CN107245540A - A kind of control strategy of blast furnace material distribution process radial direction thickness of feed layer distribution - Google Patents
A kind of control strategy of blast furnace material distribution process radial direction thickness of feed layer distribution Download PDFInfo
- Publication number
- CN107245540A CN107245540A CN201710447597.8A CN201710447597A CN107245540A CN 107245540 A CN107245540 A CN 107245540A CN 201710447597 A CN201710447597 A CN 201710447597A CN 107245540 A CN107245540 A CN 107245540A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- distribution
- blast furnace
- msubsup
- 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.)
- Granted
Links
Classifications
-
- C—CHEMISTRY; METALLURGY
- C21—METALLURGY OF IRON
- C21B—MANUFACTURE OF IRON OR STEEL
- C21B5/00—Making pig-iron in the blast furnace
- C21B5/006—Automatically controlling the process
-
- 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
-
- C—CHEMISTRY; METALLURGY
- C21—METALLURGY OF IRON
- C21B—MANUFACTURE OF IRON OR STEEL
- C21B2300/00—Process aspects
- C21B2300/04—Modeling of the process, e.g. for control purposes; CII
Landscapes
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Metallurgy (AREA)
- Organic Chemistry (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Manufacturing & Machinery (AREA)
- Materials Engineering (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Manufacture Of Iron (AREA)
- Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
Abstract
The present invention gives a kind of control strategy of the blast furnace material distribution process radial direction thickness of feed layer distribution based on iterative learning and multi-model.The present invention is distributed as by control targe with blast furnace material distribution process radial direction thickness of feed layer, using burden distribution matrix as performance variable, constructs a kind of control strategy being distributed towards blast furnace material distribution process radial direction thickness of feed layer.Write the distribution of bed of material radial thickness as weights and basic function form first, the expression method of thickness of feed layer target distribution and controlled distribution is given on the basis of basic function model formulation, a kind of control strategy of blast furnace material distribution process radial direction thickness of feed layer distribution is given on the basis of iterative learning and multi-model, multi-model is responsible for integer field cloth number of turns vectorκSelection and optimization, and iterative learning is mainly used in real number field chute dip vectorαControl.The present invention provides theoretical foundation for the adjustment and setting of burden distribution matrix in burden distribution system, process for promoting blast furnace industrial process automation, improve the blast furnace operating smelted towards high-performance, and realize that energy-conservation, emission reduction and the performance indications optimization tool of blast furnace ironmaking process are of great significance to a greater extent.
Description
Technical field
The present invention relates to a kind of control of the blast furnace material distribution process radial direction thickness of feed layer distribution based on iterative learning and multi-model
System strategy, is related to the modeling and control of distributed system, is related to metallurgy, computer science, mathematics, control science and divides at random
The intersection of the technical fields such as cloth control and fusion.
Background technology
Blast furnace material distribution is an important operation system in blast furnace operating, is blast furnace stable smooth operation, blast furnace stable yields, reduction accident
Rate and the key link for reducing fuel consumption.Practice for many years and experience have shown that, the bed of material that blast furnace material distribution process is formed
Original depth distribution not only influences the distribution in initial shape of charge level and temperature field, while being also blast furnace stable yields, blast furnace is stablized suitable
OK, blast furnace accident rate and blast furnace fuel consumption key link (Liu Yuncai, blast furnace material distribution rule [M], metallurgical industry publishing house,
2012).The valid model of shape, the system of blast furnace material distribution system are exported due to lacking the three-dimensional charge level of complex three-dimensional inside description blast furnace
Fixed and adjustment is still performed by veteran section chief, and more negative effect is brought to blast furnace steady production.
Chinese patent 201410336893.7 provides a kind of control method of blast furnace material distribution process radial direction ore coke ratio, sets up
Relationship model of the blast furnace material distribution control parameter to charge level, for description blast furnace material distribution model, blanking process model has certain
Positive role, due to lacking effective description of burden distribution matrix and thickness of feed layer output distribution, can not be realized to thickness of feed layer point
The control of cloth.Patent application document 201510586609.6 is provided between a kind of description charge level output shape and performance variable
Relationship model, but do not provide the control program of thickness of feed layer distribution.For existing blast furnace material distribution system, the present invention is proposed
It is a kind of that the blast furnace distribution control system of controlled parameter is distributed as with bed of material output thickness, and give and a kind of be based on iterative learning
The control strategy being distributed with the blast furnace material distribution process thickness of feed layer of multi-model.The present invention is for promoting blast furnace industrial process automation
Process, improve the blast furnace operating smelted towards high-performance, and realize the energy-conservation of blast furnace ironmaking process, emission reduction to a greater extent
And performance indications optimization tool is of great significance.The formulation and adjustment of existing blast furnace material distribution system are still by veteran
Section chief performs, and lacks effective theory support, more negative effect is brought to blast furnace steady production, this be in the industry generally existing and
Technical problem urgently to be resolved hurrily.
The content of the invention
Patent of the present invention is on the basis of patent application document 201510586609.6, it is proposed that a class is with blast furnace material distribution mistake
Journey radial direction thickness of feed layer is distributed as control targe, using burden distribution matrix as the distributed parameter control system of performance variable, and gives
A kind of control strategy of the blast furnace material distribution process radial direction thickness of feed layer distribution based on iterative learning and multi-model.
In order to solve the above-mentioned technical problem, the invention provides a kind of blast furnace material distribution mistake based on iterative learning and multi-model
The control strategy of journey radial direction thickness of feed layer distribution, it is characterised in that be distributed as being controlled with blast furnace material distribution process radial direction thickness of feed layer
Target processed, using burden distribution matrix as performance variable, builds the control strategy being distributed towards blast furnace material distribution process radial direction thickness of feed layer；Institute
Stating control strategy includes：
(1) distribution of controlled device bed of material radial thickness is write as weights and basic function form using separate variables；
(2) on the basis of the bed of material radial thickness distribution described by weights and basic function, weights are adjusted manually to set
The control targe of thickness of feed layer distribution；
(3) according to the control targe and the real-time radial thickness of the bed of material of the blast furnace material distribution process thickness of feed layer distribution set
Distribution define thickness of feed layer distributed controll performance indications criterion function, based on iterative learning and multi-model process to blast furnace material distribution
Process operation variable burden distribution matrix automatically adjusts.
As preferred technical scheme, in above-mentioned (1), controlled device bed of material radial thickness distribution h (y, u) meets cylindricality product
Point constraint, is a bivariate distribution function related to decision variable, performance variable burden distribution matrix include chute inclination angle sequence with
Two parts of rotating cycle sequence, wherein chute dip vector α belongs to real number field, and cloth number of turns vector κ belongs to nature number field,
It is a kind of special COMPLEX MIXED control system to cause blast furnace material distribution process radial direction thickness of feed layer distributed controll.
As preferred technical scheme, above-mentioned (1) is specially：
1) bed of material radial thickness is distributed h (y, u) under integral constraint and is write as weight wi(u) with basic function Bi(y) form,
And determine the number n+1 of basic function and weights：
Wherein, bed of material radial thickness distribution h (y, u) is the position y's and decision variable burden distribution matrix u apart from blast furnace center
Two-dimensional function, VtFor furnace charge cumulative volume, biFor basic function Bi(y) volume integral；Basic function Bi(y) it is B-spline function, n is 5-
Integer between 20；
2) relation between n dimension weight vector W (u) and the distribution of bed of material radial thickness is described using dimensionality reduction mode：
H (y, u)=C (y) W (u)+L (y),
W (u)=[w1(u),w2(u),…,wn(u)]T∈Rn×1,
Wherein, W (u) is the n dimension weight vector related to decision variable u, and C (y) is n Wikis function composition after dimension-reduction treatment
Matrix, L (y) be dimension-reduction treatment after bound variable.
As preferred technical scheme, in above-mentioned (2), controlled variable bed of material radial thickness target distribution function g (y) is by base
Function and weights are determined, on the basis of the bed of material radial thickness distribution described by the weights and basic function described in claim 3,
Weights W is adjusted manuallygTo set the control targe of thickness of feed layer distribution, it is specially：G (y)=C (y) Wg+L(y)。
As preferred technical scheme, in above-mentioned (3), the control being distributed according to the blast furnace material distribution process thickness of feed layer set
Thickness of feed layer distributed controll performance indications criterion function is determined in the distribution of target processed and the real-time radial thickness of the bed of material, and provides
A kind of blast furnace material distribution process based on iterative learning and multi-model process automatically adjusts the control strategy of performance variable burden distribution matrix,
Specially：
1) constraint followed according to cloth number of turns sequence κThe maximum number of rings m of cloth, and blast furnace material distribution
Multicenter few cloth principle in process edge determines M alternative κjFinite aggregate K={ the κ of composition1,κ2,…κM}；
2) according to target distribution g (y), cloth number of turns vector κ in limited countably infinite set is definedjCriterion function：
Wherein
3) according to cloth number of turns vector κ in limited countably infinite setjCorresponding criterion function, is slipped with the method for Gradient Iteration
Groove tilt angle vector α control law：
Wherein (k) represents the number of times of iterative learning；
4) setting maximum iteration and stopping criterion for iteration, according to the integer field cloth number of turns vector κ's having determined
Limited countably infinite set calculates cloth number of turns vector κ successivelyjCriterion function, from M limited countably infinite set K={ κ1,κ2,…κMMiddle choosing
Select the minimum value min of the criterion function corresponding to vectorial κAnd provided accordingly according to the performance indications of minimum
Decision variable α and κ.
More specifically, present invention also offers a kind of blast furnace material distribution process radial direction bed of material based on iterative learning and multi-model
The control strategy of thickness distribution, is comprised the following steps that：
Step 1：Obtain blast furnace material distribution process blast-furnace body parameter, including furnace throat radius, stockline height, chute length, larynx
Pipe height, chute fascinate away from, chute coefficient of friction, furnace charge angle of rest (repose), furnace charge heap density, charge batch weight, and provide performance variable cloth
Expect the vector description of matrix：
α=[α1,…,αm]T∈Rm×1,αi∈[αmin,αmax], (1)
U=[α, κ], (3)
Wherein αminAnd αmaxThe border of chute tilt adjustable section is represented, m represents maximum cloth number of rings, chute dip vector α
Belong to real number field, and rotating cycle vector κ belongs to nature number field.
Step 2：Obtain blast furnace material distribution process stockline radial distribution γ (y), i.e. cloth process bottom distribution shape, wherein y
Represent the distance apart from blast furnace center.
Step 3：Furnace charge volume V is calculated according to charge batch weight and furnace charge heap densityt, and assume furnace charge heap density constant, root
According to conservation of mass principle, volume and furnace charge of the furnace charge in feed bin are equal in the volume that blast furnace throat punishes cloth, and then we carry
Go out the isometric principle of blast furnace material distribution process furnace charge：
Wherein, f (y, u) represents the radial top profile that burden distribution is formed on the basis of stockline γ (y), and u is represented
Burden distribution matrix, constitutes κ by chute dip vector α and rotating cycle vector and constitutes.
Step 4：Thickness of feed layer distribution is calculated according to the distribution shape at blast furnace material distribution process radial direction bottom and top:
H (y, u)=f (y, u)-γ (y). (5)
Step 5：According to the isometric principle of blast furnace material distribution process furnace charge and the separation of variable, bed of material radial thickness is distributed h
(y, u) is write as weight wi(u) with basic function Bi(y) form, specific implementation includes following steps：
Step 5-1：Determine the number n+1 of basic function and weights：
This patent basic function Bi(y), select as B-spline basic function, n is the integer between 5-10.
Step 5-2：Determine basic function Bi(y) volume integral：
B=[b1,b2,…,bn]T∈Rn×1。 (9)
Step 5-3：Determine weights W (u) vector description：
W (u)=[w1(u),w2(u),…,wn(u)]T∈Rn×1, (10)
Wherein vector b and W (u), dimension be n.
Step 5-4：Relation between weight vector W (u) and the distribution of dynamic radial thickness of feed layer is described using dimensionality reduction mode：
H (y, u)=C (y) W (u)+L (y), (13)
Step 6：According to the description of the isometric principle of blast furnace material distribution process furnace charge, and above-mentioned weights and basic function, set
The target g (y) of thickness of feed layer distribution：
G (y)=C (y) Wg+ L (y), (15)
Step 7：According to blast furnace material distribution expertise, integer field cloth number of turns vector κ limited countably infinite set is determined：K=
{κ1,κ2,…κM, wherein M represents cardinality of a set.
Step 8：The target g (y) being distributed according to the thickness of feed layer of setting, defines cloth number of turns vector κ in limited countably infinite setj
Criterion function：
Wherein
Step 9：According to cloth number of turns vector κ in limited countably infinite setjCorresponding criterion function, is given with the method for Gradient Iteration
Slip groove tilt angle vector α control law：
Wherein (k) represents the number of times of iterative learning.
Step 10：Set maximum iteration and stopping criterion for iteration, according to the integer field cloth number of turns having determined to
Amount κ limited countably infinite set calculates cloth number of turns vector κ successivelyjCriterion function, from M limited countably infinite set K={ κ1,κ2,…κM}
The minimum value min of criterion function corresponding to middle selection vector κAnd provide corresponding decision variable α and κ.
The present invention achieves significant technique effect, is embodied in：1) class is given with blast furnace material distribution process radial direction
Thickness of feed layer is distributed as control targe, using burden distribution matrix as the distributed parameter control system of performance variable, and gives a kind of base
The control strategy being distributed in the blast furnace material distribution process radial direction thickness of feed layer of iterative learning and multi-model.2) blast furnace material distribution process, by
Adjustable parameter in performance variable burden distribution matrix belongs to different number fields, causes blast furnace material distribution process radial direction thickness of feed layer to be distributed
Control as a kind of special COMPLEX MIXED control system.This patent is distributed as being controlled with blast furnace material distribution process radial direction thickness of feed layer
Target, using burden distribution matrix as performance variable, constructs a kind of control plan being distributed towards blast furnace material distribution process radial direction thickness of feed layer
Slightly.3) present invention for burden distribution matrix in burden distribution system adjustment and set and provide theoretical foundation, and practical operation aspect
Method, gives the concrete measure of regulation thickness of feed layer distribution manually, is conducive to the reality for promoting blast furnace material distribution process control
It is existing, while the thought of this patent distributed constant control can also be used for solving the control problem of other complex objects.
Brief description of the drawings
Fig. 1 is the radial direction schematic diagram of blast furnace material distribution process：
Brief description of the drawings：VtRepresent the volume of furnace charge charge；α is chute inclination angle；ω is angular velocity of rotation；F (y, u) represents material
Distribution shape at the top of face, y represents the distance apart from stove center, and r represents furnace throat radius；γ (y) represents charge level bottom distribution shape
(also referred to as stockline distribution)；H (y, u) represents thickness distribution；
Fig. 2 is blast furnace material distribution process radial direction thickness of feed layer distribution control system structure chart；
Fig. 3 is the blast furnace material distribution process radial direction thickness of feed layer distributed controll flow chart based on iterative learning and multi-model；
Fig. 4 is the design sketch of the blast furnace material distribution process radial direction thickness of feed layer distributed controll based on iterative learning and multi-model；
Embodiment
Technical scheme is further described with specific implementation below in conjunction with the accompanying drawings.
A kind of control strategy of the blast furnace material distribution process radial direction thickness of feed layer distribution based on iterative learning and multi-model, specifically
Step is as follows：
Step 1：Obtain blast furnace material distribution process blast-furnace body parameter, including furnace throat radius, stockline height, chute length, larynx
Pipe height, chute fascinate away from, chute coefficient of friction, furnace charge angle of rest (repose), furnace charge heap density, charge batch weight, and provide performance variable cloth
Expect the vector description of matrix：
α=[α1,…,αm]T∈Rm×1,αi∈[αmin,αmax], (1)
U=[α, κ], (3)
Wherein αminAnd αmaxThe border of chute tilt adjustable section is represented, m represents maximum cloth number of rings, chute dip vector α
Belong to real number field, and rotating cycle vector κ belongs to nature number field.
Step 2：Obtain blast furnace material distribution process stockline radial distribution γ (y), i.e. cloth process bottom distribution shape, wherein y
Represent the distance apart from blast furnace center.
Step 3：Furnace charge volume V is calculated according to charge batch weight and furnace charge heap densityt, and assume furnace charge heap density constant, root
According to conservation of mass principle, volume and furnace charge of the furnace charge in feed bin are equal in the volume that blast furnace throat punishes cloth, and then we carry
Go out the isometric principle of blast furnace material distribution process furnace charge：
Wherein, f (y, u) represents the radial top profile that burden distribution is formed on the basis of stockline γ (y), and u is represented
Burden distribution matrix, constitutes κ by chute dip vector α and rotating cycle vector and constitutes.
Step 4：Thickness of feed layer distribution is calculated according to the distribution shape at blast furnace material distribution process radial direction bottom and top:
H (y, u)=f (y, u)-γ (y). (5)
Step 5：According to the isometric principle of blast furnace material distribution process furnace charge and the separation of variable, bed of material radial thickness is distributed h
(y, u) is write as weight wi(u) with basic function Bi(y) form, specific implementation includes following steps：
Step 5-1：Determine the number n+1 of basic function and weights：
This patent basic function Bi(y), select as B-spline basic function, n is the integer between 5-10.
Step 5-2：Determine basic function Bi(y) volume integral：
B=[b1,b2,…,bn]T∈Rn×1。 (9)
Step 5-3：Determine weights W (u) vector description：
W (u)=[w1(u),w2(u),…,wn(u)]T∈Rn×1, (10)
Wherein vector b and W (u), dimension be n.
Step 5-4：Relation between weight vector W (u) and the distribution of dynamic radial thickness of feed layer is described using dimensionality reduction mode：
H (y, u)=C (y) W (u)+L (y), (13)
Step 6：According to the description of the isometric principle of blast furnace material distribution process furnace charge, and above-mentioned weights and basic function, set
The target g (y) of thickness of feed layer distribution：
G (y)=C (y) Wg+ L (y), (15)
Step 7：According to blast furnace material distribution expertise, integer field cloth number of turns vector κ limited countably infinite set is determined：K=
{κ1,κ2,…κM, wherein M represents cardinality of a set.
Step 8：The target g (y) being distributed according to the thickness of feed layer of setting, defines cloth number of turns vector κ in limited countably infinite setj
Criterion function：
Wherein
Step 9：According to cloth number of turns vector κ in limited countably infinite setjCorresponding criterion function, is given with the method for Gradient Iteration
Slip groove tilt angle vector α control law：
Wherein (k) represents the number of times of iterative learning.
Step 10：Set maximum iteration and stopping criterion for iteration, according to the integer field cloth number of turns having determined to
Amount κ limited countably infinite set calculates cloth number of turns vector κ successivelyjCriterion function, from M limited countably infinite set K={ κ1,κ2,…κM}
The minimum value min of criterion function corresponding to middle selection vector κAnd provide corresponding decision variable α and κ.
For the 2500m shown in Fig. 1 and Fig. 23And tank is without clock-type steel plant blast furnace, furnace throat radius 4.3m, furnace charge volume
30m3, basic function number is your n+1=6, the weight vector W of target distributiong=[0.91,0.90,0.7,0.45,0.24]T, it is standby
The limited countably infinite set is selected to be
γ (y) is fitted by live data-oriented and obtained.
For above-mentioned specific blast furnace material distribution process, aspect is embodied：
(1) step 5 setting basic function B1(y),B2(y),…,B6(y)。
(2) according to step 6 and target distribution weight vector Wg=[0.91,0.90,0.7,0.45,0.24]TCalculate target point
Cloth g (y).
(3) blast furnace burden drop point site (Liu Yuncai, blast furnace material distribution rule [M], metallurgical work are calculated according to material flow track model
Industry publishing house, 2012), f (y, u) is calculated according to patent application document 201510586609.6, and calculate according to this patent step 4
Thickness distribution h (y, u).
(4) maximum iteration is set as 50, and the control flow chart according to Fig. 3 calculates decision variable in real timeAnd criterion functionAnd final decision variable α and κ is calculated according to step 10.
(5) initial distribution and the relativity being finally distributed are provided, as shown in Figure 4.
The present invention gives the concrete operations method of manual regulation thickness of feed layer distribution, from the ore and coke thickness of optimization
The control for the distribution of the blast furnace material distribution process radial direction thickness of feed layer based on iterative learning and multi-model that the present invention is provided is seen in degree distribution
Strategy can realize that the optimization of the performance variable burden distribution matrix of bed of material target thickness distribution is calculated, with visual strong, operation letter
Singly, all there is highly important guidance to anticipate for the characteristics of result is accurate, the realization for operation optimization of distribution and cloth process control
Justice.
This patent gives the concrete operations method of regulation thickness of feed layer distribution manually, based on iteration in terms of control effect
Given bed of material thickness can arbitrarily be tracked by practising the control strategy being distributed with the blast furnace material distribution process radial direction thickness of feed layer of multi-model
Degree distribution, with the characteristics of visuality is strong, simple to operate, result is accurate, for operation optimization of distribution and cloth process control
Realizing all has highly important directive significance.
Claims (6)
1. a kind of control strategy of the blast furnace material distribution process bed of material radial thickness distribution based on iterative learning and multi-model, its feature
Be, be distributed as with blast furnace material distribution process bed of material radial thickness by control targe, using burden distribution matrix as performance variable, build towards
The control strategy of blast furnace material distribution process bed of material radial thickness distribution, the control strategy includes：
(1) distribution of controlled device bed of material radial thickness is write as weights and basic function form using separate variables；
(2) on the basis of the bed of material radial thickness distribution described by weights and basic function, weights are adjusted manually to set the bed of material
The control targe of thickness distribution；
(3) according to the control targe and point of the real-time radial thickness of the bed of material of the blast furnace material distribution process thickness of feed layer distribution set
Cloth defines thickness of feed layer distributed controll performance indications criterion function, based on iterative learning and multi-model process to blast furnace material distribution process
Performance variable burden distribution matrix automatically adjusts.
2. a kind of blast furnace material distribution process bed of material radial thickness point based on iterative learning and multi-model according to claim 1
The control strategy of cloth, it is characterised in that：Controlled device bed of material radial thickness distribution h (y, u) meets cylindricality integral constraint, is one
The bivariate distribution function related to decision variable, performance variable burden distribution matrix includes chute inclination angle sequence and rotating cycle sequence two
Individual part, wherein chute dip vector α belongs to real number field, and cloth number of turns vector κ belongs to nature number field.
3. a kind of blast furnace material distribution process bed of material radial thickness point based on iterative learning and multi-model according to claim 1
The control strategy of cloth, it is characterised in that (1) is write as the distribution of controlled device bed of material radial thickness using separate variables
Weights and basic function form, be specially：
1) bed of material radial thickness is distributed h (y, u) under integral constraint and is write as weight wi(u) with basic function Bi(y) form, and really
Determine the number n+1 of basic function and weights：
<mrow>
<mi>h</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>,</mo>
<mi>u</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</munderover>
<msub>
<mi>w</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>u</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>B</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
</mrow>
<mrow>
<msub>
<mi>b</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<msubsup>
<mo>&Integral;</mo>
<mn>0</mn>
<mi>r</mi>
</msubsup>
<mn>2</mn>
<msub>
<mi>&pi;yB</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mi>y</mi>
<mo>,</mo>
</mrow>
<mrow>
<msub>
<mi>V</mi>
<mi>t</mi>
</msub>
<mo>=</mo>
<msubsup>
<mo>&Integral;</mo>
<mn>0</mn>
<mi>r</mi>
</msubsup>
<mn>2</mn>
<mi>&pi;</mi>
<mi>y</mi>
<mi>h</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>,</mo>
<mi>u</mi>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mi>y</mi>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</munderover>
<msub>
<mi>w</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>u</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>b</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
</mrow>
Wherein, bed of material radial thickness distribution h (y, u) is the two dimension of the position y and decision variable burden distribution matrix u apart from blast furnace center
Function, VtFor furnace charge cumulative volume, biFor basic function Bi(y) volume integral；Basic function Bi(y) be B-spline function, n be 5-20 it
Between integer；
2) relation between n dimension weight vector W (u) and the distribution of bed of material radial thickness is described using dimensionality reduction mode：
H (y, u)=C (y) W (u)+L (y),
<mrow>
<mi>C</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>B</mi>
<mn>1</mn>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mfrac>
<msub>
<mi>b</mi>
<mn>1</mn>
</msub>
<msub>
<mi>b</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
</mfrac>
<msub>
<mi>B</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>B</mi>
<mn>2</mn>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mfrac>
<msub>
<mi>b</mi>
<mn>2</mn>
</msub>
<msub>
<mi>b</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
</mfrac>
<msub>
<mi>B</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mtable>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
</mtable>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>B</mi>
<mi>n</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mfrac>
<msub>
<mi>b</mi>
<mi>n</mi>
</msub>
<msub>
<mi>b</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
</mfrac>
<msub>
<mi>B</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mi>T</mi>
</msup>
<mo>&Element;</mo>
<msup>
<mi>R</mi>
<mrow>
<mn>1</mn>
<mo>&times;</mo>
<mi>n</mi>
</mrow>
</msup>
<mo>,</mo>
</mrow>
W (u)=[w1(u),w2(u),…,wn(u)]T∈Rn×1,
<mrow>
<mi>L</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mi>V</mi>
<mi>t</mi>
</msub>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mo>&Integral;</mo>
<mn>0</mn>
<mi>r</mi>
</msubsup>
<mn>2</mn>
<msub>
<mi>&pi;yB</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mi>d</mi>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msup>
<msub>
<mi>B</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
1
Wherein, W (u) is the n dimension weight vector related to decision variable u, and C (y) is the square of n Wikis function composition after dimension-reduction treatment
Battle array, L (y) is the bound variable after dimension-reduction treatment.
4. a kind of blast furnace material distribution process bed of material radial thickness point based on iterative learning and multi-model according to claim 3
The control strategy of cloth, it is characterised in that in (2), controlled variable bed of material radial thickness target distribution function g (y) is by base letter
Number and weights are determined, on the basis of the bed of material radial thickness distribution described by weights and basic function, manually regulation weights WgWith
The control targe of thickness of feed layer distribution is set, is specially：G (y)=C (y) Wg+L(y)。
5. a kind of blast furnace material distribution process bed of material radial thickness point based on iterative learning and multi-model according to claim 1
The control strategy of cloth, it is characterised in that the control targe that described (3) are distributed according to the blast furnace material distribution process thickness of feed layer set
And the distribution of the real-time radial thickness of the bed of material defines thickness of feed layer distributed controll performance indications criterion function, based on iterative learning and
Multi-model process automatically adjusts to blast furnace material distribution process operation variable burden distribution matrix, is specially：
1) constraint followed according to cloth number of turns sequence κThe maximum number of rings m of cloth, and blast furnace material distribution process side
The few cloth principle of edge multicenter determines M alternative κjFinite aggregate K={ the κ of composition1,κ2,…κM}；
2) according to target distribution g (y), cloth number of turns vector κ in limited countably infinite set is definedjCriterion function：
<mrow>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>&alpha;</mi>
<mo>)</mo>
</mrow>
<msub>
<mo>|</mo>
<msub>
<mi>&kappa;</mi>
<mi>j</mi>
</msub>
</msub>
<mo>=</mo>
<msubsup>
<mo>&Integral;</mo>
<mn>0</mn>
<mi>r</mi>
</msubsup>
<msup>
<mrow>
<mo>(</mo>
<mi>h</mi>
<mo>(</mo>
<mrow>
<mi>y</mi>
<mo>,</mo>
<msub>
<mi>u</mi>
<mi>j</mi>
</msub>
</mrow>
<mo>)</mo>
<mo>-</mo>
<mi>g</mi>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mi>d</mi>
<mi>y</mi>
<mo>=</mo>
<msubsup>
<mi>W</mi>
<mi>j</mi>
<mi>T</mi>
</msubsup>
<msub>
<mi>&Lambda;W</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<mn>2</mn>
<msub>
<mi>&eta;W</mi>
<mi>j</mi>
</msub>
<mo>+</mo>
<mi>&Delta;</mi>
<mo>,</mo>
</mrow>
Wherein
3) according to cloth number of turns vector κ in limited countably infinite setjCorresponding criterion function, provides chute with the method for Gradient Iteration and inclines
Angular amount α control law：
<mrow>
<msup>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msup>
<mo>=</mo>
<msup>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>-</mo>
<mn>2</mn>
<mi>&mu;</mi>
<mrow>
<mo>(</mo>
<msubsup>
<mi>W</mi>
<mi>j</mi>
<mrow>
<mi>T</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</mrow>
</msubsup>
<mi>&Lambda;</mi>
<mo>-</mo>
<mi>&eta;</mi>
<mo>)</mo>
</mrow>
<mfrac>
<mrow>
<mo>&part;</mo>
<msubsup>
<mi>W</mi>
<mi>j</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msubsup>
</mrow>
<mrow>
<mo>&part;</mo>
<mi>&alpha;</mi>
</mrow>
</mfrac>
<msub>
<mo>|</mo>
<msubsup>
<mi>u</mi>
<mi>j</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msubsup>
</msub>
<mo>,</mo>
</mrow>
<mrow>
<msubsup>
<mi>u</mi>
<mi>j</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msubsup>
<mo>=</mo>
<mo>&lsqb;</mo>
<msup>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>,</mo>
<msub>
<mi>&kappa;</mi>
<mi>j</mi>
</msub>
<mo>&rsqb;</mo>
<mo>,</mo>
</mrow>
Wherein (k) represents the number of times of iterative learning；
4) setting maximum iteration and stopping criterion for iteration, according to the limited of the integer field cloth number of turns vector κ having determined
Countably infinite set calculates cloth number of turns vector κ successivelyjCriterion function, from M limited countably infinite set K={ κ1,κ2,…κMIn select to
Measure the minimum value of the criterion function corresponding to κAnd provide corresponding decision-making according to the performance indications of minimum
Variable α and κ.
6. a kind of control strategy of the blast furnace material distribution process bed of material radial thickness distribution based on iterative learning and multi-model, its feature
It is, comprises the following steps：
Step 1：Obtain blast furnace material distribution process blast-furnace body parameter, including furnace throat radius, stockline height, chute length, trunnion height
Degree, chute fascinate away from, chute coefficient of friction, furnace charge angle of rest (repose), furnace charge heap density, charge batch weight, and provide performance variable cloth square
The vector description of battle array：
α=[α1,…,αm]T∈Rm×1,αi∈[αmin,αmax], (1)
<mrow>
<mi>&kappa;</mi>
<mo>=</mo>
<msup>
<mrow>
<mo>&lsqb;</mo>
<msub>
<mi>&kappa;</mi>
<mn>1</mn>
</msub>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<msub>
<mi>&kappa;</mi>
<mi>m</mi>
</msub>
<mo>&rsqb;</mo>
</mrow>
<mi>T</mi>
</msup>
<mo>&Element;</mo>
<msup>
<mi>N</mi>
<mrow>
<mi>m</mi>
<mo>&times;</mo>
<mn>1</mn>
</mrow>
</msup>
<mo>,</mo>
<msub>
<mi>&kappa;</mi>
<mi>t</mi>
</msub>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>m</mi>
</munderover>
<msub>
<mi>&kappa;</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
U=[α, κ], (3)
Wherein αminAnd αmaxThe border of chute tilt adjustable section is represented, m represents maximum cloth number of rings, and chute dip vector α belongs to
Real number field, and rotating cycle vector κ belongs to nature number field；
Step 2：Blast furnace material distribution process stockline radial distribution γ (y), i.e. cloth process bottom distribution shape are obtained, wherein y is represented
Distance apart from blast furnace center；
Step 3：Furnace charge volume V is calculated according to charge batch weight and furnace charge heap densityt, and assume furnace charge heap density constant, according to quality
Conserva-tion principle, volume and furnace charge of the furnace charge in feed bin is equal in the volume that blast furnace throat punishes cloth, draws：
<mrow>
<msub>
<mi>V</mi>
<mi>t</mi>
</msub>
<mo>=</mo>
<msubsup>
<mo>&Integral;</mo>
<mn>0</mn>
<mi>r</mi>
</msubsup>
<mn>2</mn>
<mi>&pi;</mi>
<mi>y</mi>
<mrow>
<mo>(</mo>
<mi>f</mi>
<mo>(</mo>
<mrow>
<mi>y</mi>
<mo>,</mo>
<mi>u</mi>
</mrow>
<mo>)</mo>
<mo>-</mo>
<mi>&gamma;</mi>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mi>y</mi>
<mo>,</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, f (y, u) represents the radial top profile that burden distribution is formed on the basis of stockline γ (y), and u represents cloth
Matrix, constitutes κ by chute dip vector α and rotating cycle vector and constitutes；
Step 4：Thickness of feed layer distribution is calculated according to the distribution shape at blast furnace material distribution process radial direction bottom and top:
H (y, u)=f (y, u)-γ (y)； (5)
Step 5：According to the isometric principle of blast furnace material distribution process furnace charge and the separation of variable, bed of material radial thickness is distributed h (y, u)
Write as weight wi(u) with basic function Bi(y) form, specifically includes following sub-step：
Step 5-1：Bed of material radial thickness is distributed h (y, u) under integral constraint and is write as weight wi(u) with basic function Bi(y) shape
Formula, and determine the number n+1 of basic function and weights：
<mrow>
<mi>h</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>,</mo>
<mi>u</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</munderover>
<msub>
<mi>w</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>u</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>B</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>6</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>b</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<msubsup>
<mo>&Integral;</mo>
<mn>0</mn>
<mi>r</mi>
</msubsup>
<mn>2</mn>
<msub>
<mi>&pi;yB</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mi>y</mi>
<mo>,</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>7</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>V</mi>
<mi>t</mi>
</msub>
<mo>=</mo>
<msubsup>
<mo>&Integral;</mo>
<mn>0</mn>
<mi>r</mi>
</msubsup>
<mn>2</mn>
<mi>&pi;</mi>
<mi>y</mi>
<mi>h</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>,</mo>
<mi>u</mi>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mi>y</mi>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</munderover>
<msub>
<mi>w</mi>
<mi>i</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>u</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>b</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>8</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, bed of material radial thickness distribution h (y, u) is the two dimension of the position y and decision variable burden distribution matrix u apart from blast furnace center
Function, VtFor furnace charge cumulative volume, biFor basic function Bi(y) volume integral；Basic function Bi(y) be B-spline function, n be 5-20 it
Between integer；
Step 5-2：Relation between n dimension weight vector W (u) and the distribution of bed of material radial thickness is described using dimensionality reduction mode：
H (y, u)=C (y) W (u)+L (y), (9)
<mrow>
<mi>C</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>B</mi>
<mn>1</mn>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mfrac>
<msub>
<mi>b</mi>
<mn>1</mn>
</msub>
<msub>
<mi>b</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
</mfrac>
<msub>
<mi>B</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>B</mi>
<mn>2</mn>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mfrac>
<msub>
<mi>b</mi>
<mn>2</mn>
</msub>
<msub>
<mi>b</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
</mfrac>
<msub>
<mi>B</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mtable>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
</mtable>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>B</mi>
<mi>n</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mfrac>
<msub>
<mi>b</mi>
<mi>n</mi>
</msub>
<msub>
<mi>b</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
</mfrac>
<msub>
<mi>B</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mi>T</mi>
</msup>
<mo>&Element;</mo>
<msup>
<mi>R</mi>
<mrow>
<mn>1</mn>
<mo>&times;</mo>
<mi>n</mi>
</mrow>
</msup>
<mo>,</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>10</mn>
<mo>)</mo>
</mrow>
</mrow>
W (u)=[w1(u),w2(u),…,wn(u)]T∈Rn×1, (11)
<mrow>
<mi>L</mi>
<mrow>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mi>V</mi>
<mi>t</mi>
</msub>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mo>&Integral;</mo>
<mn>0</mn>
<mi>r</mi>
</msubsup>
<mn>2</mn>
<msub>
<mi>&pi;yB</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mi>d</mi>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msup>
<msub>
<mi>B</mi>
<mrow>
<mi>n</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>,</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>12</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, W (u) is the n dimension weight vector related to decision variable u, and C (y) is the square of n Wikis function composition after dimension-reduction treatment
Battle array, L (y) is the bound variable after dimension-reduction treatment；
Step 6：According to the description of the isometric principle of blast furnace material distribution process furnace charge, and above-mentioned weights and basic function, the bed of material is set
The target g (y) of thickness distribution：
G (y)=C (y) Wg+ L (y), (13)
Step 7：The constraint followed according to cloth number of turns sequence κThe maximum number of rings m of cloth, and blast furnace material distribution mistake
The few cloth principle of Cheng Bianyuan multicenters determines M alternative κjFinite aggregate K={ the κ of composition1,κ2,…κM}；
Step 8：The target g (y) being distributed according to the thickness of feed layer of setting, defines cloth number of turns vector κ in limited countably infinite setjStandard
Then function：
<mrow>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>&alpha;</mi>
<mo>)</mo>
</mrow>
<msub>
<mo>|</mo>
<msub>
<mi>&kappa;</mi>
<mi>j</mi>
</msub>
</msub>
<mo>=</mo>
<msubsup>
<mo>&Integral;</mo>
<mn>0</mn>
<mi>r</mi>
</msubsup>
<msup>
<mrow>
<mo>(</mo>
<mi>h</mi>
<mo>(</mo>
<mrow>
<mi>y</mi>
<mo>,</mo>
<msub>
<mi>u</mi>
<mi>j</mi>
</msub>
</mrow>
<mo>)</mo>
<mo>-</mo>
<mi>g</mi>
<mo>(</mo>
<mi>y</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mi>d</mi>
<mi>y</mi>
<mo>=</mo>
<msubsup>
<mi>W</mi>
<mi>j</mi>
<mi>T</mi>
</msubsup>
<msub>
<mi>&Lambda;W</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<mn>2</mn>
<msub>
<mi>&eta;W</mi>
<mi>j</mi>
</msub>
<mo>+</mo>
<mi>&Delta;</mi>
<mo>,</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>14</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein
Step 9：According to cloth number of turns vector κ in limited countably infinite setjCorresponding criterion function, is slipped with the method for Gradient Iteration
Groove tilt angle vector α control law：
<mrow>
<msup>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>+</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</msup>
<mo>=</mo>
<msup>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>-</mo>
<mn>2</mn>
<mi>&mu;</mi>
<mrow>
<mo>(</mo>
<msubsup>
<mi>W</mi>
<mi>j</mi>
<mrow>
<mi>T</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</mrow>
</msubsup>
<mi>&Lambda;</mi>
<mo>-</mo>
<mi>&eta;</mi>
<mo>)</mo>
</mrow>
<mfrac>
<mrow>
<mo>&part;</mo>
<msubsup>
<mi>W</mi>
<mi>j</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msubsup>
</mrow>
<mrow>
<mo>&part;</mo>
<mi>&alpha;</mi>
</mrow>
</mfrac>
<msub>
<mo>|</mo>
<msubsup>
<mi>u</mi>
<mi>j</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msubsup>
</msub>
<mo>,</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>15</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msubsup>
<mi>u</mi>
<mi>j</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msubsup>
<mo>=</mo>
<mo>&lsqb;</mo>
<msup>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>,</mo>
<msub>
<mi>&kappa;</mi>
<mi>j</mi>
</msub>
<mo>&rsqb;</mo>
<mo>,</mo>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>16</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein (k) represents the number of times of iterative learning；
Step 10：Maximum iteration and stopping criterion for iteration are set, according to the integer field cloth number of turns vector κ's having determined
Limited countably infinite set calculates cloth number of turns vector κ successivelyjCriterion function, from M limited countably infinite set K={ κ1,κ2,…κMMiddle choosing
Select criterion function corresponding to vectorial κ minimum value min (J (α) |κj), and provide corresponding determine according to the performance indications of minimum
Plan variable α and κ.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710447597.8A CN107245540B (en) | 2017-06-14 | 2017-06-14 | A kind of control strategy of blast furnace material distribution process radial direction thickness of feed layer distribution |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710447597.8A CN107245540B (en) | 2017-06-14 | 2017-06-14 | A kind of control strategy of blast furnace material distribution process radial direction thickness of feed layer distribution |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107245540A true CN107245540A (en) | 2017-10-13 |
| CN107245540B CN107245540B (en) | 2019-03-12 |
Family
ID=60018586
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710447597.8A Active CN107245540B (en) | 2017-06-14 | 2017-06-14 | A kind of control strategy of blast furnace material distribution process radial direction thickness of feed layer distribution |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107245540B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113139275A (en) * | 2021-03-22 | 2021-07-20 | 浙江大学 | Blast furnace throat temperature estimation method based on multilayer ore-coke ratio distribution model |
| CN114045369A (en) * | 2021-11-08 | 2022-02-15 | 福建三宝钢铁有限公司 | Blast furnace iron-smelting method for increasing lump ore proportion |
| CN113637819B (en) * | 2021-08-16 | 2022-10-18 | 中冶东方工程技术有限公司 | Blast furnace material distribution method and system based on deep reinforcement learning |
| CN116497166A (en) * | 2023-04-12 | 2023-07-28 | 山东钢铁股份有限公司 | Intelligent control method for blast furnace top material distribution process flow |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20000039471A (en) * | 1998-12-14 | 2000-07-05 | 이구택 | Method of controlling distribution of charging material into blast furnace |
| CN105243190A (en) * | 2015-09-14 | 2016-01-13 | 内蒙古科技大学 | Method for modeling material surface output shape in material layout process in blast furnace |
| CN106011350A (en) * | 2016-07-05 | 2016-10-12 | 东北大学 | Method and system for estimating charge surface shapes in blast furnace charge distribution process |
| CN106801114A (en) * | 2017-01-04 | 2017-06-06 | 东北大学 | A kind of blast furnace material distribution process burden distribution matrix optimization method and system |
-
2017
- 2017-06-14 CN CN201710447597.8A patent/CN107245540B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20000039471A (en) * | 1998-12-14 | 2000-07-05 | 이구택 | Method of controlling distribution of charging material into blast furnace |
| CN105243190A (en) * | 2015-09-14 | 2016-01-13 | 内蒙古科技大学 | Method for modeling material surface output shape in material layout process in blast furnace |
| CN106011350A (en) * | 2016-07-05 | 2016-10-12 | 东北大学 | Method and system for estimating charge surface shapes in blast furnace charge distribution process |
| CN106801114A (en) * | 2017-01-04 | 2017-06-06 | 东北大学 | A kind of blast furnace material distribution process burden distribution matrix optimization method and system |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113139275A (en) * | 2021-03-22 | 2021-07-20 | 浙江大学 | Blast furnace throat temperature estimation method based on multilayer ore-coke ratio distribution model |
| CN113637819B (en) * | 2021-08-16 | 2022-10-18 | 中冶东方工程技术有限公司 | Blast furnace material distribution method and system based on deep reinforcement learning |
| CN114045369A (en) * | 2021-11-08 | 2022-02-15 | 福建三宝钢铁有限公司 | Blast furnace iron-smelting method for increasing lump ore proportion |
| CN114045369B (en) * | 2021-11-08 | 2022-07-19 | 福建三宝钢铁有限公司 | Blast furnace iron-smelting method for increasing lump ore proportion |
| CN116497166A (en) * | 2023-04-12 | 2023-07-28 | 山东钢铁股份有限公司 | Intelligent control method for blast furnace top material distribution process flow |
| CN116497166B (en) * | 2023-04-12 | 2024-07-19 | 山东钢铁股份有限公司 | Intelligent control method for blast furnace top material distribution process flow |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107245540B (en) | 2019-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107245540A (en) | A kind of control strategy of blast furnace material distribution process radial direction thickness of feed layer distribution | |
| CN104531924B (en) | A kind of blast furnace burden distribution Real-time Forecasting System and method | |
| CN104765350B (en) | Cement decomposing furnace control method and system based on Combined model forecast control technology | |
| CN106801114B (en) | A kind of blast furnace material distribution process burden distribution matrix optimization method and system | |
| CN105950807B (en) | A kind of blast furnace material distribution process shape of charge level modeling method of Multi-information acquisition | |
| CN105243190A (en) | Method for modeling material surface output shape in material layout process in blast furnace | |
| CN106011350B (en) | A kind of blast furnace material distribution process shape of charge level method of estimation and system | |
| CN112906153B (en) | Soil pressure intelligent dynamic regulation and control method of soil pressure balance shield based on LSTM | |
| CN110334462A (en) | A kind of solid-liquid two-phase abrasive Flow Precision Machining molecular dynamics construction method | |
| CN108803634A (en) | A kind of underwater robot six degree of freedom dynamic positioning thrust distribution optimization method based on genetic algorithm | |
| CN104390534B (en) | A kind of tunnel photoface exploision method of quality control | |
| CN106154836A (en) | A kind of online dynamic particles group's PID optimization method | |
| CN107119159A (en) | A kind of optimization method for the blast furnace material distribution process burden distribution matrix that there is integer programming problem | |
| CN104694680A (en) | Control method for radial distribution of blast furnace burden layer structure | |
| CN111461385A (en) | Production plan optimization method based on particle swarm algorithm of L envy flight | |
| CN102636989B (en) | Design method for data-driven PID (proportional integral derivative) controller for adjusting depth of stockline of bell-less top blast furnace | |
| CN202093729U (en) | Material distributing simulation test device for melting gasifier | |
| CN112861276B (en) | Blast furnace burden surface optimization method based on data and knowledge dual drive | |
| JP2017008363A (en) | Method for estimating layer thickness distribution in blast furnace, method for operating blast furnace, and device for estimating layer thickness distribution in blast furnace | |
| CN105854998B (en) | A method of it developing abrading-ball intelligent level and matches controller | |
| CN113656966A (en) | Blast furnace bell-less top on-line material distribution model simulation method | |
| CN100575502C (en) | Use the size of new mutual relationship and mathematical model prediction packed bed system cavity | |
| Jia et al. | Research based on the improved grey wolf optimization algorithm | |
| Zhou et al. | Development of Physical Simulation Experiment Technology for Blast Furnace Coal Injection Ironmaking | |
| CN116497166B (en) | Intelligent control method for blast furnace top material distribution process flow |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |