

[0085] In one embodiment, the alloy yield prediction model adopts a long short-term memory neural network, and the specific training method is as follows: historical production data is collected from the converter production process, including historical alloy pulverization degree factors, historical alloy composition data, historical steelmaking process parameters and historical alloy yields. The collected data is cleaned to remove records containing missing values ​​or outliers, and the first record of repeated data is retained. Subsequently, the data is normalized to ensure that data of different scales will not affect the effect of model training. The pre-processed process parameters are clustered using the K-means clustering algorithm to identify production batches with similar characteristics. In order to determine the optimal number of clusters, the elbow method and the silhouette coefficient method are used. The elbow method determines the value of the optimal number of clusters by observing the downward trend of the average distance as the value of the number of clusters increases; and the silhouette coefficient rule calculates the cohesion and separation of each sample point, and then calculates the average silhouette coefficient, and selects the value of the number of clusters with the highest average silhouette coefficient as the optimal number of clusters. For each cluster, the data in the cluster is used to train the T-LSTM neural network model. T-LSTM is a special recurrent neural network (RNN) that is particularly suitable for processing time series data because it can effectively capture information over long time intervals. During the training process, hyperparameters such as the training step size, the number of hidden layers, and the learning rate are set. When the alloy yield of a new batch needs to be predicted, the cluster to which it belongs is first determined based on the process parameters of the batch. Then, the process parameters of the batch and the calculated alloy pulverization factor are input into the T-LSTM model of the corresponding cluster, and the model outputs the predicted alloy yield.

[0086] In one embodiment, the historical data source is a sensor on the production line, a laboratory test report, or a production record. The data is stored in a CSV file, an Excel spreadsheet, or a database. After acquiring the data, the records containing missing values ​​are deleted to ensure the integrity of the data. Data preprocessing includes: identifying and deleting outliers through box plot statistics; in order to ensure that data of different scales does not affect the effect of model training, a normalization method is used, and the normalization formula is: ,in To normalize the data, is the original data, is the minimum value of the original data, is the maximum value of the original data. You can directly call the pandas library in Python to process the data.

[0087] In one embodiment, the K-means clustering algorithm is used to cluster the pre-process parameters. The elbow method is used to determine the optimal number of clusters. The core idea is to evaluate the within-cluster sum of squares (WCSS) under different numbers of clusters, and plot the relationship between K value and WCSS to find the location of the "elbow" in the curve. This "elbow" corresponds to the point where the rate of reduction of WCSS begins to slow down significantly, which is considered to be the best choice for the number of clusters. Where WCSS is the sum of the squares of the distances between each data point and the center of the cluster to which it belongs. The formula is as follows:

[0088]

[0089] in, is the intra-cluster error sum of squares, For the i Number cluster, i is the cluster number. is the total number of clusters, is the sample point within the cluster, is the cluster center, is the square of the distance from the sample point in the cluster to the cluster center. Plot the relationship between the calculated sum of squared errors in the cluster and the number of clusters. Look for the point where the WCSS downward trend changes sharply, that is, the "elbow". The WCSS before this point decreases rapidly, and the rate of decrease slows down significantly after this point.

[0090] In another implementation, the silhouette coefficient method is an internal evaluation indicator for evaluating clustering effects, and is particularly suitable for determining the optimal number of clusters (K value). The silhouette coefficient combines the cohesion and separation of clustering. By calculating the silhouette coefficient of each sample point, the average silhouette coefficient of all sample points is finally obtained to evaluate the clustering effect. The formula for calculating the cohesion is:

[0091]

[0092] The formula for calculating the resolution is:

[0093]

[0094] in, For the i The separation of the clusters, For the i The cohesion of the cluster, Cluster The total number of sample points in , i , is the cluster number, For the i Cluster No. j The distance between the cluster centers of the clusters, For the i The set of sample points of the cluster, For the j The set of sample points of the number cluster;

[0095] The silhouette coefficient is used to measure the degree of match between a sample point and its cluster, and the degree of mismatch with the nearest cluster. The formula is as follows:

[0096]

[0097] in, for i The silhouette coefficient of the cluster, For the i The separation of the clusters, For the i The silhouette coefficient is close to 1, which means that the sample point is very similar to other points in the cluster to which it belongs, and has obvious differences with the data points of other clusters, and the clustering effect is good. The silhouette coefficient is close to 0, which means that the sample point is located on the boundary of two clusters and it is difficult to clearly attribute it to a specific cluster, and the clustering effect is average. The silhouette coefficient is close to -1, which means that the sample point is incorrectly assigned to the wrong cluster, and the clustering effect is poor. The value of the number of clusters with the highest average silhouette coefficient is selected as the optimal number of clusters.

[0098] In one embodiment, the corresponding T-LSTM model training includes: the process parameters and alloy pulverization degree factors in each cluster are used as input data. The corresponding alloy yield is used as output data. Setting the model parameters includes: the training step size determines the data length used for each training; the number of hidden layers determines the complexity of the model; and the learning rate controls the speed of updating the model parameters. The data is divided into a training set and a validation set. Hyperparameters are adjusted by grid search. Grid search is a commonly used hyperparameter optimization method that finds the optimal hyperparameter combination by exhaustively enumerating the specified parameter value combination. In machine learning and deep learning, the hyperparameters of the model (such as learning rate, regularization strength, hidden layer size, etc.) have an important impact on the performance of the model, and the grid search method is a commonly used method for finding the optimal hyperparameter combination. The grid search process includes: first, a parameter range or a set of candidate values ​​needs to be defined for each hyperparameter to be optimized. The combination of these parameter ranges and candidate values ​​will form a parameter grid. The grid search method will traverse each set of hyperparameter combinations in the parameter grid. For each set of hyperparameters, a model will be trained and its performance will be evaluated on the validation set. Evaluate the performance of the model on the validation set, using accuracy, recall, F1 score, cross entropy loss, etc. Compare the model performance of all hyperparameter combinations, and select the parameter combination with the best performance and the smallest validation error as the optimal parameter combination.

[0099] In another embodiment, the hyperparameters are optimized using a Bayesian optimization method. This method can effectively find the optimal solution in the parameter space while keeping the computing power within a controllable range by constructing a probabilistic model of the objective function. First, some initial points are randomly selected and the objective function values ​​corresponding to these points are calculated. These initial points and objective function values ​​will be used to initialize the proxy model. A proxy model of the objective function is constructed using a Gaussian Process (GP). The proxy model is used to predict the objective function value and uncertainty of new parameter points. Based on these predictions, an acquisition function (Acquisition Function), such as an upper confidence bound (UCB), is calculated. Based on the results of the acquisition function, the next sampling point is selected. The objective function value is calculated at the selected new point. The new data point is added to the existing data set and the proxy model is updated. The iterative process is repeated to gradually narrow the parameter space and find the optimal parameters. The iterative process ends when a preset number of iterations or convergence conditions are reached.

[0100] It is worth noting that the Gaussian process is a non-parametric model that can capture the uncertainty and smoothness of a function. It is defined by a mean function and a covariance function (kernel function). The upper confidence bound is calculated by the following formula:

[0101]

[0102] in, For coordinates x The upper confidence bound at is a Gaussian process at the point x The predicted mean of is a Gaussian process at the point x The standard deviation of . is the hyperparameter to be optimized.

[0103] In summary, the present invention discloses a steelmaking alloy yield prediction method and system based on alloy pulverization, which aims to improve the accuracy of alloy yield prediction by comprehensively considering alloy composition data, steelmaking process parameters and alloy particle size distribution data. The method first obtains alloy composition data, steelmaking process parameters and alloy particle size distribution data, wherein the alloy composition data includes alloy element types and alloy element mass ratios, and the steelmaking process parameters include steel types, heating temperatures and smelting systems. The alloy particle size distribution data is obtained by calculating parameters such as particle settling velocity, particle density, dispersion medium viscosity and density, and the particle size screening sequence and weight proportion are formed accordingly. Subsequently, the binary alloy pulverization rate is calculated based on the alloy composition data and alloy particle size distribution data in combination with the binary alloy pulverization rate formula. The formula takes into account factors such as humidity factor, alloy element particle size, temperature, empirical constants, natural constants, activation energy and gas constants. The humidity factor is calculated by relative humidity, reflecting the influence of environmental humidity on alloy pulverization. Next, the alloy pulverization degree is analyzed based on the binary alloy pulverization rate to obtain the alloy pulverization degree factor, which is calculated by weighted summation and comprehensively reflects the pulverization degree under the combined action of multiple alloy elements. Finally, the alloy pulverization degree factor, alloy composition data and steelmaking process parameters are input into the pre-trained alloy yield prediction model, and the predicted alloy yield is output. The prediction model is constructed based on the historical alloy pulverization degree factor, historical alloy composition data, historical steelmaking process parameters and historical alloy yield. The training is completed after the model is trained and the loss function is tested to meet the conditions. During the training process, the K-means clustering algorithm is used to cluster the process parameters to identify production batches with similar characteristics, and the T-LSTM neural network model is used to train each cluster to capture information over long time intervals. During the prediction, the cluster to which the new furnace belongs is determined according to the process parameters of the new furnace, and the corresponding process parameters and alloy pulverization degree factor are input into the T-LSTM model of the corresponding cluster, and the model outputs the predicted alloy yield. This method not only takes into account the two traditional factors of alloy composition and steelmaking process, but also introduces two new variables, alloy particle size distribution and degree of pulverization, thereby improving the accuracy and reliability of the prediction model. Through in-depth analysis of the degree of alloy pulverization, the amount of alloy addition in the steelmaking process can be better controlled to improve production efficiency. In addition, the use of a pre-trained model for prediction can provide more accurate prediction results than traditional empirical rules or simple statistical methods. The technical solution of the present invention also provides a detailed description of the acquisition of alloy particle size distribution data, including calculating the particle size of alloy elements through parameters such as particle settling velocity, particle density, dispersion medium viscosity and density, and obtaining the particle size screening sequence and weight percentage based on this. The calculation of the humidity factor is also an important part of this method. The value of the humidity factor is adjusted by relative humidity to more accurately simulate the humidity effect in the actual steelmaking environment.In general, the present invention provides a method for predicting the yield of steel-making alloys, which improves the accuracy of prediction of the yield of steel-making alloys.

[0104] Reference Figure 2 The second embodiment of the present invention provides a steelmaking alloy yield prediction system based on alloy pulverization, comprising:

[0105] Data acquisition module, used to obtain alloy composition data, steelmaking process parameters and alloy particle size distribution data;

[0106] A binary pulverization module, for calculating the binary alloy pulverization rate according to the alloy composition data and the alloy particle size distribution data in combination with a binary alloy pulverization rate formula;

[0107] A multi-element pulverization module, used to analyze the alloy pulverization degree according to the binary alloy pulverization rate to obtain an alloy pulverization degree factor;

[0108] The result prediction module is used to input the alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters into a pre-trained alloy yield prediction model, and output a predicted alloy yield.

[0109] Preferably, the data acquisition module is used to:

[0110] Obtain alloy composition data, steelmaking process parameters and alloy particle size distribution data;

[0111] It is worth noting that the alloy particle size distribution data includes the weight percentage and the particle size screening sequence. The alloy composition data includes the type of alloy elements and the mass ratio of alloy elements; the steelmaking process parameters include steel type, heating temperature and smelting system. The alloy composition data refers to the types and mass ratios of various elements constituting the alloy. Spectroscopic analysis methods (such as X-ray fluorescence spectrometer XRF, atomic absorption spectrometer AAS, etc.) are used in industry to determine the alloy composition. The steelmaking process parameters mainly include steel type, heating temperature and smelting system, etc. Different steel types require different smelting conditions, while heating temperature and smelting system can reduce energy consumption, improve productivity and ensure the quality of steel. These parameters are set according to the production plan, and the actual data are monitored and adjusted by a computer control system. For example, the heating temperature can be monitored in real time by a temperature sensor such as a thermocouple. The alloy particle size distribution data reflects the size of the alloy element particles and their distribution. The particle size distribution has an important influence on the processing properties (such as fluidity, compaction density, etc.) and final performance of the alloy.

[0112] Preferably, the acquisition of the alloy particle size distribution data includes:

[0113] Obtaining particle settling velocity, particle density, dispersion medium viscosity and dispersion medium density;

[0114] The alloying element particle size is calculated by the following formula:

[0115]

[0116] in, is the particle size of alloying elements, is the particle settling velocity, is the dispersion medium viscosity, is the particle density, is the density of the dispersion medium, is the acceleration due to gravity;

[0117] Sorting the alloy element particle sizes from large to small to obtain a particle size screening sequence;

[0118] Get the number of alloy element particles;

[0119] The weight percentage is calculated by dividing the number of particles of each alloy element by the total number of particles;

[0120] The alloy particle size distribution data includes the weight percentage and the particle size screening sequence;

[0121] The alloy composition data includes the types of alloy elements and the mass ratio of alloy elements;

[0122] The steelmaking process parameters include steel type, heating temperature and smelting system.

[0123] Preferably, the binary pulverization module is used for:

[0124] The binary alloy pulverization rate is calculated based on the alloy composition data and the alloy particle size distribution data in combination with the binary alloy pulverization rate formula, including:

[0125] The powdering rate of binary alloy is calculated by the following formula:

[0126]

[0127] in, is the powdering rate of binary alloy, is the humidity factor, is the alloy element particle size, is the temperature, is an empirical constant, is a natural constant, is the activation energy, is the gas constant;

[0128] The calculation of the humidity factor includes:

[0129] Get relative humidity;

[0130] The humidity factor is calculated using the following formula:

[0131]

[0132] in, is the humidity factor, is the base rate constant, is the relative humidity, It is the humidity sensitivity index.

[0133] Preferably, the multi-element pulverization module is used for:

[0134] The alloy pulverization degree analysis is performed according to the binary alloy pulverization rate to obtain the alloy pulverization degree factor, including:

[0135] The alloy powdering degree factor is calculated by the following formula:

[0136]

[0137] in, is the alloy pulverization degree factor, is the total number of alloying elements, for i The weight of alloying elements, for i The powderization rate of binary alloys of alloying elements, i The alloy element number.

[0138] Preferably, the result prediction module is used to:

[0139] The alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters are input into a pre-trained alloy yield prediction model, and the predicted alloy yield is output.

[0140] Preferably, the training process of the alloy yield prediction model includes:

[0141] Based on the historical alloy pulverization degree factor, historical alloy composition data, historical steelmaking process parameters and historical alloy yield rate, an alloy yield rate prediction model is constructed, the model is trained, and the training is determined to be complete after the loss function of the model meets the conditions, thereby obtaining a trained model;

[0142] The alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters are input into the trained model to obtain a predicted alloy yield.

[0143] It should be noted that the steelmaking alloy yield prediction system based on alloy pulverization provided in an embodiment of the present invention is used to execute all the process steps of the steelmaking alloy yield prediction method based on alloy pulverization in the above-mentioned embodiment. The working principles and beneficial effects of the two correspond one to one, and thus will not be elaborated on again.

[0144] The embodiment of the present invention also provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a data acquisition program. When the processor executes the computer program, the steps in the above-mentioned steelmaking alloy yield prediction method based on alloy pulverization are implemented, such as Figure 1 Alternatively, when the processor executes the computer program, the functions of the modules / units in the above-mentioned device embodiments are realized, such as the data acquisition module.

[0145] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the electronic device.

[0146] The electronic device may be a computing device such as a desktop computer, a notebook, a PDA, and a smart tablet. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. The electronic device may include more or fewer components than the above components, or may combine certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0147] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the electronic device, and uses various interfaces and lines to connect various parts of the entire electronic device.

[0148] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the electronic device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0149] Wherein, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0150] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0151] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for predicting steelmaking alloy yield based on alloy pulverization, characterized in that: include: Obtain alloy composition data, steelmaking process parameters and alloy particle size distribution data; According to the alloy composition data and the alloy particle size distribution data, combined with the binary alloy pulverization rate formula, the binary alloy pulverization rate is calculated; Performing alloy pulverization degree analysis according to the binary alloy pulverization rate to obtain an alloy pulverization degree factor; Inputting the alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters into a pre-trained alloy yield prediction model, and outputting a predicted alloy yield; Wherein, the acquisition of the alloy particle size distribution data includes: Obtaining particle settling velocity, particle density, dispersion medium viscosity and dispersion medium density; The alloying element particle size is calculated by the following formula: in, is the alloy element particle size, is the particle settling velocity, is the dispersion medium viscosity, is the particle density, is the density of the dispersion medium, is the acceleration due to gravity; Sorting the alloy element particle sizes from large to small to obtain a particle size screening sequence; Get the number of alloy element particles; The weight percentage is calculated by dividing the number of particles of each alloy element by the total number of particles; The alloy particle size distribution data includes the weight percentage and the particle size screening sequence; The binary alloy pulverization rate is obtained by calculating the alloy composition data and the alloy particle size distribution data in combination with the binary alloy pulverization rate formula, including: The powdering rate of binary alloy is calculated by the following formula: in, is the powdering rate of binary alloy, is the humidity factor, is the particle size of alloying elements, is the temperature, is an empirical constant, is a natural constant, is the activation energy, is the gas constant; The alloy pulverization degree analysis is performed according to the binary alloy pulverization rate to obtain the alloy pulverization degree factor, including: The alloy powdering degree factor is calculated by the following formula: in, is the alloy pulverization degree factor, is the total number of alloying elements, for i The weight of alloying elements, for i The powderization rate of binary alloys of alloying elements, i Number the alloying elements; The training process of the alloy yield prediction model includes: Based on the historical alloy pulverization degree factor, historical alloy composition data, historical steelmaking process parameters and historical alloy yield rate, an alloy yield rate prediction model is constructed, the model is trained, and the training is determined to be complete after the loss function of the model meets the conditions, thereby obtaining a trained model; Inputting the alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters into the trained model to obtain a predicted alloy yield; Wherein, the alloy yield prediction model adopts long short-term memory neural network.

2. The method for predicting steelmaking alloy yield based on alloy pulverization according to claim 1, characterized in that: The alloy composition data includes the types of alloy elements and the mass ratio of alloy elements; The steelmaking process parameters include steel type, heating temperature and smelting system.

3. The method for predicting steelmaking alloy yield based on alloy pulverization according to claim 1, characterized in that: The calculation of the humidity factor includes: Get relative humidity; The humidity factor is calculated using the following formula: in, is the humidity factor, is the base rate constant, is the relative humidity, It is the humidity sensitivity index.

4. A steelmaking alloy yield prediction system based on alloy pulverization, characterized in that: The method for predicting the steelmaking alloy yield based on alloy pulverization as claimed in any one of claims 1 to 3 comprises: Data acquisition module, used to obtain alloy composition data, steelmaking process parameters and alloy particle size distribution data; A binary pulverization module, for calculating the binary alloy pulverization rate according to the alloy composition data and the alloy particle size distribution data in combination with a binary alloy pulverization rate formula; A multi-element pulverization module, used to analyze the alloy pulverization degree according to the binary alloy pulverization rate to obtain an alloy pulverization degree factor; The result prediction module is used to input the alloy pulverization degree factor, the alloy composition data and the steelmaking process parameters into a pre-trained alloy yield prediction model, and output a predicted alloy yield.

5. An electronic device, characterized in that: The invention comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the steelmaking alloy yield prediction method based on alloy pulverization as claimed in any one of claims 1 to 3 is implemented.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the steelmaking alloy yield prediction method based on alloy pulverization according to any one of claims 1 to 3.

Images