CN108445745A - A kind of electrode lift control method of ore and system based on neural network decoupling - Google Patents
A kind of electrode lift control method of ore and system based on neural network decoupling Download PDFInfo
- Publication number
- CN108445745A CN108445745A CN201810212034.5A CN201810212034A CN108445745A CN 108445745 A CN108445745 A CN 108445745A CN 201810212034 A CN201810212034 A CN 201810212034A CN 108445745 A CN108445745 A CN 108445745A
- Authority
- CN
- China
- Prior art keywords
- electrodes
- displacement
- electrode
- rise fall
- current
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- G05B13/0265—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
- G05B13/029—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion using neural networks and expert systems
-
- 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
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D3/00—Control of position or direction
- G05D3/12—Control of position or direction using feedback
Abstract
The invention discloses a kind of electrode lift control method of ore and system based on neural network decoupling,This method is by establishing electrodes in mine hot stove Current Position Decoupled Model,The broad sense external interference item in the Decoupled Model of electrode current position is picked out based on RBF neural algorithm,Based on electrodes in mine hot stove Current Position Decoupled Model,Broad sense external interference item and tracking error,Derive rise fall of electrodes displacement,Tracking error is the deviation of electrode current instantaneous value and electrode current setting value and controls rise fall of electrodes in mine hot stove based on rise fall of electrodes displacement,It solves in mineral hot furnace production process since there are serious coupling and random disturbances,Cause electrode control unstable,And then there is the technical issues of unstable product quality and power consumption increase,And accurately three-phase electrode lifting displacement can be obtained by carrying out decoupling in real time to system,To improve efficiency,Reach energy-saving consumption-reducing,Reduce cost.
Description
Technical field
The present invention relates to a kind of electrode lift control method of ore and system based on neural network decoupling, suitable for making
It is heated with three-phase electrode and realizes the various capacity smelted, the various ore-smelting electric furnace production processes for smelting kind.
Background technology
Mineral hot furnace is a kind of capital equipment of " electrothermal way " ferroalloy smelting, and by three-phase electrode, into stove, mixed raw material is led
Enter electrical power, keep furnace charge exothermic melting, generates redox reaction and generate ferroalloy.In order to ensure three reactions in mineral hot furnace stove
The furnace charge in crucible area, which synchronizes, to be chemically reacted, and realizes high yield and low consumption, it is desirable that in-furnace temperature temperature is controlled in smelting process
It is evenly distributed.In-furnace temperature depends on the size of three-phase molten bath power, and the balance of three-phase molten bath power depends on three-phase
The balance of electrode current, the balance adjustment of three-phase electrode electric current need to realize by controlling each phase rise fall of electrodes displacement.
The mode of production traditional at present is mainly manual control three-phase electrode primary side current balance, due to transformer-supplied
The factors such as the complex relationship between mode, the special connection type and threephase load of conductive short net, there are serious couplings for system
Effect and random disturbances, cause this regulative mode to can not achieve three-phase molten bath power-balance.And different operation person operates water
Flat, operating habit difference, operates disunity and causes electrode control unstable, therefore be susceptible to unstable product quality
And the problem of power consumption increase.
Invention content
Electrode lift control method of ore and system provided by the invention based on neural network decoupling solve mine heat
Since there are serious coupling and random disturbances in stove production process, cause electrode control unstable, and then product occur
The technical issues of unstable quality and power consumption increase.
In order to solve the above technical problems, the rise fall of electrodes in mine hot stove controlling party proposed by the present invention based on neural network decoupling
Method includes：
Establish electrodes in mine hot stove electric current-position Decoupled Model；
The broad sense external interference item in electrode current-position Decoupled Model is picked out based on RBF neural algorithm；
Based on electrodes in mine hot stove electric current-position Decoupled Model, broad sense external interference item and tracking error, electrode is derived
Displacement is lifted, tracking error is the deviation of electrode current instantaneous value and electrode current setting value；
Rise fall of electrodes in mine hot stove is controlled based on rise fall of electrodes displacement.
Further, establishing electrodes in mine hot stove electric current-position Decoupled Model includes：
Based on electrode displacement and electrode current variable quantity, the electrode current-position model for establishing mineral hot furnace is：
△ I (k+1)=Φ (k) △ h (k)
Wherein △ I (k+1)=I (k+1)-I (k) is electrode current variable quantity, and △ h (k)=h (k)-h (k-1) is electrode position
It moves, I (k)=[I1(k),I2(k),I3(k)]TFor three-phase electrode electric current, h (k)=[h1(k),h2(k),h3(k)]TFor three-phase electrode
Position,For current displacement reaction coefficient matrix, and I (k)=[I1(k),I2(k),I3(k)] T is IT
(k)=[I1(k),I2(k),I3(k)] transposition, h (k)=[h1(k),h2(k),h3(k)]TFor hT(k)=[h1(k),h2(k),h3
(k)] transposition；
According to stochastic approximation algorithm, estimation current displacement reaction coefficient matrix is：
WhereinForTransposition, η is step series, and μ is weight factor, | | △ h (k-
1) | |=| △ h1(k-1)|+|△h2(k-1)|+|△h3(k-1) | it is △ h (k-1)=[△ h1(k-1),△h2(k-1),△h3
(k-1)]TEuropean norm, | △ hi(k-1) | i=1,2,3 is △ hi(k-1) absolute value, △ IT(k)=[△ I1(k),△I2
(k),△I3(k)] it is △ I (k)=[△ I1(k),△I2(k),△I3(k)]TTransposition, △ hT(k-1)=[△ h1(k-1),△
h2(k-1),△h3(k-1)] it is △ h (k-1)=[△ h1(k-1),△h2(k-1),△h3(k-1)]TTransposition；
Converting electrode current-position model to electrode current-position Decoupled Model is：
Wherein fdi' (k) be system broad sense external interference,For φ11(k) estimated value,For φ22(k)
Estimated value,For φ33(k) estimated value, and fdi' the calculation formula of (k) is：
fdi' (k)=εii(k)△hi(k)+fdi(k)
WhereinFor the estimation error of current displacement reaction coefficient, fdi(k) it is interelectrode
Coupling, and fdi(k) calculation formula is：
Further, the broad sense external world in electrode current-position Decoupled Model is picked out based on RBF neural algorithm to do
Disturbing item includes：
The RBF neural network model of the broad sense external interference amount and electrode current and electrode displacement of the i-th phase electrode is established,
In, the hidden layer node function of RBF neural selects Gaussian function, and it is H=to enable the radial base vector of RBF neural
[h1,h2,…,hm]T, m is node in hidden layer, wherein
Wherein Xk=[xk1,xk2,xk3,xk4]TFor XkT=[xk1,xk2,xk3,xk4] transposition, and xk1=△ Ii(k), xk2=
△Ii(k-1), xk3=△ hi(k-1), xk4=△ hi(k-2), the center vector of j-th of node is Cj=[cj1,cj2,cj3,cj4
]T, bjFor the sound stage width parameter of j-th of node, | | Xk-Cj| |=| xk1-cj1|+|xk2-cj2|+|xk3-cj3|+|xk4-cj4| for vector
[Xk-Cj]=[xk1-cj1,xk2-cj2,xk3-cj3,xk4-cj4]TEuropean norm, | xkn-cjn| n=1,2,3,4 xkn-cjnIt is exhausted
To value；
According to gradient descent method, obtaining output power, node sound stage width parameter and node center vector is：
wj(k)=wj(k-1)+η[fdi'(k)-fdi(k)]hj+α[wj(k-1)-wj(k-2)]
Wherein W=[w1,w2,…,wm]TTo export weight vector, η is learning rate, is α factor of momentum；
According to output weight vector, node sound stage width parameter vector and node center vector, obtaining broad sense external interference item is：
fdi' (k)=w1h1+w2h2+…+wmhm。
Further, it is based on electrodes in mine hot stove electric current-position Decoupled Model, broad sense external interference item and tracking error,
Derive that rise fall of electrodes displacement includes：
The electrode current instantaneous value that computer system detects in real time is compared operation with electrode current setting value, is obtained
Going out the i-th phase electrode current tracking error is：
ei(k+1)=Ir-Ii(k+1), i=1,2 ..., n
I in formularFor electrode current setting value, set according to expertise,
According to electrode current tracking error and electrodes in mine hot stove electric current-position Decoupled Model, derives and moved needed for electrode
Displacement be：
Wherein △ hi(k) it is the lifting displacement of the i-th phase electrode.
Further, after deriving rise fall of electrodes displacement, electrodes in mine hot stove liter is controlled based on rise fall of electrodes displacement
Further include before drop：
Rise fall of electrodes displacement is corrected according to expertise rule.
Further, correcting rise fall of electrodes displacement according to expertise rule is specially：
Rise fall of electrodes displacement is lifted into displacement threshold value with preset extreme electrode respectively and minimum electrode lifts displacement
Amount threshold value is compared, and corrects rise fall of electrodes displacement according to comparison result, and/or
By the displacement of thick stick is carried out with preset maximum secure threshold and minimum safe threshold value respectively energetically after rise fall of electrodes
Compare, and rise fall of electrodes displacement is corrected according to comparison result, energetically the displacement L of cylinderi(k+1)=Li(k)+△hi(k),
Middle Li(k+1) it is the position of the big vertical cylinder of the i-th phase after rise fall of electrodes, Li(k) it is the position of the big vertical cylinder of the i-th phase before rise fall of electrodes,
△hi(k) it is the lifting displacement of the i-th phase electrode, and/or
Furnace gas temperature value after rise fall of electrodes is compared with preset maximum furnace gas temperature threshold value, and is tied according to comparing
Fruit and rise fall of electrodes displacement correct rise fall of electrodes displacement, and/or
Power factor (PF) after rise fall of electrodes is compared with preset minimum power factor threshold value, and according to comparison result
With rise fall of electrodes displacement correct rise fall of electrodes displacement and/or
Primary current after rise fall of electrodes is compared with preset maximum primary current threshold value, and according to comparison result
Rise fall of electrodes displacement is corrected with rise fall of electrodes displacement.
Rise fall of electrodes in mine hot stove control system proposed by the present invention based on neural network decoupling includes：
Memory, processor and storage on a memory and the computer program that can run on a processor, processor
The step of above-mentioned electrode lift control method of ore based on neural network decoupling is realized when executing computer program.
Compared with the prior art, the advantages of the present invention are as follows：
Electrode lift control method of ore and system provided by the invention based on neural network decoupling, by establishing mine
Soaking furnace electrode electric current-position Decoupled Model is picked out based on RBF neural algorithm in electrode current-position Decoupled Model
Broad sense external interference item is based on electrodes in mine hot stove electric current-position Decoupled Model, broad sense external interference item and tracking error, pushes away
Leadout electrode lifts displacement, and tracking error is the deviation of electrode current instantaneous value and electrode current setting value and based on electrode
Lift displacement control rise fall of electrodes in mine hot stove, solve in mineral hot furnace production process due to there are serious coupling and with
Machine interferes, and causes electrode control unstable, and then the technical issues of unstable product quality and power consumption increase occurs, and logical
Cross the broad sense external interference item for system decouple and can pick out in real time system with RBF neural algorithm, and base
Accurately three-phase electrode, which is obtained, in the broad sense external interference item picked out lifts displacement, and the three-phase electrode liter based on acquisition
It drops displacement quickly and precisely adjusts three-phase electrode current balance type, so as to improve the degree of balance of three-phase electrode electric current, realize
The three-phase power balance in molten bath controls, and to improve efficiency, reaches energy-saving consumption-reducing, reduces cost.
Description of the drawings
Fig. 1 is the flow of the electrode lift control method of ore based on neural network decoupling of the embodiment of the present invention one
Figure；
Fig. 2 is the flow of the electrode lift control method of ore based on neural network decoupling of the embodiment of the present invention two
Figure；
Fig. 3 is the electrode displacement amount expert amendment rule flow chart of the embodiment of the present invention two；
Fig. 4 is the rise fall of electrodes in mine hot stove controller overall structure based on neural network decoupling of the embodiment of the present invention two
Figure；
Fig. 5 is the structural frames of the rise fall of electrodes in mine hot stove control system based on neural network decoupling of the embodiment of the present invention
Figure.
Reference numeral：
10, memory；20, processor.
Specific implementation mode
To facilitate the understanding of the present invention, the present invention is made below in conjunction with Figure of description and preferred embodiment more complete
Face meticulously describes, but the protection scope of the present invention is not limited to the following specific embodiments.
The embodiment of the present invention is described in detail below in conjunction with attached drawing, but the present invention can be defined by the claims
Implement with the multitude of different ways of covering.
Embodiment one
Referring to Fig.1, the electrode lift control method of ore based on neural network decoupling that the embodiment of the present invention one provides,
Including：
Step S101 establishes electrodes in mine hot stove electric current-position Decoupled Model；
It is dry to pick out the broad sense external world in electrode current-position Decoupled Model based on RBF neural algorithm by step S102
Disturb item；
Step S103 is based on electrodes in mine hot stove electric current-position Decoupled Model, broad sense external interference item and tracking error,
Derive that rise fall of electrodes displacement, tracking error are the deviation of electrode current instantaneous value and electrode current setting value；
Step S104 controls rise fall of electrodes in mine hot stove based on rise fall of electrodes displacement.
Electrode lift control method of ore provided in an embodiment of the present invention based on neural network decoupling, by establishing mine
Soaking furnace electrode electric current-position Decoupled Model is picked out based on RBF neural algorithm in electrode current-position Decoupled Model
Broad sense external interference item is based on electrodes in mine hot stove electric current-position Decoupled Model, broad sense external interference item and tracking error, pushes away
Leadout electrode lifts displacement, and tracking error is the deviation of electrode current instantaneous value and electrode current setting value and based on electrode
Lift displacement control rise fall of electrodes in mine hot stove, solve in mineral hot furnace production process due to there are serious coupling and with
Machine interferes, and causes electrode control unstable, and then the technical issues of unstable product quality and power consumption increase occurs, and logical
Cross the broad sense external interference item for system decouple and can pick out in real time system with RBF neural algorithm, and base
Accurately three-phase electrode, which is obtained, in the broad sense external interference item picked out lifts displacement, and the three-phase electrode liter based on acquisition
It drops displacement quickly and precisely adjusts three-phase electrode current balance type, so as to improve the degree of balance of three-phase electrode electric current, realize
The three-phase power balance in molten bath controls, and to improve efficiency, reaches energy-saving consumption-reducing, reduces cost.
The embodiment of the present invention passes through the coupling between fully considering three-phase electrode electric current, the decoupling control of reasonable design
Method to electrode position quickly, accurately adjust, and to keep the constant of three-phase electrode electric current, can make three-phase molten bath power in this way
Balance is finally reached the purpose of the control uniform furnace charge synchronous reaction of furnace temperature.
On the one hand, the embodiment of the present invention designs rational solution by the coupling between considering three-phase electrode electric current
Coupling device, and be based further on decoupling device identification and isolate broad sense external interference item, so as to accurately adjust electrode liter
Displacement drops；On the other hand, by being decoupled in real time to system, it can be quickly obtained broad sense external interference item, it can be disposable
Ground calculates the precise displacement amount for realizing that each phase electrode of three-phase electrode current balance type is respectively necessary for, so as to disposably rapidly adjust
Three-phase electrode position is saved, three-phase electrode electric current is made rapidly and precisely to control near setting value.Specifically, when a certain phase current
When deviateing setting value, manual control causes after adjusting the phase rise fall of electrodes, other two without the coupling influence of accurately consideration three-phase
The possible biased error of phase electrode current, then will continue to adjust other two-phase rise fall of electrodes, need repeatedly to adjust back and forth so
It can make three-phase electrode current control near setting value；And neural network decoupling method passes through the coupling work to three-phase electrode electric current
It is recognized with the part as generalized error item, is considered to it when calculating a certain item rise fall of electrodes displacement
The coupling influence of his biphase current can be disposable to which the coupling influence between electric current when three-phase electrode carries out decoupling compensation
Ground provides the displacement of three-phase electrode, by one-time electrode elevating control by three-phase electrode current control near setting value, from
And realize rapidity.
In addition, being modified to electrode displacement amount by Expert Rules and real-time furnace parameter, avoid because lifting displacement
It is unreasonable cause the working of a furnace unstable, can ensure Electrode Lift Control System operation security and stability.The embodiment of the present invention
By adjusting quickly and accurately rise fall of electrodes displacement, so as to adjust quickly and accurately the constant of three-phase electrode electric current,
And then three-phase molten bath power-balance can be made, and achieve the purpose that control the uniform furnace charge synchronous reaction of furnace temperature, it ultimately facilitates and carries
High yield quality and reduction energy consumption of unit product.
Embodiment two
Reference Fig. 2, the electrode lift control method of ore provided by Embodiment 2 of the present invention based on neural network decoupling,
Including
Step S201 establishes electrodes in mine hot stove electric current-position Decoupled Model.
Specifically, the embodiment of the present invention establishes electrodes in mine hot stove electric current-position Decoupled Model and includes：
Based on electrode displacement and electrode current variable quantity, the electrode current-position model for establishing mineral hot furnace is：
△ I (k+1)=Φ (k) △ h (k) (1)
Wherein △ I (k+1)=I (k+1)-I (k) is electrode current variable quantity, and △ h (k)=h (k)-h (k-1) is electrode position
It moves, I (k)=[I1(k),I2(k),I3(k)]TFor three-phase electrode electric current, h (k)=[h1(k),h2(k),h3(k)]TFor three-phase electrode
Position,For current displacement reaction coefficient matrix, and I (k)=[I1(k),I2(k),I3(k)] T is IT
(k)=[I1(k),I2(k),I3(k)] transposition, h (k)=[h1(k),h2(k),h3(k)]TFor hT(k)=[h1(k),h2(k),h3
(k)] transposition；
According to stochastic approximation algorithm, estimation current displacement reaction coefficient matrix Φ (k) is enabled
For the estimated matrix of Φ (k), whereinFor φij(k) estimated value, thenCalculation formula be：
WhereinForTransposition, η is step series, and μ is weight factor, | | △ h (k-
1) | |=| △ h1(k-1)|+|△h2(k-1)|+|△h3(k-1) | it is △ h (k-1)=[△ h1(k-1),△h2(k-1),△h3
(k-1)]TEuropean norm, | △ hi(k-1) | i=1,2,3 is △ hi(k-1) absolute value, △ IT(k)=[△ I1(k),△I2
(k),△I3(k)] it is △ I (k)=[△ I1(k),△I2(k),△I3(k)]TTransposition, △ hT(k-1)=[△ h1(k-1),△
h2(k-1),△h3(k-1)] it is △ h (k-1)=[△ h1(k-1),△h2(k-1),△h3(k-1)]TTransposition；
Converting electrode current-position model to electrode current-position Decoupled Model is：
Wherein fdi' (k) be system broad sense external interference,For φ11(k) estimated value,For φ22(k)
Estimated value,For φ33(k) estimated value, and fdi' the calculation formula of (k) is：
fdi' (k)=εii(k)△hi(k)+fdi(k) (4)
WhereinFor the estimation error of current displacement reaction coefficient, fdi(k) it is interelectrode
Coupling, and fdi(k) calculation formula is：
It is dry to pick out the broad sense external world in electrode current-position Decoupled Model based on RBF neural algorithm by step S202
Disturb item.Specifically, the embodiment of the present invention is picked out wide in electrode current-position Decoupled Model based on RBF neural algorithm
Adopted external interference item includes：
The RBF neural network model of the broad sense external interference amount and electrode current and electrode displacement of the i-th phase electrode is established,
In, the hidden layer node function of RBF neural selects Gaussian function, and it is H=to enable the radial base vector of RBF neural
[h1,h2,…,hm]T, m is node in hidden layer, wherein
Wherein Xk=[xk1,xk2,xk3,xk4]TFor XkT=[xk1,xk2,xk3,xk4] transposition, and xk1=△ Ii(k), xk2=
△Ii(k-1), xk3=△ hi(k-1), xk4=△ hi(k-2), the center vector of j-th of node is Cj=[cj1,cj2,cj3,cj4
]T, bjFor the sound stage width parameter of j-th of node, ‖ Xk-Cj| |=| xk1-cj1|+|xk2-cj2|+|xk3-cj3|+|xk4-cj4| for vector
[Xk-Cj]=[xk1-cj1,xk2-cj2,xk3-cj3,xk4-cj4]TEuropean norm, | xkn-cjn| n=1,2,3,4 xkn-cjnIt is exhausted
To value；
According to gradient descent method, obtaining output power, node sound stage width parameter and node center vector is：
wj(k)=wj(k-1)+η[fdi'(k)-fdi(k)]hj+α[wj(k-1)-wj(k-2)] (7)
Wherein W=[w1,w2,…,wm]TTo export weight vector, η is learning rate, is α factor of momentum；
According to output weight vector, node sound stage width parameter vector and node center vector, obtaining broad sense external interference item is：
fdi' (k)=w1h1+w2h2+…+wmhm (10)
Step S203 is based on electrodes in mine hot stove electric current-position Decoupled Model, broad sense external interference item and tracking error,
Derive that rise fall of electrodes displacement, tracking error are the deviation of electrode current instantaneous value and electrode current setting value.
Specifically, the electrode current instantaneous value first detected computer system in real time is carried out with electrode current setting value
Comparison operation show that the i-th phase electrode current tracking error is：
ei(k+1)=Ir-Ii(k+1), i=1,2 ..., n (11)
I in formularFor electrode current setting value, set according to expertise.
Then, the I in formula (3) is enabledi(k+1)=Ir, and formula (11) substitution (3) can be obtained：
To which the lifting displacement found out needed for the i-th phase electrode by formula (12) is：
Wherein △ hi(k) it is the lifting displacement of the i-th phase electrode.
Step S204 corrects rise fall of electrodes displacement according to expertise rule.
Specifically, the embodiment of the present invention corrects adjuster, the rise fall of electrodes found out to step S203 by designing displacement
Bit andits control amount is modified.Wherein, it is the rule design summarized according to expertise that displacement, which corrects adjuster, such as Fig. 3
Shown, Fig. 3 is the modification rule of the i-th phase rise fall of electrodes displacement, is as follows：
1. avoiding single rise fall of electrodes displacement excessive：
If the absolute value of the i-th phase rise fall of electrodes displacement | △ hi(k) | it is more than single lifting maximum displacement setting value △
Hmax corrects displacement according to formula (14)：
Wherein △ Hi(k) it is revised i-th phase rise fall of electrodes displacement, | △ hi(k) | it is △ hi(k) absolute value,
△ hmax are that single lifts maximum displacement setting value.
2. avoiding single rise fall of electrodes displacement too small：
If the absolute value of the i-th phase rise fall of electrodes displacement | △ hi(k) | it is less than single lifting least displacement setting value △
Hmin does not execute lifting, and displacement is corrected according to formula (15).
△Hi(k)=0 (15)
Wherein △ hmin are that single lifts least displacement setting value, △ Hi(k) it is revised i-th phase rise fall of electrodes position
Shifting amount.
3. the position of cylinder vertical greatly exceeds safety zone range after avoiding rise fall of electrodes：
The big position L for founding cylinder of the i-th phase after rise fall of electrodes is calculated according to formula (16) firsti(k+1)：
Li(k+1)=Li(k)+△hi(k) (16)
If the position L of the big vertical cylinder of the i-th phase after rise fall of electrodesi(k+1) it is more than upper safety limit setting value Lmax, or is less than
Lower safety limit setting value Lmin corrects displacement according to formula (15).
4. avoiding furnace gas temperature excessively high：
If plug for outlet furnace gas temperature T (k) is more than furnace gas temperature upper limit set value Tmax, and the lifting position of the i-th phase electrode
Shifting amount △ hi(k)>0, then stop the i-th phase electrode vertical motion.Displacement is corrected according to formula (15).
5. avoiding power factor too low：
If the power factor F of the i-th phase electrodei(k) it is less than power factor lower limit set value Fmin, and the i-th phase electrode
Lift displacement △ hi(k)<0, then stop the i-th phase electrode lowering action.Displacement is corrected according to formula (15).
6. avoiding primary current excessively high：
If the primary current C of the i-th phasei(k) it is more than primary current upper limit set value Cmax, and the lifting of the i-th phase electrode
Displacement △ hi(k)<0, then stop the i-th phase electrode lowering action.Displacement is corrected according to formula (15).
To obtain the displacement △ H of revised i-th phase electrodei(k), the previous phase of the i-th phase electrode similarly, can be obtained
That is the displacement △ H of the (i-1)-th phase electrodei-1(k) and the displacement △ H of the latter phase of the i-th phase electrode, that is, i+1 phase electrodei+1
(k)。
Step S205 controls rise fall of electrodes in mine hot stove based on rise fall of electrodes displacement.
It is that rise fall of electrodes in mine hot stove controller of the embodiment of the present invention two based on neural network decoupling is whole with reference to Fig. 4, Fig. 4
Structure chart.Wherein, the input signal of controller is the electrode current setting value I set based on expertiser, and acquisition three-phase
Electrode current negative-feedback signal I1(k)、I2(k)、I3(k) after relatively, neural network decoupling controller input error signal e is given1
(k)、e2(k)、e3(k)；Neural network decoupling controller exports the displacement under corresponding error according to current balance type control strategy
△h1(k)、△h2(k)、△h3(k)；Displacement corrects the electricity that adjuster is calculated by acquiring neural network decoupling controller
Pole lifts displacement △ h1(k)、△h2(k)、△h3(k) duty parameter (cylinder position L (k) vertical greatly, the furnace gas temperature and detected in real time
Spend T (k), power factor F (k) etc.), analysis and decision is carried out further according to adjustment rule as shown in Figure 3, to electrode displacement amount △
h1(k)、△h2(k)、△h3(k) it is modified, obtains revised electrode displacement amount △ H1(k)、△H2(k)、△H3(k), and
It is transmitted to three-phase electrode jacking system, electrode position is adjusted.
The embodiment of the present invention has the beneficial effect that：The present invention is based on current balance type control strategies, pass through computer system
The electrode current and electrode position data of on-line checking, are established electrode current-displacement Decoupled Model, are calculated using RBF neural
Method picks out the parameter of electrode current-displacement Decoupled Model, according to the tracking error of electrode current-displacement Decoupled Model and system
Electrode displacement amount is calculated, electrode displacement amount is finally corrected according to expertise, to instruct electrodes in mine hot stove to carry out rising-falling tone
Section.The degree of balance that three-phase electrode electric current can be improved realizes the three-phase power balance control in molten bath, to improve efficiency,
Achieve the purpose that energy-saving consumption-reducing, reduce cost.
With reference to the rise fall of electrodes in mine hot stove control system based on neural network decoupling that Fig. 5, the embodiment of the present invention are proposed, packet
It includes：
Memory 10, processor 20 and it is stored in the computer journey that can be run on memory 10 and on processor 20
Sequence, wherein processor 20 realizes the mineral hot furnace electricity based on neural network decoupling of embodiment of the present invention when executing computer program
The step of pole lift control method.
The specific work process and work of the rise fall of electrodes in mine hot stove control system based on neural network decoupling of the present embodiment
Can refer to as principle the electrode lift control method of ore based on neural network decoupling in the present embodiment the course of work and
Operation principle.
It these are only the preferred embodiment of the present invention, be not intended to restrict the invention, for those skilled in the art
For member, the invention may be variously modified and varied.Any modification made by all within the spirits and principles of the present invention,
Equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (7)
1. a kind of electrode lift control method of ore based on neural network decoupling, which is characterized in that the method includes：
Establish electrodes in mine hot stove electric current-position Decoupled Model；
The broad sense external interference item in the electrode current-position Decoupled Model is picked out based on RBF neural algorithm；
Based on the electrodes in mine hot stove electric current-position Decoupled Model, the broad sense external interference item and tracking error, derive
Rise fall of electrodes displacement, the tracking error are the deviation of electrode current instantaneous value and electrode current setting value；
Rise fall of electrodes in mine hot stove is controlled based on the rise fall of electrodes displacement.
2. the electrode lift control method of ore according to claim 1 based on neural network decoupling, which is characterized in that
Establishing electrodes in mine hot stove electric current-position Decoupled Model includes：
Based on electrode displacement and electrode current variable quantity, the electrode current-position model for establishing mineral hot furnace is：
△ I (k+1)=Φ (k) △ h (k)
Wherein △ I (k+1)=I (k+1)-I (k) is electrode current variable quantity, and △ h (k)=h (k)-h (k-1) is electrode displacement, I
(k)=[I1(k),I2(k),I3(k)]TFor three-phase electrode electric current, h (k)=[h1(k),h2(k),h3(k)]TFor three-phase electrode position
It sets,For current displacement reaction coefficient matrix, and I (k)=[I1(k),I2(k),I3(k)]TFor IT
(k)=[I1(k),I2(k),I3(k)] transposition, h (k)=[h1(k),h2(k),h3(k)]TFor hT(k)=[h1(k),h2(k),h3
(k)] transposition；
According to stochastic approximation algorithm, estimation current displacement reaction coefficient matrix is：
WhereinForTransposition, η is step series, and μ is weight factor, | | △ h (k-1) | |
=| △ h1(k-1)|+|△h2(k-1)|+|△h3(k-1) | it is △ h (k-1)=[△ h1(k-1),△h2(k-1),△h3(k-
1)]TEuropean norm, | △ hi(k-1) | i=1,2,3 is △ hi(k-1) absolute value, △ IT(k)=[△ I1(k),△I2
(k),△I3(k)] it is △ I (k)=[△ I1(k),△I2(k),△I3(k)]TTransposition, △ hT(k-1)=[△ h1(k-1),△
h2(k-1),△h3(k-1)] it is △ h (k-1)=[△ h1(k-1),△h2(k-1),△h3(k-1)]TTransposition；
Converting the electrode current-position model to electrode current-position Decoupled Model is：
Wherein fdi' (k) be system broad sense external interference,ForEstimated value,ForEstimated value,For φ33(k) estimated value, and fdi' the calculation formula of (k) is：
fdi' (k)=εii(k)△hi(k)+fdi(k)
WhereinFor the estimation error of current displacement reaction coefficient, fdi(k) it is interelectrode coupling
Effect, and fdi(k) calculation formula is：
3. the electrode lift control method of ore based on neural network decoupling as claimed in claim 2, which is characterized in that base
Picking out the broad sense external interference item in the electrode current-position Decoupled Model in RBF neural algorithm includes：
Establish the RBF neural network model of the broad sense external interference amount and electrode current and electrode displacement of the i-th phase electrode, wherein
The hidden layer node function of RBF neural selects Gaussian function, and it is H=[h to enable the radial base vector of RBF neural1,
h2,…,hm]T, m is node in hidden layer, wherein
Wherein Xk=[xk1,xk2,xk3,xk4]TFor Xk T=[xk1,xk2,xk3,xk4] transposition, and xk1=△ Ii(k), xk2=△ Ii
(k-1), xk3=△ hi(k-1), xk4=△ hi(k-2), the center vector of j-th of node is Cj=[cj1,cj2,cj3,cj4]T, bj
For the sound stage width parameter of j-th of node, | | Xk-Cj| |=| xk1-cj1|+|xk2-cj2|+|xk3-cj3|+|xk4-cj4| it is vector [Xk-
Cj]=[xk1-cj1,xk2-cj2,xk3-cj3,xk4-cj4]TEuropean norm, | xkn-cjn| n=1,2,3,4 xkn-cjnIt is absolute
Value；
According to gradient descent method, obtaining output power, node sound stage width parameter and node center vector is：
wj(k)=wj(k-1)+η[fdi'(k)-fdi(k)]hj+α[wj(k-1)-wj(k-2)]
Wherein W=[w1,w2,…,wm]TTo export weight vector, η is learning rate, is α factor of momentum；
According to the output weight vector, the node sound stage width parameter vector and the node center vector, it is extraneous to obtain broad sense
Distracter is：
fdi' (k)=w1h1+w2h2+…+wmhm。
4. the electrode lift control method of ore based on neural network decoupling as claimed in claim 3, which is characterized in that base
In the electrodes in mine hot stove electric current-position Decoupled Model, the broad sense external interference item and tracking error, electrode liter is derived
Dropping displacement includes：
The electrode current instantaneous value that computer system detects in real time is compared operation with electrode current setting value, obtains
I phase electrode current tracking errors are：
ei(k+1)=Ir-Ii(k+1), i=1,2 ..., n
I in formularFor electrode current setting value, set according to expertise,
According to the electrode current tracking error and electrodes in mine hot stove electric current-position Decoupled Model, derive needed for electrode
Mobile displacement is：
Wherein △ hi(k) it is the lifting displacement of the i-th phase electrode.
5. special according to any electrode lift control method of ore based on neural network decoupling of claim 1-4
Sign is, after deriving rise fall of electrodes displacement, before rise fall of electrodes displacement control rise fall of electrodes in mine hot stove
Further include：
The rise fall of electrodes displacement is corrected according to expertise rule.
6. the electrode lift control method of ore according to claim 5 based on neural network decoupling, which is characterized in that
Correcting the rise fall of electrodes displacement according to expertise rule is specially：
The rise fall of electrodes displacement is lifted into displacement threshold value with preset extreme electrode respectively and minimum electrode lifts displacement
Amount threshold value is compared, and corrects the rise fall of electrodes displacement according to comparison result, and/or
By the displacement of thick stick is compared with preset maximum secure threshold and minimum safe threshold value respectively energetically after rise fall of electrodes,
And the rise fall of electrodes displacement, the displacement L of the cylinder energetically are corrected according to comparison resulti(k+1)=Li(k)+△hi(k),
Wherein Li(k+1) it is the position of the big vertical cylinder of the i-th phase after rise fall of electrodes, Li(k) it is the position of the big vertical cylinder of the i-th phase before rise fall of electrodes
It sets, △ hi(k) it is the lifting displacement of the i-th phase electrode, and/or
Furnace gas temperature value after rise fall of electrodes is compared with preset maximum furnace gas temperature threshold value, and according to comparison result and
The rise fall of electrodes displacement corrects the rise fall of electrodes displacement, and/or
Power factor (PF) after rise fall of electrodes is compared with preset minimum power factor threshold value, and according to comparison result and institute
State rise fall of electrodes displacement correct the rise fall of electrodes displacement and/or
Primary current after rise fall of electrodes is compared with preset maximum primary current threshold value, and according to comparison result and institute
It states rise fall of electrodes displacement and corrects the rise fall of electrodes displacement.
7. a kind of rise fall of electrodes in mine hot stove control system based on neural network decoupling, including：Memory, processor and storage
On a memory and the computer program that can run on a processor, which is characterized in that the processor executes the computer
The step of any the method for the claims 1 to 6 is realized when program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810212034.5A CN108445745A (en) | 2018-03-15 | 2018-03-15 | A kind of electrode lift control method of ore and system based on neural network decoupling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810212034.5A CN108445745A (en) | 2018-03-15 | 2018-03-15 | A kind of electrode lift control method of ore and system based on neural network decoupling |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108445745A true CN108445745A (en) | 2018-08-24 |
Family
ID=63195272
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810212034.5A Pending CN108445745A (en) | 2018-03-15 | 2018-03-15 | A kind of electrode lift control method of ore and system based on neural network decoupling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108445745A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110262244A (en) * | 2019-07-02 | 2019-09-20 | 武汉科技大学 | A kind of self adaptation straightening method for improving FSRBFD |
| CN113281581A (en) * | 2021-04-14 | 2021-08-20 | 中国重型机械研究院股份公司 | LF furnace electrode testing system and testing method |
| CN114659373A (en) * | 2022-03-22 | 2022-06-24 | 宁夏昌茂祥冶炼有限公司 | Submerged arc furnace system with automatic electrode lifting function and control method thereof |
| CN116499273A (en) * | 2023-06-25 | 2023-07-28 | 成都星云智联科技有限公司 | Real-time control method and control system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102111927A (en) * | 2009-12-24 | 2011-06-29 | 深圳达实智能股份有限公司 | Method and system for controlling electrode of open-type submerged arc ore heat furnace |
| US20110224830A1 (en) * | 2006-04-25 | 2011-09-15 | Pegasus Technologies, Inc. | Control system for operation of a fossil fuel power generating unit |
| CN102331193A (en) * | 2011-09-22 | 2012-01-25 | 云南文山斗南锰业股份有限公司 | System and method for controlling electrode of ore-smelting electric furnace |
| CN102630107A (en) * | 2012-04-06 | 2012-08-08 | 南京理工大学常熟研究院有限公司 | Electrode lift control method of ore smelting electric arc furnace |
| CN103024959A (en) * | 2012-12-03 | 2013-04-03 | 北京金自天正智能控制股份有限公司 | Intelligent electrode regulator and intelligent electrode control method |
| CN103808159A (en) * | 2013-11-08 | 2014-05-21 | 西安交通大学 | Automatic control method used for submerged arc furnace electrode and based on high-precision fuzzy control |
-
2018
- 2018-03-15 CN CN201810212034.5A patent/CN108445745A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110224830A1 (en) * | 2006-04-25 | 2011-09-15 | Pegasus Technologies, Inc. | Control system for operation of a fossil fuel power generating unit |
| CN102111927A (en) * | 2009-12-24 | 2011-06-29 | 深圳达实智能股份有限公司 | Method and system for controlling electrode of open-type submerged arc ore heat furnace |
| CN102331193A (en) * | 2011-09-22 | 2012-01-25 | 云南文山斗南锰业股份有限公司 | System and method for controlling electrode of ore-smelting electric furnace |
| CN102630107A (en) * | 2012-04-06 | 2012-08-08 | 南京理工大学常熟研究院有限公司 | Electrode lift control method of ore smelting electric arc furnace |
| CN103024959A (en) * | 2012-12-03 | 2013-04-03 | 北京金自天正智能控制股份有限公司 | Intelligent electrode regulator and intelligent electrode control method |
| CN103808159A (en) * | 2013-11-08 | 2014-05-21 | 西安交通大学 | Automatic control method used for submerged arc furnace electrode and based on high-precision fuzzy control |
Non-Patent Citations (3)
| Title |
|---|
| ZHENGJIE CHEN等: "Artificial neural network modeling for evaluating the power consumption of silicon production in submerged arc furnaces", 《APPLIED THERMAL ENGINEERING》 * |
| 张传伟 等: "基于大电流检测的矿热炉电极升降自动控制系统", 《西安科技大学学报》 * |
| 邓睿: "矿热炉电极调节系统设计与控制算法研究", 《中国优秀硕士学位论文全文数据库（工程科技Ⅰ辑）》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110262244A (en) * | 2019-07-02 | 2019-09-20 | 武汉科技大学 | A kind of self adaptation straightening method for improving FSRBFD |
| CN110262244B (en) * | 2019-07-02 | 2022-04-01 | 武汉科技大学 | Self-adaptive decoupling control method for improving FSRBFD |
| CN113281581A (en) * | 2021-04-14 | 2021-08-20 | 中国重型机械研究院股份公司 | LF furnace electrode testing system and testing method |
| CN113281581B (en) * | 2021-04-14 | 2022-04-01 | 中国重型机械研究院股份公司 | Test method of LF furnace electrode test system |
| CN114659373A (en) * | 2022-03-22 | 2022-06-24 | 宁夏昌茂祥冶炼有限公司 | Submerged arc furnace system with automatic electrode lifting function and control method thereof |
| CN116499273A (en) * | 2023-06-25 | 2023-07-28 | 成都星云智联科技有限公司 | Real-time control method and control system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108445745A (en) | A kind of electrode lift control method of ore and system based on neural network decoupling | |
| CN102234819B (en) | Preheating starting method for aluminium electrolysis cell | |
| CN110247562A (en) | A kind of power supply unit and method of supplying power to of direct current electric arc furnace | |
| CN102889944B (en) | Calcium carbide production furnace temperature monitoring method | |
| CN102111927B (en) | Method and system for controlling electrode of open-type submerged arc ore heat furnace | |
| CN107045284A (en) | A kind of transformer on-load voltage regulating construction method based on electric melting magnesium furnace | |
| CN103024959B (en) | Intelligent electrode regulator and intelligent electrode control method | |
| CN101709484B (en) | Preheating starting method for inertia anode aluminum electrolyzer | |
| CN101720146A (en) | Control method of embedding depth of electrode into ore-smelting electric furnace | |
| CN109757003A (en) | Mineral hot furnace autocontrol method | |
| CN107016509B (en) | A method of reducing steel rolling process energy consumption per ton steel | |
| CN106766905B (en) | Control the method and device of electric parameter balance in mineral hot furnace | |
| CN106123768B (en) | A kind of electrodes in mine hot stove depth-measuring system | |
| CN108444280A (en) | The accuracy control method of furnace of calcium carbide rise fall of electrodes | |
| CN102297584A (en) | Electrode control system for submerged arc furnace with position inner ring and current outer ring | |
| CN103898603A (en) | Dual-power polycrystalline silicon ingot casting process | |
| CN114545865B (en) | Polycrystalline silicon growth control method | |
| CN102771183A (en) | Method for operating an arc furnace, control and/or regulating device for an arc furnace, and arc furnace | |
| CN101935851B (en) | Current strengthening and efficient energy saving method of prebaked aluminium electrolysis cell | |
| CN106931796A (en) | Based on the electrodes in mine hot stove control method for calculating electrode pay(useful) load resistance in stove | |
| CN104486859B (en) | Electrodes in mine hot stove acting point is from the method for motion tracking bath surface track | |
| CN110994669B (en) | Control method and system for centralized inverter of photovoltaic power station | |
| CN207751353U (en) | A kind of electrode regulator based on power quality dynamic self-adapting | |
| CN108574306B (en) | Virtual synchronous machine excitation adjusting method based on self-adaptive droop coefficient | |
| CN201868951U (en) | Submerged arc furnace energy-saving on-line monitoring device |
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: 20180824 |
|
| RJ01 | Rejection of invention patent application after publication |