CN115058555B - Intelligent soft measurement method and system for measuring carbon content of converter endpoint - Google Patents
Intelligent soft measurement method and system for measuring carbon content of converter endpoint Download PDFInfo
- Publication number
- CN115058555B CN115058555B CN202210848243.5A CN202210848243A CN115058555B CN 115058555 B CN115058555 B CN 115058555B CN 202210848243 A CN202210848243 A CN 202210848243A CN 115058555 B CN115058555 B CN 115058555B
- Authority
- CN
- China
- Prior art keywords
- carbon content
- predicted
- converter
- model
- converter stage
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 title claims abstract description 104
- 229910052799 carbon Inorganic materials 0.000 title claims abstract description 104
- 238000000691 measurement method Methods 0.000 title claims abstract description 24
- 238000000034 method Methods 0.000 claims abstract description 76
- 230000008569 process Effects 0.000 claims abstract description 41
- 238000005259 measurement Methods 0.000 claims abstract description 38
- 238000012549 training Methods 0.000 claims description 40
- 238000007664 blowing Methods 0.000 claims description 24
- 230000006870 function Effects 0.000 claims description 24
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 claims description 22
- 229910052760 oxygen Inorganic materials 0.000 claims description 22
- 239000001301 oxygen Substances 0.000 claims description 22
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 claims description 12
- 230000004927 fusion Effects 0.000 claims description 12
- 229910002091 carbon monoxide Inorganic materials 0.000 claims description 10
- UGFAIRIUMAVXCW-UHFFFAOYSA-N Carbon monoxide Chemical compound [O+]#[C-] UGFAIRIUMAVXCW-UHFFFAOYSA-N 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 238000005261 decarburization Methods 0.000 claims description 7
- 229910002092 carbon dioxide Inorganic materials 0.000 claims description 6
- 239000001569 carbon dioxide Substances 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 description 13
- 229910000831 Steel Inorganic materials 0.000 description 6
- 230000008859 change Effects 0.000 description 6
- 239000010959 steel Substances 0.000 description 6
- 238000006243 chemical reaction Methods 0.000 description 5
- 230000007246 mechanism Effects 0.000 description 5
- 238000009628 steelmaking Methods 0.000 description 5
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 4
- 230000003068 static effect Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000012937 correction Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000010219 correlation analysis Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000010365 information processing Effects 0.000 description 2
- 229910052742 iron Inorganic materials 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- 238000003909 pattern recognition Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000000611 regression analysis Methods 0.000 description 2
- 238000003723 Smelting Methods 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 239000002826 coolant Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 239000000428 dust Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000003546 flue gas Substances 0.000 description 1
- 238000004868 gas analysis Methods 0.000 description 1
- 230000008571 general function Effects 0.000 description 1
- RRAMGCGOFNQTLD-UHFFFAOYSA-N hexamethylene diisocyanate Chemical compound O=C=NCCCCCCN=C=O RRAMGCGOFNQTLD-UHFFFAOYSA-N 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- CSJDCSCTVDEHRN-UHFFFAOYSA-N methane;molecular oxygen Chemical compound C.O=O CSJDCSCTVDEHRN-UHFFFAOYSA-N 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000004886 process control Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
Images
Classifications
-
- C—CHEMISTRY; METALLURGY
- C21—METALLURGY OF IRON
- C21C—PROCESSING OF PIG-IRON, e.g. REFINING, MANUFACTURE OF WROUGHT-IRON OR STEEL; TREATMENT IN MOLTEN STATE OF FERROUS ALLOYS
- C21C5/00—Manufacture of carbon-steel, e.g. plain mild steel, medium carbon steel or cast steel or stainless steel
- C21C5/28—Manufacture of steel in the converter
- C21C5/30—Regulating or controlling the blowing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Materials Engineering (AREA)
- Organic Chemistry (AREA)
- Metallurgy (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Manufacturing & Machinery (AREA)
- Carbon Steel Or Casting Steel Manufacturing (AREA)
Abstract
The invention relates to an intelligent soft measurement method and system for measuring the carbon content of a converter endpoint, and belongs to the technical field of intelligent soft measurement. Firstly, input data are acquired, then, the converting process of the converter is divided into a plurality of converter stages, each converter stage corresponds to a pre-trained predictor model, the input data are respectively used as input of each predictor model, initial predicted values of the terminal carbon content corresponding to each predictor model at the predicting time are obtained, finally, the initial predicted values of the terminal carbon content at the predicting time are fused, final predicted values of the terminal carbon content at the predicting time are obtained, therefore, the characteristics of each converter stage in the converting process of the converter, the dependency relationship among each converter stage and the time correlation of auxiliary variables are considered in the soft measuring process, and the high-precision measurement of the terminal carbon content is realized.
Description
Technical Field
The invention relates to the technical field of intelligent soft measurement, in particular to an intelligent soft measurement method and system for measuring the carbon content of a converter endpoint.
Background
Soft measurement is to organically combine knowledge of the production process, apply computer technology to the variables to be estimated which are difficult to measure or cannot be measured temporarily, select other auxiliary variables which are easy to measure, infer or estimate by forming a certain mathematical relationship, and replace the function of hardware with software. The soft measurement technology is applied to realize the online detection of the content of the element components, so that the method is economical and reliable, has rapid dynamic response, can continuously give the content of the element components, and is easy to control the quality of the product. Therefore, the soft measurement technology can be widely applied to various fields of steel and chemical industry, and has great value for the measurement and application of key process parameters which are difficult to measure in various dangerous occasions.
Currently, there are three main types of methods for soft measurement of the endpoint carbon content of a converter: manual experience method, static estimation method, dynamic estimation method. The manual experience method mainly relies on means such as observation, sampling analysis and the like, the measurement accuracy is closely related to the experience level of operators, and the accuracy and the stability of measurement are difficult to ensure. The static estimation method is based on historical data statistics, and combines static data (molten iron amount, scrap steel amount, coolant amount and the like) before converting is started, and calculates the estimated value of the carbon content of each converter stage of the converter by taking the final carbon content as a target. The dynamic estimation method is the current mainstream and foremost method, but the prior art scheme has the following problems: firstly, the characteristics of each converter stage in the converting process are not considered, the whole converting process is regarded as a dynamic process to carry out single modeling and estimate the carbon content, the converter has different reaction characteristics in different converter stages of converting, the single model modeling is carried out on the long-period complex dynamic of converting, and the estimation accuracy is necessarily limited; secondly, the dependency among converter stages of the converter is not considered, and the converter can undergo reaction change of multiple converter stages in the converting process, wherein each converter stage is a precondition of the next converter stage (for example, oxygen can be supplemented in the next converter stage according to the current converting converter stage condition); third, in the existing dynamic estimation method, the time correlation of the auxiliary variable is not considered, the auxiliary variable related to the carbon content (such as the temperature of molten iron, the oxygen blowing amount of an oxygen lance, the scrap steel ratio and the like) is directly used as the modeling input, the time-dependent change characteristic of the auxiliary variable is not considered, and in the dynamic process of blowing, the auxiliary variable is often changed along with time according to the process control, and the change characteristic is not negligible for estimating the end point carbon content. Based on the problems, the dynamic estimation method has the defect of low end point carbon content measurement precision.
Based on this, there is a need for an intelligent soft measurement method and system that can measure endpoint carbon content with high accuracy.
Disclosure of Invention
The invention aims to provide an intelligent soft measurement method and system for measuring the carbon content of a converter endpoint, which are used for considering the characteristics of each converter stage in the blowing process of the converter in the soft measurement process, the dependency relationship among each converter stage and the time correlation of auxiliary variables, and realizing the high-precision measurement of the carbon content of the endpoint.
In order to achieve the above object, the present invention provides the following solutions:
an intelligent soft measurement method for measuring carbon content of a converter endpoint, comprising the following steps:
acquiring input data; the input data includes a plurality of continuous historical values of endpoint carbon content prior to a predicted time, a real-time value of each auxiliary variable at the predicted time, and a plurality of continuous historical values of each auxiliary variable prior to the predicted time; the auxiliary variables comprise carbon monoxide content, carbon dioxide content, decarburization amount, oxygen blowing intensity and oxygen blowing time;
dividing the converting process of the converter into a plurality of converter stages, wherein each converter stage corresponds to a pre-trained predictor model; respectively taking the input data as the input of each predictor model to obtain an initial predicted value of the end point carbon content corresponding to each predictor model at the predicted moment;
and fusing all the initial predicted values of the end point carbon content to obtain a final predicted value of the end point carbon content at the predicted moment.
An intelligent soft measurement system for converter endpoint carbon content measurement, the intelligent soft measurement system comprising:
the data acquisition module is used for acquiring input data; the input data includes a plurality of continuous historical values of endpoint carbon content prior to a predicted time, a real-time value of each auxiliary variable at the predicted time, and a plurality of continuous historical values of each auxiliary variable prior to the predicted time; the auxiliary variables comprise carbon monoxide content, carbon dioxide content, decarburization amount, oxygen blowing intensity and oxygen blowing time;
the converter stage prediction module is used for dividing the converting process of the converter into a plurality of converter stages, and each converter stage corresponds to a pre-trained predictor model; respectively taking the input data as the input of each predictor model to obtain an initial predicted value of the end point carbon content corresponding to each predictor model at the predicted moment;
and the fusion module is used for fusing all the initial predicted values of the end point carbon content to obtain a final predicted value of the end point carbon content at the predicted moment.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention provides an intelligent soft measurement method and system for measuring the endpoint carbon content of a converter. And then dividing the converting process of the converter into a plurality of converter stages, wherein each converter stage corresponds to a pre-trained predictive sub-model, and input data is respectively used as the input of each predictive sub-model to obtain the initial predicted value of the end carbon content corresponding to each predictive sub-model at the predicted time. And finally, fusing all initial predicted values of the end point carbon content to obtain final predicted values of the end point carbon content at the predicted moment, so that the characteristics of each converter stage in the converting process of the converter, the dependency relationship among each converter stage and the time correlation of auxiliary variables are considered in the soft measurement process, and the high-precision measurement of the end point carbon content is realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of the intelligent soft measurement method according to embodiment 1 of the present invention;
FIG. 2 is a schematic block diagram of an intelligent soft measurement method according to embodiment 1 of the present invention;
fig. 3 is a system block diagram of an intelligent soft measurement system according to embodiment 2 of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention aims to provide an intelligent soft measurement method and system for measuring the carbon content of a converter endpoint, which are used for considering the characteristics of each converter stage in the blowing process of the converter in the soft measurement process, the dependency relationship among each converter stage and the time correlation of auxiliary variables, and realizing the high-precision measurement of the carbon content of the endpoint.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
Example 1:
the basic idea of soft measurement is to select a group of auxiliary variables (also called secondary variables) which are closely related to the dominant variable (namely the variable to be measured or the variable to be estimated) and are easy to measure according to a certain optimal criterion, and estimate the dominant variable by constructing a certain mathematical relationship through computer software. The mechanisms of different production processes vary widely, and there are various methods for establishing mathematical models between an estimated process variable (i.e., a dominant variable) and other process variables related to the process variable (i.e., auxiliary variables), such as mechanism modeling, statistical regression modeling, and artificial neural network modeling.
In many converter steelmaking primary dust removal systems, the prediction of the endpoint carbon content of the converter is a dream pursued by several generations of automation engineers, and from an early static model to a dynamic model to a process computer model, the endpoint carbon content value of the converter cannot be accurately predicted, so that the endpoint carbon content value is estimated through means such as a multi-gun or flue gas analysis, and the like, which is laborious and high in cost. After the BQ-SMART dry dedusting intelligent system is developed, the converter endpoint carbon adopting the dry system is predicted conditionally through an intelligent soft measurement technology.
Based on this, the present embodiment is configured to provide an intelligent soft measurement method for measuring a carbon content at a converter endpoint, as shown in fig. 1, where the intelligent soft measurement method includes:
s1: acquiring input data; the input data includes a plurality of continuous historical values of endpoint carbon content prior to a predicted time, a real-time value of each auxiliary variable at the predicted time, and a plurality of continuous historical values of each auxiliary variable prior to the predicted time; the auxiliary variables comprise carbon monoxide content, carbon dioxide content, decarburization amount, oxygen blowing intensity and oxygen blowing time;
the expression of the input data s (k) of the present embodiment is:
wherein k is the predicted time; y (k-m) y ) Representing the dominant variable (i.e. endpoint carbon content), integer m y Representing the maximum delay order of the dominant variable in time; x is x n (k-m n ) Represents the nth auxiliary variable in the converter converting process, and the integer m n Representing the maximum delay order of the nth auxiliary variable in time. The maximum delay order in time refers to the number of consecutive history values of the variable.
Based on the input data, the embodiment can define the dynamic process of converter converting as
Because of the nonlinear relationship->
Comprising a dominant variable y (k) and an auxiliary variable x n (k) Can thus be regarded as a non-linear relation +.>
Containing elements such as x n (k-m n +1)-x n (k-m n ) Such change values reflect the change process of the main variable and the auxiliary variable, so that the time change characteristics of the main variable and the auxiliary variable in the converting process can be obtained by nonlinear relation
The method is realized, so that the problem of time correlation of the auxiliary variable is solved, wherein the problem of time correlation of the auxiliary variable is not considered in the existing method, and the time correlation of the auxiliary variable is integrated into data input of a prediction process.
Specifically, in this embodiment, the number of continuous history values of each auxiliary variable is different, that is, different maximum time delay orders are selected for different auxiliary variables.
In this embodiment, the method for determining the number of continuous history values of each auxiliary variable may include:
(1) And selecting the number of continuous historical values of each auxiliary variable as the initial number according to the characteristics of the auxiliary variable in converter steelmaking.
For example, carbon in molten steel is converted into CO and CO by chemical reaction 2 And so on, the process is continued throughout the converter steelmaking process, so that a larger number of history values can be appropriately selected. The oxygen blowing is to blow oxygen as a reaction factor, and the number of the history values can be flexibly selected according to the total duration of oxygen blowing.
(2) In the actual implementation process, based on the initial number of the first step, the number of the historical values is appropriately increased or reduced to perform a comparison training experiment, for example, the initial number n=5 is selected, the situations of n=3, 4,5,6,7 and the like can be subjected to comparison training to obtain a training result, the optimal number of the training result is selected as the optimal number of the auxiliary variables, and finally the optimal number of the historical values of each auxiliary variable is selected by utilizing the initial value and a plurality of groups of training modes.
The initial value is selected according to the process characteristics and expert experience, the times of multiple groups of experiments are reduced, and the initial value is effectively close to the optimal value. The multiple training modes are combined with the actual condition of a specific converter (the data generated by the converter is reflected), and the dynamic characteristics of dominant variables implicit in the data are mined. The mode of experience and data is more accurate, and different complex working conditions of different converters can be matched.
S2: dividing the converting process of the converter into a plurality of converter stages, wherein each converter stage corresponds to a pre-trained predictor model; respectively taking the input data as the input of each predictor model to obtain an initial predicted value of the end point carbon content corresponding to each predictor model at the predicted moment;
specifically, before the input data are respectively used as the inputs of the prediction sub-models, the intelligent soft measurement method of the embodiment further includes: training to obtain a predictor model for each converter stage may include:
for each converter stage, the following steps are performed: a training dataset is acquired. The training data set comprises a plurality of training input data and a training endpoint carbon content predicted value corresponding to each training input data, and modeling is carried out by adopting process mechanism analysis, regression analysis, artificial neural network, pattern recognition, fuzzy mathematics, state estimation, correlation analysis or nonlinear information processing based on the training data set to obtain a predictor model.
The training data sets for the respective converter stages are not identical. In this embodiment, a data set may be first established, where the data set includes a historical value of the input data shown in S1 and an endpoint carbon content predicted value corresponding to the historical value, and then the following two ways are used to determine each converter stage and a training data set of each converter stage: (1) And determining each converter stage according to the manual division of expert experience, and determining a training data set of each converter stage. (2) And clustering the data sets by using a fuzzy clustering algorithm (such as a Gustafson-Kessel algorithm), wherein the number after clustering is used as the number of each converter stage, and various data are used as training data sets of each converter stage.
The following eight modeling modes can be used for establishing a mathematical model for predicting the endpoint carbon content of the converter: soft measurement modeling based on process mechanism analysis; soft measurement modeling based on regression analysis; modeling based on soft measurement of an artificial neural network; modeling soft measurements based on pattern recognition; soft measurement modeling based on fuzzy mathematics; soft measurement modeling based on state estimation; modeling soft measurements based on correlation analysis; soft measurement modeling based on modern nonlinear information processing techniques. And a modeling mode which is most suitable for predicting the endpoint carbon content of the converter is selected by combining big data analysis and a process mechanism.
More specifically, to further obtain a nonlinear relationship
In consideration of the different degrees of influence of auxiliary variables of different converter stages on the dominant variables, the present embodiment uses fuzzy T-S (Takagi-Sugeno) rule for establishing predictive sub-models of each converter stage of converter blowing according to the model definitions given above>
The fuzzy T-S condition is expressed as follows:
R i :if s(k) is A i ,then y(k)＝f i (s(k)),i＝1,2,…,P；
wherein R is i Represents the ith converter stage; a is that i Is the front piece fuzzy set of the ith converter stage; p represents the number of converter stages of converter converting; s (k) is used as a front variable of the model, and y (k) is used as a back variable of the model.
Front proposition part "s (k) is A i "is typically expressed as a logical combination of simple propositions, and simple propositions are defined under a fuzzy set of univariate, but the fuzzy set of variables is built separately for each component in s (k), and is defined by s j To represent. The front part of the proposition can thus be modified into the following form:
sub-model function f in a back-piece i The structure of (s (k)) remains the same in the various converter stages of the converter. Due to the history of the dominant variable and the auxiliaryThe historic values of the auxiliary variables have different influence on the final predicted value, so the submodel function f in the back-piece i (s (k)) is selected as:
in order to obtain a specific expression of the rule, firstly, a fuzzy clustering operation is required to be carried out on a dominant variable and an auxiliary variable of the converter, and a fuzzy set A can be obtained from the fuzzy clustering operation ij The predictor model can be obtained by a least squares identification method. Based thereon, the step of training to obtain a predictor model for each converter stage may comprise:
for each converter stage, the following steps are performed:
(1) Acquiring a training data set, wherein the training data set comprises a plurality of training input data and a training endpoint carbon content predicted value corresponding to each training input data;
(2) Using a training data set as input, and calculating to obtain a plurality of fuzzy sets and membership functions corresponding to each fuzzy set by using a fuzzy clustering algorithm;
specifically, in this embodiment, a gustaton-Kessel fuzzy clustering algorithm may be used to obtain the fuzzy set, and the method is a known technology.
(3) And obtaining a predictor model by using a least square identification method based on the fuzzy sets and the membership function corresponding to each fuzzy set.
The least squares identification method is also known in the prior art.
S3: and fusing all the initial predicted values of the end point carbon content to obtain a final predicted value of the end point carbon content at the predicted moment.
Specifically, S3 may include: and taking all initial predicted values of the end point carbon content as input, and calculating to obtain a final predicted value of the end point carbon content at the predicted moment by using a fusion formula.
The fusion formula used in this embodiment to obtain the final predicted value of the model for the dominant variable is:
wherein y (k) is the final predicted value of the end point carbon content at the predicted time k; beta i (s) is a matching degree function of the predictor model of the ith converter stage; y is i (k) An initial predicted value of the end point carbon content output by the predicted sub-model of the ith converter stage at the predicted time k; i=1, 2..p, P is the total number of converter stages;
is a normalized value of the matching degree function.
The method for acquiring the matching degree function of the predictor model in the ith converter stage comprises the following steps: and selecting the minimum value of the membership function corresponding to each fuzzy set of the predictor model of the ith converter stage as the matching degree function of the predictor model of the ith converter stage.
Specifically, to obtain the final predicted value of the model on the dominant variable, the matching degree function beta of the front piece needs to be calculated i (s (k)), since the front part is s j So the matching degree function beta i (s (k)) can be determined by a membership function
Is described in combination of (i) that
Wherein,,
is fuzzy set A ij Is a single-sided S-type membership function, parameter alpha ij And c ij For characterizing fuzzy set A ij The parameter value of the shape of (a) is in fuzzy set A ij Can be obtained from, x j Representing the jth auxiliary variable, x is the variable representation of the general function.
Preferably, after obtaining the final predicted value of the end point carbon content, the intelligent soft measurement method of the embodiment further includes: and for each converter stage, acquiring an empirical value determined by an expert based on input data, carrying out self-adaptive adjustment on a predictor model of the converter stage by utilizing a final predicted value and the empirical value of the end point carbon content to obtain an adjusted model, and taking the adjusted model as the predictor model corresponding to the converter stage at the next predicted moment.
Specifically, for each converter stage (from fuzzy set A ij Representation) the expert can give empirical values for the dominant variable at the converter stage by means of historical values of the dominant variable and the auxiliary variable of the converter stage. The prediction sub-model of each converter stage can be adaptively adjusted by using expert experience values.
Definition:
and
ψ i (k)＝[-y i (k-1),-y i (k-2),…,-y i (k-m y ),x 1 (k),x 1 (k-1),x 1 (k-2),…,x 1 (k-m 1 ),x 2 (k),x 2 (k-1),…,x 2 (k-m 2 ),…,x n (k),x n (k-1),x n (k-2),…,x n (k-m n )] T ；
The updating rule of the parameters in the predictor model is as follows:
ω i (k)＝Γ i (k-1)ψ i (k)[ψ i (k) T Γ i (k-1)ψ i (k)+λ i ] -1 ；
wherein,,
is sigma i (k) Estimated value at the time of the kth sampling, ω i (k) Is adaptive gain Γ i (k) Is a covariance matrix lambda i And epsilon i Is a parameter for adjusting the adaptation speed of the model, < >>
Is the expert's experience value of the dominant variable of the converter stage i.
Aiming at the problems existing in the prior art, the embodiment provides an intelligent soft measurement method for measuring the endpoint carbon content of a converter, which gradually aggravates the carbon-oxygen reaction in the production stage of the converter by selecting related auxiliary variables (such as CO and CO measured by an analyzer in a dry system) 2 The method comprises the steps of carrying out a first treatment on the surface of the The decarburization amount calculated according to the target steel grade; oxygen blowing intensity, oxygen blowing time and the like), and inputting the oxygen blowing intensity, the oxygen blowing time and the like into the established predictor model for calculation and then fusion, so that a dynamic predicted value of the terminal carbon content of molten steel in converter smelting can be obtained, and the terminal carbon content can be predicted with high precision.
The method of the present embodiment may be implemented based on a BQ-ISM-T hardware system comprising: a Raspberry Pi 4B motherboard; ethernet+wifi 6 interface; HMDI or other display interface; analog input/output interface above AI8+ AO 2; using Micro Python+Java artificial intelligence programming language; F-BOX remote data acquisition module; MQTT communication protocol. As shown in fig. 2, the hardware system includes a feature selection module, a multi-stage model estimator, a matching degree reasoning module and an expert experience correction module, wherein the dotted line is an offline processing flow of the hardware system, and the solid line is an online processing flow of the hardware system.
The characteristic selection module: aiming at the problem that the time correlation of the auxiliary variable is not considered in the existing method, the embodiment provides a feature selection module, which selects different maximum time delay orders (continuous sampling points) for different auxiliary variables, and blends the time correlation of the auxiliary variables into the data input of the multi-stage model estimator, as shown in step S1.
A multi-stage model estimator: aiming at the problem that the characteristics of different stages of converter converting are not considered in the existing method, the embodiment provides a multi-stage model estimator, wherein the different stages of the converter are represented by different predictor models, the predictor models of the stages are obtained by training the historical data of the converter, and then each predictor model predicts a dominant variable according to the real-time value of an auxiliary variable of the converter, as shown in step S2.
And a matching degree reasoning module: aiming at the problem that the existing method does not consider the dependence of converter converting stages, the embodiment provides a matching degree reasoning module, which takes the output of a multi-stage model estimator as input, performs matching degree reasoning on the output of all the predictor models according to the influence degree (described by membership functions) of each predictor model on the dominant variable, and gives a final dominant variable prediction result, as shown in step S3.
Expert experience correction module: aiming at the problem that the existing method adopts an offline training or identification model and does not have self-adaptive capacity, the embodiment provides an expert experience correction module, combines the real-time data of converter blowing and the experience estimation of experts on leading variables in different stages, and carries out parameter self-adaptive adjustment on the predictive sub-model in each stage in the multi-stage model estimator so as to enable the estimation of the leading variables to be more accurate.
According to the intelligent soft measurement method for measuring the endpoint carbon content of the BQ-ISM-T converter, which is provided by the embodiment, the collected auxiliary variable (can also comprise the dominant variable) data is taken as input, the endpoint carbon content value of the converter is predicted in real time, and the prediction precision is high. Meanwhile, in combination with specific process requirements, the predicted value of the carbon content of the end point is used for related process scenes, such as abnormal alarm, wherein the alarm mode can be web pages, mobile phone small programs, mobile phone short messages and the like.
Considering that an automatic instrument for directly detecting an estimated process variable is more expensive or difficult to maintain, the technology is an intelligent core technology of which the artificial intelligence technology is applied to the high-end metallurgical industry, an endpoint carbon algorithm and a judging method are a calculation technology and a model which explore and utilize artificial intelligence, and the numerical value of endpoint carbon of a converter is predicted in real time through big data analysis of converter steelmaking process parameters, so that the technology is an intelligent control technology for applying the artificial intelligence technology to converter steelmaking for the first time in China, and can play a guiding role in the application and popularization of the AI technology in the field of modern process industry. In addition, the technical scheme provided by the embodiment solves a plurality of problems in the existing method, introduces expert experience, forms a novel soft measurement method combining data driving and domain knowledge, and is innovation in the soft measurement method.
Example 2:
the embodiment is used for providing an intelligent soft measurement system for measuring the endpoint carbon content of a converter, as shown in fig. 3, and the intelligent soft measurement system comprises:
the data acquisition module M1 is used for acquiring input data; the input data includes a plurality of continuous historical values of endpoint carbon content prior to a predicted time, a real-time value of each auxiliary variable at the predicted time, and a plurality of continuous historical values of each auxiliary variable prior to the predicted time; the auxiliary variables comprise carbon monoxide content, carbon dioxide content, decarburization amount, oxygen blowing intensity and oxygen blowing time;
the converter stage prediction module M2 is used for dividing the converting process of the converter into a plurality of converter stages, and each converter stage corresponds to a pre-trained predictor model; respectively taking the input data as the input of each predictor model to obtain an initial predicted value of the end point carbon content corresponding to each predictor model at the predicted moment;
and the fusion module M3 is used for fusing all the initial predicted values of the end point carbon content to obtain a final predicted value of the end point carbon content at the predicted moment.
Preferably, the fusing all the initial predicted values of the final carbon content at the predicted time to obtain the final predicted value of the final carbon content at the predicted time specifically includes: taking all initial predicted values of the end point carbon content as input, and calculating to obtain a final predicted value of the end point carbon content at the predicted moment by using a fusion formula;
wherein, the fusion formula is:
wherein y (k) is the final predicted value of the end point carbon content at the predicted time k; beta i (s) is a matching degree function of the predictor model of the ith converter stage; y is i (k) The initial predicted value of the end point carbon content corresponding to the predicted sub-model of the ith converter stage at the predicted time k is obtained; i=1, 2..p, P is the total number of converter stages.
Preferably, the intelligent soft measurement system further comprises: and the adjustment module is used for acquiring an experience value determined by an expert based on the input data for each converter stage, carrying out self-adaptive adjustment on the predictive sub-model of the converter stage by utilizing the final predicted value of the end point carbon content and the experience value to obtain an adjusted model, and taking the adjusted model as the predictive sub-model corresponding to the converter stage at the next predicted moment.
In this specification, each embodiment is mainly described in the specification as a difference from other embodiments, and the same similar parts between the embodiments are referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the methods of the present invention and the core ideas thereof; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.
Claims (8)
1. An intelligent soft measurement method for measuring the carbon content of a converter endpoint is characterized by comprising the following steps:
acquiring input data; the input data includes a plurality of continuous historical values of endpoint carbon content prior to a predicted time, a real-time value of each auxiliary variable at the predicted time, and a plurality of continuous historical values of each auxiliary variable prior to the predicted time; the auxiliary variables comprise carbon monoxide content, carbon dioxide content, decarburization amount, oxygen blowing intensity and oxygen blowing time;
dividing the converting process of the converter into a plurality of converter stages, wherein each converter stage corresponds to a pre-trained predictor model; respectively taking the input data as the input of each predictor model to obtain an initial predicted value of the end point carbon content corresponding to each predictor model at the predicted moment;
fusing all the initial predicted values of the final carbon content to obtain a final predicted value of the final carbon content at the predicted moment;
before the input data are respectively used as the input of each prediction sub-model, the intelligent soft measurement method further comprises the following steps: training to obtain a predictor model of each converter stage, which specifically comprises the following steps:
for each of said converter phases, the following steps are performed:
acquiring a training data set; the training data set comprises a plurality of training input data and a training endpoint carbon content predicted value corresponding to each training input data;
using the training data set as input, and calculating to obtain a plurality of fuzzy sets and membership functions corresponding to each fuzzy set by using a fuzzy clustering algorithm;
and obtaining a predictor model by using a least square identification method based on the fuzzy sets and the membership function corresponding to each fuzzy set.
2. The intelligent soft measuring method according to claim 1, wherein the number of continuous history values of each of the auxiliary variables is different.
3. The intelligent soft measurement method according to claim 1, wherein the fusing all the initial predicted values of the final carbon content at the predicted time to obtain the final predicted value of the final carbon content specifically comprises: taking all initial predicted values of the end point carbon content as input, and calculating to obtain a final predicted value of the end point carbon content at the predicted moment by using a fusion formula;
wherein, the fusion formula is:
wherein y (k) is the final predicted value of the end point carbon content at the predicted time k; beta i (s) is a matching degree function of the predictor model of the ith converter stage; y is i (k) The initial predicted value of the end point carbon content corresponding to the predicted sub-model of the ith converter stage at the predicted time k is obtained; i=1, 2..p, P is the total number of converter stages.
4. The intelligent soft measurement method according to claim 3, wherein the matching degree function of the predictor model of the ith converter stage is obtained by the following steps: and selecting the minimum value of membership functions corresponding to each fuzzy set of the predictor model of the ith converter stage as the matching degree function of the predictor model of the ith converter stage.
5. The intelligent soft measurement method of claim 1, further comprising, after obtaining the final predicted value of the endpoint carbon content: and for each converter stage, acquiring an experience value determined by an expert based on the input data, carrying out self-adaptive adjustment on a predictor model of the converter stage by utilizing the final predicted value of the end point carbon content and the experience value to obtain an adjusted model, and taking the adjusted model as the predictor model corresponding to the converter stage at the next predicted moment.
6. An intelligent soft measurement system for measuring the carbon content of a converter endpoint, which is characterized by comprising:
the data acquisition module is used for acquiring input data; the input data includes a plurality of continuous historical values of endpoint carbon content prior to a predicted time, a real-time value of each auxiliary variable at the predicted time, and a plurality of continuous historical values of each auxiliary variable prior to the predicted time; the auxiliary variables comprise carbon monoxide content, carbon dioxide content, decarburization amount, oxygen blowing intensity and oxygen blowing time;
the converter stage prediction module is used for dividing the converting process of the converter into a plurality of converter stages, and each converter stage corresponds to a pre-trained predictor model; respectively taking the input data as the input of each predictor model to obtain an initial predicted value of the end point carbon content corresponding to each predictor model at the predicted moment;
the fusion module is used for fusing all the initial predicted values of the end point carbon content to obtain a final predicted value of the end point carbon content at the predicted moment;
before the input data are respectively used as the input of each prediction sub-model, the intelligent soft measurement system further comprises: training to obtain a predictor model of each converter stage, which specifically comprises the following steps:
for each of said converter phases, the following steps are performed:
acquiring a training data set; the training data set comprises a plurality of training input data and a training endpoint carbon content predicted value corresponding to each training input data;
using the training data set as input, and calculating to obtain a plurality of fuzzy sets and membership functions corresponding to each fuzzy set by using a fuzzy clustering algorithm;
and obtaining a predictor model by using a least square identification method based on the fuzzy sets and the membership function corresponding to each fuzzy set.
7. The intelligent soft measurement system of claim 6, wherein the fusing all of the initial predicted values of the endpoint carbon content to obtain a final predicted value of the endpoint carbon content at the predicted time comprises: taking all initial predicted values of the end point carbon content as input, and calculating to obtain a final predicted value of the end point carbon content at the predicted moment by using a fusion formula;
wherein, the fusion formula is:
wherein y (k) is the final predicted value of the end point carbon content at the predicted time k; beta i (s) is a matching degree function of the predictor model of the ith converter stage; y is i (k) The initial predicted value of the end point carbon content corresponding to the predicted sub-model of the ith converter stage at the predicted time k is obtained; i=1, 2..p, P is the total number of converter stages.
8. The intelligent soft measurement system of claim 6, further comprising: and the adjustment module is used for acquiring an experience value determined by an expert based on the input data for each converter stage, carrying out self-adaptive adjustment on the predictive sub-model of the converter stage by utilizing the final predicted value of the end point carbon content and the experience value to obtain an adjusted model, and taking the adjusted model as the predictive sub-model corresponding to the converter stage at the next predicted moment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210848243.5A CN115058555B (en) | 2022-07-19 | 2022-07-19 | Intelligent soft measurement method and system for measuring carbon content of converter endpoint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210848243.5A CN115058555B (en) | 2022-07-19 | 2022-07-19 | Intelligent soft measurement method and system for measuring carbon content of converter endpoint |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115058555A CN115058555A (en) | 2022-09-16 |
| CN115058555B true CN115058555B (en) | 2023-06-30 |
Family
ID=83206114
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210848243.5A Active CN115058555B (en) | 2022-07-19 | 2022-07-19 | Intelligent soft measurement method and system for measuring carbon content of converter endpoint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115058555B (en) |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1603424A (en) * | 2003-09-29 | 2005-04-06 | 宝山钢铁股份有限公司 | Bessemerizing control method based on intelligent compound dynamic model with sublance converter |
| DE102010035411A1 (en) * | 2010-08-25 | 2012-03-01 | Sms Siemag Ag | Method for controlling the temperature of the metal bath during the blowing process in a converter |
| CN110991089B (en) * | 2019-12-26 | 2020-07-31 | 北京科技大学 | Method for forecasting carbon content in later stage of converter steelmaking |
| CN112668234B (en) * | 2020-12-07 | 2023-08-15 | 辽宁科技大学 | Intelligent control method for converter steelmaking end point |
| CN113239482B (en) * | 2021-04-23 | 2022-02-08 | 北京科技大学 | Dynamic prediction method and device for converter post-blowing carbon content |
| CN114678080B (en) * | 2022-03-28 | 2022-11-18 | 北京科技大学 | Converter end point phosphorus content prediction model, construction method and phosphorus content prediction method |
-
2022
- 2022-07-19 CN CN202210848243.5A patent/CN115058555B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN115058555A (en) | 2022-09-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108676955B (en) | A kind of BOF Steelmaking Endpoint carbon content and temprature control method | |
| CN109447346B (en) | Converter oxygen consumption prediction method based on gray prediction and neural network combined model | |
| CN109884892B (en) | Process industrial system prediction model based on cross correlation time-lag grey correlation analysis | |
| WO2022121944A1 (en) | Method and apparatus for intelligent forecasting of industrial process operation indicators, device, and storage medium | |
| CN109935280B (en) | Blast furnace molten iron quality prediction system and method based on ensemble learning | |
| CN110471276B (en) | Apparatus for creating model functions for physical systems | |
| CN114678080B (en) | Converter end point phosphorus content prediction model, construction method and phosphorus content prediction method | |
| CN113012766B (en) | Self-adaptive soft measurement modeling method based on online selective integration | |
| Yao et al. | Nonlinear Gaussian mixture regression for multimode quality prediction with partially labeled data | |
| CN116607001B (en) | Furnace temperature optimal control method and system based on multipoint feedback | |
| CN116305985A (en) | Local intelligent ventilation method based on multi-sensor data fusion | |
| CN114329347B (en) | Method and device for predicting metering error of electric energy meter and storage medium | |
| Li et al. | Dual ensemble online modeling for dynamic estimation of hot metal silicon content in blast furnace system | |
| CN115058555B (en) | Intelligent soft measurement method and system for measuring carbon content of converter endpoint | |
| JP3279250B2 (en) | Multivariable process control system | |
| KR20200013143A (en) | Apparatus and method for predicting molten iron temperature in blast furnace | |
| CN108958027A (en) | A kind of dynamic matrix control method of Interval System | |
| CN116859839A (en) | Industrial control method and device based on model training | |
| CN109725526B (en) | Multivariable semi-adaptive prediction control method | |
| Gao et al. | End-point prediction of BOF steelmaking based on wavelet transform based weighted TSVR | |
| CN114217592A (en) | Control system and method for copper top-blowing production process | |
| CN112000010A (en) | Oxygen bottom blowing copper smelting process feedback control method and device and electronic equipment | |
| CN112379601A (en) | MFA control system design method based on industrial process | |
| Shigemori | Desulphurization control system through locally weighted regression model | |
| KR970010980B1 (en) | Temperature and components in furnace bath using neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CB03 | Change of inventor or designer information |
Inventor after: Yang Di Inventor after: Yu Bing Inventor before: Yang Di Inventor before: Yu Bing Inventor before: Su Shengkui Inventor before: Liang Yupeng |
|
| CB03 | Change of inventor or designer information |