CN115186900A - Dynamic blast furnace gas production prediction method and system suitable for multiple working condition types - Google Patents
Dynamic blast furnace gas production prediction method and system suitable for multiple working condition types Download PDFInfo
- Publication number
- CN115186900A CN115186900A CN202210817240.5A CN202210817240A CN115186900A CN 115186900 A CN115186900 A CN 115186900A CN 202210817240 A CN202210817240 A CN 202210817240A CN 115186900 A CN115186900 A CN 115186900A
- Authority
- CN
- China
- Prior art keywords
- blast furnace
- prediction
- working condition
- prediction model
- furnace gas
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 105
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 99
- 238000012549 training Methods 0.000 claims description 30
- 230000035772 mutation Effects 0.000 claims description 28
- 238000004364 calculation method Methods 0.000 claims description 23
- 238000012360 testing method Methods 0.000 claims description 18
- 230000002068 genetic effect Effects 0.000 claims description 17
- 238000013528 artificial neural network Methods 0.000 claims description 16
- 230000009467 reduction Effects 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 6
- 210000000349 chromosome Anatomy 0.000 claims description 6
- 238000010219 correlation analysis Methods 0.000 claims description 6
- 230000003044 adaptive effect Effects 0.000 claims description 5
- 150000001875 compounds Chemical class 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 4
- 238000013213 extrapolation Methods 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 4
- 238000012847 principal component analysis method Methods 0.000 claims description 4
- 108091026890 Coding region Proteins 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 239000002245 particle Substances 0.000 claims description 3
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 abstract description 22
- 229910052742 iron Inorganic materials 0.000 abstract description 11
- 229910000831 Steel Inorganic materials 0.000 abstract description 10
- 239000010959 steel Substances 0.000 abstract description 10
- 230000000694 effects Effects 0.000 abstract description 5
- 238000003912 environmental pollution Methods 0.000 abstract description 3
- 239000007789 gas Substances 0.000 description 114
- 230000008569 process Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 10
- 239000003570 air Substances 0.000 description 7
- 239000003245 coal Substances 0.000 description 7
- 239000000571 coke Substances 0.000 description 7
- 230000002159 abnormal effect Effects 0.000 description 6
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 6
- 238000002347 injection Methods 0.000 description 6
- 239000007924 injection Substances 0.000 description 6
- 229910052760 oxygen Inorganic materials 0.000 description 6
- 239000001301 oxygen Substances 0.000 description 6
- 239000003034 coal gas Substances 0.000 description 5
- 238000005265 energy consumption Methods 0.000 description 3
- 239000002994 raw material Substances 0.000 description 3
- 238000003723 Smelting Methods 0.000 description 2
- 229910052799 carbon Inorganic materials 0.000 description 2
- 238000004939 coking Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000005245 sintering Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000006227 byproduct Substances 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013016 damping Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 229910052731 fluorine Inorganic materials 0.000 description 1
- 125000001153 fluoro group Chemical group F* 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 239000000446 fuel Substances 0.000 description 1
- 238000005098 hot rolling Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000009401 outcrossing Methods 0.000 description 1
- 239000008188 pellet Substances 0.000 description 1
- 238000005453 pelletization Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000010079 rubber tapping Methods 0.000 description 1
- 239000002893 slag Substances 0.000 description 1
- 238000009628 steelmaking Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 229910000859 α-Fe Inorganic materials 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biomedical Technology (AREA)
- Marketing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Physiology (AREA)
- Genetics & Genomics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a dynamic blast furnace gas production prediction method and a dynamic blast furnace gas production prediction system suitable for multiple working condition types, wherein the method judges the working condition types by using blast furnace production data; dynamically determining the duration of a prediction period based on the types of the production working conditions, acquiring blast furnace production historical data corresponding to different working conditions, selecting the production historical data within a period of time, segmenting the production historical data according to the duration of the prediction period, and using the segmented production historical data as sample data to analyze the blast furnace gas production influence factors; selecting important influence factors corresponding to different working condition types as the input of a blast furnace gas generation amount prediction model under the working condition type; and establishing a prediction model to predict the generation amount of the blast furnace gas under the working condition. By adopting the technical scheme, the blast furnace gas production amount is predicted based on the working condition type, the prediction effect is more accurate, the technical support is provided for the gas prediction of the iron and steel enterprises, the foundation is laid for the gas dynamic scheduling management, and the gas emission and the environmental pollution are reduced.
Description
Technical Field
The invention belongs to the technical field of gas prediction of steel mills, and relates to a dynamic blast furnace gas production prediction method and system suitable for multiple working condition types.
Background
The steel industry consumes huge energy and discharges carbon, at present, the steel manufacturing process in China mainly comprises a long process taking a blast furnace-converter as a main smelting method and a short process taking an electric furnace as a main smelting method, wherein the steel yield of the long process accounts for about 90 percent of the total amount. The long-flow steel manufacturing mainly comprises the working procedures of coking, pelletizing, sintering, iron making, steel making, rolling and the like. In the production process, various energy media are consumed in each process, and secondary energy is generated in part of the processes.
The energy consumption of the blast furnace ironmaking process accounts for about 60 percent of the total energy consumption of the whole long-process steel manufacturing process, so the process becomes the key for realizing low-carbon and energy-saving production in the steel industry. The blast furnace mainly uses sintered ore, pellet ore, lump ore and the like as raw materials of ferrite, uses coke, coal and the like as raw materials of energy, and generates complex chemical reaction and physical change under the action of hot air so as to finally produce molten iron, blast furnace gas, blast furnace slag and the like. Among them, blast furnace gas is one of important byproduct energy sources of iron and steel enterprises, and the amount of the blast furnace gas generated affects not only the safe operation of the whole gas pipe network including a gas tank, but also the operation pace of processes such as hot blast furnaces, coking, sintering, hot rolling and the like using the blast furnace gas as fuel, so that the amount of the blast furnace gas generated needs to be accurately predicted.
In addition, reasonable utilization and scheduling of blast furnace gas are of great significance for reducing energy consumption in the production process and reducing gas diffusion and environmental pollution, and accurate prediction of the blast furnace gas generation amount is also the basis and key for formulating blast furnace gas utilization and scheduling strategies, and has great influence on efficient gas management. The existing blast furnace gas prediction method is not tightly combined with the process under different working condition types, so that the running characteristics and the influence factors of the blast furnace are not considered sufficiently, the model prediction result has certain difference with the actual condition, the method is difficult to adapt to the complicated and changeable actual working condition types of the blast furnace, and the prediction precision needs to be improved.
Disclosure of Invention
The invention aims to provide a dynamic blast furnace gas generation amount prediction method and system suitable for multiple working condition types, which are used for predicting the blast furnace gas generation amount based on the working condition type classification and improving the prediction precision.
In order to achieve the purpose, the basic scheme of the invention is as follows: a dynamic blast furnace gas production amount prediction method suitable for multiple working condition types comprises the following steps:
s1, obtaining production data of a blast furnace and judging the type of a working condition;
s2, dynamically determining the duration of a prediction period based on the type of the production working conditions, acquiring the production historical data of the blast furnace corresponding to different working conditions, selecting the production historical data within a period of time, segmenting the production historical data according to the duration of the prediction period, and using the segmented production historical data as sample data to analyze the blast furnace gas production influence factors;
s3, selecting important influence factors corresponding to different working condition types as the input of a blast furnace gas generation amount prediction model under the working condition type;
and S4, establishing a prediction model and predicting the generation amount of the blast furnace gas under the working condition.
The operating principle and the beneficial effects of the basic scheme are as follows: according to the invention, a prediction model is established according to the type of the production working condition, and the generation amount of the blast furnace gas is predicted. Specifically, the prediction period duration is set according to the working condition types of the blast furnace, and main influence factors corresponding to the blast furnace gas generation amount under each working condition type are obtained through a correlation degree analysis model so as to determine input variables of the prediction model; and establishing a gas generation amount prediction model. The model has more accurate prediction effect, can provide certain technical support for the gas prediction of iron and steel enterprises, lays a foundation for the dynamic scheduling management of the gas, and reduces the gas emission and the environmental pollution.
Further, in step S2, the method for dynamically determining the prediction cycle duration based on the production condition category includes:
setting the prediction period duration under the forward working condition as T according to the practical requirements of gas scheduling management 1 Setting the prediction period duration in the wind reduction working condition and the compound wind working condition as T 2 Wherein T is 1 >T 2 >0, the gas generation amount is 0 when the wind stops, and the duration of a prediction period is not required to be set.
The setting is simple, the operation is convenient.
Further, the method for selecting the important influence factors corresponding to different working condition types as the input of the blast furnace gas generation amount prediction model under the working condition type in the step S3 is as follows:
and calculating the correlation degree values of the influence factors and the blast furnace gas production amount under different working condition types, and selecting the influence factors of which the correlation degree values exceed the threshold value as the input of the blast furnace gas production amount prediction model under the working condition type.
Therefore, important influence factors corresponding to different working condition classes are accurately and quickly selected, and effective prediction of the model is realized.
Further, the correlation degree is calculated by a principal component analysis method, or an analytic hierarchy method, or a fuzzy evaluation method, or a gray correlation analysis method.
And selecting a proper method for analysis according to needs, so that calculation is facilitated.
Further, when a grey correlation analysis method is adopted, the specific calculation method is as follows:
setting a reference data column X 0 And comparing data sequence X h ，X 0 For blast furnace gas production data sets, X h For the h-th influencing factor dataset related to blast furnace gas production,
{ blast pressure, blast furnace top pressure, blast furnace bottom pressure, blast furnace top-bottom pressure difference, blast humidity, molten iron yield, coal injection amount, coke charging amount, blast amount, oxygen amount }, X 0 ，X h And X is represented as:
X 0 ＝(x 0 (1),x 0 (2),…,x 0 (l)) T
X＝(X 1 ,X 2 ,…,X n )
wherein n is the total number of the comparison sequences, namely the total number of the influencing factors; l is the number of samples; x is the number of h (i) For the h-th influencing factor data set x h The number i of samples in (2) is,
x is to be 0 And X are combined to form a new data sequence X C ：
And respectively calculating a correlation coefficient xi (k) of each element corresponding to the comparison sequence and the reference sequence, wherein the calculation method comprises the following steps of:
wherein, omega is a resolution coefficient, omega is more than 0 and less than or equal to 1, the smaller omega is, the larger the difference between the correlation coefficients is, the stronger the distinguishing capability is;
calculating the correlation degree value R of the h influencing factor h The calculation method comprises the following steps:
and setting a threshold value, and selecting the influence factors of which the correlation degree values exceed the threshold value as input variables of the blast furnace gas generation quantity prediction model under the working condition type.
The influence of the influence factors on the blast furnace gas generation amount is considered, and the subsequent prediction is facilitated.
Further, in step S4, a prediction model is established, and a method of predicting the amount of blast furnace gas generated under the operating condition is as follows:
s41, establishing a prediction model, wherein the input of the prediction model is an influence factor of which the correlation degree value with the blast furnace gas generation quantity exceeds a threshold value;
s42, optimizing the parameters of the prediction model, if the parameters reach the optimization termination condition, executing the step S43, otherwise, continuing to execute the step S42;
and S43, predicting the blast furnace gas generation amount under the working condition by using the prediction model.
By selecting and inputting important influence factors of which the correlation degree value with the blast furnace gas generation quantity exceeds a threshold value, the model is predicted after being optimized, and the model prediction effect is more accurate.
Further, the prediction model is a trend extrapolation prediction model, or a regression prediction model, or a kalman filter prediction model, or a combined prediction model, or a neural network prediction model.
The flexible selection is carried out according to the use requirement, and the operation is convenient.
Further, when a neural network prediction model is employed, the neural network prediction model includes:
the input layer consists of m nodes, m influence factors with the correlation degree value exceeding a threshold value are respectively input, m is more than or equal to 1 and less than or equal to n, and n is the number of the influence factors;
the hidden layer comprises J nodes, each node is provided with a corresponding basis function, and m influence factors are calculated by utilizing each basis function;
and the output layer receives the calculation results of all the basis functions output by the hidden layer and outputs the predicted value of the gas generation amount.
Through the structure, the gas generation amount can be efficiently and accurately predicted.
Further, the base function of the jth node of the hidden layer is phi j (X)：
The predicted value of the output layer is
Wherein X is an m-dimensional input vector; c. C j Is the center of the jth node, with dimensions the same as X; sigma j Is a radial basis function phi j (X) an expansion constant; j is the total number of hidden layer unit nodes; II X-c j II denotes X and c j Of between, omega j The weight value from the jth node of the hidden layer to the output layer.
Through the setting of the functions, the prediction of the gas generation amount is efficiently and accurately realized.
Further, in step S42, the parameters of the radial basis function neural network prediction model are optimized by using an ant colony algorithm, a particle swarm algorithm, or an improved genetic algorithm.
Further, when an improved genetic algorithm is adopted, the specific steps are as follows:
s421, setting algebra of genetic algorithm;
s422, setting the total number of individuals in the population to be P, and initializing the population, namely randomly generating the population containing P chromosomes; due to c j The dimension of the input layer is the same, and the total number of variables needing to be coded is equal to J + m.J + J; each variable is coded in a binary manner, anThe code length is 10, and the total code length of each individual is 10 (m + 2) J;
s423, final decoded value F of S variable s Comprises the following steps:
wherein, F s Representing the decoded value of the s-th variable, f s Representing the value of the s-th variable converted from binary to decimal;
s424, decoding the codes of all variables on the chromosome, reducing the decoded variable values into parameter values C, delta and w, and substituting the parameters into a blast furnace gas generation prediction model; designing a fitness function of an improved genetic algorithm, and calculating the fitness value of an individual p in the G generation population
Wherein L represents the total number of training samples, L<l；
Representing a predicted value of the ith sample; y is i Representing the true value of the ith sample;
s425, if the maximum iteration number G is reached max I.e. G.gtoreq.G max Then at G max Finding out an individual with the maximum fitness from the population of the generation, decoding the coding sequence of the individual, and outputting decoded parameter values C, delta and w; turning to S427, otherwise turning to S426;
s426, designing a genetic operator, and carrying out selection, crossing and mutation operations on individuals; according to the average fitness f of all individuals in the G generation population avg And carrying out crossing and mutation operations on the larger fitness value f and the evolutionary algebra G in the two crossed individuals, wherein the method comprises the following steps:
designing a population Q capable of containing P individuals;
selecting operation: carrying out (P-2)/2 times of selection operation in total, randomly grabbing 2 individuals P ', P' from the G generation in each selection operation, removing two individuals P ', P' from the G generation, and then carrying out the next selection operation,
and (3) cross operation: after each selection operation is finished, a decimal a between 0 and 1 is randomly generated and combined with the self-adaptive crossover probability P c If a is less than or equal to P c If so, the two grabbed individuals p ', p' need to be subjected to cross operation to generate two new individuals, and the two newly generated individuals are placed into the population Q; if a>P c Then the two grabbed individuals p ', p' need no cross operation and are directly put into the population Q;
mutation operation: sequentially traversing each individual in the population Q, randomly generating a decimal b between 0 and 1 during each traversal, and combining the adaptive variation probability P m If b is less than or equal to P m If the traversed individuals need to carry out mutation operation, the mutation points are randomly generated; otherwise, the traversed individuals do not perform mutation operation;
after the three operations of selection, crossing and variation are completed, a population Q containing (P-2) individuals is generated, then 2 individuals with the maximum fitness value of the G generation are placed into the population Q, the total number of the individuals of the population Q is still P, the next generation of population is equal to the population Q, G = G +1 is updated, and the step S423 is returned;
and S427, substituting the optimized parameter values C, delta and w into the blast furnace gas generation amount prediction model, and predicting the generation amount of the blast furnace gas by using the test data set.
And key parameters of the prediction model are optimized by adopting an improved genetic algorithm, and the optimized parameters are used for constructing the prediction model, so that the optimized prediction is more accurate.
Further, in step S426, the average fitness f of all individuals in the population of the G generation avg The larger fitness value f in two intersecting individuals is calculated as follows:
the calculation steps are simple, and the operation is convenient.
Further, in step S426, an improved adaptive cross probability and mutation probability calculation method is designed, such that the cross between individuals is constrained by fitness, the mutation operation within the individual is constrained by evolution algebra, and the cross probability function P c And the variation probability function P m Comprises the following steps:
P m ＝a 3 /(ln[a 4 ·(G max +1-G)])
wherein f is the greater fitness value of the two crossed individuals; a is 1 ,a 2 As a parameter of the function, a 1 ,a 2 ∈(0,1]； f avg The average fitness value of all individuals in the population; f. of max The maximum fitness value among all individuals in the population; a is 3 ,a 4 Are all positive real numbers; g max Is the maximum evolution algebra; g is the current evolution algebra.
The operation is simple, and the use is facilitated.
The invention also provides a method for establishing the dynamic blast furnace gas production prediction model, which comprises the following steps:
s11, acquiring production data of the blast furnace and judging the type of a working condition;
s12, dynamically determining the duration of a prediction period based on the type of the production working condition;
s13, obtaining blast furnace historical production data corresponding to different working conditions, selecting multiple sections of historical production data with the time length being the duration of a prediction period as influence factor data of blast furnace gas production, dividing the influence factor data into a training set and a test set, wherein the training set is used for training a prediction model, and the test set is used for verifying the prediction performance of the prediction model;
s14, establishing an analysis model to calculate the correlation degree of the influence factors in the training set and the blast furnace gas generation amount under different working condition types;
s15, selecting the influence factors of which the correlation degree value exceeds the threshold value as the input of a blast furnace gas generation amount prediction model under each working condition type to predict the generation amount of the blast furnace gas;
s16, training the model by using a training set, testing the model by using a testing set, and ending the training if the training times reach the maximum iteration times or a termination condition; otherwise, returning to the step S1;
and S17, predicting the blast furnace gas production amount by using the trained prediction model and adopting a test set.
The invention establishes a prediction model according to the type of the production working condition and predicts the generation amount of the blast furnace gas. Specifically, the prediction period duration is set according to the working condition types of the blast furnace, main influence factors corresponding to the blast furnace gas generation amount under each working condition type are obtained through the correlation degree analysis model, and a gas generation amount prediction model is established, so that the prediction effect is more accurate.
The invention also provides a dynamic blast furnace gas generation amount prediction system suitable for multiple working condition types, which comprises a data acquisition unit and a prediction unit, wherein the data acquisition unit is used for acquiring historical data of blast furnace blast flow, the output end of the data acquisition unit is connected with the input end of the prediction unit, and the prediction unit executes the method of the invention to predict the blast furnace gas generation amount.
The blast furnace operation characteristics under different working condition types of the system are combined with the influence factors to predict the blast furnace gas production, so that the system can adapt to the complicated and changeable actual working condition types of the blast furnace and improve the prediction precision.
Drawings
FIG. 1 is a schematic diagram of the structure of the dynamic blast furnace gas generation prediction method applicable to multiple working condition classes according to the present invention;
FIG. 2 is a schematic diagram of the classification of blast furnace operating conditions for the dynamic blast furnace gas production prediction method of the present invention for multiple operating condition classes;
FIG. 3 is a schematic view of a topological structure of a blast furnace gas generation prediction model of the dynamic blast furnace gas generation prediction method applicable to multiple operating condition classes according to the present invention;
FIG. 4 is a schematic diagram of the algorithm flow for optimizing the important parameters C, δ, w in the blast furnace gas generation prediction model of the dynamic blast furnace gas generation prediction method applicable to multiple working condition categories according to the present invention;
FIG. 5 is a schematic diagram of the parameter coding structure of the parameters C, δ, w of the dynamic blast furnace gas generation prediction method applicable to multiple operating mode categories according to the present invention;
FIG. 6 is a schematic diagram showing the prediction results of blast furnace gas production under three conditions of blast furnace air reduction, air cut-off and air reblowing, which is applicable to the dynamic blast furnace gas production prediction method of multiple working condition types according to the present invention;
FIG. 7 is a schematic diagram showing the blast furnace gas production prediction results under the forward-going conditions of the blast furnace to which the dynamic blast furnace gas production prediction method of the present invention is applied in multiple operating mode classes;
FIG. 8 is a schematic diagram showing the comparison of the prediction results of models under the blast furnace air reduction condition of the dynamic blast furnace gas generation amount prediction method applicable to multiple working condition types according to the present invention;
FIG. 9 is a schematic diagram showing the comparison of the prediction results of the models under the complex wind condition of the blast furnace, which is suitable for the dynamic blast furnace gas generation amount prediction method of multiple working condition types according to the present invention;
FIG. 10 is a schematic diagram comparing the present invention for a dynamic blast furnace gas production prediction method for multiple operating modes with the MA, SES method in a forward operating mode of the blast furnace;
FIG. 11 is a schematic diagram comparing the RBFNN and BPNN method under the forward operation condition of the blast furnace suitable for the multi-condition type dynamic blast furnace gas generation amount prediction method.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.
In the description of the present invention, it is to be understood that the terms "longitudinal", "lateral", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. indicate orientations or positional relationships based on those shown in the drawings, and are merely for convenience of description and simplicity of description, but do not indicate or imply that the device or element referred to must have a particular orientation, be constructed in a particular orientation, and be operated, and thus, are not to be construed as limiting the present invention.
In the description of the present invention, unless otherwise specified and limited, it should be noted that the terms "mounted," "connected," and "connected" are to be interpreted broadly, and may be, for example, a mechanical connection or an electrical connection, a communication between two elements, a direct connection, or an indirect connection through an intermediate medium, and those skilled in the art will understand the specific meaning of the terms as they are used in the specific case.
As shown in FIG. 1, the invention discloses a dynamic blast furnace gas generation amount prediction method suitable for multiple working condition types, which comprises the following steps:
s1, acquiring production data of a blast furnace, judging the working condition type, and judging the working condition type of the blast furnace according to blast furnace blast volume data, as shown in figure 2;
s2, dynamically determining the duration of a prediction period based on the type of the production working conditions, acquiring blast furnace production historical data corresponding to different working conditions, selecting the production historical data in a period of time range (the period of time range is preferably greater than or equal to R times of the prediction period duration, R is a positive integer and is usually not less than 10, the prediction period duration is consistent with each period of the training data) by considering the correlation between the production working conditions and the gas quantity of the blast furnace, segmenting the production historical data according to the prediction period duration, and using the segmented production historical data as sample data to analyze the blast furnace gas production quantity influence factors;
s3, selecting important influence factors corresponding to different working condition types as the input of a blast furnace gas generation amount prediction model under the working condition type; the method for selecting the important influence factors corresponding to different working condition types as the input of the blast furnace gas generation amount prediction model under the working condition type in the step S3 comprises the following steps: and calculating the correlation degree values of the influence factors and the blast furnace gas generation amount under different working condition types, and selecting the influence factors of which the correlation degree values exceed the threshold value as the input of the blast furnace gas generation amount prediction model under the working condition type. And calculating the correlation degree by adopting a principal component analysis method, an analytic hierarchy process, a fuzzy evaluation method or a gray correlation analysis method. And S4, establishing a prediction model and predicting the generation amount of the blast furnace gas under the working condition.
In the preferred embodiment of the invention, the sample data set of each influence factor can be respectively subjected to normalization processing, and all sample data values are converted to be between 0 and 1, so that the training time of the prediction model is reduced, and the calculation speed is increased; because the data magnitude order difference of different factors is large, all sample data values are normalized and converted to be between 0 and 1, so that the training time of a prediction model is reduced, and the calculation speed is increased; the sample data is data collected by a real-time sensor at the minute level, and a mean value calculation method is adopted because the data such as pressure and the like cannot be integrated to obtain the total amount. For other data such as flow or flow velocity, the total amount can be calculated by integrating the time, and different kinds of data in a period of time can be calculated by adopting a mean value and an integral mode respectively. The influencing factors comprise index parameters and metering parameters; the index parameters comprise blast pressure, top pressure of the blast furnace, bottom pressure of the blast furnace, pressure difference of the top and the bottom of the blast furnace and blast humidity; the metering parameters comprise molten iron yield, coal injection quantity, coke feeding quantity, blast quantity, oxygen quantity and coal gas production quantity.
As shown in fig. 3, for the type of the production condition of the blast furnace, the influence factors corresponding to the maximum m correlation degree values exceeding the threshold are dynamically extracted, and the influence factors are used as the input of a blast furnace gas generation amount prediction model (using a radial basis function neural network prediction model) in the type of the condition to predict the generation amount of the blast furnace gas, wherein m is a positive integer. Wherein, the values of m under different working condition categories are different. And optimizing important parameters of the blast furnace gas generation amount prediction model by adopting an improved genetic algorithm, and predicting the generation amount of the blast furnace gas by using the prediction model after parameter optimization.
In a preferred embodiment of the present invention, the method further includes preprocessing the influence factor data, and the adjustment processing method of the abnormal data includes:
each sample in the data set D under the forward operation condition of the blast furnace should fluctuate within a certain small range, but due to the complexity of an industrial field, a small amount of abnormal values appear in the data under the forward operation condition. Taking the sample data of which the deviation between the acquired original sample data of the forward working condition of the blast furnace and the average value of the data set of the original sample data is more than 3 theta as abnormal data;
the Θ calculation method is as follows:
wherein D = { D 1 ,d 2 ,…,d l L is the number of samples of the data set D,
average of all samples in D;
if it is
Then d i Belonging to abnormal value, and detecting abnormal value d to reduce influence of abnormal value i First, assigning a null value and then ordering
Obtaining a new sample data set x = { d' 1 ,d′ 2 ,…,d′ l }。
In a preferred embodiment of the present invention, in step S2, the method for dynamically determining the predicted cycle duration based on the type of the production condition includes:
as shown in fig. 2, the rootAccording to the blast operation mode, the blast furnace conditions are classified into forward-running conditions, wind-reducing conditions, blowing-down (damping-down) conditions, and reblowing conditions. The cycle duration under each working condition has independence, and the cycle durations can be the same or different and are usually different, which is the meaning of prediction according to the working condition types. And determining corresponding different prediction cycle durations under different working condition categories, and performing time interval division on corresponding blast furnace historical production data, wherein the cycle durations are determined according to the actual requirements of gas scheduling management. According to the actual gas dispatching management requirement, when the type of the smooth operating condition of the blast furnace is in order to better guide the gas dispatching work, a longer prediction period (such as 1 hour) is often expected, and the prediction period duration under the smooth operating condition is set as T 1 Providing abundant time lead for the formulation of a coal gas regulation strategy; the fluctuation range of the gas production flow is large in the wind reduction working condition and the re-wind working condition, so that the uncertainty is strong, the prediction effect of the gas production in a longer period is poor, and the method is more suitable for adopting a shorter period T 2 (for example, 5 minutes) and setting the prediction period duration in the wind reduction working condition and the compound wind working condition as T 2 The prediction period duration during the damping down is set to 0, and the prediction period duration is not required to be set. Wherein T is 1 >T 2 >0。
In a preferred embodiment of the present invention, after data preprocessing and normalization, a mean value of each index parameter and a total amount of each measurement parameter in each period of time equal to a corresponding prediction period time interval in each working condition category are calculated, and the specific method is as follows:
setting the values of blast pressure, blast furnace top pressure, blast furnace bottom pressure, blast furnace top and bottom pressure difference and blast humidity equal to the average value in each period of time;
the molten iron yield, the coal injection amount, the coke feeding amount, the blast amount, the oxygen amount and the coal gas production amount are calculated by the integral of the molten iron production rate, the coal injection rate, the coke feeding amount rate, the blast flow rate, the oxygen flow rate and the coal gas production flow rate in the historical production data to the time t, and are as follows:
wherein g belongs to { molten iron, coal injection, coke, blast air, oxygen and coal gas }; t represents time; t is t s Which indicates the time of the start of the process,
represents the total amount of substance g over a period of Δ T; v g Represents the rate of production or consumption of substance g over a period of Δ T; t is 1 The predicted cycle duration under the forward working condition; t is 2 The prediction period duration in the wind reduction working condition and the compound wind working condition is set.
In a preferred embodiment of the present invention, when a gray correlation analysis method is adopted, the specific calculation method is as follows:
setting a reference data column X 0 And comparing the data sequence X h ，X 0 For blast furnace gas production data sets, X h For the h-th influence factor data set, X, related to blast furnace gas production h Comprises the mean value of all index parameters and the integral formula calculation value of other metering parameters except the gas generation amount in the metering parameter set
{ blast pressure, blast furnace top pressure, blast furnace bottom pressure, blast furnace top-bottom pressure difference, blast humidity, molten iron yield, coal injection amount, coke charging amount, blast amount, oxygen amount }, X 0 ，X h And X is represented as:
X 0 ＝(x 0 (1),x 0 (2),…,x 0 (l)) T
X＝(X 1 ,X 2 ,…,X n )
wherein n is the total number of the comparison sequences, namely the total number of influencing factors; l is the number of samples; x is a radical of a fluorine atom h (i) For the h-th influence factor data set x h The number i of samples in (2) is,
mixing X 0 And X are combined to form a new data sequence X C ：
And respectively calculating the association coefficient xi (k) of each comparison sequence and the corresponding element of the reference sequence, wherein the calculation method comprises the following steps:
wherein, omega is a resolution coefficient, omega is more than 0 and less than or equal to 1, the smaller omega is, the larger difference between the correlation coefficients is, the stronger distinguishing capability is;
calculating the relevance degree value R of the h influencing factor h The calculation method comprises the following steps:
and setting a threshold value, and selecting the influence factors of which the correlation degree values exceed the threshold value as input variables of the blast furnace gas generation quantity prediction model under the working condition type.
Besides the above methods, the present invention may also adopt the existing correlation degree calculation method, specifically, but not limited to, the principal component analysis method, the analytic models such as the analytic hierarchy process, the fuzzy evaluation method, etc., to obtain the correlation degree values of the blast furnace gas production amount affected by each affected factor and sort the values, thereby quantifying the degree of the blast furnace gas production amount affected by each factor.
The method can be used for predicting the gas production of the blast furnace by adopting a trend extrapolation prediction method, a regression prediction method, a Kalman filtering prediction method, a combined prediction method, a neural network prediction method or other methods.
In a more preferred embodiment of the present invention,
in step S4, a prediction model is established, and the method for predicting the generation amount of the blast furnace gas under the working condition comprises the following steps:
s41, establishing a prediction model, wherein the input of the prediction model is an influence factor of which the correlation degree value with the blast furnace gas generation amount exceeds a threshold value;
s42, optimizing the parameters of the prediction model, if the parameters reach the optimization termination condition, executing the step S43, otherwise, continuing to execute the step S42; in step S42, the parameters of the radial basis function neural network prediction model are optimized by using an ant colony algorithm, a particle swarm algorithm or an improved genetic algorithm;
and S43, predicting the blast furnace gas generation amount under the working condition by using the prediction model. The prediction model is a trend extrapolation prediction model, a regression prediction model, a Kalman filtering prediction model, a combined prediction model or a neural network prediction model.
More preferably, when the neural network prediction model is adopted, the neural network prediction model includes:
the input layer consists of m nodes, m influence factors with the correlation degree value exceeding a threshold value are respectively input, m is more than or equal to 1 and less than or equal to n, and n is the number of the influence factors;
the hidden layer comprises J nodes, each node is provided with a corresponding basis function, and m influence factors are calculated by using each basis function;
and the output layer receives the calculation results of all the basis functions output by the hidden layer and outputs a predicted value of the gas generation amount.
In a preferred embodiment of the invention, the base function of the jth node of the hidden layer is φ j (X)：
The predicted value of the output layer is
Wherein X is an m-dimensional input vector; c. C j Is the center of the jth node, with dimensions the same as X; sigma j Is a radial basis function phi j (X) an expansion constant; j is the total number of hidden layer unit nodes; II X-c j II denotes X and c j Of between, omega j The weight value from the jth node of the hidden layer to the output layer.
In one embodiment of the invention, as shown in fig. 4 and 5, when the improved genetic algorithm is used, the specific steps are as follows:
s421, setting algebra of genetic algorithm;
s422, setting the total number of individuals in the population to be P, and initializing the population, namely randomly generating the population containing P chromosomes; due to c j The dimension of the input layer is the same, and the total number of variables needing to be coded is equal to J + m.J + J; each variable is coded in a binary mode, the coding length is 10, and the total coding length of each individual is 10 (m + 2) J;
s423, final decoded value F of S variable s Comprises the following steps:
wherein, F s Represents the decoded value of the s-th variable, f s Representing the value of the s-th variable converted from binary to decimal;
s424, decoding the codes of all variables on the chromosome, reducing the decoded variable values into parameter values C, delta and w, and substituting the parameters into a blast furnace gas generation prediction model; designing a fitness function of an improved genetic algorithm, and calculating the fitness value of an individual p in the G generation population
Wherein L represents the total number of training samples, L<l；
Representing a predicted value of the ith sample; y is i Representing the true value of the ith sample;
s425, if the maximum iteration number G is reached max I.e. G.gtoreq.G max Then at G max Finding out an individual with the maximum fitness from the population of the generation, decoding the coding sequence of the individual, and outputting decoded parameter values C, delta and w; turning to S427, otherwise turning to S426;
s426, designing a genetic operator, and carrying out selection, crossing and mutation operations on individuals; according to the average fitness f of all individuals in the G generation population avg And carrying out crossover and mutation operations on the larger fitness value f and the evolutionary algebra G in the two crossed individuals, wherein the method comprises the following steps of:
designing a population Q capable of containing P individuals;
selecting operation: carrying out (P-2)/2 times of selection operation in total, randomly grabbing 2 individuals P ', P' from the G generation in each selection operation, removing two individuals P ', P' from the G generation, and then carrying out the next selection operation,
and (3) cross operation: after each selection operation is finished, randomly generating a decimal a between 0 and 1 and combining the self-adaptive cross probability P c If a is not more than P c If so, the two grabbed individuals p ', p' need to be subjected to cross operation to generate two new individuals, and the two newly generated individuals are put into the population Q; if a>P c Then the two grabbed individuals p ', p' need no cross operation and are directly put into the population Q;
mutation operation: sequentially traversing each individual in the population Q, and randomly generating a signal between 0 and 1 during each traversalFraction b in combination with adaptive mutation probability P m If b is less than or equal to P m If so, the traversed individuals need to carry out mutation operation, and the mutation points are randomly generated; otherwise, the traversed individuals do not perform mutation operation;
after the three operations of selection, crossing and mutation are completed, generating a population Q containing (P-2) individuals, then putting the 2 individuals with the maximum G generation fitness value into the population Q, keeping the total number of the individuals of the population Q as P, enabling the next generation population to be equal to the population Q, updating G = G +1, and returning to the step S423;
and S427, substituting the optimized parameter values C, delta and w into the blast furnace gas generation amount prediction model, and predicting the generation amount of the blast furnace gas by using the test data set.
More preferably, in step S426, the average fitness f of all individuals in the population of generation G avg The larger fitness value f in two crossing individuals is calculated as follows:
in step S426, an improved adaptive crossover probability and mutation probability calculation method is designed, such that crossover among individuals is constrained by fitness, mutation operation inside individuals is constrained by evolution algebra, and a crossover probability function P is designed c And the variation probability function P m Comprises the following steps:
P m ＝a 3 /(ln[a 4 ·(G max +1-G)])
wherein f is the greater fitness value of the two crossed individuals; a is a 1 ,a 2 As a parameter of a function 1 ,a 2 ∈(0,1]； f avg The average fitness value of all individuals in the population is obtained; f. of max The maximum fitness value among all individuals in the population; a is 3 ,a 4 Are all positive and real; g max Is the maximum evolution algebra; g is the current evolution algebra.
The invention also provides a method for establishing the dynamic blast furnace gas production prediction model, which comprises the following steps:
s11, acquiring production data of the blast furnace and judging the type of a working condition;
s12, dynamically determining the duration of a prediction period based on the type of the production working condition;
s13, obtaining blast furnace historical production data corresponding to different working conditions, selecting multiple sections of historical production data with the time length being the duration of a prediction period as influence factor data of blast furnace gas production, dividing the influence factor data into a training set and a test set, wherein the training set is used for training a prediction model, and the test set is used for verifying the prediction performance of the prediction model;
s14, establishing an analysis model to calculate the correlation degree of the influence factors in the training set and the blast furnace gas generation amount under different working condition types;
s15, selecting the influence factors of which the correlation degree value exceeds the threshold value as the input of a blast furnace gas generation amount prediction model under each working condition type to predict the generation amount of the blast furnace gas;
s16, training the model by using the training set, testing the model by using the testing set, and ending the training if the training times reach the maximum iteration times or the termination condition; otherwise, returning to the step S1;
and S17, predicting the blast furnace gas generation amount by using the trained prediction model and adopting a test set.
The invention provides a dynamic blast furnace gas generation amount prediction system suitable for multiple working condition types, which comprises a data acquisition unit and a prediction unit, wherein the data acquisition unit is used for acquiring historical data of blast furnace blast flow, the output end of the data acquisition unit is electrically connected with the input end of the prediction unit, and the prediction unit executes the method of the invention to predict the blast furnace gas generation amount. The blast furnace operation characteristics under different working condition types of the system are combined with the influence factors to predict the blast furnace gas generation amount, so that the system can adapt to the complicated and changeable actual working condition types of the blast furnace and improve the prediction precision.
In order to verify the performance of the prediction model, the prediction result of the method is compared with the prediction results of MA (Moving Average), SES (Single explicit smoothening) in a common Time Series prediction (TSF) method and RBFNN (Back prediction Neural Network) and BPNN (Radial Basis Function Neural Network) in an artificial intelligence method. Fig. 6 to 11 respectively show the BFG production prediction results of the models under the conditions of blast furnace downdraft, overfire air and forward operation. Wherein, the prediction results of each model under the wind reduction working condition are shown in FIG. 8; the prediction results of different methods under the complex wind working condition are shown in fig. 9; under the forward working condition, the prediction result of the time series method is shown in fig. 10, and the prediction result of the common intelligent method is shown in fig. 11.
As shown in FIG. 8, under the condition of reducing the wind of the blast furnace, the mean square error of the prediction result calculated by the method is 1.9111 multiplied by 10 6 m 3 Root mean square error 1382.40m 3 The mean absolute percentage error was 3.3096%. Compared with MA, SES, RBFNN and BPNN methods, the method of the invention has the best overall prediction performance.
As shown in FIG. 9, under the condition of reducing the wind of the blast furnace, the mean square error of the prediction result calculated by the method of the invention is 1.9887 multiplied by 10 6 m 3 Root mean square error of 1410.20m 3 The mean absolute percentage error was 3.2265%. Compared with MA, SES, RBFNN and BPNN methods, the method of the invention has the best overall prediction performance.
As shown in the graph 10 and the graph 11, under the working condition of the blast furnace, the blast furnace runs stably, the raw material condition and the tapping speed are stable, and the prediction accuracy of 5 methods of MA, SES, RBFNN, BPNN and the method is higher. However, the mean square error calculated by the method of the invention is 1.14 multiplied by 10 5 m 3 Root mean square error of 337.57m 3 The average absolute percentage error is 0.0383%, and the prediction performance is best.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
While embodiments of the invention have been shown and described, it will be understood by those of ordinary skill in the art that: various changes, modifications, substitutions and alterations can be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.
Claims (15)
1. A method for predicting the gas production amount of a dynamic blast furnace suitable for multiple working condition types is characterized by comprising the following steps:
s1, obtaining production data of a blast furnace and judging the type of a working condition;
s2, dynamically determining the duration of a prediction period based on the type of the production working conditions, acquiring blast furnace production historical data corresponding to different working conditions, selecting the production historical data within a period of time, segmenting the production historical data according to the duration of the prediction period, and using the segmented production historical data as sample data to analyze blast furnace gas production influence factors;
s3, selecting important influence factors corresponding to different working condition types as the input of a blast furnace gas generation amount prediction model under the working condition type;
and S4, establishing a prediction model, and predicting the generation amount of the blast furnace gas under the working condition.
2. The method for predicting the gas generation amount of the dynamic blast furnace applicable to the multi-condition types according to claim 1, wherein in the step S2, the method for dynamically determining the duration of the prediction period based on the production condition types comprises the following steps:
according to the reality of gas scheduling managementSetting the prediction period duration under the forward running working condition as T 1 Setting the prediction period duration in the wind reduction working condition and the compound wind working condition as T 2 Wherein T is 1 >T 2 >0, the gas generation amount is 0 when the wind stops, and the duration of a prediction period is not required to be set.
3. The method for predicting the gas generation amount of the dynamic blast furnace applicable to the multiple operating condition categories according to claim 1, wherein the method for selecting the important influence factors corresponding to the different operating condition categories as the input of the blast furnace gas generation amount prediction model in the operating condition category in the step S3 comprises the following steps:
and calculating the correlation degree values of the influence factors and the blast furnace gas production amount under different working condition types, and selecting the influence factors of which the correlation degree values exceed the threshold value as the input of the blast furnace gas production amount prediction model under the working condition type.
4. The method for predicting the gas generation amount of the dynamic blast furnace applicable to the multi-condition category according to claim 3, wherein the correlation degree is calculated by a principal component analysis method, an analytic hierarchy method, a fuzzy evaluation method or a gray correlation analysis method.
5. The method for predicting the gas generation amount of the dynamic blast furnace applicable to the multi-condition types according to claim 4, wherein when a grey correlation analysis method is adopted, the specific calculation method comprises the following steps:
setting a reference data column X 0 And comparing the data sequence X h ，X 0 For blast furnace gas production data sets, X h For the h-th influencing factor data set relating to blast furnace gas production,
X 0 ，X h and X is represented as:
X 0 ＝(x 0 (1),x 0 (2),…,x 0 (l)) T
X＝(X 1 ,X 2 ,…,X n )
wherein n is the total number of the comparison sequences, namely the total number of influencing factors; l is the number of samples; x is the number of h (i) For the h-th influence factor data set x h The number i of samples in (a) is,
mixing X 0 Combined with X to form a new data sequence X C ：
And respectively calculating the association coefficient xi (k) of each comparison sequence and the corresponding element of the reference sequence, wherein the calculation method comprises the following steps:
wherein, omega is a resolution coefficient, omega is more than 0 and less than or equal to 1, the smaller omega is, the larger difference between the correlation coefficients is, the stronger distinguishing capability is;
calculating the correlation degree value R of the h influencing factor h The calculation method comprises the following steps:
and setting a threshold value, and selecting the influence factors of which the correlation degree values exceed the threshold value as input variables of the blast furnace gas generation quantity prediction model under the working condition type.
6. The method for predicting the gas generation amount of the dynamic blast furnace applicable to the multiple operating conditions according to claim 1 or 3, wherein the step S4 is to establish a prediction model, and the method for predicting the gas generation amount of the blast furnace under the operating conditions comprises the following steps:
s41, establishing a prediction model, wherein the input of the prediction model is an influence factor of which the correlation degree value with the blast furnace gas generation quantity exceeds a threshold value;
s42, optimizing the parameters of the prediction model, if the parameters reach the optimization termination condition, executing the step S43, otherwise, continuing to execute the step S42;
and S43, predicting the blast furnace gas generation amount under the working condition by using the prediction model.
7. The method for predicting the gas production of the dynamic blast furnace applicable to multiple operating condition categories according to claim 6, wherein the prediction model is a trend extrapolation prediction model, a regression prediction model, a Kalman filtering prediction model, a combined prediction model or a neural network prediction model.
8. The method for predicting gas production of a dynamic blast furnace applicable to multiple operation condition classes according to claim 7, wherein when a neural network prediction model is used, the neural network prediction model comprises:
the input layer consists of m nodes, m influence factors with the correlation degree value exceeding a threshold value are respectively input, m is more than or equal to 1 and less than or equal to n, and n is the number of the influence factors;
the hidden layer comprises J nodes, each node is provided with a corresponding basis function, and m influence factors are calculated by using each basis function;
and the output layer receives the calculation results of all the basis functions output by the hidden layer and outputs the predicted value of the gas generation amount.
9. The method of claim 8, wherein the base function at the jth node of the hidden layer is φ j (X)：
The predicted value of the output layer is
Wherein X is an m-dimensional input vector; c. C j Is the center of the jth node, with dimensions the same as X; sigma j Is a radial basis function phi j (X) an expansion constant; j is the total number of hidden layer unit nodes; II X-c j II denotes X and c j Of between, omega j The weight value from the jth node of the hidden layer to the output layer.
10. The method for predicting the gas generation amount of the dynamic blast furnace applicable to multiple operating modes according to claim 4, wherein in step S42, the parameters of the prediction model of the radial basis function neural network are optimized by using an ant colony algorithm, a particle swarm algorithm or an improved genetic algorithm.
11. The method for predicting the gas generation amount of the dynamic blast furnace applicable to the multi-condition type according to claim 10, wherein when the improved genetic algorithm is adopted, the specific steps are as follows:
s421, setting algebra of genetic algorithm;
s422, setting the total number of individuals in the population to be P, and initializing the population, namely randomly generating the population containing P chromosomes; due to c j Having the same dimensions as the input layer, then need to beThe total number of variables coded is equal to J + m.J + J; each variable is coded in a binary mode, the coding length is 10, and the total coding length of each individual is 10 (m + 2) J;
s423, final decoded value F of S variable s Comprises the following steps:
wherein, F s Represents the decoded value of the s-th variable, f s Representing the value of the s-th variable converted from binary to decimal;
s424, after the codes of all the variables on the chromosome are decoded, reducing the decoded variable values into parameter values C, delta and w, and substituting the parameters into a blast furnace gas generation amount prediction model; designing a fitness function of an improved genetic algorithm, and calculating the fitness value of an individual p in the G generation population
Wherein L represents the total number of training samples, L<l；
Representing a predicted value of the ith sample; y is i Representing the true value of the ith sample;
s425, if the maximum iteration number G is reached max I.e. G.gtoreq.G max Then at G max Finding out an individual with the maximum fitness from the generation population, decoding the coding sequence of the individual, and outputting the decoded parameter values C, delta and w; turning to S427, otherwise turning to S426;
s426, designing a genetic operator, and carrying out selection, crossing and mutation operations on individuals; according to the average fitness f of all individuals in the G generation population avg Greater fitness in two intersecting individualsThe value f and the evolution algebra G are subjected to crossover and mutation operations, and the method comprises the following steps:
designing a population Q capable of accommodating P individuals;
selecting operation: carrying out (P-2)/2 times of selection operation in total, randomly grabbing 2 individuals P ', P' from the G generation in each selection operation, removing two individuals P ', P' from the G generation, and then carrying out the next selection operation,
and (3) cross operation: after each selection operation is finished, randomly generating a decimal a between 0 and 1 and combining the self-adaptive cross probability P c If a is less than or equal to P c If so, the two grabbed individuals p ', p' need to be subjected to cross operation to generate two new individuals, and the two newly generated individuals are placed into the population Q; if a>P c Then the two grabbed individuals p ', p' need no cross operation and are directly put into the population Q;
mutation operation: sequentially traversing each individual in the population Q, randomly generating a decimal b between 0 and 1 during each traversal, and combining the self-adaptive variation probability P m If b is less than or equal to P m If so, the traversed individuals need to carry out mutation operation, and the mutation points are randomly generated; otherwise, the traversed individuals do not perform mutation operation;
after the three operations of selection, crossing and mutation are completed, generating a population Q containing (P-2) individuals, then putting the 2 individuals with the maximum G generation fitness value into the population Q, keeping the total number of the individuals of the population Q as P, enabling the next generation population to be equal to the population Q, updating G = G +1, and returning to the step S423;
and S427, substituting the optimized parameter values C, delta and w into the blast furnace gas generation amount prediction model, and predicting the generation amount of the blast furnace gas by using the test data set.
12. The method of claim 11, wherein in step S426, all G generation populations are selected from the group consisting ofAverage fitness f of individual avg The larger fitness value f in two crossing individuals is calculated as follows:
13. the method of claim 11, wherein in step S426, the improved adaptive crossover probability and mutation probability calculation methods are designed such that crossover between individuals is constrained by fitness, mutation operation within individuals is constrained by evolution algebra, and a crossover probability function P c And the mutation probability function P m Comprises the following steps:
P m ＝a 3 /(ln[a 4 ·(G max +1-G)])
wherein f is the greater fitness value of the two crossed individuals; a is 1 ,a 2 As a parameter of a function 1 ,a 2 ∈(0,1]；f avg The average fitness value of all individuals in the population; f. of max The maximum fitness value among all individuals in the population; a is 3 ,a 4 Are all positive and real; g max Is the maximum evolution algebra; g is the current evolution algebra.
14. A method for establishing a dynamic blast furnace gas generation amount prediction model is characterized by comprising the following steps:
s11, acquiring production data of the blast furnace and judging the working condition type;
s12, dynamically determining the duration of a prediction period based on the type of the production working condition;
s13, obtaining blast furnace historical production data corresponding to different working conditions, selecting multiple sections of historical production data with the time length being the duration of a prediction period as influence factor data of blast furnace gas production, dividing the influence factor data into a training set and a test set, wherein the training set is used for training a prediction model, and the test set is used for verifying the prediction performance of the prediction model;
s14, establishing an analysis model to calculate the correlation degree of the influence factors in the training set and the blast furnace gas generation amount under different working condition types;
s15, under each working condition type, selecting the influence factors of which the correlation degree value exceeds a threshold value as the input of a blast furnace gas generation amount prediction model under the working condition type to predict the generation amount of the blast furnace gas;
s16, training the model by using the training set, testing the model by using the testing set, and ending the training if the training times reach the maximum iteration times or the termination condition; otherwise, returning to the step S11;
and S17, predicting the blast furnace gas generation amount by using the trained prediction model and adopting a test set.
15. A dynamic blast furnace gas generation amount prediction system suitable for multiple working condition types, which is characterized by comprising a data acquisition unit and a prediction unit, wherein the data acquisition unit is used for acquiring historical data of blast furnace blast flow, the output end of the data acquisition unit is connected with the input end of the prediction unit, and the prediction unit executes the method of one of claims 1 to 13 to predict the blast furnace gas generation amount.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2022106074200 | 2022-05-31 | ||
| CN202210607420 | 2022-05-31 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115186900A true CN115186900A (en) | 2022-10-14 |
Family
ID=83516811
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210817240.5A Pending CN115186900A (en) | 2022-05-31 | 2022-07-12 | Dynamic blast furnace gas production prediction method and system suitable for multiple working condition types |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115186900A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116612836A (en) * | 2023-07-19 | 2023-08-18 | 福建德尔科技股份有限公司 | Tail gas amount prediction method and system for trifluoromethane production |
-
2022
- 2022-07-12 CN CN202210817240.5A patent/CN115186900A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116612836A (en) * | 2023-07-19 | 2023-08-18 | 福建德尔科技股份有限公司 | Tail gas amount prediction method and system for trifluoromethane production |
| CN116612836B (en) * | 2023-07-19 | 2023-10-03 | 福建德尔科技股份有限公司 | Tail gas amount prediction method and system for trifluoromethane production |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111353656B (en) | Steel enterprise oxygen load prediction method based on production plan | |
| CN109242188B (en) | Long-term interval prediction and structure learning method for steel gas system | |
| CN110082682B (en) | Lithium battery state of charge estimation method | |
| CN102031319A (en) | Method for forecasting silicon content in blast-furnace hot metal | |
| CN115186900A (en) | Dynamic blast furnace gas production prediction method and system suitable for multiple working condition types | |
| CN113589693A (en) | Cement industry decomposing furnace temperature model prediction control method based on neighborhood optimization | |
| CN113919559A (en) | Ultra-short-term prediction method and device for equipment parameters of comprehensive energy system | |
| CN111047103B (en) | Sintering process carbon effect online prediction method and device based on multi-working-condition mode identification | |
| CN109934421B (en) | Blast furnace molten iron silicon content prediction and compensation method for fluctuating furnace condition | |
| CN116796403A (en) | Building energy saving method based on comprehensive energy consumption prediction of commercial building | |
| CN108734419B (en) | Blast furnace gas scheduling system modeling method based on knowledge migration | |
| CN111798023B (en) | Comprehensive coke ratio prediction method in steelmaking sintering production | |
| CN113762591A (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM counterstudy | |
| CN112329269A (en) | Sintering ignition temperature modeling prediction method based on working condition identification | |
| CN105302976A (en) | Aluminum electrolysis production optimization method based on SPEA2 algorithm | |
| CN112257348A (en) | Method for predicting long-term degradation trend of lithium battery | |
| CN116973761A (en) | Method for predicting remaining service life of power battery based on particle swarm optimization | |
| CN116703022A (en) | Power equipment optimal carbon emission evaluation method and system | |
| CN113222339A (en) | Energy consumption difference analysis method and system for enterprises of the same type based on measured panoramic data | |
| CN116127345B (en) | Converter steelmaking process mode design method based on deep clustering generation countermeasure network | |
| CN112541628B (en) | Method and system for predicting blast furnace gas production capacity driven by multi-model coupling | |
| CN111696003B (en) | Intelligent identification and early warning method for abnormal period of line loss rate of distribution room under drive of mass data | |
| CN117831672A (en) | Fuming furnace smelting end point prediction method based on gray connotation GM (1, 1) | |
| CN114912335B (en) | Coal gas generation amount prediction method based on missing data | |
| Wang et al. | Endpoint Prediction of Copper Converter Blowing Based on Artificial Bee Colony Algorithm and Regularized Extreme Learning Machine |
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 |