CN112986491A - Mixture water detection value correction method based on feedback adaptive prediction model - Google Patents
Mixture water detection value correction method based on feedback adaptive prediction model Download PDFInfo
- Publication number
- CN112986491A CN112986491A CN201911295597.6A CN201911295597A CN112986491A CN 112986491 A CN112986491 A CN 112986491A CN 201911295597 A CN201911295597 A CN 201911295597A CN 112986491 A CN112986491 A CN 112986491A
- Authority
- CN
- China
- Prior art keywords
- value
- mixture
- water content
- neural network
- water
- 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
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 136
- 239000000203 mixture Substances 0.000 title claims abstract description 110
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000001514 detection method Methods 0.000 title claims abstract description 40
- 238000012937 correction Methods 0.000 title claims abstract description 22
- 230000003044 adaptive effect Effects 0.000 title claims description 17
- 238000013528 artificial neural network Methods 0.000 claims abstract description 57
- 238000004364 calculation method Methods 0.000 claims abstract description 49
- 239000002994 raw material Substances 0.000 claims abstract description 49
- 238000005245 sintering Methods 0.000 claims abstract description 30
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 28
- 230000008569 process Effects 0.000 claims abstract description 27
- 238000012549 training Methods 0.000 claims description 41
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 claims description 32
- ODINCKMPIJJUCX-UHFFFAOYSA-N Calcium oxide Chemical compound [Ca]=O ODINCKMPIJJUCX-UHFFFAOYSA-N 0.000 claims description 27
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 claims description 24
- LIVNPJMFVYWSIS-UHFFFAOYSA-N silicon monoxide Chemical compound [Si-]#[O+] LIVNPJMFVYWSIS-UHFFFAOYSA-N 0.000 claims description 24
- 239000000292 calcium oxide Substances 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 17
- 239000004615 ingredient Substances 0.000 claims description 17
- 239000000463 material Substances 0.000 claims description 17
- 238000012360 testing method Methods 0.000 claims description 14
- 229910052742 iron Inorganic materials 0.000 claims description 13
- BRPQOXSCLDDYGP-UHFFFAOYSA-N calcium oxide Chemical compound [O-2].[Ca+2] BRPQOXSCLDDYGP-UHFFFAOYSA-N 0.000 claims description 12
- 239000000395 magnesium oxide Substances 0.000 claims description 12
- CPLXHLVBOLITMK-UHFFFAOYSA-N magnesium oxide Inorganic materials [Mg]=O CPLXHLVBOLITMK-UHFFFAOYSA-N 0.000 claims description 12
- AXZKOIWUVFPNLO-UHFFFAOYSA-N magnesium;oxygen(2-) Chemical compound [O-2].[Mg+2] AXZKOIWUVFPNLO-UHFFFAOYSA-N 0.000 claims description 12
- 235000012255 calcium oxide Nutrition 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 8
- 235000019738 Limestone Nutrition 0.000 claims description 6
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N Silicium dioxide Chemical compound O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 claims description 6
- 239000000571 coke Substances 0.000 claims description 6
- 239000010459 dolomite Substances 0.000 claims description 6
- 229910000514 dolomite Inorganic materials 0.000 claims description 6
- 230000005284 excitation Effects 0.000 claims description 6
- 239000006028 limestone Substances 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 238000004088 simulation Methods 0.000 claims description 6
- 239000000126 substance Substances 0.000 claims description 5
- 239000000843 powder Substances 0.000 claims description 4
- 238000005303 weighing Methods 0.000 claims description 4
- FYYHWMGAXLPEAU-UHFFFAOYSA-N Magnesium Chemical compound [Mg] FYYHWMGAXLPEAU-UHFFFAOYSA-N 0.000 claims description 3
- NINIDFKCEFEMDL-UHFFFAOYSA-N Sulfur Chemical compound [S] NINIDFKCEFEMDL-UHFFFAOYSA-N 0.000 claims description 3
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 claims description 3
- 229910052782 aluminium Inorganic materials 0.000 claims description 3
- 238000012854 evaluation process Methods 0.000 claims description 3
- 239000011777 magnesium Substances 0.000 claims description 3
- 229910052749 magnesium Inorganic materials 0.000 claims description 3
- TWNQGVIAIRXVLR-UHFFFAOYSA-N oxo(oxoalumanyloxy)alumane Chemical compound O=[Al]O[Al]=O TWNQGVIAIRXVLR-UHFFFAOYSA-N 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 239000000377 silicon dioxide Substances 0.000 claims description 3
- 235000012239 silicon dioxide Nutrition 0.000 claims description 3
- 239000011593 sulfur Substances 0.000 claims description 3
- 229910052717 sulfur Inorganic materials 0.000 claims description 3
- 238000011160 research Methods 0.000 abstract description 5
- 238000003908 quality control method Methods 0.000 abstract description 2
- 239000002245 particle Substances 0.000 description 14
- 238000004519 manufacturing process Methods 0.000 description 7
- 230000006872 improvement Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 238000007726 management method Methods 0.000 description 5
- 238000003062 neural network model Methods 0.000 description 4
- 238000010606 normalization Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 230000003111 delayed effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005469 granulation Methods 0.000 description 2
- 230000003179 granulation Effects 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000035699 permeability Effects 0.000 description 2
- 229910000831 Steel Inorganic materials 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 238000013329 compounding Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 235000010755 mineral Nutrition 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000008188 pellet Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 238000002945 steepest descent method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/0003—Composite materials
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Manufacture And Refinement Of Metals (AREA)
Abstract
The invention relates to a method for correcting a moisture detection value of a mixture based on a feedback self-adaptive prediction model, which comprises the following steps: step 1: collecting water content information of the raw materials, and step 2: calculating the water content of the mixture through a batching calculation system, and step 3: the MIV algorithm screens model input quantity, a BP neural network is trained by using a collected sample, a trained self-adaptive neural network system predicts a water content value of a mixture and corrects the water content value, a predicted value and a target value are compared at regular time, and if deviation exceeds a certain value, feedback correction is carried out. The technical scheme breaks through the research range of the prior detection process of the sintering mixture, creatively applies a feedback self-adaptive prediction model to train according to the known sintering mixture raw material and then predicts and corrects the water content value of the mixture, so that the detection precision can be greatly improved, the influence of 3-5 minutes of delay detection on quality control can be conveniently avoided by combining with feedforward control, and an advanced adjustment process is provided.
Description
Technical Field
The invention relates to a dredging component, in particular to a method for correcting a moisture detection value of a mixture based on a feedback self-adaptive prediction model, and belongs to the technical field of moisture and humidity detection.
Background
The moisture of the sintering mixture is one of the parameters which need to be strictly controlled in the sintering production, and directly influences the quality of the sintering production. The moisture content of the sintering mixture is too low, the bonding force among particles is small, the ground mineral powder and other additive components cannot be agglomerated into pellets with certain particle size, and the air permeability of a sinter bed is poor, so that the efficiency of the sinter in blast furnace iron-making production is reduced. The sintering mixture has too high water content, although the granulation performance of the particles is good, granulation adhesion is caused due to too high viscosity, and excessive free water is separated out on a cold material layer of a sintering material bed, so that local materials are overhumid, the air permeability is poor, the energy consumption is increased, and the production efficiency is also influenced. Effective control of the moisture of the sinter mix is a constant goal pursued by the sinter production sector.
At present, the detection means of the iron and steel enterprises for the moisture of the sintering mixture mainly depends on the detection by using an online moisture detector or a discontinuous manual detection mode after the sintering raw materials are mixed. The current commonly used on-line moisture meters comprise a near-infrared moisture detector, a microwave moisture monitor, a capacitance moisture detector and the like, which have advantages and disadvantages respectively. In the aspect of detection precision, a bottleneck period seems to be reached, and the biggest problem of the detection technology is that water is added after the sintering raw materials are mixed once, a water detector is used for detecting real-time water during the process of conveying the sintering raw materials to the next procedure on a belt conveying system, the time delay is usually 3-5 minutes, and at the moment, the water content of a considerable part of mixed materials is not in a standard range, so that the water data is easy to lose control.
With the development of neural networks and intelligent detection technologies, more and more factories and enterprises improve production technologies in a dispute, especially sintering key process procedures such as a sintering process site, a mixture moisture detection and control process, a sintering end point temperature control and the like are widely concerned and researched, but because the working condition environment of the sintering site is complex and external factors greatly influence the sintering process procedures, the improvement of the technologies is usually only stopped at a theoretical research stage, and the problems in the prior art can not be completely solved when the technology is applied to an actual site. Therefore, a new solution to solve the above technical problems is urgently needed.
Disclosure of Invention
The invention provides a method for correcting the moisture detection value of a mixture based on a feedback self-adaptive prediction model aiming at the problems in the prior art, the technical scheme breaks through the research range of the prior detection process of the sintered mixture, and creatively applies the feedback self-adaptive prediction model to train according to the known sintered mixed raw materials and then predict and correct the moisture value of the mixture, thereby not only greatly improving the detection precision, but also conveniently combining with feedforward control to avoid the influence on quality control caused by delayed detection for 3-5 minutes, and having an advanced adjustment process,
in order to achieve the above object, the technical solution of the present invention is a method for correcting a moisture detection value of a mixture based on a feedback adaptive prediction model, comprising the steps of:
step 1: the information of the water content of the raw materials is collected,
step 2: the water content of the mixture is calculated by the ingredient calculation system,
and step 3: the MIV algorithm screens model input quantity, a BP neural network is trained by using a collected sample, a trained self-adaptive neural network system predicts a water content value of a mixture and corrects the water content value, a predicted value and a target value are compared at regular time, and if deviation exceeds a certain value, feedback correction is carried out.
As an improvement of the invention, step 1: the method comprises the following steps of collecting water content information of raw materials:
providing sintering raw material component information of the sinter by a sintering site worker, wherein the sintering raw material component information comprises the water content and the content of the mixed ore containing iron ore, dolomite, quicklime, limestone, coke powder and return finesHas corresponding material components (such as Fe, CaO, MgO, SiO)2) The percentage of (2) each mixed raw material is placed in the bin, the outlet is provided with a disc weighing device, the weight of the mixed raw material can be read when the mixed raw material is discharged, and the supply of all the raw materials of the bin is completed by a belt flow connected with a raw material yard.
As an improvement of the present invention, step 2: the water content of the mixture is calculated by a batching calculation system, and the method comprises the following steps:
the formula for calculating the mixture ratio is as follows:
in the formula: hkiThe proportion r of each component is the return fine proportion and the dry weight;
c-ratio of coke powder, dry weight miThe mixture is prepared by mixing the components according to the proportion and the dry weight;
pimoisture content of each component of the mix i-batch tank number, 1-n
The formula for calculating the discharge amount of each component of the mixture is as follows: wsi＝Wt×Hki；
The water content of each ingredient component is calculated by the formula: mi＝Pi×Wsi
The coke powder proportioning coefficient calculation formula is as follows:
Wsi-the amount of each component delivered; wt-total delivery (t/h); p is a radical ofc-rate of coke powder delivery;
coke powder discharge amount calculation formula Wsc＝Wt×Hkc；
Calculation formula M for moisture content of coke powdersc＝Psc×Wsc；
psc-moisture content of the coke powder;
the return mine proportioning coefficient calculation formula is as follows:
return ore discharge calculation formula Wsr＝Wt×Hkr；
The water content of the mixed material after the mixing is as follows
Let the predicted water content of the mix be pdCorresponding to a water content of WdThen, there are:
Wd＝pd×(Mt-Wmix+Wd)；
the desired moisture content can be calculated as:
the supplementary water adding amount is F ═ Wd-Wmix；
The water adding amount formula for distributing the supplementary water adding amount to the primary mixing and the secondary mixing is respectively as follows;
F1＝F×80％ F2＝F×20％
wherein F1-one-time mixing water addition amount (t/h) F2-secondary mixing water addition (t/h)
And the calculated supplementary water addition amount is also used as an input parameter of the feedback self-adaptive prediction model to predict the water content of the mixture.
As an improvement of the present invention, the step 3 specifically includes the following steps: selecting a BP neural network as a core of a model, wherein the established network model comprises an input layer, an intermediate layer and an output layer, and the training process comprises the following steps: determining the number of network layers and the number of nodes in each layer; defining an excitation function and a loss function; normalizing the data; initializing a network; forward calculation; calculating errors and calculating error correction learning errors; and correcting the network weight and the bias. The improved BP neural network fuses an MIV (mean influence value) algorithm and can screen all independent variables, and the independent variables with the influence degree on the prediction result reaching a preset standard are screened out; and taking the screened independent variables with the influence degree on the prediction result reaching the preset standard as input parameters of the prediction model.
As an improvement of the present invention, the step 3 specifically includes the following steps: for the BP neural network training process, firstly determining the number of layers and the number of nodes, generally determining a three-layer BP neural network according to experience, wherein the number of hidden layer nodes is 10, selecting input parameters meeting the influence standard on the result through an MIV algorithm in order to establish a data set for model training and testing, and selecting objects of iron-containing grade (TFe) and silicon dioxide (SiO)2) Calcium oxide (CaO), aluminum oxide (Al)2O3) The material proportion and water content of the sintering raw materials of 6 elements of magnesium oxide (MgO) and sulfur (S), and the alkalinity control (CaO/SiO)2) The screening result shows that the mixed ore containing iron ore, dolomite, quicklime, limestone, coke powder and return ore contain the water contents and the components (including Fe, CaO, MgO and SiO) of corresponding substances2) The percentage of the total water content and the water content of each raw material meet the influence standard, 1000 groups of raw material samples of the influence factors under stable working conditions are collected on site, 900 groups of data are randomly selected as a training set to train the established BP neural network, and the other 100 groups of data are used as a test set to test whether the predicted value of the BP neural network to the water content of the mixture is close to a target value. And comparing the calculated predicted value of the water content with the target value, taking the difference value of the two as an input value of neural network training, and determining the intermediate weight and the threshold value of the neural network by training until the difference between the predicted value and the target value is small enough.
As an improvement of the present invention, the step 3 is specifically as follows, and is used for evaluating the influence degree of each factor on the predicted water content value of the mixture for the MIV algorithm. The evaluation process of the MIV algorithm mainly comprises the following steps: (1) the original sample matrix Input is eight influence factors of the predicted value of the water of the mixture, and Output is the predicted value of the water of the mixture; (2) completing model training on the basis of the original matrix; (3) adding and subtracting 10% on the basis of Input to form new samples Input1 and Input2 for model simulation, and obtaining simulation results Output1 and Output 2; (4) calculating the difference value between Output1 and Output2, namely an influence change value (IV), and calculating the average influence value (MIV) of the network input to the network Output according to the sample size; (5) and (3) calculating MIV values of 8 predicted value influence factors of the mixture water according to the steps, sequencing and analyzing, wherein the absolute value represents the influence degree of each factor, the symbol represents the influence direction of each factor, finally determining the iron ore containing ratio and the water content thereof, the calcium oxide containing raw material ratio and the components thereof, the magnesium and aluminum containing ore ratio and the components thereof, the coke oven return ore ratio, and calculating the water adding amount as an input variable of a neural network.
Compared with the prior art, the method has the advantages that 1) the technical scheme adopts the MIV algorithm to screen all independent variables, and the independent variables with the influence degree on the prediction result reaching the preset standard are screened out; the screened independent variables with the influence degree on the prediction result reaching the preset standard are used as input parameters, so that the problem of complex calculation caused by excessive input parameters of the neural network can be solved; 2) the method breaks through the research range of the prior detection process of the sintering mixture, creatively applies a feedback self-adaptive prediction model to predict and correct the online detection moisture value of the mixture after training according to the known proportion of the sintering mixture raw materials, the moisture content and the calculated water addition amount, and can avoid the influence caused by the large deviation of the online moisture detection value of the mixture and the moisture detection feedback when the mixture is conveyed to a moisture detector for 3-5 minutes on a belt conveyor, because if the moisture content of the mixture in the conveying process exceeds the standard range, the influence is caused when the mixture is conveyed to the moisture detector and can be detected, and corresponding adjustment measures are taken, and the influence is caused at the moment; 3) according to the method, a batching calculation system is added before a self-adaptive prediction model, all material tanks are numbered on the basis of the original batching calculation model, the relation among batching components is summarized by analyzing sample data, return fine proportioning and coke powder proportioning influence factors are innovatively added, all raw material proportioning is obtained, and the discharge amount and the moisture content of all components of a mixture are calculated; and analyzing sample data of the coke powder proportioning and the return mine proportioning, obtaining a coke powder proportioning coefficient calculation formula and a return mine proportioning coefficient calculation formula by referring to the proportioning coefficient calculation formulas of all the components, obtaining the water content rate of the mixture, reading the feeding amount through a disc weighing device, finally obtaining the expected water feeding amount, and realizing the advanced water regulation. The water adding amount of the mixture calculated theoretically is innovatively used as an input parameter of the self-adaptive prediction model, and whether the water adding amount meets the requirement of the real-time state of the mixture or not can be judged after calculation in advance.
Drawings
FIG. 1 is a flow chart of the correction of the moisture content of the mixture based on the feedback adaptive prediction model according to the present invention.
FIG. 2 is a data flow diagram of the ingredient calculation system of the present invention.
FIG. 3 is a diagram of a feedback adaptive prediction model of the MIV algorithm of the present invention.
FIG. 4 is a BP neural network training process of the present invention.
FIG. 5 is a system for correcting moisture prediction of a mix in accordance with the present invention.
Wherein: in fig. 2:
the specific implementation mode is as follows:
for the purpose of enhancing an understanding of the present invention, the present embodiment will be described in detail below with reference to the accompanying drawings.
Example 1: referring to fig. 1-5, a method for correcting a moisture detection value of a mixture based on a feedback adaptive prediction model includes the following steps:
step 1: the information of the water content of the raw materials is collected,
step 2: the water content of the mixture is calculated by the ingredient calculation system,
and step 3: the MIV algorithm screens model input quantity, a BP neural network is trained by using a collected sample, a trained self-adaptive neural network system predicts a water content value of a mixture and corrects the water content value, a predicted value and a target value are compared at regular time, and if deviation exceeds a certain value, feedback correction is carried out.
Step 1: the method comprises the following steps of collecting water content information of raw materials:
providing sintering material component information of sintered ore by sintering site worker, including the water content of mixed ore containing iron ore, dolomite, quicklime, limestone, coke powder and return fines, and the content of corresponding substances (such as Fe, CaO, MgO and SiO)2) The percentage of (2) each mixed raw material is placed in the bin, the outlet is provided with a disc weighing device, the weight of the mixed raw material can be read when the mixed raw material is discharged, and the supply of all the raw materials of the bin is completed by a belt flow connected with a raw material yard.
Step 2: the water content of the mixture is calculated by a batching calculation system, and the method comprises the following steps:
the formula for calculating the mixture ratio is as follows:
in the formula: hkiThe proportion r of each component is the return fine proportion and the dry weight;
c-ratio of coke powder, dry weight miThe mixture is prepared by mixing the components according to the proportion and the dry weight;
pimoisture content of each component of the mix i-batch tank number, 1-n
The formula for calculating the discharge amount of each component of the mixture is as follows: wsi＝Wt×Hki；
The water content of each ingredient component is calculated by the formula: mi＝Pi×Wsi
The coke powder proportioning coefficient calculation formula is as follows:
Wsi-ComponentsThe amount of delivery; wt-total delivery (t/h); p is a radical ofc-rate of coke powder delivery;
coke powder discharge amount calculation formula Wsc＝Wt×Hkc；
Calculation formula M for moisture content of coke powdersc＝Psc×Wsc；
psc-moisture content of the coke powder;
the return mine proportioning coefficient calculation formula is as follows:
return ore discharge calculation formula Wsr＝Wt×Hkr；
The water content of the mixed material after the mixing is as follows
Let the predicted water content of the mix be pdCorresponding to a water content of WdThen, there are:
Wd＝pd×(Mt-Wmix+Wd)；
the desired moisture content can be calculated as:
the supplementary water adding amount is F ═ Wd-Wmix；
The water adding amount formula for distributing the supplementary water adding amount to the primary mixing and the secondary mixing is respectively as follows;
F1＝F×80％ F2＝F×20％
wherein F1-one-time mixing water addition amount (t/h) F2-secondary mixing water addition (t/h)
And the calculated supplementary water addition amount is also used as an input parameter of the feedback self-adaptive prediction model to predict the water content of the mixture.
The step 3 is specifically as follows: selecting a BP neural network as a core of a model, wherein the established network model comprises an input layer, an intermediate layer and an output layer, and the training process comprises the following steps: determining the number of network layers and the number of nodes in each layer; defining an excitation function and a loss function; normalizing the data; initializing a network; forward calculation; calculating errors and calculating error correction learning errors; and correcting the network weight and the bias. The improved BP neural network fuses an MIV (mean influence value) algorithm and can screen all independent variables, and the independent variables with the influence degree on the prediction result reaching a preset standard are screened out; and taking the screened independent variables with the influence degree on the prediction result reaching the preset standard as input parameters of the prediction model.
The step 3 is specifically as follows: for the BP neural network training process, firstly determining the number of layers and the number of nodes, generally determining a three-layer BP neural network according to experience, wherein the number of hidden layer nodes is 10, selecting input parameters meeting the influence standard on the result through an MIV algorithm in order to establish a data set for model training and testing, and selecting objects of iron-containing grade (TFe) and silicon dioxide (SiO)2) Calcium oxide (CaO), aluminum oxide (Al)2O3) The material proportion and water content of the sintering raw materials of 6 elements of magnesium oxide (MgO) and sulfur (S), and the alkalinity control (CaO/SiO)2) The burning loss ratio, and the screening result shows that the water content of the mixed ore containing iron ore, dolomite, quicklime, limestone, coke powder and return ore and the content of corresponding substance components (such as Fe, CaO, MgO and SiO)2) The percentage of the total water content and the water content of each raw material meet the influence standard, 1000 groups of raw material samples of the influence factors under stable working conditions are collected on site, 900 groups of data are randomly selected as a training set to train the established BP neural network, and the other 100 groups of data are used as a test set to test whether the predicted value of the BP neural network to the water content of the mixture is close to a target value. And comparing the calculated predicted value of the water content with the target value, taking the difference value of the two as an input value of neural network training, and determining the intermediate weight and the threshold value of the neural network by training until the difference between the predicted value and the target value is small enough.
The step 3 is specifically as follows, and for the MIV algorithm, the influence degree of each factor on the water prediction value of the mixture is evaluated. The evaluation process of the MIV algorithm mainly comprises the following steps: (1) the original sample matrix Input is eight influence factors of the predicted value of the water of the mixture, and Output is the predicted value of the water of the mixture; (2) completing model training on the basis of the original matrix; (3) adding and subtracting 10% on the basis of Input to form new samples Input1 and Input2 for model simulation, and obtaining simulation results Output1 and Output 2; (4) calculating the difference value between Output1 and Output2, namely an influence change value (IV), and calculating the average influence value (MIV) of the network input to the network Output according to the sample size; (5) and (3) calculating MIV values of 8 predicted value influence factors of the mixture water according to the steps, sequencing and analyzing, wherein the absolute value represents the influence degree of each factor, the symbol represents the influence direction of each factor, finally determining the iron ore containing ratio and the water content thereof, the calcium oxide containing raw material ratio and the components thereof, the magnesium and aluminum containing ore ratio and the components thereof, the coke oven return ore ratio, and calculating the water adding amount as an input variable of a neural network.
Under different weather conditions (clear days, cloudy days, rainy days and the like), the information of the water content of the same mixture is collected, the change of the air humidity caused by the weather change is compared to find that the water content of the mixture has certain influence, the change is defined as a weather influence factor eta, the range of the weather influence factor eta is between 0.9 and 1, the weather influence factor eta is generally close to 1 in a clear day, the weather influence factor eta is close to 0.9 in other weather conditions, and the weather influence factor eta is also used as the input quantity of the self-adaptive prediction model. Considering the influence of weather change on the humidity of the mixture, defining the influence as a weather influence factor, and generally placing the weather influence factor on the predicted moisture fraction p of the mixturedIn the past, the influence on the water content is negligible in sunny days, so the influence is omitted, and in other conditions such as rainy days and the like, the water content rate calculated by a prediction model is lower than the actual water content rate, and the weather influence factor is considered, so the water content rate is pdThe coefficient {1+ (1-eta) } is added before, and the influence factor is not added because the prediction model is taken in the case of common sunny days.
Further, in the neural network training process, the excitation function is used for calculating linear or nonlinear functions of output of each layer of nodes, and the loss function is used for measuring the output result of the output layer anderror in target outcome. The output of the hidden layer is selected to be excited by a Sigmoid function, and a loss function is introduced to be gamma2And evaluating the model prediction result by using the performance index. The basic idea of the BP network algorithm is as follows: the weight values of all layers are continuously adjusted through the forward propagation of signals and the backward propagation of errors so as to achieve the purpose of network learning, and therefore the connection weight values among all nodes are determined, and the expected state can be achieved through the self-learning function. The inputs sorted by fitness as described above are used as a training set of a three-layer BP neural network, and the Sigmoid function adopted by the hidden layer is Sigmoid (x) 1/(1+ e)-x) And the value of each neuron of the hidden layer is the weighted sum of the value of each node of the input layer multiplied by the connection weight of the neuron, the weight factor is adjusted by a reverse gradient steepest descent method, if the sum is greater than a threshold value, the output of the hidden node is 1, otherwise, the output is 0. And after the training of the BP neural network is finished, the prediction performance of the water content is tested by using 100 groups of test set test models.
And the predicted value of the moisture content of the BP neural network model, the actual value of the moisture content of the mixture and the average value of the actual values of the moisture content of the mixture are represented, and m represents the number of samples. When gamma is2When the index is close to 0.0, the BP neural network model is shown to be the best in the prediction performance of the water content of the mixture; when gamma is2When the performance index is close to 1.0, the predicted performance is the worst. Therefore, to improve the gamma of the prediction model2Optimizing BP neural network model by using performance index, and optimizing model parameters such as excitation function, hidden node number and the like in the model by using a grid search method.
Data normalization and network initialization in the neural network training process, because various variable dimensions are different, the data normalization processing is adopted, the amplitude of network weight value change can be reduced, and calculation is easy; the method has the advantages that the parameters of each node of the network are initialized, the model training speed can be improved, the situation that the local optimal solution is trapped is prevented, and the network weight parameters are searched and initialized by adopting a PSO algorithm in a heuristic algorithm.
The Particle Swarm Optimization (Particle Swarm Optimization) is a Swarm intelligence algorithm, which was proposed in 1995 and comes from the random predation behavior of a bird Swarm, each Particle in the algorithm represents a potential optimal solution of the extremum Optimization problem, the Particle characteristics are represented by three indexes, namely position, speed and an adaptive value, the adaptive value is calculated by a fitness function, and the quality of the value represents the quality of the Particle. The particles move in the solution space, and the individual positions are updated by tracking individual extremum Pbest and group extremum Gbest, wherein the individual extremum Pbest refers to the optimal position of the fitness value obtained by calculation in the positions where the individuals experience, and the group extremum Gbest refers to the optimal position of the fitness value obtained by searching all the particles in the group. And calculating a fitness value once every time the particle updates the position, and updating the individual extremum Pbest and the group extremum Gbest by comparing the fitness value of the new particle with the individual extremum and the fitness value of the group extremum.
Xi＝(xi1，xi2，···,xin)T-the position of the ith particle in the search space;
Vi＝(vi1，Vi2，···ViD)T-the velocity of the ith particle;
Pi＝(pi1，pi2，···piD)T-an individual extremum;
Pg＝(pg1，pg2，···pgD)T-a population extremum;
omega is inertia, c1And c2Is an acceleration factor, r1And r2Is distributed in [0, 1 ]]A random number in between.
The forward calculation, calculation error and calculation error correction learning error of the neural network training process are that normalized data are input into a network model, and the output of each layer of nodes is calculated to obtain the output result of the neural network calculation; calculating the error between the result of the training set and the output result, and if the error is within a set range, ending the training process; obtaining error correction learning errors of each layer of node parameters by a gradient descent method; and modifying the parameters of each layer of nodes of the network through the error correction learning error and the learning rate, and jumping to forward calculation.
And (4) detecting periodically every three months, and when the deviation between the predicted value and the actual value of the water content of the mixture exceeds a certain value, acquiring related parameter input information and retraining the neural network. After a series of pretreatment is carried out on the mixture raw materials by the method, the self-adaptive prediction model trained by the data set predicts the water content of the mixture according to the input variable, compares the water content with the standard water content, and then carries out water adding regulation and advanced regulation, thereby solving the problems of water adding delay and lag.
The specific application embodiment is as follows:
the invention provides a new method for correcting a moisture detection value of a mixture, which is applied to the detection and correction of the moisture value of a sintering mixture in a sintering compounding process, for example, firstly, a large amount of data of a component table of a mixed raw material, the water content of the raw material under different weather conditions and the water content of the mixture in a stable output state in a certain period of time are taken from a sintering site, data preprocessing, filtering, noise reduction, normalization and the like are carried out, the data are divided into a training set and a test set (for example, 100 sets of data), the training set is used for training a BP neural network model, then weight factors of each layer of the model are optimized through a PSO algorithm, and finally, a; after the training of the BP neural network is finished, the water content is subjected to prediction testing by using 100 groups of test set data testing models. Compared with the target value calculated by the formula derived from expert experience and theory, and used for correction and retraining. When the precision requirement is met, determining parameters of the BP neural network for water detection, correction and compensation in production; the method comprises the following specific steps:
fig. 1 is a flowchart illustrating a method for correcting a moisture detection value of a mixture based on a feedback adaptive prediction model according to the present invention, and as shown in fig. 1, the method for correcting a moisture detection value of a mixture based on a feedback adaptive prediction model according to embodiment 1 of the present invention includes the following steps:
firstly, collecting water content information of raw materials, calculating the water adding amount of a mixture through a batching calculation system, screening model input amount through an MIV algorithm, training a BP neural network by using a collected sample, predicting and correcting the water content value of the mixture through a trained adaptive neural network system, comparing a predicted value with a target value at regular time, and performing feedback correction if the deviation exceeds a certain value.
The system can be roughly divided into a batching calculation module, an MIV algorithm screening module, a sample acquisition module, a sample training module, a self-adaptive neural network prediction module and a mixture moisture prediction and correction module.
The ingredient calculation module is as shown in the data flow chart of the ingredient calculation system of fig. 2, and comprises raw material component classification data, target value classification data, current ingredient data, ingredient calculation data, date, historical ingredient data and historical ingredient data maintenance, ingredient calculation is carried out according to the raw material component data and the target value classification data, the calculated ingredient data is stored in a corresponding database, the current ingredient data is stored in the historical ingredient database and is maintained, and the date is recorded.
The sample collection refers to selecting the proportion and water content information of each mixed raw material under different weather conditions (such as raining, sunny days and cloudy days), and selecting the information of the water adding amount and the actual water content of the mixed material within 12 hours when the working condition is stable as a training sample of the neural network. After the sample is collected, the neural network needs to be trained, as shown in fig. 4, the number of layers and the number of nodes in each layer of the BP neural network are determined, an excitation function and a loss function are defined to measure errors of an output result and an expected result, data are normalized, the network is initialized to improve the training speed of the model, the BP neural network is prevented from falling into a local optimal solution, a calculation result of the neural network is obtained through forward calculation, errors are calculated and error correction learning errors are calculated, if the errors are within a set range, the training process is ended, otherwise, parameters of each node are modified through correcting the weight of the network and the bias, and forward calculation is carried out again.
And evaluating the influence degree of each factor on the predicted value of the water content of the mixture by using an MIV algorithm, and screening out variables with the influence degrees meeting the requirements as input information.
Example 2
As shown in fig. 5, the system for predicting and correcting the moisture of the mixture mainly comprises three parts, namely data display, moisture prediction and correction and system management. The data display part comprises the display of real-time data, water adding amount data and sample data and the query and download of historical data. The moisture prediction and correction part is a process of training a neural network through a sample, inputting variables screened by an MIV algorithm into the trained neural network, outputting a predicted value of the moisture of the mixture, comparing the predicted value with a target value, feeding back a difference value to input, and readjusting. The system management part is used for managing data of the moisture prediction correction process, and comprises algorithm management, log management and authority management.
The invention breaks through the research range of the prior sintered mixture detection process, creatively applies the feedback self-adaptive prediction model to predict and correct the moisture detection value of the mixture after training according to the known sintered mixed raw materials, thus avoiding the influence caused by the larger deviation of the online moisture detection value of the mixture and the delayed detection of the mixture for 3-5 minutes, predicting the moisture content of the mixture by the self-adaptive prediction model before the moisture detector detects the moisture value of the mixture, adjusting the water adding module in advance, greatly reducing the reject ratio of sintered products, reducing ore return and saving cost.
It should be noted that the above-mentioned embodiments are not intended to limit the scope of the present invention, and all equivalent modifications and substitutions based on the above-mentioned technical solutions are within the scope of the present invention as defined in the claims.
Claims (6)
1. A method for correcting a moisture detection value of a mixture based on a feedback adaptive prediction model is characterized by comprising the following steps:
step 1: the information of the water content of the raw materials is collected,
step 2: the water content of the mixture is calculated by the ingredient calculation system,
and step 3: the MIV algorithm screens model input quantity, a BP neural network is trained by using a collected sample, a trained self-adaptive neural network system predicts a water content value of a mixture and corrects the water content value, a predicted value and a target value are compared at regular time, and if deviation exceeds a certain value, feedback correction is carried out.
2. The method for correcting the moisture detection value of the mixture based on the feedback adaptive prediction model according to claim 1, characterized in that the method comprises the following steps: the method comprises the following steps of collecting water content information of raw materials:
comprises the water contents of the mixed ore containing iron ore, dolomite, quicklime, limestone, coke powder and return fines and the components containing corresponding substances (including Fe, CaO, MgO and SiO)2) The percentage of (2) each mixed raw material is placed in the feed bin, and the exit is equipped with the disc weighing device, can read its weight when the mixed raw material ejection of compact.
3. The method for correcting the moisture detection value of the mixture based on the feedback adaptive prediction model according to claim 2, wherein the step 2 comprises the following steps: the water content of the mixture is calculated by a batching calculation system, and the method comprises the following steps:
the formula for calculating the mixture ratio is as follows:
in the formula: hkiThe proportion r of each component is the return fine proportion and the dry weight;
c-ratio of coke powder, dry weight miThe mixture is prepared by mixing the components according to the proportion and the dry weight;
pimoisture content of each component of the mix i-batch tank number, 1-n
The formula for calculating the discharge amount of each component of the mixture is as follows: wsi＝Wt×Hki；
The water content of each ingredient component is calculated by the formula: mi＝Pi×Wsi
The coke powder proportioning coefficient calculation formula is as follows:
Wsi-the amount of each component delivered; wt-total delivery (t/h); p is a radical ofc-rate of coke powder delivery;
coke powder discharge amount calculation formula Wsc＝Wt×Hkc；
Calculation formula M for moisture content of coke powdersc＝Psc×Wsc；
psc-moisture content of the coke powder;
the return mine proportioning coefficient calculation formula is as follows:
return ore discharge calculation formula Wsr＝Wt×Hkr；
The water content of the mixed material after the mixing is as follows
Let the predicted water content of the mix be pdCorresponding to a water content of WdThen, there are:
Wd＝pd×(Mt-Wmix+Wd)；
the desired moisture content can be calculated as:
the supplementary water adding amount is F ═ Wd-Wmix；
The water adding amount formula for distributing the supplementary water adding amount to the primary mixing and the secondary mixing is respectively as follows;
F1＝F×80％ F2＝F×20％
wherein F1——The first mixing water adding amount (t/h) F2-secondary mixing water addition (t/h)
And the calculated supplementary water addition amount is also used as an input parameter of the feedback self-adaptive prediction model to predict the water content of the mixture.
4. The method for correcting the moisture detection value of the mixed material based on the feedback adaptive prediction model according to claim 3, wherein the step 3 is as follows: selecting a BP neural network as a core of a model, wherein the established network model comprises an input layer, an intermediate layer and an output layer, and the training process comprises the following steps: determining the number of network layers and the number of nodes in each layer; defining an excitation function and a loss function; normalizing the data; initializing a network; forward calculation; calculating errors and calculating error correction learning errors; and correcting the network weight and the bias.
5. The method for correcting the moisture detection value of the mixed material based on the feedback adaptive prediction model according to claim 3, wherein the step 3 is as follows:
for the training process of the BP neural network, firstly determining the number of layers and the number of nodes of the BP neural network, generally determining a three-layer BP neural network, wherein the number of hidden layer nodes is 10, selecting input parameters meeting the standard influencing the result through an MIV algorithm, and selecting objects of iron-containing grade (TFe) and silicon dioxide (SiO)2) Calcium oxide (CaO), aluminum oxide (Al)2O3) The material proportion and water content of the sintering raw materials of 6 elements of magnesium oxide (MgO) and sulfur (S), and the alkalinity control (CaO/SiO)2) The screening result shows that the mixed ore containing iron ore, dolomite, quicklime, limestone, coke powder and return ore contain the water contents and the components (including Fe, CaO, MgO and SiO) of corresponding substances2) The percentage of the total water content of each raw material and the water content of each raw material meet the influence standard, 1000 groups of raw material samples of the influence factors under stable working conditions are collected on site, 900 groups of data are randomly selected as a training set to train the established BP neural network, and the other 100 groups of data are used as a test set to test the BP neural network pairWhether the predicted value of the mixture water is close to the target value. And comparing the calculated predicted value of the water content with the target value, taking the difference value of the two as an input value of neural network training, and determining the intermediate weight and the threshold value of the neural network by training until the difference between the predicted value and the target value is small enough.
6. The method for correcting the moisture detection value of the mixed material based on the feedback adaptive prediction model according to claim 3, wherein the step 3 is as follows, and for the MIV algorithm, the MIV algorithm mainly comprises the following evaluation processes: (1) the original sample matrix Input is eight influence factors of the predicted value of the water of the mixture, and Output is the predicted value of the water of the mixture; (2) completing model training on the basis of the original matrix; (3) adding and subtracting 10% on the basis of Input to form new samples Input1 and Input2 for model simulation, and obtaining simulation results Output1 and Output 2; (4) calculating the difference value between Output1 and Output2, namely an influence change value (IV), and calculating the average influence value (MIV) of the network input to the network Output according to the sample size; (5) and (3) calculating MIV values of 8 predicted value influence factors of the mixture water according to the steps, sequencing and analyzing, wherein the absolute value represents the influence degree of each factor, the symbol represents the influence direction of each factor, finally determining the iron ore containing ratio and the water content thereof, the calcium oxide containing raw material ratio and the components thereof, the magnesium and aluminum containing ore ratio and the components thereof, the coke oven return ore ratio, and calculating the water adding amount as an input variable of a neural network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911295597.6A CN112986491A (en) | 2019-12-16 | 2019-12-16 | Mixture water detection value correction method based on feedback adaptive prediction model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911295597.6A CN112986491A (en) | 2019-12-16 | 2019-12-16 | Mixture water detection value correction method based on feedback adaptive prediction model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112986491A true CN112986491A (en) | 2021-06-18 |
Family
ID=76343400
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911295597.6A Pending CN112986491A (en) | 2019-12-16 | 2019-12-16 | Mixture water detection value correction method based on feedback adaptive prediction model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112986491A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113406313A (en) * | 2021-06-28 | 2021-09-17 | 浙江邦业科技股份有限公司 | Method for predicting f-CaO of clinker in real time based on data of full-automatic free calcium oxide analyzer |
| CN114842918A (en) * | 2022-04-11 | 2022-08-02 | 青岛恒拓环保科技有限公司 | Automatic sintering mixture water adding method based on machine learning |
| CN115430347A (en) * | 2022-07-14 | 2022-12-06 | 上海荣美农业科技有限公司 | Controller, stirring apparatus, control method, and computer-readable storage medium |
| CN116626080A (en) * | 2023-07-24 | 2023-08-22 | 四川省石棉县恒达粉体材料有限责任公司 | Marble screening method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102270265A (en) * | 2010-06-07 | 2011-12-07 | 鞍钢股份有限公司 | Online detection method for moisture of mixture |
| CN108133106A (en) * | 2017-12-21 | 2018-06-08 | 山东诺德能源科技有限公司 | A kind of method for the moisture that mixture of sintering machine is estimated according to amount of water |
| CN109376863A (en) * | 2018-11-02 | 2019-02-22 | 国网浙江省电力有限公司宁波供电公司 | Photovoltaic power based on MIV-BP neural network is classified prediction technique in short term |
| CN110533082A (en) * | 2019-08-09 | 2019-12-03 | 东北大学秦皇岛分校 | A kind of sintering, mixing and water adding control method based on dual model collaborative forecasting |
-
2019
- 2019-12-16 CN CN201911295597.6A patent/CN112986491A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102270265A (en) * | 2010-06-07 | 2011-12-07 | 鞍钢股份有限公司 | Online detection method for moisture of mixture |
| CN108133106A (en) * | 2017-12-21 | 2018-06-08 | 山东诺德能源科技有限公司 | A kind of method for the moisture that mixture of sintering machine is estimated according to amount of water |
| CN109376863A (en) * | 2018-11-02 | 2019-02-22 | 国网浙江省电力有限公司宁波供电公司 | Photovoltaic power based on MIV-BP neural network is classified prediction technique in short term |
| CN110533082A (en) * | 2019-08-09 | 2019-12-03 | 东北大学秦皇岛分校 | A kind of sintering, mixing and water adding control method based on dual model collaborative forecasting |
Non-Patent Citations (2)
| Title |
|---|
| 刘伟东: "神经网络PID在烧结混合料水分控制中的应用", 《仪器仪表与分析监测》 * |
| 唐选觉等: "《现代钢铁工业技术 烧结》", 31 October 1984, 冶金工业出版社 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113406313A (en) * | 2021-06-28 | 2021-09-17 | 浙江邦业科技股份有限公司 | Method for predicting f-CaO of clinker in real time based on data of full-automatic free calcium oxide analyzer |
| CN114842918A (en) * | 2022-04-11 | 2022-08-02 | 青岛恒拓环保科技有限公司 | Automatic sintering mixture water adding method based on machine learning |
| CN114842918B (en) * | 2022-04-11 | 2024-05-24 | 青岛恒拓环保科技有限公司 | Automatic water adding method for sintering mixture based on machine learning |
| CN115430347A (en) * | 2022-07-14 | 2022-12-06 | 上海荣美农业科技有限公司 | Controller, stirring apparatus, control method, and computer-readable storage medium |
| CN115430347B (en) * | 2022-07-14 | 2024-04-19 | 上海荣美农业科技有限公司 | Controller, stirring apparatus, control method, and computer-readable storage medium |
| CN116626080A (en) * | 2023-07-24 | 2023-08-22 | 四川省石棉县恒达粉体材料有限责任公司 | Marble screening method |
| CN116626080B (en) * | 2023-07-24 | 2023-09-26 | 四川省石棉县恒达粉体材料有限责任公司 | Marble screening method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112986491A (en) | Mixture water detection value correction method based on feedback adaptive prediction model | |
| CN106802977B (en) | Method for predicting performance index of sinter and evaluating comprehensive quality | |
| CN101226377B (en) | Robust control method for asphalt mixing plant batching error | |
| CN100401213C (en) | Intelligent optimized control method for comprehensive production index in ore dressing process | |
| Wu et al. | An intelligent integrated optimization system for the proportioning of iron ore in a sintering process | |
| CN102059071A (en) | Automatic blending material control system for sintering production | |
| CN1831087A (en) | Method for calculating thermal property of coke and control method therefor | |
| CN110070217A (en) | A kind of Forcasting Sinter Quality method of Kernel-based methods parameter | |
| CN100546791C (en) | Asphalt mixture stirring equipment cold aggregate feeding control method | |
| CN105023061B (en) | Iron and steel enterprise's gas balance based on prediction model result dispatches system and method | |
| CN110533082B (en) | Sintering mixed water adding control method based on dual-model collaborative prediction | |
| CN115982178A (en) | Intelligent formula batching method and system for autoclaved aerated concrete product | |
| CN109342279B (en) | Mixed soft measurement method based on grinding mechanism and neural network | |
| CN104680012A (en) | Calculating model for sintering and burdening | |
| CN107368125A (en) | A kind of blast furnace temperature control system and method based on CBR Yu the parallel mixed inferences of RBR | |
| CN114091871A (en) | Blast furnace iron-making ore blending method and system | |
| CN103454395A (en) | Coke quality prediction method in coal silo mode | |
| CN114265885A (en) | Automatic control method and system for moisture of sintering return powder | |
| CN104913639A (en) | Data integration based sintering end-point control system and control method | |
| CN114216349B (en) | Sintering end point forecasting method based on coding and decoding network | |
| CN102054125B (en) | Method for stabilizing chemical constituents of charging agglomerate | |
| CN108800967A (en) | Ring cold machine temprature control method based on neural network and system | |
| CN113962150A (en) | Method and system for predicting granularity of sintering mixture | |
| CN113299352B (en) | Material layer thickness dynamic prediction system and method of sintering trolley | |
| CN114842918B (en) | Automatic water adding method for sintering mixture based on machine learning |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210618 |
|
| RJ01 | Rejection of invention patent application after publication |