Intelligent control system for optimizing iron ore blending structure

Abstract

The present invention provides an intelligent control system for optimizing the mixed structure of iron ore, which uses a storage unit to store the data of each mixed smelting process of multiple ore varieties in history, and performs function fitting based on the historical data, and obtains the extreme value of the fitted function, so as to obtain the optimal solution of the content of each ore variety for the next mixed smelting. The whole system can be digitally fully automatically controlled, and can be continuously iterated over time, so that the function and the related optimal solution are constantly approaching the ideal value.

Description

Technical Field

[0001] The invention relates to an intelligent control system for optimizing iron ore blending structure. Background Art

[0002] Iron ore blending refers to the mixing of iron ores of different types and properties in a specific proportion and method to achieve the purpose of optimizing smelting effects, increasing metal recovery rates and reducing production costs. Iron ore blending plays an important role in metallurgical industries such as steel production.

[0003] Traditional iron ore blending methods often have problems such as low efficiency, large errors, and waste of resources. Optimizing and controlling the iron ore blending process through intelligent means can significantly improve smelting efficiency and product quality while reducing production costs. Summary of the invention

[0004] The present invention provides an intelligent control system for optimizing the iron ore blending structure, which effectively solves the problems existing in the prior art mentioned above.

[0005] Specifically, the present invention provides an intelligent control system for optimizing the mixed structure of iron ore, which is used to control the mixed smelting process of iron ore, wherein the iron ore includes M types of ore, M is an integer and M≥2, The system includes a storage unit, a computing unit, and a decision unit. The storage unit stores the content of each ore variety in each of the N mixed smelting processes of the iron ore in history, N is an integer and N≥3, and stores a parameter vector composed of T smelting index parameter values ​​of each mixed smelting process, T is an integer and T≥2. The computing unit first executes a vector distance calculation step to calculate the parameter vector distance between the parameter vector of each mixed smelting process and a preset reference vector, and then executes a function fitting step to fit an M-dimensional fitting function with the content of the M ore varieties as independent variables and the parameter vector distance as dependent variables based on the parameter vector distance in each mixed smelting process and the content of each ore variety, and then executes a function solving step to find the lower limit value of the M-dimensional fitting function and the optimal solution of the independent variable corresponding to the lower limit value; the decision unit takes the content of the M ore varieties as the optimal solution, thereby starting the N+1th mixed smelting process.

[0006] In particular, M=3, and iron ore includes three ore varieties, which are hematite, limonite, and magnetite.

[0007] In particular, in the vector distance calculation step, the T types of smelting index parameters are D 1 , D 2 ,…D T, set an integer k, k is taken from 1 to N in sequence, after the kth iron ore blending smelting process is completed, the T smelting index parameter values ​​are calculated as D k1 , D k2 ,…D kT , forming a parameter vector [D k1 , D k2 ,…D kT ], the reference values ​​of the T smelting index parameters are D 01 , D 02 ,…D 0T , thus forming the reference vector [D 01 , D 02 ,…D 0T ], based on the parameter vector and the reference vector, calculate the kth parameter vector distance L k .

[0008] Specifically, the k-th parameter vector distance L k The calculation formula is as follows:

[0009] .

[0010] Preferably, in the function fitting step, as k is sequentially changed from 1 to N, the contents of the M ore varieties in the first mixed smelting process are [C 11 , C 12 ,…,C 1M ], the corresponding vector distance L 1 , and the contents in the second mixing and smelting process are [C 21 , C 22 ,…,C 2M ], the corresponding vector distance L 2 , and so on, in the Nth mixing and smelting process, the contents are [C N1 , C N2 ,…,C NM ], the corresponding vector distance L N , set the common content of M ore varieties in each mixed smelting process as C 1 , C 2 ,…,C M , then the content of these M kinds of ore varieties is taken as the independent variable, and the universal vector distance is L, which is taken as the dependent variable, and the M-variable function L=f(C 1 , C 2 ,…,C M ).

[0011] In particular, in the function solving step, for the M-ary function L=f(C 1 , C 2 ,…,CM ) to obtain the lower limit value, that is, Lmin, the solution of the lower limit value corresponding to the independent variable of M yuan is C 0 1 , C 0 2 ,…,C 0 M , Lmin = f(C 0 1 , C 0 2 ,…,C 0 M ).

[0012] In particular, in the decision-making unit, during the N+1th mixed smelting process, the contents of the M ore varieties will be respectively taken as the solution C of the lower limit value. 0 1 , C 0 2 ,…,C 0 M , and thus start the N+1th mixed smelting process.

[0013] Optionally, T types of smelting index parameter values ​​include four parameters: phosphorus content, sulfur content, energy consumption, and melting temperature, that is, T=4.

[0014] In summary, the present invention provides an intelligent control system for optimizing the mixed structure of iron ore, which uses a storage unit to store the data of each mixed smelting process of multiple ore varieties in history, and performs function fitting based on the historical data, and obtains the extreme value of the fitted function, so as to obtain the optimal solution of the content of each ore variety for the next mixed smelting. The entire system can be digitally fully automatically controlled, and can be continuously iterated over time, so that the function and the related optimal solution continue to approach the ideal value. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be discussed below. Obviously, the technical solutions described in conjunction with the drawings are only some embodiments of the present invention. For ordinary technicians in this field, other embodiments and their drawings can be obtained based on the embodiments shown in these drawings without paying creative work.

[0016] Figure 1 A general flow chart of the intelligent control system for optimizing the iron ore blending structure according to the present invention is shown. DETAILED DESCRIPTION

[0017] The technical solutions of various embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments described in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Figure 1 The following is a general flow chart of the intelligent control system for optimizing the iron ore blending structure according to the present invention. Figure 1 The basic structure and operation of the intelligent control system for optimizing the iron ore blending structure according to the present invention are described in detail.

[0019] In summary, the present invention provides an intelligent control system for optimizing the structure of iron ore blending. As mentioned above, iron ore blending is to mix raw iron ores of different types and properties in a certain proportion to achieve the best smelting effect. Therefore, this system aims to explore the influence of the proportion of various types of raw iron ores on the final smelting effect. The current challenge is that there is no ready-made rule or formula that can be applied to the law between the mixing ratio of each component of ore blending and the final smelting effect. Therefore, the present invention attempts to explore the relationship between the two based on historical data.

[0020] It is assumed that the iron ore to be mixed includes M ore varieties, where M is an integer and M≥2. For example, among the M ore varieties, some may be hematite, the main component of which is Fe 2 O 3 , usually containing 60% to 70% iron, some may be magnetite, the main component of which is Fe 3 O 4 , usually containing 72% to 74% iron, some may be limonite, the main component of which is 2Fe 2 O 3 ·3H 2 O, the iron content is low, usually between 40% and 60%.

[0021] It is assumed that the iron ore has gone through N times of blending and smelting in history, where N is an integer and N ≥ 3. The reason why N is set to be no less than 3 is that only when N is no less than 3 can the most basic statistical significance be formed.

[0022] In any k-th iron ore blending and smelting process, the content of each of the M ore varieties in the iron ore to be blended is C k1 , C k2 ,…,C kM, where k ranges from 1 to N. For example, in the second iron ore blending process, M=3, which means there are three types of ores, namely hematite, limonite, and magnetite. The content of hematite in this blending process is 30%, then C 21 =30%, the content of limonite in this mixed smelting is 40%, then C 22 =40%, the content of magnetite in this mixed smelting is 30%, then C 23 =30%. As k changes from 1 to N, the contents of various ore varieties involved in the first mixed smelting process to the Nth mixed smelting process are stored in the storage unit.

[0023] After the kth iron ore blending smelting process is completed, the smelting result index parameter value is also produced and stored in the storage unit accordingly. The smelting result index parameter includes T parameter values, which can be calculated as D k1 , D k2 ,…D kT , where T is an integer and T ≥ 2, thus forming the parameter vector [D k1 , D k2 ,…D kT ]. Parameter benchmarks D are established for these smelting result indicators. 01 , D 02 ,…D 0T , thus forming the reference vector [D 01 , D 02 ,…D 0T ] The operation unit further starts the vector distance calculation step to calculate the kth parameter vector distance L k .

[0024] The parameter vector distance L k The calculation formula is as follows:

[0025] ,

[0026] Different product structures may affect smelting parameters and ultimately affect smelting costs. Therefore, in practical applications, indicator parameters often need to select several indicators that are most sensitive to smelting costs. For example, the indicator parameter values ​​after smelting include: phosphorus content, sulfur content, energy consumption, and melting temperature, that is, T=4. The four indicator parameters selected in this example can be considered to be the factors that are most sensitive to smelting costs in the actual situation of this example (of course, this is not a fixed rule. Other indicator parameters that are extremely sensitive to smelting costs can also be introduced or replaced). The parameter benchmarks for these four parameters are phosphorus content of 0.2%, sulfur content of 0.3%, energy consumption of 2MWh / ton, and melting temperature of 1400℃. After the second iron ore was mixed and smelted, the measured parameters formed after testing were: phosphorus content of 0.22%, sulfur content of 0.25%, energy consumption of 2.2MWh / ton, and melting temperature of 1450℃. Then the parameter vector distance L 2 The calculation is:

[0027] ,

[0028] Calculate L 2 =0.222.

[0029] Thus, the k-th iron ore blending smelting process forms the input data, that is, the above parameter vector [D k1 , D k2 ,…D kT ], and also forms the result data, that is, the kth parameter vector distance L k .

[0030] Here, the kth parameter vector distance L k is considered as the dependent variable.

[0031] The content of each of the M types of ore in the iron ore to be mixed is C k1 , C k2 ,…,C kM The content vector [C k1 , C k2 ,…,C kM ], in the kth mixed smelting process, the contents of these M iron ore varieties can be used as M independent variables to input into the mixed smelting system, and after the mixed smelting is completed, the smelting index parameters are formed, thereby outputting the dependent variable L k .

[0032] Next, the calculation unit starts the function fitting step. As k changes from 1 to N, the content vectors input in each mixing and smelting process in history are [C 11 , C 12 ,…,C 1M ]、[C 21 , C22 ,…,C 2M ], …, [C N1 , C N2 ,…,C NM ], the vector distance of the one-to-one output as the dependent variable is L 1 , L 2 ,…,L N .

[0033] In other words, the common content of M ore varieties in each mixed smelting process is set as C 1 , C 2 ,…,C M , then the content of these M ore varieties can be used as independent variables, and the vector distance of the common smelting index parameters is calculated as L as the dependent variable.

[0034] In other words, the content of the first variety C 1 The actual content of C in the N mixed smelting processes in history is 11 , C 21 ,…,C N1 ; The content of the second variety C 2 The actual content of C in the N mixed smelting processes in history is 12 , C 22 ,…,C N2 ; Similarly, the content of the Mth variety C M The actual content of C in the N mixed smelting processes in history is 1M , C 2M ,…,C NM .

[0035] Then based on the historical data of the previous N times, the independent variable is C 1 , C 2 ,…,C M The change of and the change of the parameter vector distance L as the dependent variable can fit the multivariate function L=f(C 1 , C 2 ,…,C M ).

[0036] Regarding the fitting of multivariate functions based on historical data, I will not go into details here. For example, it can be implemented using statistical analysis software using linear regression or least squares method, as shown in the following code:

[0037] import numpy as np

[0038] import pandas as pd

[0039] from sklearn.model_selection import train_test_split

[0040] from sklearn.linear_model import LinearRegression

[0041] # Load data (assuming data is stored in a DataFrame)

[0042] df = pd.DataFrame({

[0043] 'X1': [values],

[0044] 'X2': [values],

[0045] 'Y': [values]

[0046] })

[0047] # Define independent and dependent variables

[0048] X = df[['X1', 'X2']]

[0049] Y = df['Y']

[0050] X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.2, random_state=42)

[0051] # Fit the model

[0052] model = LinearRegression()

[0053] model.fit(X_train, Y_train)

[0054] # Get regression coefficients

[0055] coefficients = model.coef_

[0056] intercept = model.intercept_

[0057] Then, the computing unit starts the function solving step. For the multivariate function L=f(C 1 , C 2 ,…,C M) can be used to obtain the lower limit value, that is, Lmin, the solution of the lower limit value corresponding to the independent variable of M yuan is C 0 1 , C 0 2 ,…,C 0 M , in other words, Lmin = f(C 0 1 , C 0 2 ,…,C 0 M ).

[0058] It should be explained here that the lower limit value Lmin represents the value closest to the reference vector in the current function, and the reference vector represents the most reasonable smelting parameter in the ideal. Therefore, the solution C corresponding to the lower limit value Lmin is 0 1 , C 0 2 ,…,C 0 M It represents the optimal solution whose function value is closest to the most reasonable smelting result parameters.

[0059] Therefore, in the next current N+1th mixed smelting process, the decision-making unit makes relevant decisions, and the contents of the M ore varieties will be respectively taken as the solution of the lower limit value C 0 1 , C 0 2 ,…,C 0 M , and thus start the N+1th mixed smelting process.

[0060] After the N+1th mixed smelting process is completed, the optimal solution for the content of each ore type in the first N+1 mixed smelting process can be obtained based on the numerical values ​​extracted from the first N+1 mixed smelting process, and the optimal solution can be used for the proposed content of the ore types in the N+2th mixed smelting process.

[0061] Of course, preferably, the above optimal solution C 0 1 , C 0 2 ,…,C 0 M Originally, a limiting condition needs to be met, that is, the M optimal solutions represent the content of all ore types, so the sum of the M optimal solutions should be equal to 1 (i.e., 100%). In actual applications, the sum of the optimal solutions calculated may deviate from 1 due to errors, so the above optimal solution can be improved.

[0062] First, find the sum MC of the M optimal solutions:

[0063] ,

[0064] If MC deviates from 1, that is, MC≠1, then normalization is performed and C 1 1 = C 0 1 / MC,C 1 2 = C 0 2 / MC, and so on, until C 1 M = C 0 M / MC. And the value C 1 1 , C 1 2 ,…,C 1 M Replacement Solution C 0 1 , C 0 2 ,…,C 0 M Assigned to the content of each ore type in the N+1th mixed smelting process.

[0065] It can be seen that the iron ore blending structure optimization intelligent control system provided by the present invention is essentially a specific iterative system, and in the process of continuous iteration, the content of each ore variety is constantly approaching the theoretical optimal solution. For example, when N=3, that is, only three blending smelting processes have been experienced in history, and the historical data that can be obtained are relatively limited, so the fitting of the multivariate function is relatively rough, and the optimal solution obtained to guide the input of the fourth ore content is as close to the ideal value as possible, but it is still relatively biased from the ideal value; and when the fourth blending smelting process is completed, it is updated to N=4, and the input and output data in the fourth blending smelting process are also included in the historical data. Due to the addition of historical data that is closer to the ideal value than the first three blending smeltings, the four historical data obtained in the case of N=4 are closer to the ideal value than the first three historical data obtained in the case of N=3, so the decision on the ore content made in the case of N=4 is also more accurate. In this way, as time goes by and the number of mixed smelting increases, the value of N increases progressively, and the decision data made becomes more reasonable as the actual mixed smelting operation progresses.

[0066] The above description is only an exemplary embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An intelligent control system for optimizing the mixing structure of iron ore, used to control the mixing and smelting process of iron ore, characterized in that: Iron ore includes M types of ore, M is an integer and M≥2. The system includes a storage unit, a computing unit, and a decision unit. The storage unit stores the content of each ore type in each of the N mixed smelting processes of the iron ore in history, N is an integer and N≥3, and stores a parameter vector composed of T smelting index parameter values ​​of each mixed smelting process, T is an integer and T≥2, The operation unit first executes a vector distance calculation step to calculate the parameter vector distance between the parameter vector of each mixed smelting process and a preset reference vector, then executes a function fitting step to fit an M-dimensional fitting function with the content of M ore varieties as independent variables and the parameter vector distance as dependent variable based on the parameter vector distance in each mixed smelting process and the content of each ore variety, and then executes a function solving step to find the lower limit value of the M-dimensional fitting function and the optimal solution of the independent variable corresponding to the lower limit value; The decision unit takes the content of the M types of ore as the optimal solution, thereby starting the N+1th mixed smelting process. In the vector distance calculation step, the T types of smelting index parameter values ​​are D1, D2, ...D T , set an integer k, k is taken from 1 to N in sequence, after the kth iron ore blending smelting process is completed, the T smelting index parameter values ​​are calculated as D k1 , D k2 ,…D kT , forming a parameter vector [D k1 , D k2 ,…D kT ], the reference values ​​of the T smelting index parameter values ​​are D 01 , D 02 ,…D 0T , thus forming the reference vector [D 01 , D 02 ,…D 0T ], based on the parameter vector and the reference vector, calculate the kth parameter vector distance L k , The k-th parameter vector distance L k The calculation formula is as follows: ， In the function fitting step, as k changes from 1 to N, the contents of the M ore varieties in the first mixed smelting process are [C 11 , C 12 ,…,C 1M ], the corresponding vector distance L1, the contents in the second mixing and smelting process are [C 21 , C 22 ,…,C 2M ], the corresponding vector distance L2, and so on, in the Nth mixing and smelting process, the contents are [C N1 , C N2 ,…,C NM ], the corresponding vector distance L N , Set the common content of the M types of ores in each mixed smelting process as C1, C2, ..., C M , then the contents of these M ore varieties are taken as independent variables, and the universal vector distance is L, which is taken as the dependent variable, and the M-element fitting function L=f(C1, C2, ..., C M ).

2. The system according to claim 1, characterized in that M=3, iron ore includes three kinds of ore varieties, and the three kinds of ore varieties are hematite, limonite and magnetite.

3. The system according to claim 1, characterized in that In the function solving step, for the M-element fitting function L=f(C1, C2, ..., C M ) to obtain the lower limit value, that is, Lmin, the solution of the lower limit value corresponding to the M-element independent variable in the M-element fitting function is C 0 1. C 0 2. ..., C 0 M , Lmin = f(C 0 1. C 0 2. ..., C 0 M ).

4. The system according to claim 3, characterized in that In the decision-making unit, during the N+1th mixed smelting process, the contents of the M ore varieties will be respectively taken as the solution of the lower limit value C 0 1. C 0 2. ..., C 0 M , and thus start the N+1th mixed smelting process.

5. The system according to claim 1, characterized in that T types of smelting index parameter values ​​include: phosphorus content, sulfur content, energy consumption, and melting temperature, that is, T=4.

Images