Method for evaluating furnace condition of blast furnace based on data self-learning strategy

Abstract

The invention belongs to the technical field of metallurgical blast furnace control, and provides a method for evaluating the furnace condition of a blast furnace based on a data self-learning strategy. Statistical processing of historical data; setting boundary parameters of the evaluation model; according to the characteristics of the furnace condition parameters, the furnace condition parameters are evaluated and scored according to the three modes; and performing modular scoring and comprehensive evaluation according to production characteristics and furnace condition indexes of the blast furnace. According to the method, statistics is taken as a mechanism, the furnace condition monitoring management process is perfected, quantitative furnace condition evaluation indexes can be provided for furnace operators, the furnace operators can conveniently and comprehensively master furnace condition information, stable and efficient operation of the blast furnace is facilitated, and the method has remarkable guiding significance and practical value for blast furnace operation.

Description

Technical Field

[0001] The invention belongs to the technical field of metallurgical blast furnace control and relates to a method for evaluating blast furnace conditions, in particular to a method for evaluating blast furnace conditions based on a data self-learning strategy. Background Art

[0002] For blast furnace operation, blast furnace evaluation and diagnosis are the core of blast furnace operation. With the continuous improvement of blast furnace monitoring methods, the monitoring parameters of blast furnaces are constantly increasing. How to use a large number of monitoring parameters to achieve a systematic and scientific evaluation of blast furnaces has become a problem that needs to be solved. Regarding the evaluation of blast furnace conditions, different blast furnaces have different responses. Most blast furnaces are mainly based on extensive and qualitative evaluations, and have not formed a scientific and systematic evaluation system. In addition, the evaluation is mainly based on manual judgment. The evaluation method is not scientific enough and needs further optimization. Summary of the invention

[0003] The purpose of the present invention is to provide a method for evaluating blast furnace conditions based on data self-learning strategies, a data self-learning furnace condition evaluation model based on statistics, and a data mining method to systematically evaluate the blast furnace, providing quantitative evaluation indicators for furnace conditions for furnace operators, so as to facilitate furnace operators to fully grasp furnace condition information, deepen furnace condition monitoring and management, and provide a quantitative basis for furnace condition control.

[0004] The technical solution adopted by the present invention to achieve the above-mentioned purpose is:

[0005] A method for evaluating blast furnace conditions based on a data self-learning strategy comprises the following steps:

[0006] S1. Data preprocessing: Select key operating indicators of the blast furnace, eliminate data items that are too high or too low, and retain data within a reasonable range.

[0007] S2. Historical data statistical processing: obtain the 5% quantile P5 and 95% quantile P of the parameter within a certain period of time 95 , median m and standard deviation σ.

[0008] S3. Set the limit parameters of the evaluation model: According to the characteristics of furnace parameters, three modes and five types are used to set the limit parameters. 95 , m and σ are used to calculate the limit parameters.

[0009] Mode 1: When the parameter index value X is as low as possible, the lower limit L1 is P5+σ×ε and the upper limit is P 95

[0010] Mode 2: When the parameter index value X is higher, the lower limit L1 is P 95 , the upper limit is P 95 -σ×ε

[0011] Mode 3, when the parameter index value X is centered, the lower limit L1 is P5, the lower limit L2 is m×a+p5×(1-a), and the upper limit U2 is m×a+p 95 ×(1-a), the upper limit U1 is p 95 ;

[0012] When the parameter index value X is biased towards the middle and lower, the lower limit L1 is p5-σ×ε, the lower limit L2 is m×b+p5×(1-b), the upper limit U2 is m, and the upper limit U1 is p 95 -σ×ε;

[0013] When the parameter index value X is biased towards the middle, the lower limit L1 is p5+σ×ε, the lower limit L2 is m, and the upper limit U2 is m×b+p 95 ×(1-b), the upper limit U1 is p 95 +σ×ε;

[0014] Among them, ε is the standard deviation coefficient, generally ranging from 0.2 to 0.4; a is ranging from 0.75 to 0.9; and b is ranging from 0.55 to 0.7.

[0015] The limit parameters L1, L2, U1, and U2 can be automatically and dynamically updated by analyzing historical data based on the data self-learning strategy.

[0016] S4. According to the characteristics of furnace condition parameters, the furnace condition parameters are evaluated and scored in three modes:

[0017] Mode 1: The lower the parameter index value X, the better. The score below the lower limit L1 is 100, and the score above the upper limit U1 is 0. The evaluation scoring formula between L1 and U1 is: Single item score R = (U1-X) / (U1-L1)×100

[0018] Mode 2: The higher the parameter index value X, the better. The score below the lower limit L1 is 0, and the score above the upper limit U1 is 100. The evaluation scoring formula between L1 and U1 is: Single item score R = (X-U1) / (U1-L1)×100

[0019] Mode 3: The parameter index X value is better in the middle, that is, the score is 100 when it is between the lower limit L2 and the upper limit U2. The farther away from this interval, the lower the score. If it exceeds the lower limit L1 or the upper limit U1, the score is 0. The evaluation scoring formula between L1~L2 or U2~U1 is: single item score R=(X-L1) / (L2-L1)×100 or R=(U1-X) / (U1-U2)×100

[0020] S5. Comprehensive evaluation of furnace conditions: According to the production characteristics of blast furnaces, furnace condition indicators are divided into five aspects: pressure-volume relationship, airflow distribution, material feeding, furnace heat level and technical indicators for module scoring and comprehensive evaluation;

[0021] The comprehensive evaluation steps include:

[0022] S51. According to the characteristics of blast furnace production process, the total weights λ of the five furnace condition indicators are specified. 压量关系 , 气流分布 , 下料情况 , 炉热水平 and λ 技术指标 , keeping the sum of the five items at 100%.

[0023] S52, determine the weight of specific parameter indicators. According to the blast furnace production and smelting principles, production process rules and the importance of parameters in the overall furnace condition indicators, a certain weight η is assigned to each parameter. The weight coefficient φ of each parameter can be obtained by the formula φ=λ×η;

[0024] S53, the individual scores of the n furnace condition parameters are accumulated according to the assigned weights to obtain the comprehensive score of the furnace condition.

[0025]

[0026] The beneficial effects of the present invention are as follows: the method can provide furnace operators with quantitative evaluation indicators for furnace conditions, which is convenient for furnace operators to fully grasp furnace condition information, is beneficial to the stable and efficient operation of the blast furnace, and has significant guiding significance and practical value for blast furnace operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flow chart for evaluating blast furnace conditions based on data self-learning strategy;

[0028] Figure 2 The weight setting and data statistics table of furnace condition parameters of the embodiment;

[0029] Figure 3 It is the limit parameter setting table of the furnace condition evaluation model;

[0030] Figure 4 The furnace condition parameter index limit parameters and evaluation result statistics table of the embodiment;

[0031] In the figure, type 3-1 has better parameters centered, type 3-2 has better parameters in the lower middle, and type 3-3 has better parameters in the upper middle. DETAILED DESCRIPTION

[0032] The present invention will be further described below in conjunction with embodiments:

[0033] Example

[0034] Take the method of the present invention to evaluate the condition of the company's 6# blast furnace as an example:

[0035] The process of evaluating furnace condition is as follows Figure 1 As shown,

[0036] S1. Data preprocessing

[0037] Since abnormal data will affect the rationality of the quantitative evaluation of blast furnaces, it is necessary to set rules for deleting abnormal data to ensure data quality. The method of deleting abnormal data is to select key operating indicators of blast furnaces, such as coke ratio, fuel ratio and output, remove data items that are too high or too low, and retain data within a reasonable range.

[0038] In actual production, data items with coke ratio ≥ 390kg / tFe, fuel ratio ≥ 535kg / tFe and output ≤ 9000t / day are first eliminated.

[0039] S2. Furnace parameter weight setting and data statistical processing

[0040] Obtain the historical data of the parameter to be analyzed in the last 1 to 2 years, and obtain the 5% quantile P5 of the parameter (that is, after sorting the historical data from small to large, delete the first 5% of the data and take the minimum value), the 95% quantile P 95 (That is, after the historical data is sorted from large to small, the first 5% of the data is deleted and the maximum value is taken), as well as the median m and standard deviation σ of the parameter.

[0041] In actual production, according to the production characteristics of 6# blast furnace, the λ and η of each blast furnace condition parameter raw fuel are determined, and the weight coefficient φ of each parameter is calculated; the percentile parameter P of each parameter in the last two years is counted. 5％ , P 95 , m and σ. Furnace condition parameter weight setting and data statistics are as follows Figure 2 As shown in the table.

[0042] S3. Limit parameter setting

[0043] According to the characteristics of furnace parameters and statistical results, Figure 2 The evaluation model boundary parameter setting method in the evaluation model boundary parameter setting table can obtain the boundary parameters of various indicators such as Figure 3 As shown in the table.

[0044] S4. Furnace condition parameter evaluation and scoring

[0045] According to the characteristics of furnace parameters, the evaluation and scoring are carried out in three modes, as follows:

[0046] Mode 1: The lower the parameter index value X, the better. The score below the lower limit L1 is 100, and the score above the upper limit U1 is 0. The evaluation scoring formula between L1 and U1 is:

[0047] R=(U1-X) / (U1-L1)×100 (Formula 1)

[0048] Mode 2: The higher the parameter index value X, the better. The score below the lower limit L1 is 0, and the score above the upper limit U1 is 100. The evaluation scoring formula between L1 and U1 is:

[0049] R = (X-L1) / (U1-L1) × 100 (Formula 2)

[0050] Mode 3: The parameter index value is better in the middle, that is, the score is 100 when it is between the lower limit L2 and the upper limit U2. The farther away from this interval, the lower the score. If it exceeds the lower limit L1 or the upper limit U1, the score is 0. The evaluation scoring formula between L1 and L2 or U2 and U1 is:

[0051] R = (X-L1) / (L2-L1)×100 or R = (U1-X) / (U1-U2)×100 (Formula 3)

[0052] Substituting the limit parameters of each indicator obtained by S3 into the evaluation scoring formula 1, formula 2 or formula 3, the single score R of each raw material and fuel indicator can be obtained, such as Figure 4 As shown in the table;

[0053] S5. Comprehensive evaluation of furnace conditions

[0054] First, according to the characteristics of blast furnace production process, the total weights λ of the five furnace condition indicators are specified. 压量关系 , 气流分布 , 下料情况 , 炉热水平 and λ 技术指标 , keeping the sum of the five items at 100%.

[0055] Then, determine the weight of specific parameter indicators. According to the blast furnace production and smelting principles, production process rules and the importance of parameters in the overall furnace condition indicators, a certain weight η is given to each parameter.

[0056] Thus, the weight coefficient of each parameter can be obtained: φ = λ × η (Formula 4)

[0057] The comprehensive score of the furnace condition can be obtained by adding up the individual scores of the n furnace condition parameters according to the assigned weights:

[0058]

[0059] The single scores of the furnace condition parameters obtained in S4 are accumulated according to the assigned weights to obtain the module score of the furnace condition, among which the pressure-volume relationship is 69.5 points, the airflow distribution is 62.5 points, the material feeding situation is 79.9 points, the furnace heat level is 84.6 points, and the technical index is 59.0 points. Finally, the comprehensive weighting is used to obtain the comprehensive score of the furnace condition, which is 67.5 points. In this way, the overall level of the 6# blast furnace condition can be determined, providing a reliable quantitative basis for the blast furnace condition control.

[0060] The method of the present invention is applied to the company's No. 6 blast furnace, and the daily output of molten iron in the blast furnace is increased from 9559t / d to 9957t / d. According to the normal production and operation of the steel enterprise, the economic benefit brought to the enterprise by each ton of molten iron is at least 200 yuan / tFe. The annual economic benefit generated after the output is increased is: (9957-9557)×360×200 / 10000×40%=11.462 million yuan.

[0061] The method of the present invention combines the actual blast furnace production, technically uses the data quantitative analysis method, and establishes a method for evaluating the blast furnace condition based on the data self-learning strategy, which can provide quantitative evaluation furnace condition indicators for furnace operators, facilitate furnace operators to fully grasp the furnace condition information, and is conducive to the stable and efficient operation of the blast furnace, and has significant guiding significance and practical value for blast furnace operation. The method of the present invention can bring better economic benefits and has extremely high promotion value.

Claims

1. A method for evaluating blast furnace conditions based on data self-learning strategy, characterized in that: The steps include: S1. Data preprocessing: Select key operating indicators of the blast furnace, eliminate data items that are too high or too low, and retain data within a reasonable range. S2. Historical data statistical processing: obtain the 5% quantile P5 and 95% quantile P of the parameter within a certain period of time 95 , median m and standard deviation σ. S3. Set the limit parameters of the evaluation model: According to the characteristics of furnace parameters, it is divided into three modes and five types for setting the limit parameters. 95 , m and σ are used to calculate the limit parameters. S4. According to the characteristics of furnace condition parameters, the furnace condition parameters are evaluated and scored in three modes: Mode 1: The lower the parameter index value X, the better. The score below the lower limit L1 is 100, and the score above the upper limit U1 is 0. The evaluation scoring formula between L1 and U1 is: single item score R = (U1-X) / (U1-L1)×100; Mode 2: The higher the parameter index value X, the better. The score below the lower limit L1 is 0, and the score above the upper limit U1 is 100. The evaluation scoring formula between L1 and U1 is: single item score R = (X-U1) / (U1-L1)×100; Mode 3: The parameter index X value is better in the middle, that is, the score is 100 when it is between the lower limit L2 and the upper limit U2. The farther away from this interval, the lower the score. If it exceeds the lower limit L1 or the upper limit U1, the score is 0. The evaluation scoring formula between L1~L2 or U2~U1 is: single item score R=(X-L1) / (L2-L1)×100 or R=(U1-X) / (U1-U2)×100. S5. Comprehensive evaluation of furnace conditions: According to the production characteristics of blast furnaces, furnace condition indicators are divided into five aspects: pressure-volume relationship, airflow distribution, material feeding conditions, furnace heat level and technical indicators for module scoring and comprehensive evaluation.

2. A method for evaluating blast furnace conditions based on data self-learning strategy according to claim 1, characterized in that: The three modes and five types of limit parameter settings are: Mode 1: When the parameter index value X is as low as possible, the lower limit L1 is P5+σ×ε and the upper limit is P 95 ; Mode 2: When the parameter index value X is higher, the lower limit L1 is P 95 , the upper limit is P 95 -σ×ε; Mode 3, when the parameter index value X is centered, the lower limit L1 is P5, the lower limit L2 is m×a+p5×(1-a), and the upper limit U2 is m×a+p 95 ×(1-a), the upper limit U1 is p 95 ; When the parameter index value X is biased towards the middle and lower, the lower limit L1 is p5-σ×ε, the lower limit L2 is m×b+p5×(1-b), the upper limit U2 is m, and the upper limit U1 is p 95 -σ×ε; When the parameter index value X is biased towards the middle, the lower limit L1 is p5+σ×ε, the lower limit L2 is m, and the upper limit U2 is m×b+p 95 ×(1-b), the upper limit U1 is p 95 +σ×ε; in ε is the standard deviation coefficient, generally ranging from 0.2 to 0.4; a ranges from 0.75 to 0.9; and b ranges from 0.55 to 0.

7.

3. A method for evaluating blast furnace conditions based on data self-learning strategy according to claim 2, characterized in that: The limit parameters L1, L2, U1, and U2 can be automatically and dynamically updated by analyzing historical data based on a data self-learning strategy.

4. The method for evaluating blast furnace conditions based on data self-learning strategy according to claim 1, characterized in that: The comprehensive evaluation steps described in S5 include: S51. According to the characteristics of blast furnace production process, the total weights λ of the five furnace condition indicators are specified. 压量关系 , 气流分布 , 下料情况 , 炉热水平 and λ 技术指标 , keeping the sum of the five items at 100%. S52. According to the blast furnace production and smelting principle, the production process law and the importance of the parameter in the overall furnace condition index, a certain weight η is assigned to each parameter, and the weight coefficient φ of each parameter is obtained by the formula: φ=λ×η; S53, summing up the individual scores of the n furnace condition parameters according to the assigned weights, to obtain a comprehensive score of the furnace condition

Images