CN106200379B - A kind of distributed dynamic matrix majorization method of Nonself-regulating plant - Google Patents
A kind of distributed dynamic matrix majorization method of Nonself-regulating plant Download PDFInfo
- Publication number
- CN106200379B CN106200379B CN201610539559.0A CN201610539559A CN106200379B CN 106200379 B CN106200379 B CN 106200379B CN 201610539559 A CN201610539559 A CN 201610539559A CN 106200379 B CN106200379 B CN 106200379B
- Authority
- CN
- China
- Prior art keywords
- intelligent body
- moment
- cor
- matrix
- model
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a kind of distributed dynamic matrix majorization methods of Nonself-regulating plant.The present invention passes through the matrix model vector that acquisition step response data establishes the multivariable process containing Nonself-regulating plant first, then the on-line optimization implementation issue of multivariable process is converted to the optimal enforcement problem of each small-scale subsystem.Then suitable performance indicator is chosen, the Nash optimization solution of each intelligent body is obtained by continuous iteration, and then obtain the parameter of each intelligent body dynamic matrix controller, implement the instant control law at the moment to each intelligent body again, and time domain is rolled to subsequent time, above-mentioned optimization process is repeated, to complete the optimization task of whole system.The present invention can effectively compensate for deficiency of traditional DDMC method in the multivariable process control containing Nonself-regulating plant under the premise of guaranteeing compared with high control precision and stability, and meet the needs of actual industrial process.
Description
Technical field
The invention belongs to fields of automation technology, are related to a kind of distributed dynamic matrix majorization (DDMC) of Nonself-regulating plant
Method.
Background technique
The large scale system of large amount of complex higher-dimension is widely present in real process, using the integrated solution of centralization to meter
Performance and processing speed of calculation machine etc. often require that very high, disagree with the economy that must be taken into consideration in actual industrial system.Point
A Main Branches of the cloth dynamic matrix control (DDMC) as Distributed Predictive Control (DMPC), comprehensive utilization computer are logical
Letter technology and control theory are distributed to the line solver problem of a complicated large scale system in subsystems and go distribution real
It is existing, the scale and complexity of problem are effectively reduced, can control that there are multivariables, close coupling, controlled pair uncertain well
As improving system control performance.However in actual industrial process, there are many multivariable processes containing Nonself-regulating plant,
Such as a part of storage tank, boiler drum level, rectifying column liquid level.It is typical due to containing in the transmission function of Nonself-regulating plant
Integral element, and then response of the controlled device under definite value step is caused to tend to be infinite, this allows for traditional DDMC algorithm can not
Directly apply.If can improve in the actual process to traditional DDMC method, traditional DDMC method just can be effectively made up
Deficiency in the multivariable process control containing Nonself-regulating plant, so that DMPC is further extended and sent out in practical applications
Exhibition.
Summary of the invention
Object of the present invention is to the deficiencies for tradition DDMC method in the multivariable process control containing Nonself-regulating plant
Place, proposes a kind of DDMC method of Nonself-regulating plant.
This method passes through the matrix model that acquisition step response data establishes the multivariable process containing Nonself-regulating plant first
Vector excavates basic plant characteristic, then the on-line optimization implementation issue of multivariable process is converted to each small-scale son
The optimal enforcement problem of system, and theory and thought in multiple agent are combined, each subsystem under network environment is regarded as
Pass through network for an intelligent body, between each intelligent body and carries out substance, energy and information communication.Then non-by the way that one kind to be directed to
Self regulating plant improves the method for transfer matrix and a kind of new error calibration method combines, and chooses suitable performance indicator,
The Nash optimization solution of each intelligent body is obtained by continuous iteration based on the thought of assorted optimization is received, and then obtains each intelligent body dynamic square
The parameter of battle array controller, then implement the instant control law at the moment to each intelligent body, and time domain is rolled to subsequent time, weight
Multiple above-mentioned optimization process, to complete the optimization task of whole system.
The technical scheme is that establishing one by means such as data acquisition, model foundation, prediction mechanism, optimizations
The distributed dynamic matrix majorization method of kind Nonself-regulating plant, using this method before guaranteeing compared with high control precision and stability
It puts, can effectively compensate for deficiency of traditional DDMC method in the multivariable process control containing Nonself-regulating plant, and meet
The demand of actual industrial process.
The step of the method for the present invention includes：
Step 1. establishes corresponding dynamic matrix model vector by the real-time step response data of Nonself-regulating plant, specifically
Method is：
1.1, according to Distributed Predictive Control thought, the large scale system of a N input N output Nonself-regulating plant are dispersed
For N number of intelligent body subsystem；
1.2 under steady state operating conditions, is that input carries out step to i-th of intelligent body output quantity with j-th of intelligent body control amount
Response experiment records jth (1≤j≤N) a input to the step response curve of i-th (1≤i≤N) a output respectively；
1.3 step response curves for obtaining step 1.2 are filtered, and are then fitted to a smooth curve, note
The corresponding step response data of each sampling instant on smooth curve is recorded, first sampling instant is Ts, when two neighboring sampling
The interval time at quarter is Ts, sampling instant sequence is Ts、2Ts、3Ts……；The step response data of controlled device will be at some
Moment tL=IijTsStart to present and determine slope rising, with the data at the momentFor starting point, data before are denoted as respectivelyJ-th of input is established to the step response model vector a between i-th of outputij：
Wherein T be matrix transposition symbol, δ be step response data in constant-slope rise after adjacent two data it
Between constant difference, LijFor the model length that j-th of input of setting exports i-th, Lij≥Iij+1。
Step 2. designs the dynamic matrix controller of i-th of intelligent body, and specific method is：
The 2.1 model vector a obtained using step 1ijThe dynamic matrix of controlled device is established, form is as follows：
Wherein AijP × M rank the dynamic matrix exported to i-th of intelligent body, a are inputted for j-th of intelligent bodyijIt (k) is jth
The step response data that a input exports i-th, P are the optimization time domain of Dynamic array control algorithm, and M is dynamic matrix control
The control time domain of algorithm, Lij=L (1≤i≤3,1≤j≤3), M<P<L, N are input and output number；
2.2 obtain the model prediction initial communication value y at i-th of intelligent body current k momenti,0(k)
Firstly, controlling increment △ u is added at the k-1 moment1(k-1),△u2(k-1),…,△un(k-1), i-th of intelligence is obtained
The model predication value y of energy bodyi,P(k-1)：
Wherein,
yi,P(k-1)=[yi,1(k|k-1),yi,1(k+1|k-1),…,yi,1(k+L-1|k-1)]T
yi,0(k-1)=[yi,0(k|k-1),yi,0(k+1|k-1),…,yi,0(k+L-1|k-1)]T,
Aii,0=[aii(1),aii(2),…,aii(L)]T,Aij,0=[aij(1),aij(2),…,aij(L)]T
yi,1(k|k-1),yi,1(k+1|k-1),…,yi,1(k+L-1 | k-1) i-th of intelligent body is respectively indicated at the k-1 moment
To k, k+1 ..., the model predication value at k+L-1 moment, yi,0(k|k-1),yi,0(k+1|k-1),…,yi,0(k+L-1 | k-1) table
Show the k-1 moment to k, k+1 ..., the initial prediction at k+L-1 moment, Aii,0,Aij,0Respectively i-th of intelligent body and j-th of intelligence
The matrix that the step response data of i-th of intelligent body output is established in the input of energy body, △ u1(k-1),△u2(k-1),…,△un
It (k-1) is the input control quantity of k-1 moment each intelligent body；
It is then possible to obtain the model predictive error value e of i-th of intelligent body of k momenti(k)：
ei(k)=yi(k)-yi,1(k|k-1)
Wherein yi(k) real output value for i-th of intelligent body that the k moment measures is indicated；
Further obtain k moment revised model output value yi,cor(k)：
yi,cor(k)=yi,0(k-1)+h1*ei(k)+h2*ei(k)
Wherein,
yi,cor(k)=[yi,cor(k|k),yi,cor(k+1|k),…,yi,cor(k+L-1|k)]T,
h1=[1, α ..., α]T,h2=[0,1 ..., L-1]T
yi,cor(k|k),yi,cor(k+1|k),…,yi,cor(k+L-1 | k) i-th of intelligent body is respectively indicated in k moment model
Correction value, h1And h2For the weight matrix of error compensation, α is error correction coefficient, 0<α≤1；
Finally obtain the initial communication value y of the model prediction at i-th of intelligent body k momenti,0(k)：
yi,0(k)=Syi,cor(k)
Wherein, S is the new state-transition matrix of L × L rank,
2.3 obtain i-th of intelligent body in M continuous controlling increment △ u according to step 2.1i(k),△ui(k+1),…,
△ui(k+M-1) the prediction output valve y underi,PM, specific method is：
Wherein,
yi,PM(k)=[yi,M(k+1|k),yi,M(k+2|k),…,yi,M(k+P|k)]T
yi,P0(k)=[yi,0(k+1|k),yi,0(k+2|k),…,yi,0(k+P|k)]T
△ui,M(k)=[△ ui(k),△ui(k+1),…,△ui(k+M-1)]T
△uj,M(k)=[△ uj(k),△uj(k+1),…,△uj(k+M-1)]T
yi,P0It (k) is yi,0(k) preceding P, yi,0(k+1|k),yi,0(k+2|k),…,yi,0(k+P | k) it is the k moment to k+
1, k+2 ..., the model prediction output valve at k+P moment；
2.4 establish the performance indicator J of i-th of intelligent body dynamic matrix controller of Nonself-regulating planti(k) and reference locus ωi
(k), form is as follows：
minJi(k)=(ωi(k)-yi,PM(k))TQi(ωi(k)-yi,PM(k))+△ui,M(k)TRi△ui,M(k)
ωi(k)=[ωi(k+1),ωi(k+2),…,ωi(k+P)]T
ωi(k+ ε)=βεy(k)+(1-βε) c (k) (ε=1,2 ..., P)
WhereinFor error weighting matrix,Square is weighted for control
Battle array,WithRespectively Qi,RiIn weight coefficient, ωiIt (k) is the reference locus of i-th of intelligent body, β
For the softening coefficient of reference locus；
2.5 thought according to Nash optimization obtains receiving for i-th of intelligent body current k moment by performance indicator in step 2.4
Assorted optimal solution：
Wherein：
2.6 are by the available new round iteration optimal solution in k moment intelligent body i of step 2.2 to 2.5：
Whole system is further obtained in the optimal control law at k moment：
Wherein：
ω (k)=[ω1(k),ω2(k),…,ωn(k)]T, yP0(k)=[y1,P0(k),y2,P0(k),…,yn,P0(k)]T
2.7 using the Nash optimization solution first term at i-th of intelligent body k moment as instant control law △ ui(k), intelligent body is obtained
The practical control amount u of ii(k)=ui(k-1)+△ui(k) i-th of intelligent body is acted on；
2.8, in subsequent time, repeat step 2.2 to 2.7 and continue to solve the instant control law △ u of i-th of intelligent bodyi(k+
1), and then the optimal solution △ u (k+1) of whole system is obtained, and circuited sequentially.
The invention proposes a kind of DDMC methods of Nonself-regulating plant.This method, will on the basis of traditional DDMC method
A kind of a kind of method and new error calibration method combination for improving transfer matrix for Nonself-regulating plant, is guaranteeing higher control
Under the premise of precision and stability, traditional DDMC method is effectively compensated in the multivariable process control containing Nonself-regulating plant
Deficiency, and meet the needs of actual industrial process.
Specific embodiment
By taking general predictive control as an example：
Drum Water Level Control System for Boiler is the typically multivariable Nonself-regulating plant with integral element, regulating measure
Using control water-supply valve valve opening.
Step 1. by the real-time step response data of boiler drum level object establish corresponding dynamic matrix model to
Amount, specific method are：
1.1 according to Distributed Predictive Control thought, by the extensive system of one 33 output boiler drum level object of input
System is separated into 3 subsystems；
1.2 under steady state operating conditions, with j-th boiler feed valve valve opening be input to i-th of boiler drum level into
The experiment of row step response records jth (1≤j≤3) a input to the step response curve of i-th (1≤i≤3) a output respectively；
1.3 step response curves for obtaining step 1.2 are filtered, and are then fitted to a smooth curve, note
The corresponding step response data of each sampling instant on smooth curve is recorded, first sampling instant is Ts, when two neighboring sampling
The interval time at quarter is Ts, sampling instant sequence is Ts、2Ts、3Ts……；The step response data of boiler drum level will be at certain
One moment tL=IijTsStart to present and determine slope rising, with the data at the momentFor starting point, data before are remembered respectively
It doesJ-th of boiler input is established to the step response model vector a between i-th of boiler outputij：
Wherein T be matrix transposition symbol, δ be step response data in constant-slope rise after adjacent two data it
Between constant difference, LijFor the model length that j-th of input of setting exports i-th, Lij≥Iij+1。
Step 2. designs the dynamic matrix controller of i-th of boiler, specifically：
The 2.1 model vector a obtained using step 1ijThe dynamic matrix of boiler drum level is established, form is as follows：
Wherein AijP × M rank the dynamic matrix exported to i-th of boiler, a are inputted for j-th of boilerijIt (k) is j-th pot
Furnace inputs the step response data exported to i-th of boiler, and P is the optimization time domain of Dynamic array control algorithm, and M is dynamic matrix
The control time domain of control algolithm, Lij=L (1≤i≤3,1≤j≤3), M<P<L, N=3 are input and output number；
2.2 obtain the model prediction initial communication value y at i-th of boiler current k momenti,0(k)
Firstly, controlling increment △ u is added at the k-1 moment1(k-1),△u2(k-1),…,△un(k-1) (n=3), obtains
The model predication value y of i-th of boileri,P(k-1)：
Wherein,
yi,P(k-1)=[yi,1(k|k-1),yi,1(k+1|k-1),…,yi,1(k+L-1|k-1)]T
yi,0(k-1)=[yi,0(k|k-1),yi,0(k+1|k-1),…,yi,0(k+L-1|k-1)]T,
Aii,0=[aii(1),aii(2),…,aii(L)]T,Aij,0=[aij(1),aij(2),…,aij(L)]T
yi,1(k|k-1),yi,1(k+1|k-1),…,yi,1(k+L-1 | k-1) i-th of boiler is respectively indicated at the k-1 moment pair
K, k+1 ..., the model predication value at k+L-1 moment, yi,0(k|k-1),yi,0(k+1|k-1),…,yi,0(k+L-1 | k-1) it indicates
The k-1 moment is to k, k+1 ..., the initial prediction at k+L-1 moment, Aii,0,Aij,0Respectively i-th of boiler and j-th of boiler are defeated
Enter the matrix established to the step response data of i-th of boiler output, △ u1(k-1),△u2(k-1),…,△unIt (k-1) is k-
The input water-supply valve valve opening increment of 1 moment each boiler；
It is then possible to obtain the model predictive error value e of i-th of boiler of k momenti(k)：
ei(k)=yi(k)-yi,1(k|k-1)
Wherein yi(k) real output value for i-th of boiler that the k moment measures is indicated；
Further obtain k moment revised model output value yi,cor(k)：
yi,cor(k)=yi,0(k-1)+h1*ei(k)+h2*ei(k)
Wherein,
yi,cor(k)=[yi,cor(k|k),yi,cor(k+1|k),…,yi,cor(k+L-1|k)]T,
h1=[1, α ..., α]T,h2=[0,1 ..., L-1]T
yi,cor(k|k),yi,cor(k+1|k),…,yi,cor(k+L-1 | k) i-th of boiler is respectively indicated in k moment model
Correction value, h1And h2For the weight matrix of error compensation, α is error correction coefficient, 0<α≤1；
Finally obtain the initial communication value y of the model prediction at i-th of boiler k momenti,0(k)：
yi,0(k)=Syi,cor(k)
Wherein, S is the new state-transition matrix of L × L rank,
2.3 obtain i-th of boiler in M continuous controlling increment △ u according to step 2.1i(k),△ui(k+1),…,△
ui(k+M-1) the prediction output valve y underi,PM, specific method is：
Wherein,
yi,PM(k)=[yi,M(k+1|k),yi,M(k+2|k),…,yi,M(k+P|k)]T
yi,P0(k)=[yi,0(k+1|k),yi,0(k+2|k),…,yi,0(k+P|k)]T
△ui,M(k)=[△ ui(k),△ui(k+1),…,△ui(k+M-1)]T
△uj,M(k)=[△ uj(k),△uj(k+1),…,△uj(k+M-1)]T
yi,P0It (k) is yi,0(k) preceding P, yi,0(k+1|k),yi,0(k+2|k),…,yi,0(k+P | k) it is the k moment to k+
1, k+2 ..., the model prediction output valve at k+P moment；
2.4 establish the performance indicator J of i-th of boiler dynamic matrix controller of boiler drum level objecti(k) and with reference to rail
Mark ωi(k), form is as follows：
minJi(k)=(ωi(k)-yi,PM(k))TQi(ωi(k)-yi,PM(k))+△ui,M(k)TRi△ui,M(k)
ωi(k)=[ωi(k+1),ωi(k+2),…,ωi(k+P)]T
ωi(k+ ε)=βεy(k)+(1-βε) c (k) (ε=1,2 ..., P)
WhereinFor error weighting matrix,Square is weighted for control
Battle array,WithRespectively Qi,RiIn weight coefficient, ωiIt (k) is the reference locus of i-th of boiler, β is
The softening coefficient of reference locus；
It is assorted to obtain receiving for i-th of boiler current k moment by performance indicator in step 2.4 for 2.5 thought according to Nash optimization
Optimal solution：
Wherein：
2.6 are by the available new round iteration optimal solution in k moment boiler i of step 2.2 to 2.5：
Whole system is further obtained in the optimal control law at k moment：
Wherein：
ω (k)=[ω1(k),ω2(k),…,ωn(k)]T, yP0(k)=[y1,P0(k),y2,P0(k),…,yn,P0(k)]T
2.7 using the Nash optimization solution first term at i-th of boiler k moment as instant control law △ ui(k), obtain boiler i's
Practical water-supply valve valve opening ui(k)=ui(k-1)+△ui(k) i-th of boiler is acted on；
2.8, in subsequent time, repeat step 2.2 to 2.7 and continue to solve the instant control law △ u of i-th of boileri(k+1),
And then the optimal control law △ u (k+1) of whole system is obtained, and circuit sequentially.
Claims (1)
1. a kind of distributed dynamic matrix majorization method of Nonself-regulating plant, it is characterised in that this approach includes the following steps：
Step 1. establishes corresponding dynamic matrix model vector by the real-time step response data of Nonself-regulating plant, specifically：
1.1, according to Distributed Predictive Control thought, the large scale system of a N input N output Nonself-regulating plant are separated into N number of
Intelligent body subsystem；
1.2 under steady state operating conditions, is that input carries out step response to i-th of intelligent body output quantity with j-th of intelligent body control amount
Experiment records the step response curve that j-th of input exports i-th, wherein 1≤j≤N, 1≤i≤N respectively；
1.3 step response curves for obtaining step 1.2 are filtered, and are then fitted to a smooth curve, recording light
The corresponding step response data of each sampling instant on sliding curve, first sampling instant is Ts, two neighboring sampling instant
Interval time is Ts, sampling instant sequence is Ts、2Ts、3Ts……；The step response data of controlled device will be at some moment
tL=IijTsStart to present and determine slope rising, with the data at the momentFor starting point, data before are denoted as respectivelyJ-th of input is established to the step response model vector a between i-th of outputij：
Wherein T is the transposition symbol of matrix, and δ is step response data between adjacent two data after constant-slope rising
Constant difference, LijFor the model length that j-th of input of setting exports i-th, Lij≥Iij+1；
Step 2. designs the dynamic matrix controller of i-th of intelligent body, specifically：
The 2.1 model vector a obtained using step 1ijThe dynamic matrix of controlled device is established, form is as follows：
Wherein AijP × M rank the dynamic matrix exported to i-th of intelligent body, a are inputted for j-th of intelligent bodyijIt (k) is j-th of input
The step response data exported to i-th, P are the optimization time domain of Dynamic array control algorithm, and M is Dynamic array control algorithm
Control time domain, Lij=L, M<P<L, N are input and output number；
2.2 obtain the model prediction initial communication value y at i-th of intelligent body current k momenti,0(k)
Firstly, controlling increment Δ u is added at the k-1 moment1(k-1),Δu2(k-1),…,Δun(k-1), i-th of intelligent body is obtained
Model predication value yi,P(k-1)：
Wherein,
yi,P(k-1)=[yi,1(k|k-1),yi,1(k+1|k-1),…,yi,1(k+L-1|k-1)]T
yi,0(k-1)=[yi,0(k|k-1),yi,0(k+1|k-1),…,yi,0(k+L-1|k-1)]T,
Aii,0=[aii(1),aii(2),…,aii(L)]T,Aij,0=[aij(1),aij(2),…,aij(L)]T
yi,1(k|k-1),yi,1(k+1|k-1),…,yi,1(k+L-1 | k-1) respectively indicates i-th of intelligent body at the k-1 moment to k,
The model predication value at k+1 ..., k+L-1 moment, yi,0(k|k-1),yi,0(k+1|k-1),…,yi,0(k+L-1 | k-1) indicate k-1
Moment is to k, k+1 ..., the initial prediction at k+L-1 moment, Aii,0,Aij,0Respectively i-th of intelligent body and j-th of intelligent body are defeated
Enter the matrix established to the step response data of i-th of intelligent body output, Δ u1(k-1),Δu2(k-1),…,Δun(k-1) it is
The input control quantity increment of k-1 moment each intelligent body；
Then, the model predictive error value e of i-th of intelligent body of k moment is obtainedi(k)：
ei(k)=yi(k)-yi,1(k|k-1)
Wherein yi(k) real output value for i-th of intelligent body that the k moment measures is indicated；
Further obtain k moment revised model output value yi,cor(k)：
yi,cor(k)=yi,0(k-1)+h1*ei(k)+h2*ei(k)
Wherein,
yi,cor(k)=[yi,cor(k|k),yi,cor(k+1|k),…,yi,cor(k+L-1|k)]T,
h1=[1, α ..., α]T,h2=[0,1 ..., L-1]T
yi,cor(k|k),yi,cor(k+1|k),…,yi,cor(k+L-1 | k) respectively indicate i-th of intelligent body repairing in k moment model
Positive value, h1And h2For the weight matrix of error compensation, α is error correction coefficient, 0<a≤1；
Finally obtain the initial communication value y of the model prediction at i-th of intelligent body k momenti,0(k)：
yi,0(k)=Syi,cor(k)
Wherein, S is the new state-transition matrix of L × L rank,
2.3 obtain i-th of intelligent body in M continuous input control quantity increment Delta u according to step 2.1i(k),Δui(k+
1),…,Δui(k+M-1) the prediction output valve y underi,PM：
Wherein,
yi,PM(k)=[yi,M(k+1|k),yi,M(k+2|k),…,yi,M(k+P|k)]T
yi,P0(k)=[yi,0(k+1|k),yi,0(k+2|k),…,yi,0(k+P|k)]T
Δui,M(k)=[Δ ui(k),Δui(k+1),…,Δui(k+M-1)]T
Δuj,M(k)=[Δ uj(k),Δuj(k+1),…,Δuj(k+M-1)]T
yi,P0It (k) is yi,0(k) preceding P, yi,0(k+1|k),yi,0(k+2|k),…,yi,0(k+P | k) it is the k moment to k+1, k+
2 ..., the model prediction output valve at k+P moment；
2.4 establish the performance indicator J of i-th of intelligent body dynamic matrix controller of Nonself-regulating planti(k) and reference locus ωi(k),
Form is as follows：
min Ji(k)=(ωi(k)-yi,PM(k))TQi(ωi(k)-yi,PM(k))+Δui,M(k)TRiΔui,M(k)
ωi(k)=[ωi(k+1),ωi(k+2),…,ωi(k+P)]T
ωi(k+ ε)=βεy(k)+(1-βε) c (k) (ε=1,2 ..., P)
WhereinFor error weighting matrix,To control weighting matrix,WithRespectively Qi,RiIn weight coefficient, ωiIt (k) is the reference locus of i-th of intelligent body, β is
The softening coefficient of reference locus；
2.5 according to Nash optimizations thought, by performance indicator in step 2.4 obtain i-th of intelligent body current k moment receive it is assorted most
Excellent solution：
Wherein：
2.6 obtained by step 2.2 to 2.5 be in the new round iteration optimal solution of k moment intelligent body i：
Whole system is further obtained in the optimal control law at k moment：
Wherein：
ω (k)=[ω1(k),ω2(k),…,ωn(k)]T, yP0(k)=[y1,P0(k),y2,P0(k),…,yn,P0(k)]T
2.7 using the Nash optimization solution first term at i-th of intelligent body k moment as instant control law Δ ui(k), the reality of intelligent body i is obtained
Border control amount ui(k)=ui(k-1)+Δui(k) i-th of intelligent body is acted on；
2.8, in subsequent time, repeat step 2.2 to 2.7 and continue to solve the instant control law Δ u of i-th of intelligent bodyi(k+1), into
And the optimal solution Δ u (k+1) of whole system is obtained, and circuit sequentially.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610539559.0A CN106200379B (en) | 2016-07-05 | 2016-07-05 | A kind of distributed dynamic matrix majorization method of Nonself-regulating plant |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610539559.0A CN106200379B (en) | 2016-07-05 | 2016-07-05 | A kind of distributed dynamic matrix majorization method of Nonself-regulating plant |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106200379A CN106200379A (en) | 2016-12-07 |
| CN106200379B true CN106200379B (en) | 2018-11-16 |
Family
ID=57473231
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610539559.0A Active CN106200379B (en) | 2016-07-05 | 2016-07-05 | A kind of distributed dynamic matrix majorization method of Nonself-regulating plant |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106200379B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106483853A (en) * | 2016-12-30 | 2017-03-08 | 杭州电子科技大学 | The fractional order distributed dynamic matrix majorization method of Heat Loss in Oil Refining Heating Furnace furnace pressure |
| CN109725532B (en) * | 2018-12-24 | 2021-11-16 | 杭州电子科技大学 | Relative distance control and self-adaptive correction method applied to multiple intelligent agents |
| CN111123708B (en) * | 2019-12-30 | 2022-10-18 | 杭州电子科技大学 | Coking furnace hearth pressure control method based on distributed dynamic matrix control optimization |
| CN111506037A (en) * | 2020-05-26 | 2020-08-07 | 杭州电子科技大学 | Dynamic matrix optimization distributed control method for industrial heating furnace system |
| CN112286043B (en) * | 2020-10-13 | 2022-04-19 | 国网浙江省电力有限公司电力科学研究院 | PID parameter setting method based on controlled object step response characteristic data |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103116283A (en) * | 2013-01-18 | 2013-05-22 | 杭州电子科技大学 | Method for controlling dynamic matrix of non-self-balance object |
| CN103605284A (en) * | 2013-11-14 | 2014-02-26 | 杭州电子科技大学 | Dynamic matrix control optimization-based waste plastic cracking furnace pressure controlling method |
| CN103616815A (en) * | 2013-11-14 | 2014-03-05 | 杭州电子科技大学 | Control method for waste plastic oil refining cracking furnace chamber temperature based on dynamic matrix control optimization |
-
2016
- 2016-07-05 CN CN201610539559.0A patent/CN106200379B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103116283A (en) * | 2013-01-18 | 2013-05-22 | 杭州电子科技大学 | Method for controlling dynamic matrix of non-self-balance object |
| CN103605284A (en) * | 2013-11-14 | 2014-02-26 | 杭州电子科技大学 | Dynamic matrix control optimization-based waste plastic cracking furnace pressure controlling method |
| CN103616815A (en) * | 2013-11-14 | 2014-03-05 | 杭州电子科技大学 | Control method for waste plastic oil refining cracking furnace chamber temperature based on dynamic matrix control optimization |
Non-Patent Citations (4)
| Title |
|---|
| Design of dynamic matrix control based PID for residual oil outlet temperature in a coke furnace;Wu S,et al.;《Chemometrics and Intelligent Laboratory Systems》;20141231;第134卷;全文 * |
| Distributed model predictive control of an experimental four-tank system;Mercangöz M,et al.;《Journal of Process Control》;20071231;第17卷(第3期);全文 * |
| Nash-optimization enhanced distributed model predictive control applied to the Shell benchmark problem;Li S,et al.;《Information Sciences》;20051231;第170卷(第2-4期);全文 * |
| 啤酒发酵温度SMITH补偿分布式预测控制算法研究;窦秀华;《中国优秀硕士学位论文全文数据库工程科技I辑》;20091031;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106200379A (en) | 2016-12-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106200379B (en) | A kind of distributed dynamic matrix majorization method of Nonself-regulating plant | |
| Mi et al. | Short‐term power load forecasting method based on improved exponential smoothing grey model | |
| CN106249724B (en) | A kind of polynary molten steel quality forecast Control Algorithm of blast furnace and system | |
| CN105910169B (en) | District heating system regulating of heating net method and system based on mechanism model PREDICTIVE CONTROL | |
| CN105805822A (en) | Heat supply energy saving control method and system based on neural network prediction | |
| CN109143853B (en) | Self-adaptive control method for liquid level of fractionating tower in petroleum refining process | |
| CN117270483B (en) | Full-flow dynamic optimization control method and device for chemical production device and electronic equipment | |
| Lu et al. | Data augmentation strategy for short-term heating load prediction model of residential building | |
| CN106483853A (en) | The fractional order distributed dynamic matrix majorization method of Heat Loss in Oil Refining Heating Furnace furnace pressure | |
| CN108333933A (en) | A kind of single order pure delay system closed-loop identification method | |
| Gao et al. | Design of PID controller for greenhouse temperature based on Kalman | |
| CN105573123A (en) | Thermal power generating unit mechanical furnace coordination control method based on improved T-S fuzzy prediction modeling | |
| CN109656138A (en) | A kind of brewed spirit process temperature intelligent control method based on fuzzy reasoning | |
| CN113361214A (en) | Open channel control model parameter identification method based on water level flow data | |
| Qiao et al. | Decoupling control for wastewater treatment process based on recurrent fuzzy neural network | |
| CN115313403A (en) | Real-time voltage regulation and control method based on deep reinforcement learning algorithm | |
| CN105955014A (en) | Method for controlling coke furnace chamber pressure based on distributed dynamic matrix control optimization | |
| CN116306911A (en) | Distributed machine learning-based thermodynamic station load prediction and optimization control method | |
| CN106444388A (en) | Distributed PID type dynamic matrix control method for furnace pressure of coke furnace | |
| CN113422371B (en) | Distributed power supply local voltage control method based on graph convolution neural network | |
| Yan et al. | Study on prediction model of dissolved oxygen about water quality monitoring system based on BP neural network | |
| Zhu et al. | Structural Safety Monitoring of High Arch Dam Using Improved ABC‐BP Model | |
| CN116415728B (en) | Carbon emission prediction method, system, electronic equipment and storage medium | |
| CN107168066A (en) | A kind of greenhouse self-adaptation control method | |
| Zhang et al. | Economic model predictive control for the operation optimization of water distribution networks with risks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |