CN117851816A - Method for predicting steelmaking end point component of converter - Google Patents
Method for predicting steelmaking end point component of converter Download PDFInfo
- Publication number
- CN117851816A CN117851816A CN202410261108.XA CN202410261108A CN117851816A CN 117851816 A CN117851816 A CN 117851816A CN 202410261108 A CN202410261108 A CN 202410261108A CN 117851816 A CN117851816 A CN 117851816A
- Authority
- CN
- China
- Prior art keywords
- converter
- data
- end point
- predicting
- variables
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000009628 steelmaking Methods 0.000 title claims abstract description 45
- 238000013528 artificial neural network Methods 0.000 claims abstract description 28
- 229910000831 Steel Inorganic materials 0.000 claims abstract description 19
- 239000010959 steel Substances 0.000 claims abstract description 19
- 230000009467 reduction Effects 0.000 claims abstract description 12
- 238000004519 manufacturing process Methods 0.000 claims description 27
- 238000010079 rubber tapping Methods 0.000 claims description 16
- 230000002159 abnormal effect Effects 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 14
- 229910045601 alloy Inorganic materials 0.000 claims description 13
- 239000000956 alloy Substances 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 13
- 239000011159 matrix material Substances 0.000 claims description 13
- 239000000203 mixture Substances 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 11
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 claims description 10
- 239000013598 vector Substances 0.000 claims description 10
- 241000283153 Cetacea Species 0.000 claims description 9
- 238000003723 Smelting Methods 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 238000007664 blowing Methods 0.000 claims description 8
- 230000001186 cumulative effect Effects 0.000 claims description 8
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 claims description 7
- 229910052799 carbon Inorganic materials 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 7
- 229910000640 Fe alloy Inorganic materials 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000004091 panning Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 229910052742 iron Inorganic materials 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 230000036284 oxygen consumption Effects 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 3
- 238000011425 standardization method Methods 0.000 claims description 3
- 229910000805 Pig iron Inorganic materials 0.000 claims description 2
- 238000009825 accumulation Methods 0.000 claims description 2
- 238000003556 assay Methods 0.000 claims description 2
- CSJDCSCTVDEHRN-UHFFFAOYSA-N methane;molecular oxygen Chemical compound C.O=O CSJDCSCTVDEHRN-UHFFFAOYSA-N 0.000 claims description 2
- 239000002893 slag Substances 0.000 claims description 2
- 238000010200 validation analysis Methods 0.000 claims description 2
- 230000009286 beneficial effect Effects 0.000 abstract description 6
- 238000012847 principal component analysis method Methods 0.000 abstract description 6
- 229910001021 Ferroalloy Inorganic materials 0.000 abstract description 5
- 238000004134 energy conservation Methods 0.000 abstract description 3
- 238000009851 ferrous metallurgy Methods 0.000 abstract description 3
- 238000000513 principal component analysis Methods 0.000 description 9
- 239000004615 ingredient Substances 0.000 description 5
- 238000005275 alloying Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 241000251468 Actinopterygii Species 0.000 description 1
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 229910052760 oxygen Inorganic materials 0.000 description 1
- 239000001301 oxygen Substances 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
Landscapes
- Carbon Steel Or Casting Steel Manufacturing (AREA)
Abstract
The invention belongs to the technical field of ferrous metallurgy, in particular to a method for predicting a converter steelmaking end point component, which combines a principal component analysis method (PCA) with a WOA-BP neural network, analyzes the correlation among variables in initial data, changes the variables with compact relations into new variables as few as possible, ensures that the new variables are uncorrelated two by two, respectively represents various information existing in each variable by using fewer comprehensive indexes, realizes the rapid and accurate prediction of the converter steelmaking end point component, is beneficial to improving the quality of molten steel, efficiently utilizes ferroalloy and achieves the purposes of energy conservation and cost reduction.
Description
Technical Field
The invention belongs to the technical field of ferrous metallurgy, and particularly relates to a method for predicting a steelmaking end point component of a converter.
Background
The iron alloy is used as a raw material for deoxidization alloying in the steelmaking process, the consumption is huge, the production of the iron alloy is a high-energy-consumption and high-emission process, the iron and steel industry aims at reducing the consumption of the iron alloy, and the resource saving is realized at the downstream of the iron alloy production line.
In the actual smelting process, the alloy batching needs to be completed before tapping of the converter, the ferroalloy is added into a ladle during tapping, and the alloy batching process needs to be participated in by terminal component data, so that the accurate control of the terminal component is the basis of scientific alloy batching. The end point components are generally controlled by operators through operating experiences such as flame, oxygen supply time and the like, but the variable factors in the converter are many, unknown factors exist, the end point components are inaccurately controlled only through manual experience, certain steel grade requirements with strict quality requirements are difficult to meet, and errors are large, so that alloy ingredients are inaccurate, and the quality of molten steel and the stable running of a subsequent process are affected. At present, the informatization technology is more and more developed, and meanwhile, the development of the steel industry is also driven. In the aspect of prediction of converter smelting parameters, the prediction research on oxygen consumption and molten steel temperature is more, although the prediction precision is high, the prediction method is not used in large scale in actual production, and meanwhile, a scientific method for predicting end point components is not available.
Disclosure of Invention
The invention aims to solve the problems in the prior art, and mainly aims to provide a method for predicting the endpoint components of converter steelmaking, which aims to solve the problem that the endpoint components are difficult to predict in the current converter smelting and tapping process.
In order to solve the technical problems, according to one aspect of the present invention, the following technical solutions are provided:
a method for predicting the endpoint component of converter steelmaking comprises the following steps:
s1, collecting a converter production data set, and establishing a prediction model database;
s2, carrying out data screening and panning on the acquired converter production data set, and preprocessing the screened and panned data;
s3, performing main component analysis (PCA) dimension reduction treatment on the preprocessed data;
s4, establishing a WOA-BP neural network (whale algorithm optimized BP neural network) converter molten steel component prediction model;
s5, training and testing a prediction model;
s6, collecting real-time data of a field smelting process;
s7, performing main component analysis (PCA) dimension reduction treatment on the converter production process data;
s8, substituting a WOA-BP neural network converter molten steel composition prediction model to predict a converter steelmaking end point composition.
As a preferable scheme of the method for predicting the steelmaking end point component of the converter, the invention comprises the following steps: s9, adding ferroalloy in the tapping process according to the converter steelmaking end point component prediction result, storing data after tapping into a prediction model database, and periodically updating a prediction model.
As a preferable scheme of the method for predicting the steelmaking end point component of the converter, the invention comprises the following steps: in the step S1, the converter production dataset includes: steel grade, furnace age, sublance age, ladle temperature, molten iron temperature, tapping temperature, number of times of blowing, tapping amount, molten iron weight, scrap weight, pig iron weight, carbon drawing temperature, total oxygen consumption, blowing time, slag splashing time, TSC carbon, TSC temperature, TSO carbon, TSO temperature, carbon oxygen accumulation, number of times of sliding plates, bottom blowing mode, alloy addition amount, production flow number, tapping number, scrap type and tapping time.
As a preferable scheme of the method for predicting the steelmaking end point component of the converter, the invention comprises the following steps: in the step S2, data screening and panning are performed on the collected converter production dataset, which specifically includes:
duplicate and abnormal data were deleted and the abnormal data were evaluated using Grubbs double sided test as in formula (1):
(1)
wherein:is the average value of the samples, S is the standard deviation of the samples, x n For the n-th data after sorting from small to large, < >>For the upper statistics, +.>Is the following statistic.
Determining the detection level alpha, and determining the critical value G by looking up a table 1-α/2 (n) whenAnd G is n ＞G 1-α/2 (n) at the time of determining x n Abnormal value, when->And->＞G 1-α/2 (n) at the time of judgment of x 1 If the abnormal value is not found, judging that the abnormal value is not found.
As a preferable scheme of the method for predicting the steelmaking end point component of the converter, the invention comprises the following steps: in the step S2, preprocessing the screened and panned data specifically includes: the standardization of the historical data set adopts a Z standardization method, and the specific processing method is shown as a formula (2):
(2)
where i denotes the ith parameter, j denotes the jth sample point,for acquisition of the raw data obtained +.>The average value calculation method for the ith parameter is shown as formula (3)>The standard deviation calculation method for the ith parameter is shown in the formula (4),
(3)
(4)
in the method, in the process of the invention,sample points for the i-th parameter.
As a preferable scheme of the method for predicting the steelmaking end point component of the converter, the invention comprises the following steps: in the step S2, the converter production dataset is represented by 8: the scale of 2 distinguishes between training and validation sets.
As a preferable scheme of the method for predicting the steelmaking end point component of the converter, the invention comprises the following steps: in the step S3, the principal component analysis method is a statistical method for converting a plurality of variables into a few principal components by a dimension reduction technique, and the larger the contribution rate of the principal components, the more sample feature information can be reflected.
As a preferable scheme of the method for predicting the steelmaking end point component of the converter, the invention comprises the following steps: in the step S3, the principal component analysis method includes the following steps:
1) Normalizing raw data: for example m original variables X 1 、X 2 、X 3... X m And n objects, carrying out standardization processing on the original variables to eliminate the size and dimension difference among the variables, and obtaining a normalized coefficient matrix as shown in a formula (5).
(5)
Wherein:for normalizing the coefficient matrix, m represents the m-th original variable, n represents the n-th object under the original variable, and x nm Representing reasonable raw data after panning.
2) Establishing a correlation coefficient matrix, namely a covariance matrix R, and calculating characteristic roots and characteristic vectors of the covariance matrix R, wherein the characteristic roots and the characteristic vectors are shown in a formula (6):
(6)
wherein: r is (r) ij As the original variable X i 、X j Is of the correlation coefficient of (2)
Obtaining characteristic root
And a corresponding unit feature vector as shown in formula (7):
(7)
wherein: u (u) m Feature vector corresponding to mth feature root
3) Determining the number of main component variables: the number of selected principal components depends on the contribution rate with respect to the cumulative variance; when the cumulative variance contribution rate before the P-th principal component variable is not lower than 85%, the first P principal component variables can well reflect the information of the original variables. The variance contribution rate and the cumulative variance contribution rate are respectively shown in the formula (8):
(8)
wherein: alpha i The contribution rate is the variance; beta i In order to accumulate the contribution rate of the variance,is the characteristic root.
4) The eigenvectors of the P principal components arePrincipal component variable composition of n samples +.>As shown in formula (9):
(9)
wherein:for normalizing the coefficient matrix, < >>Main component variable composition for n samples, +.>Is the feature vector of the P principal components.
As a preferable scheme of the method for predicting the steelmaking end point component of the converter, the invention comprises the following steps: in the step S4, the WOA iterative algorithm is as shown in formulas (10) - (13):
1) Surrounding prey
(10)
(11)
Wherein:for the number of iterations,is the best whaleA fish position;is the current whale position; a is a coefficient variable; a is a convergence factor; along withIncreasing from 2 to 0; r is (r) 1 And r 2 Is [0,1]Random numbers in between.
2) Bubble network attack
(12)
Wherein: b is a logarithmic spiral shape constant; l is a random number between [ -1,1 ]; p is a random number between [0,1 ].
3) Searching for prey
(13)
Wherein:the whale positions to be approached were selected randomly.
As a preferable scheme of the method for predicting the steelmaking end point component of the converter, the invention comprises the following steps: in the step S4, the activation function of the neurons of the hidden layer of the BP neural network is a tan sig function and a purelin function, the training function is a tranlm function, and the neurons of the BP neural network are selected and determined by adopting an orthogonal least square method, so as to determine the hidden layer of the BP neural network.
As a preferable scheme of the method for predicting the steelmaking end point component of the converter, the invention comprises the following steps: in the step S6, the method for collecting real-time data in the on-site smelting process includes: obtained from the Oracle database of the assay system and the secondary system by establishing an ODBC connection.
The beneficial effects of the invention are as follows:
the invention provides a prediction method of converter steelmaking end point components, which provides a data-driven prediction model based on data, combines a Principal Component Analysis (PCA) with a WOA-BP neural network, analyzes the correlation among variables in initial data, changes the variables with compact relations into new variables as few as possible, makes the new variables independent of each other, and uses fewer comprehensive indexes to respectively represent various information existing in each variable, thereby realizing rapid and accurate prediction of converter steelmaking end point components, being beneficial to improving molten steel quality, efficiently utilizing ferroalloy and achieving the purposes of energy conservation and cost reduction.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to the structures shown in these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a method for predicting the endpoint ingredients of converter steelmaking of the present invention;
FIG. 2 shows the main component dimension reduction result of example 1 of the present invention;
FIG. 3 shows the WOA-BP prediction result of example 1 of the present invention;
FIG. 4 shows the prediction results of PCA-WOA-BP of example 1 of the present invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
The following description will be made clearly and fully with reference to the technical solutions in the embodiments, and it is apparent that the described embodiments are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, the method for predicting the steelmaking end point component of the converter can rapidly and accurately predict the steelmaking end point component of the converter, is beneficial to high-efficiency alloy batching, reduces the production cost, improves the molten steel quality, ensures the normal operation of the subsequent working procedures, and has good application prospect in the field of ferrous metallurgy. The reaction is multiple in the converter steelmaking process, the condition in the converter is complex, the accurate acquisition of the end point components is difficult, and the accurate alloy ingredients are not facilitated. The principal component analysis method is combined with a WOA-BP neural network (whale algorithm optimized BP neural network), and the related relations among variables in initial data are analyzed to change the variables with compact relations into new variables as few as possible, so that the new variables are uncorrelated two by two, and the fewer comprehensive indexes are used for representing various information existing in each variable respectively, thereby realizing the rapid and accurate prediction of the converter steelmaking end point components. The main component analysis (PCA) is adopted to carry out dimension reduction processing on the converter production process data, and meanwhile, the WOA-BP neural network (whale algorithm optimized BP neural network) overcomes the inherent defects that the traditional neural network is slow in learning speed and easy to fall into overfitting and local optimal solution, and has the advantages of high learning speed, strong global searching capability, easiness in algorithm realization and the like.
The number of layers and the number of units of the hidden layer of the traditional BP neural network are not guided theoretically temporarily, and are generally determined according to a large number of experiments or manual experience, so that great redundancy exists, a large amount of training time is required for complex problems, the training time is very sensitive to initial weights, and different training results can be obtained by different initial weights. Based on the characteristics of WOA (whale algorithm) optimization BP neural network, the characteristics of WOA high-speed iteration and global optimization are mainly applied, and the characteristics of the system are optimized by adjusting the initial weight and the threshold of the BP neural network, so that the problem of sinking into a local optimal solution is prevented. The method can realize high-speed processing on the processing of industrial production process data, and is suitable for solving the problems of large volume and multiple variables of converter steelmaking production data.
The technical scheme of the invention is further described below by combining specific embodiments.
Example 1
The top-bottom combined blown converter of a certain steel plant 120t has smelting steel of HRB400E and average end point C content of 0.2436wt%, and the end point C content of each furnace is distributed within the range of 0.2142-0.2603 wt%, so that the fluctuation range is unfavorable for alloy batching. Meanwhile, the alloying worker needs to weigh the iron alloy according to the blowing end point before tapping of the converter, but the smelting end point is controlled by a converter master control worker through production experience, so that larger errors can occur in the numerical value of the component of the blowing end point, and the alloying worker can be influenced to perform alloy batching. Thus, inaccuracy of the end point component causes inaccuracy of the alloy ingredients, which can bring about a certain influence on alloying, so that the subsequent production process is difficult to be carried out smoothly, and the production time is prolonged. Therefore, the accurate prediction of the converter steelmaking end point components is significant for the converter production, the alloy is efficiently mixed, the production cost is reduced, and the molten steel quality is improved.
The production data of 532 groups of the factory 120t converter in 2023 is collected, the repeated data and the abnormal data are deleted, and the abnormal data are judged by using a Grubbs double-side test method, wherein the following formula is as follows:
wherein:is the average value of the samples, S is the standard deviation of the samples, x n For the n-th data after sorting from small to large, < >>For the upper statistics, +.>Is the following statistic.
Determining the detection level alpha, and determining the critical value G by looking up a table 1-α/2 (n) whenAnd G is n ＞G 1-α/2 (n) at the time of determining x n Abnormal value, when->And->＞G 1-α/2 (n) at the time of judgment of x 1 If the abnormal value is not found, judging that the abnormal value is not found.
340 groups of effective data are obtained after screening. The Z standardization method is adopted for the standardization of the historical data set, and the specific processing method is shown as the following formula:
where i denotes the ith parameter, j denotes the jth sample point,for acquisition of the raw data obtained +.>The mean value calculation method for the i-th parameter is shown in the following formula,
the standard deviation calculation method for the i-th parameter is shown in the following formula,
in the method, in the process of the invention,sample points for the i-th parameter.
Wherein,is a characteristic variable of the input, wherein->And->Maximum and minimum values for each individual sample data.
The principal component analysis dimension reduction processing is performed on all acquired data variables, the data set in the example contains 28 influencing factors, if all values are used as input data, the data dimension is high, the model training time is long, and therefore the 28×340 data matrix obtained after the preprocessing is used as the input of PCA, and the input dimension of the prediction model is reduced. Fig. 2 is a principal component dimension reduction result, which is the first 6 principal component interpretation rates and cumulative interpretation rates after PCA processing. It can be seen that the first 6 principal components have a cumulative interpretation rate of 92.5% greater than 85% and thus may contain the primary information of the converter endpoint carbon content influencing factors. Taking the dimensionality reduced data variable as an input variable of the WOA-BP neural network, establishing a converter molten steel component prediction model, taking a tan sig function and a purelin function as activation functions of hidden layer neurons, and selecting neurons for determining the BP neural network by adopting an orthogonal least square method so as to determine the hidden layer of the BP neural network. The 340 groups of data 272 groups obtained after screening are used for training the model, the 68 groups are used for model testing to verify the effect of the model, and parameters of the network are continuously adjusted to obtain the prediction model with the best prediction effect.
Using four performance indicators to evaluate the performance of a molten steel composition prediction model, including determining coefficients (R 2 ) Mean Absolute Error (MAE), mean square error (RMSE), root Mean Square Error (RMSE), and comparing with WOA-BP neural network, the detailed results are shown in Table 1.
Table 1 test dataset prediction error
From fig. 3 and 4 and table 1, the PCA-WOA-BP neural network is able to fit the raw data well and better than the WOA-BP neural network, while having good predictive performance on the test set.
According to the invention, a principal component analysis method is combined with a WOA-BP neural network, and the problems that the reaction is multiple in the converter steelmaking process, the condition in the converter is complex, the accurate acquisition of the end point component is difficult and the accurate alloy ingredients are not beneficial are considered. The PCA can reduce the loss of the information contained in the original index as much as possible while reducing the index to be analyzed, so as to achieve the purpose of comprehensively analyzing the collected data. The WOA algorithm has the advantages of few required parameters, simple and easily understood principle, and better convergence speed and precision in the optimizing process than those of Particle Swarm (PSO), gravitation Search Algorithm (GSA), differential Evolution (DE) and other optimizing algorithms. The method can rapidly and accurately realize the prediction of the steelmaking end point component of the converter, improves the hit rate of the steelmaking end point component of the converter and the molten steel quality, is favorable for realizing the efficient utilization of ferroalloy, and promotes the energy conservation and the emission reduction. The method is verified by the actual production data on site, and the result shows that the method has better accuracy and applicability, and can provide beneficial guidance for the production process of the converter steelmaking site.
The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the invention, and all equivalent structural changes made by the content of the present invention or direct/indirect application in other related technical fields are included in the scope of the present invention.
Claims (10)
1. A method for predicting the endpoint component of converter steelmaking is characterized by comprising the following steps:
s1, collecting a converter production data set, and establishing a prediction model database;
s2, carrying out data screening and panning on the acquired converter production data set, and preprocessing the screened and panned data;
s3, performing main component analysis method dimension reduction treatment on the preprocessed data;
s4, establishing a WOA-BP neural network converter steelmaking end point component prediction model;
s5, training and testing a prediction model;
s6, collecting real-time data of a smelting process of the converter on site;
s7, performing main component analysis method dimension reduction treatment on the acquired real-time data;
s8, substituting a WOA-BP neural network converter molten steel composition prediction model to predict a converter steelmaking end point composition.
2. The method for predicting the end point composition of converter steelmaking according to claim 1, wherein said method further comprises S9. Adding iron alloy during tapping according to the end point composition prediction result of converter steelmaking, storing the tapping end point data in a prediction model database, and periodically updating the prediction model.
3. The method for predicting a converter steelmaking end point component according to claim 1, wherein in step S1, the converter production dataset includes: steel grade, furnace age, sublance age, ladle temperature, molten iron temperature, tapping temperature, number of times of blowing, tapping amount, molten iron weight, scrap weight, pig iron weight, total oxygen consumption, blowing time, slag splashing time, carbon drawing temperature, TSC carbon, TSC temperature, TSO carbon, TSO temperature, carbon oxygen accumulation, number of times of sliding plates, bottom blowing mode, alloy addition amount, production flow number, tapping number, scrap type and tapping time.
4. The method for predicting a converter steelmaking end point component according to claim 1, wherein in step S2, the collected converter production dataset is subjected to data screening and panning, and specifically comprises:
duplicate and abnormal data were deleted and the abnormal data were evaluated using Grubbs double sided test as in formula (1):
(1)
wherein:as an average value of the samples,s is the standard deviation of the sample, x n For the n-th data after sorting from small to large, < >>For the upper statistics, +.>Is the following statistic;
determining the detection level alpha, and determining the critical value G through a Chagrans table 1-α/2 (n) whenAnd G is n ＞G 1-α/2 (n) at the time of determining x n Abnormal value, when->And->＞G 1-α/2 (n) at the time of judgment of x 1 If the abnormal value is not found, judging that the abnormal value is not found.
5. The method for predicting a final component in converter steelmaking according to claim 1, wherein in step S2, the screened and panned data is preprocessed, specifically comprising: the standardization of the historical data set adopts a Z standardization method, and the specific processing method is shown as a formula (2):
(2)
where i denotes the ith parameter, j denotes the jth sample point,for acquisition of the raw data obtained +.>Mean value calculation method for ith parameterThe method is shown in formula (3)>The standard deviation calculation method for the ith parameter is shown in the formula (4),
(3)
(4)
in the method, in the process of the invention,sample points for the i-th parameter.
6. The method for predicting a converter steelmaking end point component according to claim 1, wherein in said step S2, said converter production dataset is prepared by a method comprising: the scale of 2 distinguishes between training and validation sets.
7. The method for predicting a final component in converter steelmaking according to claim 1, wherein in step S3, the main component analysis method comprises the steps of:
1) Normalizing raw data: for example m original variables X 1 、X 2 、X 3... X m And n objects, carrying out standardization processing on the original variables to eliminate the size and dimension difference among the variables, and obtaining a normalized coefficient matrix, as shown in a formula (5),
(5)
wherein:for normalizing the coefficient matrix, m represents the m-th original variable, n representsN-th object under original variable, x nm Representing reasonable raw data after panning;
2) Establishing a correlation coefficient matrix, namely a covariance matrix R, and calculating characteristic roots and characteristic vectors of the covariance matrix R, wherein the characteristic roots and the characteristic vectors are shown in a formula (6):
(6)
wherein: r is (r) ij As the original variable X i 、X j Is used for the correlation coefficient of (a),
obtaining characteristic root
And a corresponding unit feature vector as shown in formula (7):
(7)
wherein: u (u) m The feature vector corresponding to the mth feature root;
3) Determining the number of main component variables: the number of selected principal components depends on the contribution rate with respect to the cumulative variance; when the cumulative variance contribution rate before the P-th principal component variable is not lower than 85%, the first P principal component variables can well reflect the information of the original variables; the variance contribution rate and the cumulative variance contribution rate are respectively shown in the formula (8):
(8)
wherein: alpha i For variance contribution rate, beta i In order to accumulate the contribution rate of the variance,is a characteristic root;
4) The eigenvectors of the P principal components areThe principal component variable composition of the n samples is shown in formula (9):
(9)
wherein:for normalizing the coefficient matrix, < >>Main component variable composition for n samples, +.>Is the feature vector of the P principal components.
8. The method according to claim 1, wherein in the step S4, the iterative algorithm of WOA is as shown in formulas (10) to (13):
1) Surrounding prey
(10)
(11)
Wherein:for the number of iterations->Is the best whale position; />Is the current whale position; a is a coefficient variable; a is a convergence factor; along with->Increasing from 2 to 0; r is (r) 1 And r 2 Is [0,1]Random numbers in between;
2) Bubble network attack
(12)
Wherein: b is a logarithmic spiral shape constant; l is a random number between [ -1,1 ]; p is a random number between [0,1 ];
3) Searching for prey
(13)
Wherein:the whale positions to be approached were selected randomly.
9. The method for predicting the endpoint component of converter steelmaking according to claim 1, wherein in the step S4, the activation function of neurons of an hidden layer of the BP neural network is a tan sig function and a purelin function, the training function is a tranlm function, and the neurons of the BP neural network are determined by selecting the neurons by an orthogonal least square method, so as to determine the hidden layer of the BP neural network.
10. The method for predicting a converter steelmaking end point component according to claim 1, wherein in step S6, the method for collecting real-time data of the on-site smelting process comprises: obtained from the Oracle database of the assay system and the secondary system by establishing an ODBC connection.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410261108.XA CN117851816A (en) | 2024-03-07 | 2024-03-07 | Method for predicting steelmaking end point component of converter |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410261108.XA CN117851816A (en) | 2024-03-07 | 2024-03-07 | Method for predicting steelmaking end point component of converter |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117851816A true CN117851816A (en) | 2024-04-09 |
Family
ID=90533036
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410261108.XA Pending CN117851816A (en) | 2024-03-07 | 2024-03-07 | Method for predicting steelmaking end point component of converter |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117851816A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112668234A (en) * | 2020-12-07 | 2021-04-16 | 辽宁科技大学 | Intelligent control method for steelmaking endpoint of converter |
| CN114875196A (en) * | 2022-07-01 | 2022-08-09 | 北京科技大学 | Method and system for determining converter tapping quantity |
| CN117612651A (en) * | 2023-11-30 | 2024-02-27 | 东北大学 | Method for predicting manganese content of converter endpoint |
-
2024
- 2024-03-07 CN CN202410261108.XA patent/CN117851816A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112668234A (en) * | 2020-12-07 | 2021-04-16 | 辽宁科技大学 | Intelligent control method for steelmaking endpoint of converter |
| CN114875196A (en) * | 2022-07-01 | 2022-08-09 | 北京科技大学 | Method and system for determining converter tapping quantity |
| CN117612651A (en) * | 2023-11-30 | 2024-02-27 | 东北大学 | Method for predicting manganese content of converter endpoint |
Non-Patent Citations (1)
| Title |
|---|
| 张习康 等: "基于 WOA－BP 神经网络的 25CrMo4 钢本构关系研究", 塑性工程学报, vol. 30, no. 8, 31 August 2023 (2023-08-31), pages 182 - 187 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111353656B (en) | Steel enterprise oxygen load prediction method based on production plan | |
| CN114611844B (en) | Method and system for determining alloy addition amount in converter tapping process | |
| CN112100745B (en) | Automobile girder steel mechanical property prediction method based on LDA theory | |
| CN109359723A (en) | Based on the converter terminal manganese content prediction technique for improving regularization extreme learning machine | |
| CN111444942B (en) | Intelligent forecasting method and system for silicon content of blast furnace molten iron | |
| CN105740984A (en) | Product concept performance evaluation method based on performance prediction | |
| CN106600037B (en) | Multi-parameter auxiliary load prediction method based on principal component analysis | |
| CN104899425A (en) | Variable selection and forecast method of silicon content in molten iron of blast furnace | |
| CN114897227A (en) | Multi-steel-grade mechanical property forecasting method based on improved random forest algorithm | |
| CN110472349A (en) | A kind of hot-rolled steel performance prediction method based on EEMD and depth convolutional network | |
| Feng et al. | Endpoint temperature prediction of molten steel in RH using improved case-based reasoning | |
| CN116520772A (en) | Sheet part numerical control machining quality prediction method based on multitasking transfer learning | |
| CN108830405B (en) | Real-time power load prediction system and method based on multi-index dynamic matching | |
| US11987855B2 (en) | Method and system for determining converter tapping quantity | |
| CN117851816A (en) | Method for predicting steelmaking end point component of converter | |
| CN116484745A (en) | Design method of hot-rolled low-alloy steel for physical metallurgy guidance industry big data mining | |
| CN110648023A (en) | Method for establishing data prediction model based on quadratic exponential smoothing improved GM (1,1) | |
| CN100371938C (en) | Quality design method under minute new aluminium sample | |
| Liu et al. | Establishment and Application of Steel Composition Prediction Model Based on t-Distributed Stochastic Neighbor Embedding (t-SNE) Dimensionality Reduction Algorithm | |
| CN117744503A (en) | Prediction method for tapping temperature of converter | |
| Chen et al. | An intelligent online detection approach based on big data for mechanical properties of hot-rolled strip | |
| CN117711516A (en) | Method for predicting molten steel composition at converter end point | |
| Chen et al. | Endpoint Temperature Prediction of Molten Steel in VD Furnace Based on AdaBoost. RT-ELM | |
| CN116127345A (en) | Converter steelmaking process mode design method based on deep clustering generation countermeasure network | |
| CN116108733A (en) | Method for predicting mechanical properties of steel |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |