CN106681146A - Blast furnace multi-target optimization control algorithm based on BP neural network and genetic algorithm - Google Patents

Blast furnace multi-target optimization control algorithm based on BP neural network and genetic algorithm Download PDF

Info

Publication number
CN106681146A
CN106681146A CN201611269284.XA CN201611269284A CN106681146A CN 106681146 A CN106681146 A CN 106681146A CN 201611269284 A CN201611269284 A CN 201611269284A CN 106681146 A CN106681146 A CN 106681146A
Authority
CN
China
Prior art keywords
output
neural network
input
layer
value
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
Application number
CN201611269284.XA
Other languages
Chinese (zh)
Inventor
周恒�
杨春节
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN201611269284.XA priority Critical patent/CN106681146A/en
Publication of CN106681146A publication Critical patent/CN106681146A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Manufacture Of Iron (AREA)

Abstract

The invention discloses a blast furnace multi-target optimization control algorithm based on a BP neural network and a genetic algorithm, and belongs to the field of industrial process monitoring, modeling and simulation. The algorithm comprises the following steps: selecting input and output variables of a model, setting seven state variables such as air blast kinetic energy, hot air pressure, hot air temperature, cold air flow, total column pressure difference, oxygen enrichment rate and coal ration as input variables, and setting three target parameters such as molten iron sulfur content, carbon dioxide discharge capacity and coke ratio as output variables; establishing a BP neural network, and training the neural network after initialization; constructing an NSGA-II multi-target optimization algorithm, and taking the prediction output of the trained BP neural network as a fitness function of an NSGA-II; and utilizing the algorithm to optimize the target variables, and finding Pareto optimality of targets and values of corresponding control variables. Therefore, the operating efficiency of a blast furnace can be improved.

Description

Based on BP neural network and the blast furnace multiobjective optimal control algorithm of genetic algorithm
Technical field
The invention belongs to industrial process monitoring, modeling and simulation field, more particularly to a kind of to be based on BP neural network and something lost The blast furnace multiobjective optimal control algorithm of propagation algorithm.
Technical background
Because complicated mass-and heat-transfer, chemical reaction and phase change in blast furnace, it is special that ironmaking processes are presented non-linear and dynamic Property.In view of the extreme condition in blast furnace, it is difficult to blast furnace held stationary safe operation is controlled, therefore the consumption of unnecessary crude fuel is May be appreciated.But, blast furnace ironmaking account for percentage 40 of whole Steel Production Flow Chart energy consumption, and the energy consumption of steel industry Global 8 percent are account for, even the small lifting of the high efficiency of furnace can also save the very big energy, huge is made to environmental protection Big contribution.
Coke ratio, cost, the core of molten iron quality is affected in blast furnace to be " transports ", and it is mass transfer, heat transfer, dynamic Amount transmission and chemical reaction are affected by factors such as such as hot blast temperature, the grade of ore, cooling conditions again, and these influence factors are still Height correlation.Researcher constructs a large amount of models to find out the potential contact between variable, it is desirable to which it is suitable to select and optimize Parameter can make the stable operation of blast furnace safety.The genetic algorithm and adaptive weighted method having been proposed that has respective advantage, But show not fully up to expectations under multivariate and high-dimensional complex situations.
In view of the complexity and closure of blast furnace, operator carry out operating blast furnaces generally according to conventional experience.Ensure high The stable smooth operation of stove is top priority, so people can stay very big surplus capacity, causes the extra consumption of Ore and fuel.In recent years Come, researchers model to find the optimum point of blast furnace operating all the time by mechanism or data-driven.Conventional single goal analysis Method is difficult to solve the problems, such as multiple-objection optimization, because the various variables in blast furnace are height being mutually related, it is difficult to individually divide Analyse the Changing Pattern of certain element.
The content of the invention
It is difficult to solve the problems, such as multiple-objection optimization for conventional single goal analysis method, and universal model is not suitable for height Dimension, multivariable situation.In order to optimize conflicting target, such as minimum molten steel sulfur content, CO simultaneously2Discharge Amount and coke ratio, we have proposed a kind of based on BP neural network and the blast furnace multiobjective optimal control algorithm of genetic algorithm.It is this Algorithm utilizes a kind of mapping relations of input and output of neural network, is then found out under multi-target condition with genetic algorithm The value of the optimum and corresponding input variables of Pareto.
It is a kind of based on BP neural network and the blast furnace multiobjective optimal control algorithm of genetic algorithm, step is as follows：
Step one：The input/output variable of Selection Model, the specific requirement and the practical situation of production process according to factory It is determined that the object function for needing optimization and the control variable being operable to；
Step 2：Set up, initialize, train BP neural network, the mapping ruler of input and output is made it have by study；
Step 3：NSGA-II multi-objective optimization algorithms are set up, after initialization of population, by selection, variation, crossover operation Calculate fitness function；
Step 4：Using the output of BP neural network as the fitness function of NSGA-II genetic algorithms, Pareto is found most The value of excellent and corresponding input variable.
The choosing method of the parameter of the model described in step one is as follows：
Select seven variables：Blast energy, hot-blast pressure, hot blast temperature, cold flow, full tower pressure reduction, oxygen enrichment percentage and coal Than as the |input paramete of model；
Select coke ratio, sulfur content and CO2Discharge capacity is optimization aim.
Foundation, initialization, training BP neural network step described in step 2 is as follows：
1. neutral net is initialized：According to the variable of input and output, need to pre-set input layer, hidden layer and output layer Nodes, learning rate and activation primitive, in addition it is also necessary to set the threshold value in the weights and layer of interlayer；
2. hidden layer is calculated：Hidden layer output H is calculated according to below equation, wherein f is activation primitive,
ω in formulaijIt is the connection weight of input layer and hidden layer, and a is the threshold value of hidden layer；
3. output layer is calculated：Output H, weights ω according to BP neural networkjkPredictive value O is calculated with threshold value bk,
4. Error Calculation：According to actual value and the predictor calculation error of neutral net,
ek=Yk-Ok, k=1 ... m (1.3)
5. right value update：Update the weights ω of input layer and hidden layerij, hidden layer and output layer weights ωjk,
ωjk=ωjk+ηHjek, (j=1 ... l；K=1 ... m)
6. threshold value updates：Threshold value a, threshold value b of output layer of hidden layer are updated,
bk=bk+ek, k=1 ... m
The specific algorithm step of the NSGA-II genetic algorithms described in step 3 is as follows：
1. initialize：In genetic algorithm, different data types can represent that one group of data is just with binary number It is item chromosome, and the solution space for including chromosome is referred to as population, the size of population is given at the very start；
2. select：Chromosome can relay by the way that selection is generation upon generation of in nature, by quick non-bad in NSGA-II algorithms Sort method and crowding distance operator select solution dominant in data and constitute new disaggregation, carry out the calculating of next step；
3. operate：Outstanding subspace is filtered out by selection operation, is also needed further through the operation such as intersection and variation Calculate；Wherein intersection is the core procedure in operation, and by crossover operation the key property of former generation can be at utmost inherited, in Between population y=(y1,...yn) in solution ykX of the DE crossing operations from former generation can be passed through1,x2,x3Obtain,
F and CR are control parameters in formula；
Additionally, new populationObtained by index mutation operator,
Rand represents the random number between 0 to 1 in formula, and β represents profile exponent, pmIt is aberration rate, ak、bkIt is optimization The lower limit of variate-value and reach the standard grade.
In described step four：
BP neural network study obtains the mapping relations of input and output, exports the fitness as NSGA-II genetic algorithms Value, the Pareto that genetic algorithm finds out target is optimum, and the value of corresponding input variable, contains so as to obtain target variable molten iron sulfur The minima of amount, CO2 emissions and coke ratio, corresponding input variable blast energy, hot blast also under Target Min The numerical value of pressure, hot blast temperature, cold flow, full tower pressure reduction, oxygen enrichment percentage and coal ratio.
The present invention has the advantage that：
For environment and multivariate severe, extreme in blast furnace, high-dimensional data, it is proposed that one kind is based on BP nerve net The blast furnace multiobjective optimal control algorithm of network and genetic algorithm.This algorithm can simultaneously optimize various interrelated or contradiction mesh Mark, by BP neural network the corresponding relation of input/output variable is learnt, and then finds target ginseng by NSGA-II genetic algorithms The value of the optimum solution spaces of several Pareto and corresponding input variable, reaches the purpose of optimal control, lifts operation of blast furnace Efficiency.
Description of the drawings
Fig. 1 is BP neural network structural representation,
Fig. 2 is genetic algorithm flow chart,
Fig. 3 is blast furnace Optimized model,
Fig. 4 is blast furnace historical data.
Specific implementation method
The present invention proposes a kind of based on BP neural network and the blast furnace multiobjective optimal control algorithm of genetic algorithm, step It is as follows：
Step one：The input/output variable of Selection Model, the specific requirement and the practical situation of production process according to factory It is determined that the object function for needing optimization and the control variable being operable to.
Step 2：Set up, initialize, train BP neural network, the mapping ruler of input and output is made it have by study.
Step 3：NSGA-II multi-objective optimization algorithms are set up, after initialization of population, by selection, variation, crossover operation Calculate fitness function.
Step 4：Using the output of BP neural network as the fitness function of NSGA-II genetic algorithms, Pareto is found most The value of excellent and corresponding input variable.
The choosing method of the model parameter described in step one is as follows：
Recognize from field engineer, operation of blast furnace is affected maximum state variable be blast energy, hot-blast pressure, Hot blast temperature, cold flow, full tower pressure reduction, oxygen enrichment percentage and coal ratio, therefore we select this 7 variables to join as the input of model Number.
Coke ratio is one of most important technical-economic index of blast furnace, and it represents the coke that the production pig iron per ton is consumed Amount, relatively low coke ratio not only represents the stable smooth of operation of blast furnace, is also the guarantee that interests are earned by enterprise.In iron and steel production, Sulfur is a kind of harmful element, and it can cause steel to produce " red brittleness ", therefore low-level sulfur content is quality product in molten iron Symbol.Additionally, ironmaking is the industry of a highly energy-consuming, and the energy expenditure of blast furnace account for 7 the percent of whole production procedure Ten, it is also CO2 emission rich and influential family that this represents blast furnace.Go from bad to worse in environment, greenhouse effect ever-increasing today, energy-conservation Reduction of discharging is particularly important, and a big target of environmental conservation is exactly to reduce CO2 emissions.For the above-mentioned reasons, we will Coke ratio, sulfur content and CO2Discharge capacity is set as optimization aim.
The structure of the BP neural network described in step 2 is as follows：
As shown in figure 1, BP neural network is a kind of reverse transmittance nerve network of multilayer feedforward, it is by input layer, hidden layer With output layer constitute, signal wherein back-propagation and error is propagated forward.During study, if output layer can not be obtained To the value for needing, then force input signal back propagation to update weights and threshold value, make output signal move closer to setting Target.Giving a forecast using BP neural network should first complete following structure, initialization and training step：
1. neutral net is initialized：According to the variable of input and output, need to pre-set input layer, hidden layer and output layer Nodes, learning rate and activation primitive.In addition it is also necessary to set the threshold value in the weights and layer of interlayer.
2. hidden layer is calculated：Hidden layer output H is calculated according to below equation, wherein f is activation primitive.
ω in formulaijIt is the connection weight of input layer and hidden layer, and a is the threshold value of hidden layer.
3. output layer is calculated：Output H, weights ω according to BP neural networkjkPredictive value O is calculated with threshold value bk。
4. Error Calculation：According to actual value and the predictor calculation error of neutral net.
ek=Yk-Ok, k=1 ... m (1.3)
5. right value update：Update the weights ω of input layer and hidden layerij, hidden layer and output layer weights ωjk。
ωjk=ωjk+ηHjek, (j=1 ... l；K=1 ... m)
6. threshold value updates：Update threshold value a, threshold value b of output layer of hidden layer.
bk=bk+ek, k=1 ... m
The structure of the NSGA-II genetic algorithms described in step 3 is as follows：
As shown in Fig. 2 NSGA-II is a kind of evolution algorithm of similar " natural selection ", environment is most adapted in data set Solution can survive in the evolutionary process such as selection, variation, intersection, and specific algorithm steps are as follows：
1. initialize：In genetic algorithm, different data types can represent that one group of data is just with binary number It is item chromosome, and the solution space for including chromosome is referred to as population, the size of population is given at the very start.
2. select：Chromosome can relay by the way that selection is generation upon generation of in nature, by quick non-bad in NSGA-II algorithms Sort method and crowding distance operator select solution dominant in data and constitute new disaggregation, carry out the calculating of next step.
3. operate：Outstanding subspace is filtered out by selection operation, is also needed further through the operation such as intersection and variation Calculate.Wherein intersection is the core procedure in operation, and by crossover operation the key property of former generation can be at utmost inherited.Example Such as, middle population y=(y1,...yn) in solution ykX of the DE crossing operations from former generation can be passed through1,x2,x3Obtain.
F and CR are control parameters in formula.
Additionally, new populationCan be obtained by index mutation operator.
Rand represents the random number between 0 to 1 in formula, and β represents profile exponent, pmIt is aberration rate, ak、bkIt is optimization The lower limit of variate-value and reach the standard grade.
Only it is provided with suitable crossover probability, variation rate and Population Size, NSGA-II genetic algorithms competence exertion reason The function of thinking.
Step described in step 4 is explained in detail below：
BP neural network study obtains the mapping relations of input and output, adaptation of its output as NSGA-II genetic algorithms Angle value, the Pareto that genetic algorithm finds out target is optimum, and the value of corresponding input variable.So as to obtain target variable molten iron sulfur The minima of content, CO2 emissions and coke ratio, corresponding input variable blast energy, heat also under Target Min The numerical value of wind pressure, hot blast temperature, cold flow, full tower pressure reduction, oxygen enrichment percentage and coal ratio.
Embodiment
In Steel Production Flow Chart, ironmaking is a most important link, not only decides that whole flow process can be transported smoothly OK, also it is related to the productivity effect of whole enterprise.In order to optimize the operation of blast furnace, the purpose of multiple-objection optimization is reached.I Have chosen blast energy, hot-blast pressure, hot blast temperature, cold flow, full tower pressure reduction, oxygen enrichment percentage and coal compare 7 operation ginseng Count the input as model, coke ratio, sulfur content and CO2This 3 indexs of discharge capacity as model output (Fig. 3 be blast furnace optimization Model), neutral net (Fig. 4 is blast furnace historical data) is trained by the 300 of Liu Gang groups of historical datas, allow BP neural network to learn The mapping relations of input and output, then allow NSGA-II genetic algorithms to find the minima of output variable, and corresponding input The value of variable, reaches the purpose of optimization operating parameter.
By above modeling and simulating process, we have obtained the value of following optimization：
Variable Unit Numerical value
Coke ratio kg/t 240.92
CO2Discharge capacity t/d 9263.36
Molten steel sulfur content Wt% 0.01344
Coal ratio kg/t 194.19
Oxygen enrichment percentage Vol% 2.816
Full tower pressure reduction kPa 177.46
Cold flow 104m3/h 28.03
Hot blast temperature ℃ 1202.81
Hot-blast pressure MPa 0.3958
Blast energy kW 105.99
The present invention has advantages below：
1st, blast furnace multiobjective optimal control is solved the problems, such as, it is proposed that one kind determines control by optimization aim index The method of variable.
2nd, model calculates that coke ratio is minimum to can reach 240.92kg/t, and this is from not up in actual production, to reality Production operation has directive significance.
3rd, model is simple, it is easy to understand.
Above-described embodiment is used for illustrating the present invention, rather than limits the invention, the present invention spirit and In scope of the claims, any modifications and changes made to the present invention belong to protection scope of the present invention.

Claims (5)

1. it is a kind of based on BP neural network and the blast furnace multiobjective optimal control algorithm of genetic algorithm, it is characterised in that step is such as Under：
Step one：The input/output variable of Selection Model, determines according to the specific requirement of factory and the practical situation of production process The object function for needing optimization and the control variable being operable to；
Step 2：Set up, initialize, train BP neural network, the mapping ruler of input and output is made it have by study；
Step 3：NSGA-II multi-objective optimization algorithms are set up, after initialization of population, by selecting, making a variation, crossover operation is calculated Fitness function；
Step 4：Using the output of BP neural network as NSGA-II genetic algorithms fitness function, find Pareto it is optimum and The value of corresponding input variable.
2. method according to claim 1, it is characterised in that the choosing method of the parameter of the model described in step one is such as Under：
Select seven variables：Blast energy, hot-blast pressure, hot blast temperature, cold flow, full tower pressure reduction, oxygen enrichment percentage and coal ratio, make For the |input paramete of model；
Select coke ratio, sulfur content and CO2Discharge capacity is optimization aim.
3. method according to claim 1, it is characterised in that the foundation, initialization, training BP nerve net described in step 2 Network step is as follows：
1. neutral net is initialized：According to the variable of input and output, the node for pre-setting input layer, hidden layer and output layer is needed Number, learning rate and activation primitive, in addition it is also necessary to set the threshold value in the weights and layer of interlayer；
2. hidden layer is calculated：Hidden layer output H is calculated according to below equation, wherein f is activation primitive,
H j = f ( Σ i = 1 n ω i j x i - a j ) , j = 1 , ... l f ( x ) = 2 1 + e - 2 x - 1 - - - ( 1.1 )
ω in formulaijIt is the connection weight of input layer and hidden layer, and a is the threshold value of hidden layer；
3. output layer is calculated：Output H, weights ω according to BP neural networkjkPredictive value O is calculated with threshold value bk,
O k = Σ j = 1 l H j ω j k - b k , k = 1 , ... m - - - ( 1.2 )
4. Error Calculation：According to actual value and the predictor calculation error of neutral net,
ek=Yk-Ok, k=1 ... m (1.3)
5. right value update：Update the weights ω of input layer and hidden layerij, hidden layer and output layer weights ωjk,
ω i j = ω i j + ηH j ( 1 - H j ) x ( i ) Σ k = 1 m ω j k e k , ( i = 1 , ... , n ; j = 1 , ... l ) ω j k = ω j k + ηH j e k , ( j = 1 , ... l ; k = 1 , ... m ) - - - ( 1.4 )
6. threshold value updates：Threshold value a, threshold value b of output layer of hidden layer are updated,
a j = a j + ηH j ( 1 - H j ) Σ k = 1 m ω j k e k , j = 1 , ... l b k = b k + e k , k = 1 , ... m - - - ( 1.5 ) .
4. method according to claim 1, it is characterised in that the concrete calculation of the NSGA-II genetic algorithms described in step 3 Method step is as follows：
1. initialize：In genetic algorithm, different data types can represent that one group of data is exactly one with binary number Bar chromosome, and the solution space for including chromosome is referred to as population, the size of population is given at the very start；
2. select：Chromosome can relay by the way that selection is generation upon generation of in nature, by quick non-bad sequence in NSGA-II algorithms Method and crowding distance operator select solution dominant in data and constitute new disaggregation, carry out the calculating of next step；
3. operate：Outstanding subspace is filtered out by selection operation, also needs further to be calculated through the operation such as intersection and variation； Wherein intersection is the core procedure in operation, and by crossover operation the key property of former generation, among can be at utmost inherited Group y=(y1,...yn) in solution ykX of the DE crossing operations from former generation can be passed through1,x2,x3Obtain,
y k = x 1 + F ( x 2 - x 3 ) , p c < C R x 1 , p c > C R - - - ( 2 , 1 )
F and CR are control parameters in formula；
Additionally, new populationObtained by index mutation operator,
y k - = y k + &sigma; k ( b k - a k ) , p m < a y k , a < p m < 1 - - - ( 2 , 2 )
&sigma; k = ( 2 &times; r a n d ) 1 &beta; + 1 - 1 , r a n d < 0.5 1 - 2 ( 2 &times; r a n d ) 1 &beta; + 1 , r a n d > 0.5 - - - ( 2 , 3 )
Rand represents the random number between 0 to 1 in formula, and β represents profile exponent, pmIt is aberration rate, ak、bkIt is optimized variable The lower limit of value and reach the standard grade.
5. method according to claim 1, it is characterised in that in described step four：
BP neural network study obtains the mapping relations of input and output, exports the fitness value as NSGA-II genetic algorithms, loses The Pareto that propagation algorithm finds out target is optimum, and the value of corresponding input variable, so as to obtain target variable molten steel sulfur content, two Oxidation carbon emission amount and coke ratio minima, also under Target Min corresponding input variable blast energy, hot-blast pressure, The numerical value of hot blast temperature, cold flow, full tower pressure reduction, oxygen enrichment percentage and coal ratio.
CN201611269284.XA 2016-12-31 2016-12-31 Blast furnace multi-target optimization control algorithm based on BP neural network and genetic algorithm Pending CN106681146A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201611269284.XA CN106681146A (en) 2016-12-31 2016-12-31 Blast furnace multi-target optimization control algorithm based on BP neural network and genetic algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201611269284.XA CN106681146A (en) 2016-12-31 2016-12-31 Blast furnace multi-target optimization control algorithm based on BP neural network and genetic algorithm

Publications (1)

Publication Number Publication Date
CN106681146A true CN106681146A (en) 2017-05-17

Family

ID=58848925

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201611269284.XA Pending CN106681146A (en) 2016-12-31 2016-12-31 Blast furnace multi-target optimization control algorithm based on BP neural network and genetic algorithm

Country Status (1)

Country Link
CN (1) CN106681146A (en)

Cited By (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108845501A (en) * 2018-09-11 2018-11-20 东北大学 A kind of blast-melted quality adaptation optimal control method based on Lazy learning
CN109102109A (en) * 2018-07-16 2018-12-28 四川大学 A kind of piece of exploitation of stone explosion gradation prediction technique
CN109143870A (en) * 2018-10-23 2019-01-04 宁波溪棠信息科技有限公司 A kind of control method of multiple target task
CN109190846A (en) * 2018-10-08 2019-01-11 华东理工大学 A kind of new dry process rotary kiln furnace calcination process Multipurpose Optimal Method
CN109284818A (en) * 2018-09-07 2019-01-29 北方爆破科技有限公司 A kind of blasting vibration control prediction technique based on accident tree and genetic algorithm
CN109598092A (en) * 2018-12-28 2019-04-09 浙江工业大学 Merge the air source heat pump multi-objective optimization design of power method of BP neural network and more parent genetic algorithms
CN109829244A (en) * 2019-02-25 2019-05-31 浙江工业大学 The blower optimum design method of algorithm optimization depth network and three generations's genetic algorithm
CN109858093A (en) * 2018-12-28 2019-06-07 浙江工业大学 The air source heat pump multi-objective optimization design of power method of the non-dominated sorted genetic algorithm of SVR neural network aiding
CN109886448A (en) * 2019-03-14 2019-06-14 浙江工业大学 Using learning rate changing BP neural network and the heat pump multiobjective optimization control method of NSGA-II algorithm
CN109918749A (en) * 2019-02-25 2019-06-21 北京妙微科技有限公司 The fan design two generations algorithm Multipurpose Optimal Method of learning rate changing network modelling
CN109918702A (en) * 2019-01-03 2019-06-21 上海交通大学 A kind of blast-furnace burden cooperates with Multipurpose Optimal Method with operation
CN109932903A (en) * 2019-02-25 2019-06-25 北京妙微科技有限公司 The air-blower control Multipurpose Optimal Method of more parent optimization networks and genetic algorithm
CN110033118A (en) * 2019-02-25 2019-07-19 浙江工业大学 Elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm
CN110400007A (en) * 2019-07-05 2019-11-01 浙江大学 Based on the molten steel quality forecasting procedure for improving gating cycle neural network
CN110400009A (en) * 2019-07-09 2019-11-01 浙江大学 Blast furnace ironmaking Multiobjective Intelligent optimization method based on self-adapted genetic algorithm
US10604814B2 (en) 2017-09-27 2020-03-31 International Business Machines Coporation Manufacturing process control with deep learning-based predictive model for hot metal temperature of blast furnace
CN116837422A (en) * 2023-07-24 2023-10-03 扬中凯悦铜材有限公司 Production process and system of high-purity oxygen-free copper material

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3054521B2 (en) * 1993-10-14 2000-06-19 株式会社東芝 Induction motor control device
EP1363005B1 (en) * 2002-05-15 2009-07-15 Caterpillar Inc. Engine control system using a cascaded neural network
CN101634459A (en) * 2009-08-24 2010-01-27 陶晓鹏 Thermal power generation boiler intelligent combustion optimizing system and realizing method thereof
CN103488085A (en) * 2013-09-22 2014-01-01 上海交通大学 Multi-objective optimization control method of methyl alcohol four-tower rectification system
CN104573869A (en) * 2015-01-16 2015-04-29 河海大学常州校区 Optimization method and system for achieving dredging operations based on BP neural network and NSGA-II
CN104598686A (en) * 2015-01-24 2015-05-06 安徽大学 Water pump motor modeling and optimization method based on electromagnetic calculation and neural network
CN105760571A (en) * 2016-01-08 2016-07-13 浙江工业大学 Electric-heating variable mold temperature injection mold heating system design method
CN106202731A (en) * 2016-07-12 2016-12-07 南京理工大学 Bridge crane multi-flexibl e dynamics structural optimization method

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3054521B2 (en) * 1993-10-14 2000-06-19 株式会社東芝 Induction motor control device
EP1363005B1 (en) * 2002-05-15 2009-07-15 Caterpillar Inc. Engine control system using a cascaded neural network
CN101634459A (en) * 2009-08-24 2010-01-27 陶晓鹏 Thermal power generation boiler intelligent combustion optimizing system and realizing method thereof
CN103488085A (en) * 2013-09-22 2014-01-01 上海交通大学 Multi-objective optimization control method of methyl alcohol four-tower rectification system
CN104573869A (en) * 2015-01-16 2015-04-29 河海大学常州校区 Optimization method and system for achieving dredging operations based on BP neural network and NSGA-II
CN104598686A (en) * 2015-01-24 2015-05-06 安徽大学 Water pump motor modeling and optimization method based on electromagnetic calculation and neural network
CN105760571A (en) * 2016-01-08 2016-07-13 浙江工业大学 Electric-heating variable mold temperature injection mold heating system design method
CN106202731A (en) * 2016-07-12 2016-12-07 南京理工大学 Bridge crane multi-flexibl e dynamics structural optimization method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
华长春等: "基于NSGA-Ⅱ算法的高炉生产配料多目标优化模型建立", 《化工学报》 *
王小川: "《MATLAB神经网络43个案例分析》", 30 June 2013, 北京航空航天大学出版社 *

Cited By (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10604814B2 (en) 2017-09-27 2020-03-31 International Business Machines Coporation Manufacturing process control with deep learning-based predictive model for hot metal temperature of blast furnace
US11242575B2 (en) 2017-09-27 2022-02-08 International Business Machines Corporation Manufacturing process control with deep learning-based predictive model for hot metal temperature of blast furnace
US10633716B2 (en) 2017-09-27 2020-04-28 International Business Machines Corporation Manufacturing process control with deep learning-based predictive model for hot metal temperature of blast furnace
CN109102109A (en) * 2018-07-16 2018-12-28 四川大学 A kind of piece of exploitation of stone explosion gradation prediction technique
CN109284818A (en) * 2018-09-07 2019-01-29 北方爆破科技有限公司 A kind of blasting vibration control prediction technique based on accident tree and genetic algorithm
CN108845501A (en) * 2018-09-11 2018-11-20 东北大学 A kind of blast-melted quality adaptation optimal control method based on Lazy learning
CN108845501B (en) * 2018-09-11 2021-07-20 东北大学 Blast furnace molten iron quality self-adaptive optimization control method based on lazy learning
CN109190846A (en) * 2018-10-08 2019-01-11 华东理工大学 A kind of new dry process rotary kiln furnace calcination process Multipurpose Optimal Method
CN109143870A (en) * 2018-10-23 2019-01-04 宁波溪棠信息科技有限公司 A kind of control method of multiple target task
CN109143870B (en) * 2018-10-23 2021-08-06 宁波溪棠信息科技有限公司 Multi-target task control method
CN109598092A (en) * 2018-12-28 2019-04-09 浙江工业大学 Merge the air source heat pump multi-objective optimization design of power method of BP neural network and more parent genetic algorithms
CN109858093A (en) * 2018-12-28 2019-06-07 浙江工业大学 The air source heat pump multi-objective optimization design of power method of the non-dominated sorted genetic algorithm of SVR neural network aiding
CN109918702A (en) * 2019-01-03 2019-06-21 上海交通大学 A kind of blast-furnace burden cooperates with Multipurpose Optimal Method with operation
CN109932903A (en) * 2019-02-25 2019-06-25 北京妙微科技有限公司 The air-blower control Multipurpose Optimal Method of more parent optimization networks and genetic algorithm
CN110033118A (en) * 2019-02-25 2019-07-19 浙江工业大学 Elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm
CN109918749A (en) * 2019-02-25 2019-06-21 北京妙微科技有限公司 The fan design two generations algorithm Multipurpose Optimal Method of learning rate changing network modelling
CN109829244A (en) * 2019-02-25 2019-05-31 浙江工业大学 The blower optimum design method of algorithm optimization depth network and three generations's genetic algorithm
CN109886448A (en) * 2019-03-14 2019-06-14 浙江工业大学 Using learning rate changing BP neural network and the heat pump multiobjective optimization control method of NSGA-II algorithm
CN110400007A (en) * 2019-07-05 2019-11-01 浙江大学 Based on the molten steel quality forecasting procedure for improving gating cycle neural network
CN110400009A (en) * 2019-07-09 2019-11-01 浙江大学 Blast furnace ironmaking Multiobjective Intelligent optimization method based on self-adapted genetic algorithm
CN116837422A (en) * 2023-07-24 2023-10-03 扬中凯悦铜材有限公司 Production process and system of high-purity oxygen-free copper material
CN116837422B (en) * 2023-07-24 2024-01-26 扬中凯悦铜材有限公司 Production process and system of high-purity oxygen-free copper material

Similar Documents

Publication Publication Date Title
CN106681146A (en) Blast furnace multi-target optimization control algorithm based on BP neural network and genetic algorithm
CN111651931B (en) Blast furnace fault diagnosis rule derivation method based on deep neural network
CN103064289A (en) Multiple-target operation optimizing and coordinating control method and device of garbage power generator
CN104630410A (en) Real-time dynamic converter steelmaking quality prediction method based on data analysis
CN102540879A (en) Multi-target evaluation optimization method based on group decision making retrieval strategy
CN109447346A (en) Based on gray prediction and neural network ensemble model converter oxygen consumption prediction technique
CN106845012A (en) A kind of blast furnace gas system model membership function based on multiple target Density Clustering determines method
CN106019940A (en) UKF (Unscented Kalman Filter) neural network-based converter steelmaking process cost control method and system
CN111311577A (en) Intelligent water seepage detection method based on generation of confrontation network and reinforcement learning
CN106096724B (en) Converter steelmaking process cost control method and system based on ELM neural network
Yan et al. Multi-objective evaluation-based hybrid intelligent control optimization for shaft furnace roasting process
CN116307149A (en) Blast furnace performance optimization method based on attention LSTM and KBNSGA
CN105955014A (en) Method for controlling coke furnace chamber pressure based on distributed dynamic matrix control optimization
CN106444388A (en) Distributed PID type dynamic matrix control method for furnace pressure of coke furnace
CN109359320B (en) Blast furnace index prediction method based on multiple sampling rate autoregressive distribution hysteresis model
CN107092188A (en) A kind of GPC algorithm of CSTR system
Wang et al. Whale optimization end-point control model for 260 tons BOF steelmaking
Zagoskina et al. Intelligent Control System of Blast-Furnace Melting Efficiency
CN108287474A (en) Based on the probabilistic catalytic reforming reactor robust operation optimization method of raw material
Yang et al. A multi-objective optimization model based on long short-term memory and non-dominated sorting genetic algorithm II
Zhou et al. Multi-objective optimization of operating parameters based on neural network and genetic algorithm in the blast furnace
CN111897217B (en) Time domain decomposition acceleration method of model prediction controller
CN106444362A (en) Distributed PID (Proportion Integration Differentiation) predictive function control method for furnace box temperature of waste plastic cracking furnace
CN103645633B (en) Furnace temperature self-learning control method of conversion furnace system
CN113900379A (en) Neural network-based predictive control algorithm

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: 20170517

RJ01 Rejection of invention patent application after publication