CN105759606A - Steady state target robust optimization method and device in view of model mismatch - Google Patents
Steady state target robust optimization method and device in view of model mismatch Download PDFInfo
- Publication number
- CN105759606A CN105759606A CN201510991026.1A CN201510991026A CN105759606A CN 105759606 A CN105759606 A CN 105759606A CN 201510991026 A CN201510991026 A CN 201510991026A CN 105759606 A CN105759606 A CN 105759606A
- Authority
- CN
- China
- Prior art keywords
- mismatch
- gain
- model
- determined
- steady
- 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
-
- 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
- G05B13/048—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 using a predictor
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 steady state target robust optimization method and a device in view of model mismatch. The method comprises steps: firstly, a transfer function mismatch model for describing a relationship among a prediction error for a controlled variable, a control variable increment and a mismatch gain is built, the mismatch gain in the mismatch model is identified so as to determine a range set for the mismatch gain; then, according to the range set for the mismatch gain and a steady state gain for the control model, a range set for overall gains for the control model is determined, and thus, based on the range set for the overall gains, a preset optimization algorithm is used for determining the optimal steady state target of the controller. Compared with the prior art, the method and the device of the invention consider a mismatch condition exists in an object model and the controller model when the optimal steady state target is optimized, and thus, the stability and the robustness of the model prediction controller are thus improved.
Description
Technical field
The application relates to Model Predictive Control field, more particularly, it relates to a kind of steady-state target robust Optimal methods for model mismatch and device.
Background technology
Model predictive controller, as typical case's realization of Dynamic matrix control, is used widely in industries such as oil refining, chemical industry, papermaking, is brought considerable economic benefit and social benefit to enterprise.The model predictive controller of current main-stream adopts the double-decker solution that steady-state optimization layer combines with dynamic key-course.Referring to seeing that Fig. 1 illustrates the structural representation of a kind of model predictive controller.Controller model adopts linear model, such as transfer function model, state-space model and step response model.Wherein, steady-state optimization layer is according to controller model steady-state gain and current open loop steady-state value, optimization meets the optimum steady-state target of current constraints, control target as the dynamic key-course of lower floor, dynamic key-course is according to controlling target and open-loop prediction, optimization meets the controlled quentity controlled variable of current control constraints, is applied to control object finally by dcs.
Current model predictive controller is when optimizing optimum steady-state target, controller model steady-state gain is used as active procedure object real gain and carrys out Optimization Solution, do not account for object model and controller model and there is the situation of mismatch, thus the fluctuation of optimization aim can be caused, reduce stability and the robustness of model predictive controller.
Summary of the invention
In view of this, the application provides a kind of steady-state target robust Optimal methods for model mismatch and device, to improve stability and the robustness of model predictive controller.
To achieve these goals, it is proposed that scheme as follows:
A kind of steady-state target robust Optimal methods for model mismatch, including:
Set up the transmission function mismatch model being used for the relation between the forecast error of controlled variable, control variable increment and mismatch gain that describes;
Utilize default identification algorithm that the mismatch gain in described transmission function mismatch model is carried out identification, it is determined that mismatch gain estimated value;
Using described mismatch gain estimated value as the independent random variable meeting standard normal distribution, it is determined that the scope set of mismatch gain, wherein said range set is combined into convex set;
Steady-state gain according to controller model and the scope set of described mismatch gain, it is determined that the overall gain scope set of described controller model;
Based on the scope set of described overall gain, default optimizing algorithm is utilized to determine the optimum steady-state target of controller.
Preferably, the expression formula of described transmission function mismatch model is:
I=1 ..., M
J=1 ..., N
Wherein, EiFor the forecast error of current controlled variable, Δ ujControl variable increment, Δ gijFor control variable increment Delta ujTo controlled variable increment Delta yiMismatch gain, A (s) transmits function denominator, and M is the output number of controlled variable, and N is the input number of control variable.
Preferably, described utilization is preset identification algorithm and the mismatch gain in described transmission function mismatch model is carried out identification, including:
Adopt RECURSIVE DAMPED LEAST SQUARE ALGORITHM that the mismatch gain in described transmission function mismatch model is carried out identification.
Preferably, described employing RECURSIVE DAMPED LEAST SQUARE ALGORITHM also includes before the mismatch gain in described transmission function mismatch model is carried out identification:
Obtain described controlled variable forecast error and described control variable increment;
It is filtered described controlled variable forecast error and described control variable increment processing.
Preferably, the mismatch gain in described transmission function mismatch model is carried out identification by described employing RECURSIVE DAMPED LEAST SQUARE ALGORITHM, it is determined that after mismatch gain estimated value, also include:
Identification described mismatch gain estimated value out is carried out bound protection.
Preferably, described using described mismatch gain estimated value as the independent random variable meeting standard normal distribution, it is determined that the scope set of described mismatch gain, including:
Using described mismatch gain as the independent random variable meeting standard normal distribution, it is determined that the oval scope set of described mismatch gain.
Preferably, described utilization is preset optimizing algorithm and is determined the optimum steady-state target of controller, including:
Based on second order cone constraints, steady-state target is carried out optimizing, it is determined that described optimum steady-state target.
Preferably, described using described mismatch gain estimated value as the independent random variable meeting standard normal distribution, it is determined that the scope set of described mismatch gain, including:
Just described mismatch gain is as independent random variable, it is determined that the square scope set of described mismatch gain.
Preferably, the described forecast error to described controlled variable and described control variable increment are filtered processing, including:
One order inertia filter method is adopted to be filtered processing to forecast error and the described control variable increment of described controlled variable.
A kind of steady-state target robust for model mismatch optimizes device, including:
Mismatch model sets up unit, for setting up the transmission function mismatch model for describing relation between the forecast error of controlled variable, control variable increment and mismatch gain;
Mismatch gain identification unit, for utilizing default identification algorithm that the mismatch gain in described transmission function mismatch model is carried out identification, it is determined that mismatch gain estimated value;
Unit is determined in first scope set, for using described mismatch gain estimated value as the independent random variable meeting standard normal distribution, it is determined that the scope set of described mismatch gain, wherein said range set is combined into convex set;
Unit is determined in second scope set, for the scope set of the steady-state gain according to controller model and described mismatch gain, it is determined that the overall gain scope set of described controller model;
Optimizing unit, for the scope set based on described overall gain, utilizes default optimizing algorithm to determine the optimum steady-state target of controller.
Via technique scheme it can be seen that this application discloses a kind of steady-state target robust Optimal methods for model mismatch and device.The method initially sets up the transmission function mismatch model for describing relation between the forecast error of controlled variable, control variable increment and mismatch gain, and then the mismatch gain in mismatch model is carried out identification, to determine the scope set of mismatch gain.Then, the steady-state gain according to the scope set of mismatch gain and Controlling model, it is determined that the scope set of Controlling model overall gain, thus based on the scope set of overall gain, utilizing default optimizing algorithm to determine the optimum steady-state target of controller.Compared with prior art, the present invention considers that when to optimum steady-state target optimizing object model and controller model exist the situation of mismatch, thus improves stability and the robustness of model predictive controller.
Accompanying drawing explanation
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, the accompanying drawing used required in embodiment or description of the prior art will be briefly described below, apparently, accompanying drawing in the following describes is only embodiments of the invention, for those of ordinary skill in the art, under the premise not paying creative work, it is also possible to obtain other accompanying drawing according to the accompanying drawing provided.
Fig. 1 illustrates the structural representation of a kind of model predictive controller;
Fig. 2 illustrates that one embodiment of the invention discloses the schematic flow sheet of a kind of steady-state target robust Optimal methods for model mismatch;
Fig. 3 illustrates the schematic flow sheet of a kind of mismatch gain discrimination method disclosed by the invention;
Fig. 4 illustrates the structural representation of the device for identifying of mismatch gain disclosed in another embodiment of the present invention.
Detailed description of the invention
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is only a part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, the every other embodiment that those of ordinary skill in the art obtain under not making creative work premise, broadly fall into the scope of protection of the invention.
The steady-state optimization layer of model predictive controller is the layer of structure of the multiple-input and multiple-output of a N number of M output of input, and its steady-state equation is:
Wherein, Δ us∈RMFor control variable stable state increment, Δ ys∈RMFor controlled volume stable state increment.gijRepresent the jth controlled quentity controlled variable steady-state gain to i-th controlled variable, G ∈ RM×NFor steady state gain matrix.
This steady-state equation generally carrying out LP (LinearProgramming, linear programming) at present optimize, its optimization method is:
s.t.
ΔuLj≤Δusj≤ΔuHj(1)
ΔyLi-εLi≤Δysi≤ΔyHi+εHi(2)
εLi,εHi,≥0
Δysi=GiΔus
Gi=[gi1,...,giN]
Δus=[Δ us1,...,ΔusN]T
J=1 ..., M
I=1 ..., N
Wherein, wherein, c ∈ RNFor each controlled quentity controlled variable Δ usjUnit optimizes cost, wH∈RM、wL∈RMFor slack variable εH∈RM、εL∈RMPunishment weights, constraints (1) is current control variable ujStable state increment restriction, constraints (2) is current controlled variable yiStable state increment restriction.So, Optimization Solution Δ u under current constraints is carried out by calling LP Optimization Solution devicesWith Δ ysSteady-state value, i.e. optimal objective steady-state value.
By above-mentioned optimization problem it can be seen that about Δ ysiConstraints carrys out Optimization Solution steady-state target increment Delta u by controller model steady-state gain as active procedure object real gainsWith Δ ys, it does not have considering model gain mismatch problems, stability and the robustness of its controller are relatively low.For this, the invention discloses kind the steady-state target robust Optimal methods for model mismatch and device, to improve stability and the robustness of controller.
Referring to Fig. 2, the present invention illustrates that one embodiment of the invention discloses the schematic flow sheet of a kind of steady-state target robust Optimal methods for model mismatch.
As shown in Figure 2, the method includes:
S1: set up the transmission function mismatch model being used for the relation between the forecast error of controlled variable, control variable increment and mismatch gain that describes.
Set between controlled variable separate, controlled variable yiMultiple input single output transmission function mismatch model be:
I=1 ..., M
J=1 ..., N
Wherein, EiFor the forecast error of current controlled variable, Δ ujControl variable increment, Δ gijFor control variable increment Delta ujTo controlled variable increment Delta yiMismatch gain, A (s) transmits function denominator, and M is the output number of controlled variable, and N is the input number of control variable.
Described A (s)=s2+a1s+a2, wherein, a1And a2For undetermined coefficient, and mismatch gain delta gijParticipate in identification together to estimate;S is transmission letter numerical representation symbol.
S2: utilize default identification algorithm that the mismatch gain in described transmission function mismatch model is carried out identification, it is determined that mismatch gain estimated value.
It should be noted that and can adopt multiple identification algorithm that the mismatch gain in transmission function mismatch model is carried out identification in the present invention.
S3: using described mismatch gain estimated value as the independent random variable meeting standard value normal distribution, it is determined that the scope set of mismatch gain, wherein said range set is combined into convex set.
Due to these impacts of time variation of the noise of engineering site, not measurable disturbance and process object, the mismatch gain estimated according to above-mentioned identification algorithm is current true mismatch gain approximate evaluation, and its confidence probability is α ∈ (0.5,1).From the angle of probability statistics, we can the mismatch gain delta g estimatedijRegarding the independent random variable meeting standard normal distribution as, this stochastic variable changes in a scope set, such as oval scope set or square scope set.
Oval range set cooperation is the scope set describing mismatch gain by we in the present embodiment.
Wherein, mismatch gain delta gijOval scope set be described as:
Δgij∈ Θ={ Ω δijξ,||ξ||2≤1}
Ω=Φ-1(α)
Wherein, δijIt is Δ gijStandard variance, Φ (x) is the cumulative distribution function of Standard Normal Distribution, Φ-1X () is the inverse function of this cumulative distribution function, α ∈ (0.5,1) is Δ gijConfidence probability.Ω is relevant with this ellipse radii, R=Ω δijHaving together decided on the radius of this ellipse, stochastic variable ξ ∈ [-1,1], confidence probability α ∈ (0.5,1) ensure that the set of this ellipse is a convex set.
S4: the scope set according to the steady-state gain of controller model and described mismatch gain, it is determined that the overall gain scope set of described controller model.
It should be noted that overall gain equal to controller model steady-state gain and mismatch gain and.
S5: based on the scope set of described overall gain, utilizes default optimizing algorithm to determine the optimum steady-state target of controller.
Optionally, in the present embodiment based on second order cone constraints, steady-state target is carried out optimizing, it is determined that described optimum steady-state target.
Detailed process is as follows:
Consider single variable linear constraint:
ax≤b
Assuming that b is the definite value of bounded, a has uncertainty, it is possible to describe with the stochastic variable of normal distribution, and its average isVariance is δ2, the oval range set at place is combined into:
Ω=Φ-1(α)
Wherein, α ∈ (0.5,1) is to ensure that this constraint is a convex set, and corresponding optimization problem is convex optimization.
To above-mentioned constraints, owing to a has uncertainty, it is desirable to constraints to meet a institute's likely value in its oval uncertain set.This constraints is equivalent in this elliptic region and maximizes following quadratic constraints condition:
Thus, Linearly constrained problem is converted into second order cone constraints problem.
Thus, the overall gain ellipse scope set of controller model can be described as:
Here gijIt is controller model steady-state gain, is overall gainNominal value.This uncertain set can be construed to overall gainExist with probability α with steady-state gain gijFor the center of circle, with Ω δijFor in the elliptic region of radius.
Owing to the steady-state optimization layer of controller model is one M the process object exporting N number of input, it is assumed that separate between controlled variable and forecast error, mismatch model gain delta gijBetween Normal Distribution and separate.Through so simplifying, for a multiple-input and multiple-output process object, it is considered to the steady state relation of mismatch gain is:
Δysi=GiΔus+Ω||diΔus||2
Wherein diCan represent with diagonal matrix:
The LP optimization problem of the process object of such a multiple-input and multiple-output is converted to robust LP optimization problem:
s.t.
ΔuLj≤Δusj≤ΔuHj(1)
ΔyLi-εLi≤Δysi≤ΔyHi+εHi(2)
Δysi=GiΔus+Ω||diΔus||2(3)
s.t.
ΔuLj≤Δusj≤ΔuHj(1)
ΔyLi-εLi≤Δysi≤ΔyHi+εHi(2)
εLi,εHi,≥0
Wherein,
Δysi=GiΔus+Ω||diΔus||2
J=1 ..., M
I=1 ..., N
This robust LP optimization problem is a kind of Second-order cone programming (second-orderconeprogram, SOCP) problem, it is possible to carry out Optimization Solution with the relevant solver of SOCP.
As seen from the above embodiment, the control instability problem that this patent causes for model mismatch, has improved model predictive controller, has improved control stability and robustness.At steady-state optimization layer, transmission function mismatch model carrys out estimation mismatch gain.From the angle of probability statistics, using independent random variable as the standard normal distribution meeting confidence degree of the mismatch gain that estimates, the oval uncertain collection of its range of indeterminacy describes.The convex restricted problem of linear space of this stochastic variable bounded is converted into second order cone constraint, thus original linear programming (LP) optimization problem is changed into Second-order cone programming (SOCP) optimization problem, forming robust LP to optimize, last Optimization Solution goes out the robust solution of steady-state target.So, at the steady-state optimization layer of controller it is contemplated that model mismatch, controller robustness and stability are added.
The schematic flow sheet of a kind of mismatch gain discrimination method disclosed by the invention is illustrated referring to Fig. 3.
From the figure 3, it may be seen that the method includes:
The method specifically includes:
S31: obtain described controlled variable forecast error and described control variable increment.
S32: be filtered described controlled variable forecast error and described control variable increment processing.
Optionally, in order to weaken influence of noise and smooth Identification Data, one order inertia is adopted to filter current time forecast error ei(k) and Δ ujK () is filtered.
eif(k)=(1-α) eif(k-1)+αei(k)
Δujf(k)=(1-α) Δ ujf(k-1)+αΔuj(k)
Wherein.K represents current discrete time point.K-1 represents previous time point.K=0,1,2 ..., α ∈ (0,0.5) is filter coefficient.
On-the-spot at engineer applied, due to the impact of not measurable disturbance, inevitable containing not measurable disturbance composition in model predictive error, need for this in forecast error, weaken this impact.Utilize the recurrence formula averaged with variance, first calculate average and the variance of forecast error, it is assumed that forecast error meets Gauss distribution condition, current prediction error value is limited in one times of difference of two squares of average last time.
Wherein M is the E that averagese(k) and varianceUsed sliding window data length, span is 5 to 20.
S33: adopt RECURSIVE DAMPED LEAST SQUARE ALGORITHM that the mismatch gain in described transmission function mismatch model is carried out identification.
In order to ensure the robustness of identification result, meet engineering on-line identification demand, adopting the damped least square method with forgetting factor to carry out identification "current" model mismatch gain, this identification algorithm is under Persistent Excitation, and identification result is the stochastic variable converging on zero-mean, variance bounded.
This identification algorithm at k moment recurrence formula is:
P′0(k-1)=P (k-1)
Wherein, Identification Data vector
Identified parameters vector θ (k)=[a1,a2,Δgi1,...,ΔgiN]。
0.9 < β≤1 is forgetting factor, and its value is more little, forgets speed more fast.
μ >=0.5 is damping factor, adjusts its value and can control the rate of change of θ.
riIt is r1Follow-up vector, r1=[1,0 ..., 0]T。
μ '=(1-β) μ/β, P (0)=IL×LUnit matrix.
L=2+N is the length of identified parameters.
In this discrimination method, P (k) is the covariance matrix estimating parameter, and its initial value is the diagonal matrix of given bigger numerical (such as 1000).In each iterative computation, utilize value and the current data vector of last P (k-1)Calculate the P (k) of current time.After obtaining this value, calculate identified parameters θ (k), and mismatch gain is included in identified parameters.
S34: identification described mismatch gain estimated value out is carried out bound protection.
In engineer applied, engineering staff, according to process object technological requirement, can estimate the gain ranging between controlled quentity controlled variable and controlled volume.According to controller model gain and its bound scope, it is possible to calculate the bound of mismatch gainWithUtilize this bound that mismatch gain is carried out amplitude limit so that it is to meetRequirement.Wherein, when mismatch gain estimated value is if greater than the upper limit then capping value, limit value is then taken off if less than lower limit.
The structural representation of the device for identifying of mismatch gain disclosed in another embodiment of the present invention is illustrated referring to Fig. 4.
As shown in Figure 4, this device includes: mismatch model sets up unit 41, the 42, the first scope set of mismatch gain identification unit determines that unit 44 and optimizing unit 45 are determined in unit the 43, second scope set.
Wherein, described mismatch model sets up unit, for setting up the transmission function mismatch model for describing relation between the forecast error of controlled variable, control variable increment and mismatch gain
Described mismatch gain identification unit, for utilizing default identification algorithm that the mismatch gain in described transmission function mismatch model is carried out identification, it is determined that mismatch gain estimated value.
Unit is determined in described first scope set, for using described mismatch gain estimated value as the independent random variable meeting standard normal distribution, it is determined that the scope set of described mismatch gain, wherein said range set is combined into convex set.
Unit is determined in described second scope set, for the scope set of the steady-state gain according to controller model and described mismatch gain, it is determined that the overall gain scope set of described controller model.
Described optimizing unit, for the scope set based on described overall gain, utilizes default optimizing algorithm to determine the optimum steady-state target of controller.
It should be noted that this device embodiment and said method embodiment adapt, the executive mode of unit is identical with the concrete execution flow process in embodiment of the method, and therefore not to repeat here.
Finally, it can further be stated that, in this article, the relational terms of such as first and second or the like is used merely to separate an entity or operation with another entity or operating space, and not necessarily requires or imply the relation that there is any this reality between these entities or operation or sequentially.And, term " includes ", " comprising " or its any other variant are intended to comprising of nonexcludability, so that include the process of a series of key element, method, article or equipment not only include those key elements, but also include other key elements being not expressly set out, or also include the key element intrinsic for this process, method, article or equipment.When there is no more restriction, statement " including ... " key element limited, it is not excluded that there is also other identical element in including the process of described key element, method, article or equipment.
In this specification, each embodiment adopts the mode gone forward one by one to describe, and what each embodiment stressed is the difference with other embodiments, between each embodiment identical similar portion mutually referring to.
Described above to the disclosed embodiments, makes professional and technical personnel in the field be capable of or uses the present invention.The multiple amendment of these embodiments be will be apparent from for those skilled in the art, and generic principles defined herein can without departing from the spirit or scope of the present invention, realize in other embodiments.Therefore, the present invention is not intended to be limited to the embodiments shown herein, and is to fit to the widest scope consistent with principles disclosed herein and features of novelty.
Claims (10)
1. the steady-state target robust Optimal methods for model mismatch, it is characterised in that including:
Set up the transmission function mismatch model being used for the relation between the forecast error of controlled variable, control variable increment and mismatch gain that describes;
Utilize default identification algorithm that the mismatch gain in described transmission function mismatch model is carried out identification, it is determined that mismatch gain estimated value;
Using described mismatch gain estimated value as the independent random variable meeting standard normal distribution, it is determined that the scope set of mismatch gain, wherein said range set is combined into convex set;
Steady-state gain according to controller model and the scope set of described mismatch gain, it is determined that the overall gain scope set of described controller model;
Based on the scope set of described overall gain, default optimizing algorithm is utilized to determine the optimum steady-state target of controller.
2. method according to claim 1, it is characterised in that the expression formula of described transmission function mismatch model is:
I=1 ..., M
J=1 ..., N
Wherein, EiFor the forecast error of current controlled variable, Δ ujControl variable increment, Δ gijFor control variable increment Delta ujTo controlled variable increment Delta yiMismatch gain, A (s) transmits function denominator, and M is the output number of controlled variable, and N is the input number of control variable.
3. method according to claim 1, it is characterised in that described utilization is preset identification algorithm and the mismatch gain in described transmission function mismatch model is carried out identification, including:
Adopt RECURSIVE DAMPED LEAST SQUARE ALGORITHM that the mismatch gain in described transmission function mismatch model is carried out identification.
4. method according to claim 3, it is characterised in that described employing RECURSIVE DAMPED LEAST SQUARE ALGORITHM also includes before the mismatch gain in described transmission function mismatch model is carried out identification:
Obtain described controlled variable forecast error and described control variable increment;
It is filtered described controlled variable forecast error and described control variable increment processing.
5. method according to claim 3, it is characterised in that the mismatch gain in described transmission function mismatch model is carried out identification by described employing RECURSIVE DAMPED LEAST SQUARE ALGORITHM, it is determined that after mismatch gain estimated value, also include:
Identification described mismatch gain estimated value out is carried out bound protection.
6. method according to claim 1, it is characterised in that described using described mismatch gain estimated value as the independent random variable meeting standard normal distribution, it is determined that the scope set of described mismatch gain, including:
Using described mismatch gain as the independent random variable meeting standard normal distribution, it is determined that the oval scope set of described mismatch gain.
7. method according to claim 6, it is characterised in that described utilization is preset optimizing algorithm and determined the optimum steady-state target of controller, including:
Based on second order cone constraints, steady-state target is carried out optimizing, it is determined that described optimum steady-state target.
8. method according to claim 1, it is characterised in that described using described mismatch gain estimated value as the independent random variable meeting standard normal distribution, it is determined that the scope set of described mismatch gain, including:
Just described mismatch gain is as independent random variable, it is determined that the square scope set of described mismatch gain.
9. method according to claim 4, it is characterised in that the described forecast error to described controlled variable and described control variable increment are filtered processing, including:
One order inertia filter method is adopted to be filtered processing to forecast error and the described control variable increment of described controlled variable.
10. the steady-state target robust for model mismatch optimizes device, it is characterised in that including:
Mismatch model sets up unit, for setting up the transmission function mismatch model for describing relation between the forecast error of controlled variable, control variable increment and mismatch gain;
Mismatch gain identification unit, for utilizing default identification algorithm that the mismatch gain in described transmission function mismatch model is carried out identification, it is determined that mismatch gain estimated value;
Unit is determined in first scope set, for using described mismatch gain estimated value as the independent random variable meeting standard normal distribution, it is determined that the scope set of described mismatch gain, wherein said range set is combined into convex set;
Unit is determined in second scope set, for the scope set of the steady-state gain according to controller model and described mismatch gain, it is determined that the overall gain scope set of described controller model;
Optimizing unit, for the scope set based on described overall gain, utilizes default optimizing algorithm to determine the optimum steady-state target of controller.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510991026.1A CN105759606B (en) | 2015-12-24 | 2015-12-24 | A kind of steady-state target robust Optimal methods and device for model mismatch |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510991026.1A CN105759606B (en) | 2015-12-24 | 2015-12-24 | A kind of steady-state target robust Optimal methods and device for model mismatch |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105759606A true CN105759606A (en) | 2016-07-13 |
CN105759606B CN105759606B (en) | 2019-01-25 |
Family
ID=56342199
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510991026.1A Active CN105759606B (en) | 2015-12-24 | 2015-12-24 | A kind of steady-state target robust Optimal methods and device for model mismatch |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105759606B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109870924A (en) * | 2019-01-18 | 2019-06-11 | 昆明理工大学 | A kind of distributed system control method of the Model Predictive Control using multilayer steady-state target calculation |
CN112068431A (en) * | 2020-09-03 | 2020-12-11 | 广州大学 | Control method, system and device for double time scales and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1652051A (en) * | 2003-12-03 | 2005-08-10 | 费舍-柔斯芒特系统股份有限公司 | Adaptive multivariable process controller using model switching and attribute interpolation |
US20070276512A1 (en) * | 2006-05-25 | 2007-11-29 | Honeywell Asca Inc. | Automatic tuning method for multivariable model predictive controllers |
CN101925866A (en) * | 2008-01-31 | 2010-12-22 | 费希尔-罗斯蒙特系统公司 | Robust adaptive model predictive controller with tuning to compensate for model mismatch |
CN102183699A (en) * | 2011-01-30 | 2011-09-14 | 浙江大学 | Method for model mismatching detection and positioning of multivariate predictive control system in chemical process |
CN102825603A (en) * | 2012-09-10 | 2012-12-19 | 江苏科技大学 | Network teleoperation robot system and time delay overcoming method |
CN104932488A (en) * | 2015-06-30 | 2015-09-23 | 南京工业大学 | Model predictive control performance evaluation and diagnosis method |
-
2015
- 2015-12-24 CN CN201510991026.1A patent/CN105759606B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1652051A (en) * | 2003-12-03 | 2005-08-10 | 费舍-柔斯芒特系统股份有限公司 | Adaptive multivariable process controller using model switching and attribute interpolation |
US20070276512A1 (en) * | 2006-05-25 | 2007-11-29 | Honeywell Asca Inc. | Automatic tuning method for multivariable model predictive controllers |
CN101925866A (en) * | 2008-01-31 | 2010-12-22 | 费希尔-罗斯蒙特系统公司 | Robust adaptive model predictive controller with tuning to compensate for model mismatch |
US20120221124A1 (en) * | 2008-01-31 | 2012-08-30 | Fisher-Rosemount Systems, Inc. | Using autocorrelation to detect model mismatch in a process controller |
CN102183699A (en) * | 2011-01-30 | 2011-09-14 | 浙江大学 | Method for model mismatching detection and positioning of multivariate predictive control system in chemical process |
CN102825603A (en) * | 2012-09-10 | 2012-12-19 | 江苏科技大学 | Network teleoperation robot system and time delay overcoming method |
CN104932488A (en) * | 2015-06-30 | 2015-09-23 | 南京工业大学 | Model predictive control performance evaluation and diagnosis method |
Non-Patent Citations (2)
Title |
---|
梁腾伟: "多变量预测控制系统模型失配评价方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
王学剑: "预测控制模型失配的性能评价与监视", 《中国优秀硕士学位论文全文数据库 基础科学辑》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109870924A (en) * | 2019-01-18 | 2019-06-11 | 昆明理工大学 | A kind of distributed system control method of the Model Predictive Control using multilayer steady-state target calculation |
CN112068431A (en) * | 2020-09-03 | 2020-12-11 | 广州大学 | Control method, system and device for double time scales and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN105759606B (en) | 2019-01-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11921505B2 (en) | Unmanned surface vehicle control method based on switching T-S fuzzy system under DoS attack | |
Weng et al. | Data-driven robust output tracking control for gas collector pressure system of coke ovens | |
Deng et al. | Output-feedback stabilization of stochastic nonlinear systems driven by noise of unknown covariance | |
Zhang et al. | Adaptive output feedback tracking control of stochastic nonlinear systems with dynamic uncertainties | |
Thenozhi et al. | Stability analysis of active vibration control of building structures using PD/PID control | |
CN103064282A (en) | Nonlinear parameter variation (NPV) model identification method | |
Zhu et al. | On a new adaptive sliding mode control for MIMO nonlinear systems with uncertainties of unknown bounds | |
Xia et al. | Adaptive output feedback dynamic surface control of stochastic nonlinear systems with state and input unmodeled dynamics | |
CN106325075A (en) | H-infinity control method for time-delay linear parameter variation discrete system | |
CN105759606A (en) | Steady state target robust optimization method and device in view of model mismatch | |
Mukaidani | Dynamic games for stochastic systems with delay | |
Moon et al. | Control over lossy networks: A dynamic game approach | |
Galuppini et al. | A gain scheduling approach to improve pressure control in water distribution networks | |
He et al. | Lexicographic multi-objective MPC for constrained nonlinear systems with changing objective prioritization | |
CN107272416A (en) | One class Linear Parameter-Varying Systems dynamic quantization H ∞ control methods | |
Liu et al. | Adaptive fuzzy controller design with observer for a class of uncertain nonlinear MIMO systems | |
Van den Broeck et al. | Time optimal MPC for mechatronic applications | |
Laflamme et al. | Application of self-tuning Gaussian networks for control of civil structures equipped with magnetorheological dampers | |
CN114859725A (en) | Self-adaptive event trigger control method and system for nonlinear system | |
Ugrinovskii et al. | Robust filtering of stochastic uncertain systems on an infinite time horizon | |
Ogura et al. | H-infinity control design considering packet loss as a disturbance for networked control systems | |
Kheawhom et al. | Constrained robust model predictive control based on polyhedral invariant sets by off-line optimization | |
Bozer et al. | Hybrid tracking controller with attached tuned mass damper | |
Diamantis et al. | Identification and adaptive control for open channel water flow systems | |
El Heraiki et al. | Robust $ H_ {\infty} $ Controller Design for Active Vibration Suppression with Input Constraints |
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 | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20190514 Address after: 310053 4th Floor, Zone D, No. 309 Liuhe Road, Binjiang District, Hangzhou City, Zhejiang Province Co-patentee after: Zhejiang University Patentee after: Zhejiang SUPCON Software Co., Ltd. Address before: 310053 4th Floor, D District, No. 309 Liuhe Road, Binjiang District, Hangzhou City, Zhejiang Province (High-tech Zone) Patentee before: Zhejiang SUPCON Software Co., Ltd. |