A blast furnace key furnace condition parameter optimization method based on industrial big data

Abstract

The present application relates to a kind of blast furnace key furnace condition parameter optimization method based on industrial big data, its method includes: the full process data in the obtained blast furnace smelting process is preprocessed operation;Based on the data after being preprocessed, the selected blast furnace key furnace condition parameter is carried out feature screening by correlation analysis to obtain influence parameter;Based on blast furnace key furnace condition parameter and influence parameter, the multivariate linear fitting function of target and constraint function is structured, and optimization operation solution set is obtained by solving;Optimization operation solution set is classified and sorted by using preset case library to carry out multistage matching and combined with feasibility analysis, and the optimal operation solution is obtained.The present application collects, analyzes and processes blast furnace parameter by big data technology, obtains optimization operation solution by optimization algorithm based on machine learning and integrated learning fusion metallurgical theory and expert experience, and finally selects the solution of minimum operation deviation and lowest economic cost by multistage matching analysis and feasibility analysis, to be used to guide blast furnace production.

Description

Technical Field

[0001] This invention relates to the field of blast furnace control technology, and in particular to a method for optimizing key blast furnace condition parameters based on industrial big data. Background Technology

[0002] Because the blast furnace production and smelting process is extremely complex, there is a complex and strongly coupled nonlinear relationship between the blast furnace process parameters and production targets, and many parameters affect the fluctuation of furnace conditions, making the optimal control of the blast furnace extremely difficult.

[0003] Existing optimization methods for key blast furnace parameters mostly employ single-objective or traditional multi-objective optimization, failing to explain how to apply the optimized solution set in the field. Incomplete parameter considerations result in the ineffective utilization of a large amount of blast furnace production process data, leading to low correlation between optimization results and actual production, and a lack of universality. Furthermore, existing optimization methods for key blast furnace parameters do not comprehensively consider both intelligent algorithms and metallurgical theory, raising questions about the applicability and rationality of the resulting optimization schemes. The optimization results also lack flexibility by not linking them to actual production conditions. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] In view of the above-mentioned shortcomings and deficiencies of the existing technology, the present invention provides a method for optimizing key blast furnace condition parameters based on industrial big data, which solves the technical problems of low correlation, lack of universality, low practicality, questionable rationality, and lack of automatic adjustment capability of existing optimization schemes for key blast furnace condition parameters.

[0006] (II) Technical Solution

[0007] To achieve the above objectives, the main technical solutions adopted by the present invention include:

[0008] In a first aspect, embodiments of the present invention provide a method for optimizing key blast furnace condition parameters based on industrial big data, including:

[0009] The acquired data from all processes in the blast furnace smelting process are classified, integrated, and cleaned.

[0010] Based on the classified, integrated, and cleaned data, the influencing parameters were obtained by feature screening of the selected key blast furnace condition parameters through correlation analysis.

[0011] Based on the key blast furnace condition parameters and the influencing parameters, a multivariate linear fitting function for the objective and constraint functions is constructed, and the optimized operation solution set is obtained by solving the multivariate linear fitting function;

[0012] By using a pre-defined case library for multi-level matching and combining it with feasibility analysis, the solutions in the optimized operation solution set are classified and sorted to obtain the optimal operation solution with the minimum operation deviation and the lowest economic cost.

[0013] Optionally, the acquired data from the entire blast furnace smelting process may be classified, integrated, and cleaned, including:

[0014] Obtain full-process data of the blast furnace smelting process;

[0015] All process data are categorized into different types based on the same process. These different types include raw material data, pulverized coal injection data, air supply data, cooling data, and slag and iron data at the blast furnace site.

[0016] Data from different processes are linked using a defined time-series index to integrate multi-source data, followed by the data cleaning process:

[0017] For duplicate data in the process data, simply delete the duplicate data.

[0018] For null values ​​in process data, if the proportion of null values ​​in the process data is less than the first threshold, the null value part will be filled; if the proportion is not less than the first threshold, the process data will be deleted directly.

[0019] For outliers in the process data, if the proportion of outliers in the process data is greater than the second threshold, the process data is directly deleted. If the proportion of outliers in the process data is not greater than the second threshold, linear interpolation is used to replace non-continuous outliers, and spline interpolation is used to replace multiple consecutive outliers.

[0020] Optionally, based on the classified, integrated, and cleaned data, the influencing parameters are obtained by feature screening of the selected key blast furnace condition parameters through correlation analysis, including:

[0021] Coke ratio, permeability index, and gas utilization rate were selected as key furnace condition parameters for the blast furnace.

[0022] Standardize the cleaned data to eliminate the impact of differences in units and ranges of values ​​between data.

[0023] Based on standardized data, correlation analysis was used to screen the features of the coke ratio, the permeability index and the gas utilization rate, respectively, to obtain the influence features corresponding to these three parameters.

[0024] Among the influencing characteristics corresponding to coke ratio, permeability index, and gas utilization rate, the influencing factors are sorted from largest to smallest, and the top N are taken and their union is used as the influencing parameters of coke ratio, permeability index, and gas utilization rate.

[0025] Optionally, based on the key blast furnace condition parameters and the influencing parameters, a multivariate linear fitting function for the objective and constraint functions is constructed. The optimized operational solution set is obtained by solving the multivariate linear fitting function, including:

[0026] Based on the key blast furnace condition parameters and the influencing parameters, a multivariate linear fitting function for the objective and constraint functions is constructed.

[0027] The following process is used to solve the multivariate linear fitting function using the ε-constraint optimization algorithm: the coke ratio is selected as the main optimization objective, and the permeability index and gas utilization rate are used as objective constraints. By solving the thresholds of the permeability index and gas utilization rate, objective constraints are added to the coke ratio. Finally, the solution of the coke ratio under the constraint conditions is obtained, which is the optimization operation solution set.

[0028] in,

[0029] The multivariate linear fitting function is:

[0030] f1(x)=228+4.63x1+5.57x2+10.28x3-0.00006x4-0.39x5+0.000035x6-3.3x7

[0031] -3.4x8 -3.8x9

[0032] f2(x)=73.65-0.12x1-0.13x2-0.2x3-0.44x4+0.87x5+0.0004x6+1.0x7-0.45x8

[0033] -0.25x9

[0034] f3(x)=0.56-0.00004x1-0.00004x2-0.00075x3-0.00009x4-0.0012x5

[0035] +0.0000001x6+0.007x7+0.01x8-0.026x9

[0036] The constraints include:

[0037] Constraints for each influencing parameter: x1(49.6, 52.8), x2(1.4, 3.5), x3(1.39, 2.23), x4(322.9, 343), x5(184.5, 186.5), x6(16666, 19073), x7(1.22, 1.99), x8(12.9, 13.5), x9(0.215, 0.423);

[0038] Constraints between target coke ratio, permeability index, gas utilization rate and their respective influencing parameters:

[0039] y1=106.1+4.58x1+5.39x2+11.76x3

[0040] y2=65.5-0.45x4+0.88x5+0.0003x6

[0041] y3=0.314+0.007x7+0.01x8-0.026x9

[0042] In the above formula, f1(x), f2(x), and f3(x) represent the coke ratio, permeability index, and gas utilization rate, respectively. y1, y2, and y3 are the constraint equations for the coke ratio, permeability index, and gas utilization rate, respectively. x1-x9 are the influencing parameters, which are the actual value of the material flow valve opening, the feeding speed, the sintering Al2O3, the hot air pressure, the average top pressure, the pulverized coal injection rate, the pellet FeO, the coke Ad (%), and the coke Mad (%).

[0043] Optionally, before classifying and sorting the solutions in the optimized operation solution set using a preset case library for multi-level matching and combining feasibility analysis to obtain the optimal operation solution with the minimum operation deviation and lowest economic cost, the method further includes:

[0044] The selected key blast furnace condition parameters are used as operating condition indicators, and the operating condition indicator function is determined based on the operating condition indicators.

[0045] Based on the operating condition index function, determine whether the called blast furnace historical data is an excellent value that meets the conditions, and establish a blast furnace case library based on the excellent values.

[0046] The operating condition index function is as follows:

[0047]

[0048] In the formula, α is the weighting coefficient of each quantity, which is set to 0.4, 0.3, and 0.3 respectively based on the priority ranking of the working condition indicators on site; i is the number of working condition indicators selected; and k is the actual value of the working condition indicator. * These are the expected values ​​for each operating condition indicator.

[0049] Optionally, by using a preset case library for multi-level matching and combining it with feasibility analysis, the solutions in the optimized operation solution set are classified and sorted to obtain the optimal operation solution with the minimum operation deviation and the lowest economic cost, including:

[0050] Cluster analysis is performed on the blast furnace case library to obtain the number of subclasses and the cluster center of each subclass; the subclasses are prioritized by the working condition index function, and then the distance between each optimization operation solution in the optimization operation solution set and the cluster center of each subclass is calculated and converted into a similarity value to determine the subclass to which each optimal solution belongs, thus completing the initial matching.

[0051] Calculate the similarity value between each optimal solution and all data in its subclass, select the case database data corresponding to the maximum similarity value, and sort the selected case database data using the working condition index function and its subclass order to complete the secondary matching;

[0052] Feasibility analysis is used to screen each optimal operation solution in the optimal operation solution set based on the ease of operation and the operation cost. The ease of operation includes changes in the blast furnace operation system and the magnitude of operation deviations, and the operation cost includes changes in raw materials and fuels and the impact data on production.

[0053] Based on the on-site operation data, the parameters in the influencing parameters are prioritized to determine the adjustable parameters and their order of priority.

[0054] The data deviation between the optimal solution set after secondary matching is calculated and the actual value of blast furnace data under the current operating mode is obtained. The data deviation is sorted according to the adjustable parameters and the order of the adjustable parameters to obtain the deviation sort of the solutions in the optimal solution set after secondary matching is completed.

[0055] Calculate the cost of the solutions in the optimized solution set after completing the secondary matching;

[0056] Based on the deviation ranking and the cost consumption, the optimal operation solution with the minimum operation deviation and the lowest economic cost is obtained.

[0057] Optionally, after classifying and sorting the solutions in the optimized operation solution set using a preset case library for multi-level matching and combining feasibility analysis to obtain the optimal operation solution with the minimum operation deviation and the lowest economic cost, the method further includes:

[0058] The solution with the minimum operational deviation and the lowest economic cost is pushed to the blast furnace operator's control terminal to optimize blast furnace production operations.

[0059] Secondly, embodiments of the present invention provide a blast furnace key furnace condition parameter optimization system based on industrial big data, including: a data management subsystem and a multi-objective optimization dynamic control subsystem;

[0060] The data management subsystem includes:

[0061] The data acquisition module is used to collect data from all processes during the blast furnace smelting process.

[0062] The preprocessing module is used to classify, integrate, and clean the acquired data from all processes in the blast furnace smelting process; and,

[0063] Case library, used to store cases containing blast furnace data and operating condition indicators;

[0064] The multi-objective optimization dynamic control subsystem includes:

[0065] The multi-objective optimization module is used to obtain influencing parameters by performing feature screening on selected key blast furnace condition parameters through correlation analysis based on classified, integrated and cleaned data. Then, it constructs a multivariate linear fitting function of objective and constraint functions based on the key blast furnace condition parameters and the influencing parameters, and obtains the optimization operation solution set by solving the multivariate linear fitting function.

[0066] The feedback module is used to perform multi-level matching using a preset case library and combine it with feasibility analysis to classify and sort the solutions of the optimized operation solution set, so as to obtain the optimal operation solution with the minimum operation deviation and the lowest economic cost.

[0067] Thirdly, embodiments of the present invention provide a device for optimizing key blast furnace condition parameters based on industrial big data, comprising:

[0068] At least one database;

[0069] And a memory that is communicatively connected to the at least one database;

[0070] The memory stores instructions that can be executed by the at least one database, which are then executed by the at least one database to enable the at least one database to perform the above-described method for optimizing key blast furnace condition parameters based on industrial big data.

[0071] Fourthly, embodiments of the present invention provide a computer-readable medium storing computer-executable instructions, characterized in that, when the executable instructions are executed by a processor, they implement the above-described method for optimizing key blast furnace condition parameters based on industrial big data.

[0072] (III) Beneficial Effects

[0073] The beneficial effects of this invention are as follows: This invention establishes a data management system with blast furnace process parameters as the core, and uses big data technology to comprehensively collect, process, and analyze blast furnace data resources, so that blast furnace production process data can be effectively utilized; furthermore, based on machine learning and ensemble learning, it integrates metallurgical theory and expert experience, and uses optimization algorithms to solve the objective function. The optimization results obtained are highly relevant to actual production and have universality; based on the solution results, it uses multi-level matching of case libraries and feasibility analysis to classify and sort the optimal operation solution set, and finally selects the solution with the minimum operation deviation and the lowest economic cost, and pushes suggestions to blast furnace operators to guide blast furnace production. Attached Figure Description

[0074] Figure 1 A flowchart illustrating a method for optimizing key blast furnace parameters based on industrial big data, provided in an embodiment of the present invention;

[0075] Figure 2 This is a schematic diagram illustrating the specific process of step S1 in a method for optimizing key blast furnace condition parameters based on industrial big data, as provided in an embodiment of the present invention.

[0076] Figure 3 This is a schematic diagram illustrating step S2 of a method for optimizing key blast furnace condition parameters based on industrial big data, as provided in an embodiment of the present invention.

[0077] Figure 4 This is a schematic diagram illustrating step S3 of a method for optimizing key blast furnace condition parameters based on industrial big data, as provided in an embodiment of the present invention.

[0078] Figure 5 This is a schematic diagram illustrating the specific process of step S4 in a method for optimizing key blast furnace condition parameters based on industrial big data, as provided in an embodiment of the present invention.

[0079] Figure 6 This is a schematic diagram of the overall structure of a blast furnace key furnace condition parameter optimization system based on industrial big data, provided for an embodiment of the present invention. Detailed Implementation

[0080] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0081] like Figure 1As shown in the embodiment of the present invention, a method for optimizing key blast furnace condition parameters based on industrial big data is proposed. This method includes: classifying, integrating, and cleaning the acquired full-process data of the blast furnace smelting process; based on the classified, integrated, and cleaned data, using correlation analysis to perform feature screening on the selected key blast furnace condition parameters to obtain influencing parameters; constructing a multivariate linear fitting function for the objective and constraint functions based on the key blast furnace condition parameters and influencing parameters; obtaining the optimized operation solution set by solving the multivariate linear fitting function; and classifying and sorting the solutions in the optimized operation solution set using a preset case library and combining it with feasibility analysis to obtain the optimal operation solution with the minimum operational deviation and the lowest economic cost.

[0082] This invention establishes a data management system centered on blast furnace process parameters and employs big data technology to comprehensively collect, process, and analyze blast furnace data resources, enabling effective utilization of blast furnace production process data. Furthermore, based on machine learning and ensemble learning, it integrates metallurgical theory and expert experience, and uses optimization algorithms to solve the objective function. The optimization results obtained are highly relevant to actual production and have universal applicability. Based on the solution results, the optimal operation solution set is classified and sorted using multi-level matching of case libraries and feasibility analysis. Finally, the solution with the minimum operational deviation and the lowest economic cost is selected to provide suggestions and guidance for blast furnace operation.

[0083] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0084] Specifically, the present invention provides a method for optimizing key blast furnace condition parameters based on industrial big data, comprising:

[0085] S1. Perform preprocessing on the acquired data from all processes in the blast furnace smelting process, including classification, integration, and cleaning.

[0086] like Figure 2 As shown, step S1 includes:

[0087] S11. Obtain full-process data of the blast furnace smelting process.

[0088] S12. Classify all process data into different categories, including raw material data, pulverized coal injection data, air supply data, cooling data, and slag and iron data at the blast furnace site.

[0089] Taking raw material data as an example, the various raw material data are not placed in the same table but are given in the form of daily reports. For example, the data on coke, pulverized coal, sinter, and pellets are not in the same table but are given in the form of daily reports. It is necessary to organize the data of coke, pulverized coal, sinter, and pellets and then classify them into the major category of raw material data. Similarly, the data on pulverized coal injection, air supply, cooling, and slag and iron are also classified in the same way. Therefore, the obtained full-process data also needs to be classified into these major categories according to the same process: blast furnace on-site raw material data, pulverized coal injection data, air supply data, cooling data, and slag and iron data, etc.

[0090] S13. Link data from different processes using a set time series index to achieve the integration of multi-source data.

[0091] In step S1, the collected data includes raw material data, pulverized coal injection data, air supply data, cooling data, slag and iron data, and other data from the entire process at the blast furnace site; different types of collected data are classified according to the same process; then, the data from different processes are associated with a unique time series index to integrate multi-source data; finally, data cleaning is performed, including the identification and processing of null values, duplicate values, and outliers.

[0092] S14. Perform the following data cleaning operations:

[0093] For duplicate data in the process data, simply delete the duplicate data.

[0094] For null values ​​in the process data, if the proportion of null values ​​in the process data is less than the first threshold, the null value part will be filled; if the proportion is not less than the first threshold, the process data will be deleted directly.

[0095] For outliers in the process data, if the proportion of outliers in the process data is greater than the second threshold, the process data is directly deleted. If the proportion of outliers in the process data is not greater than the second threshold, linear interpolation is used to replace non-continuous outliers, and spline interpolation is used to replace multiple consecutive outliers.

[0096] For duplicate data, duplicate data is deleted directly. For null values, if the proportion of null values ​​in the feature column is small, the KNN method is used to fill them; if the proportion is large, the feature column is deleted directly. For outlier identification, box plots are usually used. Similarly, feature columns with a large proportion of outliers are deleted directly. If the proportion is small, non-continuous outliers are replaced by linear interpolation, and multiple consecutive outliers are replaced by spline interpolation.

[0097] S2. Based on the preprocessed data that includes classification, integration and cleaning, the influencing parameters are obtained by feature screening of the selected key blast furnace condition parameters through correlation analysis.

[0098] like Figure 3 As shown, step S2 includes:

[0099] S21. Select coke ratio, permeability index, and gas utilization rate as key blast furnace operating parameters. Based on the criteria of high output, low consumption, high quality, smooth operation, and long service life, select key blast furnace operating parameters.

[0100] S22. Standardize the data after cleaning to eliminate the impact of differences in units and ranges of values ​​between data.

[0101] S23. Based on standardized data, correlation analysis is used to screen features for the coke ratio, permeability index, and gas utilization rate, respectively, to obtain the influence characteristics corresponding to these three parameters. Pearson regression, MIC regression, and stepwise regression methods are then used to screen features for the coke ratio, permeability index, and gas utilization rate, respectively.

[0102] S24. Among the influencing features corresponding to coke ratio, permeability index, and gas utilization rate, sort them from largest to smallest influencing factor, and take the union of the top N as the influencing parameters of coke ratio, permeability index, and gas utilization rate. The influencing parameters of each feature are sorted from largest to smallest influencing factor, and the union of the top three influencing parameters of each feature is taken as the final influencing parameter. Specifically, the influencing features of target coke ratio are (actual value of material flow valve opening_ore, feeding speed_coke, sintered Al2O3; it is worth noting that the subscripts including "_ore" and "_coke" are feature names, representing only the feature names), the influencing features of permeability index are (hot air pressure, average top pressure, pulverized coal injection rate), and the influencing features of gas utilization rate are (pelletized FeO, coke Ad (%), coke Mad (%)). Then, the influence characteristics of each target are combined and used as the final influence parameters, which are denoted as X1-9 (actual value of material flow valve opening_ore, feeding speed_coke, sintering Al2O3, hot air pressure, average top pressure, pulverized coal injection rate, pellet FeO, coke Ad (%), coke Mad (%)).

[0103] S3. Construct a multivariate linear fitting function for the objective and constraint functions based on the key furnace condition parameters and influencing parameters of the blast furnace, and obtain the optimized operation solution set by solving the multivariate linear fitting function.

[0104] like Figure 4 As shown, step S3 includes:

[0105] S31. Based on the key furnace condition parameters and influencing parameters of the blast furnace, construct a multivariate linear fitting function for the objective and constraint functions.

[0106] S32. The following process is used to solve the multivariate linear fitting function using the ε-constraint optimization algorithm: select the coke ratio as the main optimization objective, and the permeability index and gas utilization rate as objective constraints. By solving the thresholds of the permeability index and gas utilization rate, objective constraints are added to the coke ratio. Finally, the solution of the coke ratio under the constraint conditions is obtained, which is the optimization operation solution set.

[0107] The ε-constraint method is based on the principle of using a certain objective function as the main optimization objective and other objective functions as constraints to solve a multi-objective optimization problem. According to the on-site operation policy and actual conditions, the coke ratio is selected as the main optimization objective, while the permeability index and gas utilization rate are used as objective constraints. By solving the threshold values ​​of the permeability index and gas utilization rate, objective constraints are added to the coke ratio, and finally the solution of the coke ratio under the constraint conditions is obtained.

[0108] Specifically, the ε-constraint method is used to solve multi-objective constrained problems. The principle is to use one objective function as the optimization objective and constrain other objective functions to solve the multi-objective optimization problem. The specific steps are as follows:

[0109] The first step is to determine the objective function and constraints;

[0110] Objective function: A multiple linear fitting function was established using Python with the target focal ratio, air permeability index, and influencing parameters. The fitting function achieved a hit rate of 85%, and the expression is as follows:

[0111] f1(x)=228+4.63x1+5.57x2+10.28x3-0.00006x4-0.39x5+0.000035x6-3.3x7

[0112] -3.4x8 -3.8x9

[0113] f2(x)=73.65-0.12x1-0.13x2-0.2x3-0.44x4+0.87x5+0.0004x6+1.0x7-0.45x8

[0114] -0.25x9

[0115] f3(x)=0.56-0.00004x1-0.00004x2-0.00075x3-0.00009x4-0.0012x5

[0116] +0.0000001x6+0.007x7+0.01x8-0.026x9

[0117] The constraints include:

[0118] 1. Constraints on each influencing parameter: The lower limit constraint value for each parameter is the 2% quantile of the historical data, and the upper limit constraint value is the 98% quantile of the historical data; specifically: x1(49.6, 52.8), x2(1.4, 3.5), x3(1.39, 2.23), x4(322.9, 343), x5(184.5, 186.5), x6(16666, 19073), x7(1.22, 1.99), x8(12.9, 13.5), x9(0.215, 0.423).

[0119] 2. Constraints between the target coke ratio, permeability index, gas utilization rate, and their respective influencing parameters; constraints on the target coke ratio, permeability index, gas utilization rate, and their respective influencing parameters using linear fitting with Python; the constraint intervals for each parameter and equation are set according to on-site operating policies and actual conditions, and the constraint expressions and thresholds are as follows:

[0120] y1=106.1+4.58x1+5.39x2+11.76x3

[0121] y2=65.5-0.45x4+0.88x5+0.0003x6

[0122] y3=0.314+0.007x7+0.01x8-0.026x9

[0123] In the above formula, f1(x), f2(x), and f3(x) represent the coke ratio, permeability index, and gas utilization rate, respectively. y1, y2, and y3 are the constraint equations for the coke ratio, permeability index, and gas utilization rate, respectively. x1-x9 are the influencing parameters, which are the actual value of the material flow valve opening (ore), the feeding speed (coke), the sintering Al2O3, the hot air pressure, the average top pressure, the pulverized coal injection rate, the pellet FeO, the coke Ad (%), and the coke Mad (%).

[0124] The second step is to solve for the single objective;

[0125] Based on the on-site operational guidelines and actual conditions, all objectives are prioritized. Then, the coke ratio is selected as the main optimization objective. The optimal solutions for each objective function under the above constraints are solved, with the coke ratio and permeability index to be minimized, and the gas utilization rate to be maximized. The following solutions are obtained:

[0126] X 1 (49.6, 1.4, 1.39, 343, 186.5, 16666, 1.99, 13.5, 0.423) T f1(x) = 353.5X 2(52.8, 2.9, 1.39, 343, 184.5, 16666, 1.22, 13.5, 0.423) T f2(x) = 77.96X 3 (49.6, 1.4, 1.39, 322.9, 184.5, 19073, 1.99, 13.5, 0.215) T f3(x) = 0.452. Then, the solutions obtained for each objective are substituted into the other objective functions to calculate the objective function values ​​for each solution. The maximum value M and minimum value m of the objective function values ​​for each solution are then selected. The corresponding results are shown in the table below:

[0127] <![CDATA[f1(x)]]> <![CDATA[f2(x)]]> <![CDATA[f3(x)]]> <![CDATA[X 1 ]]> 353.5 81.05 0.442 <![CDATA[X 2 ]]> 380.1 72.96 0.439 <![CDATA[X 3 ]]> 355.1 89.17 0.452 M 380.1 89.17 0.452 m 353.5 72.96 0.439

[0128] The third step is to determine other target thresholds;

[0129] The step size t is determined by r, where r is any integer greater than 1, and t = 0, 1, ..., r-1. Using the maximum value M and minimum value m of each objective function value from step three, the threshold of each objective is solved using the following formula:

[0130]

[0131] In the formula, i is the number of objective functions, r is any integer greater than 1, and t is the step size determined by r. Choosing r = 7, we obtain the following table:

[0132] t 0 1 2 3 4 5 6 <![CDATA[ε2(x)]]> 72.96 75.66 78.36 81.07 83.77 86.47 89.17 <![CDATA[ε3(x)]]> 0.439 0.441 0.443 0.446 0.448 0.450 0.452

[0133] Based on the optimization objectives set on-site, the cases of t=0, 1, and 2 are initially discarded.

[0134] The fourth step is to solve the problem.

[0135] Using the threshold values ​​of each objective obtained in step three, constraints are added to the target permeability index and gas utilization rate: f i (x)≤ε i i = 2, ..., n, serving as the objective constraints, mathematical programming methods are used to solve for the minimum focal ratio of the main objective under the above constraints and objective constraints, ultimately yielding 4 sets of solutions, as follows:

[0136] t=3, x1 (49.6, 1.4, 1.39, 343, 186.5, 16666, 1.99, 13.5, 0.423) T

[0137] f1(x)=353.49, f2(x)=81.04, f3(x)=0.442

[0138] t=4, x1 (51.6, 1.4, 1.39, 343, 186.5, 16666, 1.99, 13.5, 0.423) T

[0139] f1(x)=362.87, f2(x)=80.54, f3(x)=0.442

[0140] t=5, x1 (50.2, 1.4, 2.23, 343, 186.2, 17645, 1.99, 13.5, 0.28) T

[0141] f1(x)=365.8, f2(x)=80.96, f3(x)=0.445

[0142] t=6, x1 (50.6, 2.4, 1.89, 343, 186.2, 17645, 1.46, 13.5, 0.28) T

[0143] f1(x)=371.3, f2(x)=80.33, f3(x)=0.442.

[0144] S4. Using a pre-defined case library, perform multi-level matching and combine it with feasibility analysis to classify and sort the solutions in the optimized operation solution set, thereby obtaining the optimal operation solution with the minimum operation deviation and the lowest economic cost.

[0145] Before step S4, the process also includes: using the selected key blast furnace condition parameters as operating condition indicators, and determining the operating condition indicator function based on the operating condition indicators; judging whether the called blast furnace historical data is an excellent value that meets the conditions based on the operating condition indicator function, and establishing a blast furnace case library based on the excellent values.

[0146] The operating condition index function is as follows:

[0147]

[0148] In the formula, α is the weighting coefficient of each quantity, which is set to 0.4, 0.3, and 0.3 respectively based on the priority ranking of the working condition indicators on site; i is the number of working condition indicators selected; and k is the actual value of the working condition indicator. * These are the expected values ​​for each operating condition indicator.

[0149] Based on the comprehensive operating condition evaluation index, the weighted coefficient is based on a P value greater than 0 and less than 0.03 as the excellent index. The P value corresponding to the blast furnace data under different operating modes is calculated to determine whether the data under the current operating mode is excellent, and a blast furnace case library is established.

[0150] Furthermore, such as Figure 5 As shown, step S4 includes:

[0151] S41. Perform cluster analysis on the blast furnace case library to obtain the number of subclasses and the cluster center of each subclass; prioritize each subclass using the operating condition index function, and then calculate the distance between each optimized operation solution in the optimized operation solution set and the cluster center of each subclass to convert it into a similarity value, which is used to determine the subclass to which each optimal solution belongs, thus completing the initial matching.

[0152] Based on a blast furnace case library, the number of subclasses is first set, and the FCM clustering algorithm is used to perform cluster analysis on the case library to obtain the cluster centers of each subclass. Next, the operating condition index function is used to prioritize each subclass. The data from each subclass are substituted into the operating condition index function, and its mean P is calculated. The subclasses are then prioritized based on the mean P. Then, the distance between each optimal solution and each cluster center is calculated and converted into a similarity value. The subclass to which each optimal solution belongs is determined based on the maximum similarity value of each optimal solution, completing the initial matching process.

[0153] S42. Calculate the similarity value between each optimal solution and all data in its subclass, select the case database data corresponding to the maximum similarity value, and sort the selected case database data using the working condition index function and its subclass order to complete the secondary matching.

[0154] After performing initial matching of the case library based on the previous optimal solution, the subclass to which the optimal solution belongs is determined. Then, the similarity value between each optimization operation solution and all data in its subclass is calculated. The case library data corresponding to the maximum similarity value is selected as the similar case of each optimization operation solution (that is, a blast furnace historical data that is closest to each optimal solution is selected from the corresponding subclass of each optimal solution. Assuming there are five optimal solutions, the five blast furnace historical data that are closest to these five solutions are selected from the case library).

[0155] Secondary matching is achieved by sorting similar cases of each optimization operation solution according to the working condition index function and the order of each subclass. (Secondary matching involves further prioritizing the matched optimal solutions, based on the working condition index function and the order of each subclass. The working condition index function is as described above. The case library data that best matches the optimal solution is substituted into the working condition index function to calculate its P-value. The smaller the P-value, the higher the ranking. The order of each subclass is to avoid an extreme case: if some optimal solutions have the same P-value (this rarely happens since the values ​​are different, but the possibility must be considered). In this case, we sort according to their subclass. Since the subclasses were prioritized using the working condition index function, if the case library data have the same P-value, the case library data is sorted according to the order of its subclass (if the P-value is the same, the higher the ranking of its subclass, the higher the ranking of this case library data). That is, secondary matching is achieved by sorting similar cases of each optimization operation solution according to the working condition index function and the order of each subclass.

[0156] S43. Through feasibility analysis, each optimal operation solution in the optimal operation solution set is screened based on the ease of operation and the operation cost. The ease of operation includes changes in the blast furnace operation system and the magnitude of operation deviations. The operation cost includes changes in raw materials and fuels and the impact data on production.

[0157] Feasibility analysis mainly screens the optimal solution set based on the ease of operation and the operating cost; the ease of operation includes changes in the blast furnace operating system and the magnitude of operational deviations; the operating cost includes changes in raw materials and fuels and their impact on production.

[0158] S44. Based on the on-site operation data, prioritize each parameter in the influencing parameters to determine the adjustable parameters and their order.

[0159] S45. Calculate the data deviation between the optimal solution set after secondary matching and the actual value of blast furnace data under the current operating mode, and sort the data deviation according to the adjustable parameters and the order of the adjustable parameters to obtain the deviation sort of the solutions in the optimal solution set after secondary matching.

[0160] S46. Calculate the cost of solutions in the optimized solution set after completing the secondary matching.

[0161] S47. Based on the deviation sorting and the cost consumption, obtain the optimal operation solution with the minimum operation deviation and the lowest economic cost.

[0162] Based on the filtered impact parameters, priority is assigned according to on-site operations. The data deviation between the actual blast furnace data values ​​and the optimal solutions in the current operating mode is calculated after matching the case library. The deviations are then ranked according to their magnitude. Simultaneously, cost consumption under the optimal solutions, such as changes in coke, ore, and blast furnace costs, is calculated. Finally, the solution with the smallest operational deviation and lowest economic cost is selected, and suggestions are provided to the blast furnace foreman to guide blast furnace production.

[0163] Furthermore, after step S4, the method also includes: pushing the solution with the minimum operational deviation and the lowest economic cost to the blast furnace operator's terminal to complete the optimization of blast furnace production operations.

[0164] like Figure 6 As shown, this embodiment of the invention provides a blast furnace key furnace condition parameter optimization system based on industrial big data, including: a data management subsystem and a multi-objective optimization dynamic control subsystem.

[0165] The data management subsystem includes:

[0166] The data acquisition module is used to collect data from all processes during the blast furnace smelting process.

[0167] The preprocessing module is used to perform preprocessing on the acquired data from all processes in the blast furnace smelting process, including classification, integration, and cleaning; and,

[0168] A case library is used to store cases containing blast furnace data and operating parameters. A raw database storing historical blast furnace data is also included.

[0169] The multi-objective optimization dynamic control subsystem includes:

[0170] The multi-objective optimization module is used to select key blast furnace condition parameters based on preprocessed data, and then use correlation analysis to screen the features of the selected key blast furnace condition parameters to obtain the influencing parameters. Then, based on the key blast furnace condition parameters and the influencing parameters, a multivariate linear fitting function of objective and constraint functions is constructed, and the optimization operation solution set is obtained by solving the multivariate linear fitting function.

[0171] The feedback module is used to perform multi-level matching using a preset case library and combine it with feasibility analysis to classify and sort the solutions in the optimized operation solution set, so as to obtain the optimal operation solution with the minimum operation deviation and the lowest economic cost.

[0172] Meanwhile, this invention also provides a blast furnace key condition parameter optimization device based on industrial big data, comprising: at least one database; and a memory communicatively connected to at least one database; wherein the memory stores instructions that can be executed by at least one database, and the instructions are executed by at least one database to enable at least one database to execute the blast furnace key condition parameter optimization method based on industrial big data as described above.

[0173] Furthermore, embodiments of the present invention also provide a computer-readable medium storing computer-executable instructions, characterized in that the executable instructions, when executed by a processor, implement the above-described method for optimizing key blast furnace condition parameters based on industrial big data.

[0174] In summary, this invention provides a method, system, equipment, and medium for optimizing key blast furnace parameters. The data sources include comprehensive data from the entire process, such as raw material and fuel parameters, charging parameters, blast parameters, slag and iron parameters, and key furnace condition parameters. This data is closely integrated with actual on-site production, resulting in more comprehensive data. Rigorous data preprocessing, including data reduction, data cleaning, correlation analysis, and feature filtering, ensures the accuracy and reliability of data stored in the database and input to the model. Using the feature-filtered blast furnace parameters and key furnace condition parameters (coke ratio, permeability index, and gas utilization rate), a multivariate linear fit is constructed using Python. Then, the ε-constraint optimization algorithm is used to solve the multi-objective problem, yielding reliable calculation results. Based on the solution results, the optimal operation solution set is classified and sorted using multi-level matching and feasibility analysis in a case library. Finally, the solution with the minimum operational deviation and lowest economic cost is selected, providing suggestions to the blast furnace foreman to guide blast furnace production.

[0175] Therefore, compared with the prior art, the present invention has the following beneficial effects:

[0176] (1) The data on blast furnaces is extensive and closely integrated with the field, ensuring the comprehensiveness and reliability of the data.

[0177] (2) It can accurately judge and optimize furnace condition parameters, thereby optimizing blast furnace operation and achieving high output, low consumption, high quality, long service life and stable operation of blast furnace.

[0178] (3) It provides theoretical support for blast furnace operators to judge the furnace condition, which has great significance for production practice.

[0179] (4) Update the model in real time based on blast furnace data to improve model accuracy.

[0180] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0181] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0182] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0183] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0184] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0185] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0186] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A blast furnace key furnace condition parameter optimization method based on industrial big data, characterized in that, include: The acquired data from all processes in the blast furnace smelting process are classified, integrated, and cleaned. Based on the classified, integrated, and cleaned data, the key blast furnace condition parameters were selected and their influencing parameters were obtained through correlation analysis. These parameters included: selecting coke ratio, permeability index, and gas utilization rate as key blast furnace condition parameters; standardizing the cleaned data; and using correlation analysis including Pearson, MIC, and stepwise regression to screen the coke ratio, permeability index, and gas utilization rate, respectively, to obtain the influencing characteristics corresponding to these three parameters. Among the influencing characteristics corresponding to each of the three parameters (coke ratio, permeability index, and gas utilization rate), they were sorted from largest to smallest influencing factor, and the union of the top N values ​​was taken as the influencing parameters for coke ratio, permeability index, and gas utilization rate. Based on the key blast furnace condition parameters and the influencing parameters, a multivariate linear fitting function of objective and constraint functions is constructed. The optimized operation solution set is obtained by solving the multivariate linear fitting function, including: constructing a multivariate linear fitting function of objective and constraint functions based on the key blast furnace condition parameters and the influencing parameters; and using the ε-constraint optimization algorithm to solve the multivariate linear fitting function as follows: selecting coke ratio as the main optimization objective, and permeability index and gas utilization rate as objective constraints; adding objective constraints to the coke ratio by solving the thresholds of permeability index and gas utilization rate; and finally obtaining the solution of coke ratio under the constraint conditions, which is used as the optimized operation solution set. By using a pre-set case library for multi-level matching and combining it with feasibility analysis, the solutions of the optimized operation solution set are classified and sorted to obtain the optimal operation solution with the minimum operation deviation and the lowest economic cost. The feasibility analysis is based on the analysis of the difficulty of operation and the analysis of the operation cost. The difficulty of operation includes changes in the blast furnace operation system and the magnitude of the operation deviation. The operation cost includes changes in raw materials and fuels and the impact data on production.

2. The blast furnace key condition parameter optimization method based on industrial big data according to claim 1, characterized in that, The acquisition of full-process data in the blast furnace smelting process involves classification, integration, and cleaning, including: Obtain full-process data of the blast furnace smelting process; All process data are categorized into different types based on the same process. These different types include raw material data, pulverized coal injection data, air supply data, cooling data, and slag and iron data at the blast furnace site. Data from different processes are linked using a defined time-series index to integrate multi-source data, followed by the data cleaning process: For duplicate data in the process data, simply delete the duplicate data. For null values ​​in process data, if the proportion of null values ​​in the process data is less than the first threshold, the null value part will be filled; if the proportion is not less than the first threshold, the process data will be deleted directly. For outliers in the process data, if the proportion of outliers in the process data is greater than the second threshold, the process data is directly deleted. If the proportion of outliers in the process data is not greater than the second threshold, linear interpolation is used to replace non-continuous outliers, and spline interpolation is used to replace multiple consecutive outliers.

3. The method for optimizing key blast furnace condition parameters based on industrial big data as described in claim 1, characterized in that, The multivariate linear fitting function is: ； The constraints include: Constraints on various influencing parameters: x 1 (49.6, 52.8), x 2 (1.4, 3.5) x 3 (1.39, 2.23). x 4 (322.9, 343). x 5 (184.5, 186.5), x 6 (16666, 19073). x 7 (1.22, 1.99). x 8 (12.9, 13.5) x 9 (0.215, 0.423); Constraints between target coke ratio, permeability index, gas utilization rate and their respective influencing parameters: ； In the above formula, f 1( x ), f 2( x ), f 3( x These represent the coke ratio, permeability index, and gas utilization rate, respectively. y 1. y 2. y 3 are the constraint equations for coke ratio, permeability index, and gas utilization rate, respectively. x 1- x 9 refers to the following influencing parameters: actual value of material flow valve opening, feeding speed, sintering Al2O3, hot air pressure, average top pressure, pulverized coal injection rate, pellet FeO, coke Ad (%), and coke Mad (%).

4. The blast furnace key condition parameter optimization method based on industrial big data according to claim 1, characterized in that, Before classifying and sorting the solutions in the optimized operation solution set using a preset case library for multi-level matching and combining feasibility analysis to obtain the optimal operation solution with the minimum operation deviation and lowest economic cost, the process also includes: The selected key blast furnace condition parameters are used as operating condition indicators, and the operating condition indicator function is determined based on the operating condition indicators. Based on the operating condition index function, determine whether the called blast furnace historical data is an excellent value that meets the conditions, and establish a blast furnace case library based on the excellent values. The operating condition index function is as follows: ； In the formula The weighting coefficients for each quantity are 0.4, 0.3, and 0.3, respectively, based on the priority ranking of the working condition indicators on site. i Select the number of operating condition indicators; k These are the actual values ​​of the operating condition indicators. These are the expected values ​​for each operating condition indicator.

5. The blast furnace key condition parameter optimization method based on industrial big data according to claim 4, characterized in that, By using a pre-defined case library for multi-level matching and combining it with feasibility analysis, the solutions in the optimized operation solution set are classified and sorted to obtain the optimal operation solution with the minimum operation deviation and the lowest economic cost, including: Cluster analysis is performed on the blast furnace case library to obtain the number of subclasses and the cluster center of each subclass; the subclasses are prioritized by the working condition index function, and then the distance between each optimization operation solution in the optimization operation solution set and the cluster center of each subclass is calculated and converted into a similarity value to determine the subclass to which each optimal solution belongs, thus completing the initial matching. Calculate the similarity value between each optimal solution and all data in its subclass, select the case database data corresponding to the maximum similarity value, and sort the selected case database data using the working condition index function and its subclass order to complete the secondary matching; Feasibility analysis is used to filter each optimized operation solution in the optimized operation solution set based on the ease of operation and the cost of operation; Based on the on-site operation data, the parameters in the influencing parameters are prioritized to determine the adjustable parameters and their order of priority. The data deviation between the optimal solution set after secondary matching is calculated and the actual value of blast furnace data under the current operating mode is obtained. The data deviation is sorted according to the adjustable parameters and the order of the adjustable parameters to obtain the deviation sort of the solutions in the optimal solution set after secondary matching is completed. Calculate the cost of solutions in the optimized solution set after completing the secondary matching; Based on the deviation ranking and the cost consumption, the optimal operation solution with the minimum operation deviation and the lowest economic cost is obtained.

6. The blast furnace key parameters optimization method based on industrial big data according to claim 1, characterized in that, After classifying and sorting the solutions in the optimized operation solution set using a preset case library for multi-level matching and combining feasibility analysis to obtain the optimal operation solution with the minimum operation deviation and the lowest economic cost, the process further includes: The solution with the minimum operational deviation and the lowest economic cost is pushed to the blast furnace operator's control terminal to optimize blast furnace production operations.

7. A blast furnace key condition parameter optimization system based on industrial big data, characterized in that, The system, which is described by any one of claims 1-6, comprises: a data management subsystem and a multi-objective optimization dynamic control subsystem; The data management subsystem includes: The data acquisition module is used to collect data from all processes during the blast furnace smelting process. The preprocessing module is used to classify, integrate, and clean the acquired data from all processes in the blast furnace smelting process; and, Case library, used to store cases containing blast furnace data and operating condition indicators; The multi-objective optimization dynamic control subsystem includes: The multi-objective optimization module is used to obtain influencing parameters by feature screening of selected key blast furnace condition parameters based on classified, integrated and cleaned data through correlation analysis. Then, based on the key blast furnace condition parameters and the influencing parameters, a multivariate linear fitting function of objective and constraint functions is constructed, and the optimization operation solution set is obtained by solving the multivariate linear fitting function. The feedback module is used to perform multi-level matching using a preset case library and combine it with feasibility analysis to classify and sort the solutions of the optimized operation solution set, so as to obtain the optimal operation solution with the minimum operation deviation and the lowest economic cost.

8. A blast furnace key condition parameter optimization device based on industrial big data, characterized in that, include: At least one database; And a memory that is communicatively connected to the at least one database; The memory stores instructions that can be executed by the at least one database, which are then executed by the at least one database to enable the at least one database to perform the method for optimizing key blast furnace condition parameters based on industrial big data as described in any one of claims 1-6.

9. A computer readable medium having stored thereon computer- executable instructions, characterized in that, When the executable instructions are executed by the processor, they implement the method for optimizing key blast furnace condition parameters based on industrial big data as described in any one of claims 1-6.

Images