CN105807611B - A kind of model of closed-loop control system and the unmatched detection method of object - Google Patents
A kind of model of closed-loop control system and the unmatched detection method of object Download PDFInfo
- Publication number
- CN105807611B CN105807611B CN201610124989.6A CN201610124989A CN105807611B CN 105807611 B CN105807611 B CN 105807611B CN 201610124989 A CN201610124989 A CN 201610124989A CN 105807611 B CN105807611 B CN 105807611B
- Authority
- CN
- China
- Prior art keywords
- model
- closed
- control system
- loop control
- interference
- 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.)
- Expired - Fee Related
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
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 models of closed-loop control system and the unmatched detection method of object, the closed-loop data of current system is obtained in industrial process normal operation, using the Data Acquisition Model quality index, target value is referred to come the matching degree of detection model and object according to model quality；For model quality index closer to 1, the matching degree of model and object is higher；The influence of the change of the adjustment parameter of the uncontrolled device of model quality index and interference model variation；This method can effectively mismatch model with object and be separated from other factors for influencing control performance, more clearly analyze model and influence of the object mismatch to control system performance.In addition, method using the present invention carries out model and mismatches detection with object, does not need to add in any dynamic excitation signal to the industrial process of normal operation, can detect that model is mismatched with object under industrial process accidental conditions, system maintenance cost is reduced, improves security of system.
Description
Technical field
The invention belongs to technical field of industrial control, more particularly, to the model and object of a kind of closed-loop control system
Unmatched detection method.
Background technology
The performance of closed-loop control system plays critical effect to the product quality and security of system of industrial process,
In past 20 years, a large amount of scholar proposes the performance methodology of different evaluation control systems, for example：With minimum variance index
Assess the performance of closed-loop control system；Using Linear-Quadratic-Gauss (Linear Quadratic Gaussian, LQG) benchmark
The performance of rating system；Using the performance of generalized Hurst index evaluation system.
In System design based on model technology, process model is that reduction control system performance is most important with object mismatch
One of factor.In the prior art, following several method is mismatched for detection model and object：(1) using partial Correlation Analysis
Detection model is mismatched with object；(2) method based on subspace is detected model and the unmatched method of object；(3) it adopts
With the thought of comment, then the transmission function the ratio between model and object being expressed as first in frequency domain removes detection mould with Bode figures
The situation of type and object mismatch；(4) it will be assumed to examine and model combine with the transmission function of the ratio between object, propose a set of reason
It is mismatched by detection model is removed with object；However, the above method be required for adding in into the industrial process of normal operation it is a certain amount of
Dynamic excitation, the normal operation of control system will necessarily be influenced.
Invention content
For the disadvantages described above or Improvement requirement of the prior art, the present invention provides a kind of model of closed-loop control system with
The unmatched detection method of object, its object is to using the routine operation data of closed-loop control system come detection process model with
The mismatch of real process.
To achieve the above object, one side according to the invention, provide a kind of model of closed-loop control system with it is right
It is specific as follows as unmatched detection method：
(1) close loop maneuver data are acquired, and centralization processing is carried out to collected close loop maneuver data；Wherein, closed loop
Operation data includes the output of process of closed-loop system and process input；
(2) according to closed-loop control system structure, by rectangular projection method, interference update is obtained；
(3) it is exported according to reference signal and the above process, obtains the tracking error of closed-loop control system, and the tracking is missed
Difference carries out centralization processing；
(4) it according to the tracking error after above-mentioned interference update and centralization, shrinks and selects by adaptive least absolute value
Operator (adaptive Lasso) method is selected, establishes the interference model of closed-loop control system；
(5) according to the interference model of close loop maneuver data and above-mentioned closed-loop control system, the mould of closed-loop control system is obtained
Type quality variable；
(6) the model quality variable obtained according to the interference update that above-mentioned steps (2) obtain with above-mentioned steps (5) obtains
For detecting the model quality index of the model of closed-loop control system and the unmatched index of object, i.e. closed-loop control system；
(7) according to closed-loop control system structure, the model quality index obtained using above-mentioned steps (6) detects closed loop control
Whether the model of system processed matches with object.
Preferably, above-mentioned steps (1) include following sub-step：
(1.1) setting value is added in into closed-loop control system, which is constant；
(1.2) process input and the output of process of closed-loop control system are obtained；
(1.3) centralization processing is carried out with the output of process to the process input that above-mentioned steps (1.2) obtain.
Preferably, process is inputted according to the following formula and carries out centralization processing：
Centralization processing is carried out to the output of process according to the following formula：
Wherein, t represents t-th of sampling instant, and y (t) represents the output of process of closed-loop control system, and u (t) represents closed loop control
The process input of system processed, N represent sampled data number, and u ' (t) represents the process input after centralization, and y ' (t) represents center
The output of process after change.
Preferably, above-mentioned steps (2) include following sub-step：
(2.1) according to closed-loop control system structure, the high-order autoregression model of the output of process is established, it is specific as follows：
Wherein, M is the order of high-order autoregression model；
(2.2) according to centralization treated the output of process y ' (t), following data matrix is established：
y′P(t)=[y ' (t) ... y ' (t-P+1)] Y 'P(t)=[y 'P(t-1)T … y′P(t-M)T]T
Wherein, P represents that newer data window size, y ' are interfered in estimationP(t) centralization treated the output of process is represented
Tie up matrix, Y ' in the P that y ' (t) is formed × 1P(t) M × P dimension matrixes that centralization treated the output of process y ' (t) is formed are represented；
(2.3) according to above-mentioned high-order autoregression model, by rectangular projection method, the interference renewal vector of estimation is obtained：
Wherein,Represent the interference renewal vector for the estimation that P × 1 is tieed up, wherein,
Preferably, above-mentioned steps (2.3) obtain newer method of interfering, specific as follows：
Preferably, above-mentioned steps (3) include following sub-step：
(3.1) tracking error of closed-loop control system is obtained：E (t)=y (t)-r (t)；
(3.2) centralization processing is carried out to tracking error：
Wherein, e (t) represents tracking error, and e ' (t) represents centralization treated tracking error, and r (t) is setting value.
Preferably, above-mentioned steps (4) include following sub-step：
(4.1) according to closed-loop control system structure, the ARMA model of the tracking error after centralization is obtained,
It represents as follows：
Wherein, z-1Represent delay factor, φ1..., φIRepresent the coefficient of autoregression part, φI+1..., φI+JIt represents to move
The coefficient of dynamic average portion, I represent the maximum order of autoregression part in ARMA model, and J represents that autoregression moves
The maximum order of rolling average part in dynamic averaging model；
Wherein, interference update is the driving noise of the ARMA model；
(4.2) according to above-mentioned interference update and centralization treated tracking error, following data matrix is built：
EW(t)=[e ' (t) ... e ' (t-W+1)]T
Wherein, W represents the window size of sampled data, EW(t) the dimension data matrix of W × 1, H are representedW(t) W × (I+J) is represented
Dimension data matrix；
(4.3) it is shunk and selection opertor method, acquisition ARMA model system according to adaptive least absolute value
Number vector：
Wherein,Represent oneself for the above-mentioned steps (4.1) that adaptive least absolute value is shunk and selection opertor method is estimated
The coefficient vector of moving average model(MA model) is returned, i.e., Represent the above-mentioned step of estimation
Suddenly the coefficient of the ARMA model of (4.1), φ represent the coefficient of practical but unknown auto regressive moving average type to
Amount, φ=[φ1,…,φI+J], λ represents that adaptive least absolute value is shunk and control penalty factor in selection opertor method
Adjustable parameter,Represent j-th of coefficient φ in the coefficient of ARMA modeljWeight factor；
(4.4) according to closed-loop system structure and above-mentioned ARMA model, the interference mould of closed-loop system is obtained
Type, discrete transfer function are expressed as：
Wherein,WithRepresent the coefficient for the ARMA model that above-mentioned steps (4.3) are estimated,
Represent the discrete transfer function of the interference model of estimation.
Preferably, above-mentioned steps (5) are specific as follows：
According to closed-loop control system structure, the model quality variable of closed-loop control system is obtained, is expressed as：
Wherein, ν (t) represents the model quality variable of closed-loop control system,Represent above-mentioned closed-loop control system
Inverse, the d expression interference of the discrete transfer function of the interference model of middle estimation, Gm(z-1) represent the model of above-mentioned closed-loop control system
Discrete transfer function, m represent model.
Preferably, above-mentioned steps (6) are specific as follows：
According to above-mentioned interference update and model quality variable, model quality index η is obtained：
Wherein, η represents model quality index,Represent the newer variance of interference of estimation,Represent the variance of model quality variable, N represents unmatched for detection model and object
Sampled data length.
Preferably, above-mentioned steps (7) include following sub-step：
(7.1) according to closed-loop control system structure, model quality variable ν (t) and the estimation of closed-loop control system are obtained
Interference updateRelationship：
Wherein, Gd(z-1) represent the discrete transfer function of interference model practical in above-mentioned closed-loop system, Gp(z-1) represent
The discrete transfer function of object in above-mentioned closed-loop control system, p represent process, Q (z-1) be closed-loop control system internal model control
The discrete transfer function of device；
Wherein,It is the discrete transfer function for the interference model that above-mentioned steps (4) obtain；Mould after above-mentioned simplification
Type quality variable is with interfering newer relationship to show：Model quality variable and the model of process interference are unrelated；When closed-loop control system
When the model and object matching of system, i.e. Gp(z-1)=Gm(z-1), model quality variable is approximately equal to the interference update of estimation, internal model
The variation of controller on model quality variable in estimation the newer relationship of interference without influence；
It (7.3), will be above-mentioned according to the structure of closed-loop control system
(1+Q(z-1)[Gp(z-1)-Gm(z-1)]) it is transformed to following multinomial：
Wherein, above-mentioned 1+Q (z-1)[Gp(z-1)-Gm(z-1)] first term be 1, fiRepresentative polynomial 1+Q (z-1)[Gp(z-1)-
Gm(z-1)] in i-th delay item coefficient；Wherein, model quality variable ν (t) is an infinite order moving average process, interference
UpdateIt is driving noise, specially：
(7.4) it is updated according to the model quality variable ν (t) that above-mentioned steps (7.3) obtain and interferenceRelationship, obtain
The value range of model quality index η, η ∈ (0,1]；Model quality index η closer to 1, the model of closed-loop control system and
The matching degree of object is higher；Model quality quality index closer to 0, get over by the model of closed-loop control system and the matching degree of object
It is low.
In general, by the above technical scheme conceived by the present invention compared with prior art, it can obtain down and show
Beneficial effect：
(1) model of closed-loop control system provided by the invention and the unmatched detection method of object, in industrial process just
Often in the case of operation, the close loop maneuver data of current system are acquired；Utilize close loop maneuver Data Acquisition Model quality index；Root
Refer to target value according to model quality and come whether detection model matches with object, and the process model quality of current system can be characterized to control
The influence of system performance processed；
Using method provided by the invention in the matching degree of detection model and object, since model quality variable is with estimating
The shadow of the newer relationship of interference of meter, the change of the adjustment parameter of the uncontrolled device of model quality index and interference model variation
It rings；Relative to the performance indicator of other closed-loop control systems, due to model quality index proposed by the present invention and the tune of controller
The variation for saving the factor and interference model is unrelated, and method proposed by the present invention can be effectively the mismatch of model and object from other
It influences to separate in the factor of control performance, more accurately knows influence of the mismatch of model and object to system performance；
(2) model of closed-loop control system provided by the invention and the unmatched detection method of object, due to only with normal
The close loop maneuver data of rule, using adapting to, least absolute value is shunk and selection opertor method directly acquires the dry of general process interference
Model is disturbed, the model quality index of closed-loop control system is directly acquired further according to the structure of closed-loop control system, without right
Current closed-loop control system makees any adjustment, in industrial process normal course of operation, does not need to add in any dynamic excitation letter
Number, just can detect whether model matches with object under industrial process accidental conditions, so as to reduce system maintenance into
This, improves the product quality and security of system of industrial process.
Description of the drawings
Fig. 1 is the overall flow schematic diagram that model provided in an embodiment of the present invention mismatches detection with object；
Fig. 2 is the closed-loop control system structure diagram of the embodiment of the present invention；
Fig. 3 is that the model of the embodiment of the present invention mismatches the sub-process schematic diagram of detection with object；
Fig. 4 is the agitator tank liquid mixed process schematic diagram of the embodiment of the present invention.
Specific embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to the accompanying drawings and embodiments, it is right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.As long as in addition, technical characteristic involved in the various embodiments of the present invention described below
It does not constitute a conflict with each other and can be combined with each other.
The model of closed-loop control system provided by the invention and the unmatched detection method of object, flow are as shown in Figure 1
Meaning, it is specific as follows：
(1) close loop maneuver data are acquired, and centralization is carried out to collected process input data and the output of process data
Processing；
(2) according to closed-loop control system structure, by rectangular projection method, interference update is obtained；
(3) it is exported according to reference signal and the above process, obtains the tracking error of closed-loop control system, and the tracking is missed
Difference carries out centralization processing；
(4) it according to the tracking error after above-mentioned interference update and centralization, shrinks and selects by adaptive least absolute value
Operator Method is selected, establishes the interference model of closed-loop control system process；
(5) according to the interference model of above-mentioned close loop maneuver data and closed-loop system, the model matter of closed-loop control system is obtained
Quantitative change amount；
(6) according to above-mentioned interference update and the relationship of model quality variable, the mould for detecting closed-loop control system is obtained
Type and the unmatched index of object, i.e. model quality index；
(7) according to closed-loop control system structure, using above-mentioned model quality index, detect the model of closed-loop control system with
Whether object matches；
Wherein, the flow of detection as schematically shown in Figure 2, if first determining whether model quality index is equal to 1 model quality
Index is equal or close to 1, then it is assumed that the model and object of closed-loop control system are matched, conversely, being then unmatched.
It is closed-loop control system structure chart of the present invention shown in Fig. 3, u (t), y (t) represent closed-loop control respectively
The process input of system and the output of process, d (t) represent the external interference of closed-loop control system；According to closed-loop control system structure
Figure, obtains the controller of general closed-loop control system and the relationship of internal mode controller, specific as follows：
Wherein, Gc(z-1) represent general closed-loop control system controller discrete transfer function, can be based on model
PID controller or MPC controller；
Below using agitator tank mixed process as embodiment, to model provided by the invention and the unmatched detection method of object
It is further described.
In embodiment, as schematically shown in Figure 4, detailed process is as follows for the mixed process of agitator tank：Liquid passes through one first
Pipeline enters agitator tank, is then sufficiently mixed in agitator tank, is finally flowed out from the exit of agitator tank.
It is according to mass conservation law, the mixed process differential equation of liquid is as follows：
Wherein, cout(t) concentration in tank diameter exit, c are representedin(t) concentration at entrance is represented, q (t) is represented
Flow in pipeline, τ represent that liquid flows into the delay of agitator tank.
When the flow in pipeline is constant, the above-mentioned differential equation is considered as single order time delay process, the mixing of liquid
Journey is specially：
Wherein, Gp(s) the continuous transmission function of agitator tank mixed process is represented, s represents Laplace operator.
In embodiment, sampling time 1min, the process transmission function after discretization is：
The concentration c in agitator tank exitout(t) it can be disturbed by a coloured noise, coloured noise is as follows：
Wherein, d (t) represents the coloured noise in agitator tank exit,The parameter of interference model, real in the interference of expression process
It applies in example,It is 0.95；It is zero that interference update ε (t), which is mean value, and variance is 0.25 white noise.
In embodiment, the discrete transfer function of interference model is：
In embodiment, agitator tank exit concentration cout(t) reference signal is r (t)=10.
In embodiment, the mixed process of agitator tank is controlled using the PI controllers based on model.According to the process of acquisition
The discrete transfer function of model obtains the PI controllers of agitator tank mixed process, and the operation for stablizing it is specific as follows：
First, the continuous transmission function of acquisition process model, it is specific as follows：
Wherein, Gm(s) the continuous transmission function of process model, K are representedmRepresent the gain of model transfer function, TmRepresent mould
The time constant of type transmission function, τmRepresent the delay of model transfer function；
Secondly, the optimal approximation model of process model is established, it is specific as follows：
Wherein,Represent process model Gm(s) the continuous transmission function of optimal approximation model；
According to internal model control principle and the optimal approximation model of process modelEstablish the Discrete PI control based on model
Device processed, specially：
Wherein, Gc(z-1) represent the Discrete PI controller based on model transmission function, γ represent PI controllers adjusting
The factor；
The system performance of above-mentioned closed-loop control system is evaluated using minimum variance index, minimum variance index calculation formula
It is as follows：
Wherein, η0It is minimum variance index,It is the output variance under LMS control,It is the defeated of real process
Go out variance.
Using model provided by the invention and the unmatched detection method of object to the agitator tank mixed process of embodiment into
The method that row model mismatches detection with object, it is specific as follows：
(1) close loop maneuver data are acquired, obtain process input data and the output of process data after centralization
In closed-loop control system normal operation, the sampling time is set as 1 minute, acquires closed-loop data, the sample of acquisition
Number N is 2000；According to the 2000 groups of process input datas collected, 2000 groups of process input datas after centralization are obtained,
It obtains according to the following formula：
According to the 2000 of acquisition groups of the output of process data, 2000 groups of the output of process data after centralization are obtained, under
Formula obtains：
(2) interference update is obtained
According to centralization treated the output of process data y ' (t), two data matrixes are established, it is as follows：
y′P(t)=[y ' (t) ... y ' (t-P+1)] Y 'P(t)=[y 'P(t-1)T … y′P(t-M)T]T
(3) tracking error after the tracking error and centralization of system is obtained
According to the structure of closed-loop system, the tracking error of system is obtained：E (t)=y (t)-r (t), wherein r (t) are settings
Value；According to the tracking error of acquisition, the tracking error after centralization is calculated：
Wherein, after t=1 ..., 2000, r (t)=10, N=2000, e (t) represent that tracking error, e ' (t) represent centralization
Tracking error；
(4) interference model of closed-loop control system is established
First, using the tracking error after the newer estimated value of interference and centralization of above-mentioned steps (3) acquisition, construction is such as
Lower data matrix：
EW(t)=[e ' (t) ... e ' (t-W+1)]T
Wherein, W represents the window size of sampled data, EW(t) the dimension data matrix of W × 1, H are representedW(t) W × (I+J) is represented
Dimension data matrix；Wherein, t=1950, I=20, J=20, W=1930；
Finally, formula is utilizedThe discrete transfer function of interference model is obtained, wherein,WithRepresent the coefficient of the ARMA model of estimation；
(5) the model quality variable of closed-loop control system is obtained
Utilize formulaFirst obtain the discrete transfer function G of modelm(z-1)；
(6) model quality index η is obtained
Utilize formulaThe model quality index of closed-loop control system is obtained, wherein, N=2000；
(7) whether the model of detection closed-loop control system matches with object
Using the structure of closed-loop control system, the value range of model quality index is first obtained：η∈(0,1]；By model matter
For figureofmerit η compared with 1, model quality index η then shows between the model of closed-loop control system and object closer to 1
It is higher with spending；Conversely, then matching degree is lower；
On the other hand, the model quality index of two closed-loop control systems can also be compared, model quality index is bigger
System, the matching degree between model and object are higher；
(8) performance indicator of closed-loop control system is obtained
Obtain the output of process variance under LMS control
Utilize formulaObtain the performance indicator of current closed-loop control system；
According to the performance of the size evaluation system of minimum variance index, if performance indicator η0Smaller, the performance of system is got over
Difference, if performance indicator η0Closer to 1, the performance of system is better.
Then according to regulatory factor γ=2, five different Discrete PI controllers are obtained；Coloured noise d (t) is kept not
Become；Method proposed by the present invention forms closed-loop control system using above-mentioned five different process models, obtains respectively corresponding
Model quality index and system performance index, the results are shown in Table 1：
The model quality index and system performance index of 1 five closed-loop control systems of table
In table 1,Represent the interference mould shunk using adaptive least absolute value and selection opertor method obtains
Type；As shown in Table 1, interference model can correctly be estimated with selection opertor method by being shunk by adaptive least absolute value.
According to the result of table 1 it is found that when mismatching, model quality index proposed by the present invention can be examined effectively when process model exists
Measure the mismatch of process model and object.
When process model isChange closed-loop control system Discrete PI controller adjusting because
Sub- γ, wherein, the value of γ is γ ∈ { 1,2,3,4 }, and method proposed by the present invention has detected closed loop when regulatory factor changes
The model quality index of control system, while the system performance of this five systems is had rated, the results are shown in Table 2：
Model quality index and system performance index when 2 regulatory factor of table changes
| γ | η | η0 |
| 1 | 0.99 | 0.73 |
| 2 | 0.99 | 0.76 |
| 3 | 0.99 | 0.70 |
| 4 | 0.99 | 0.63 |
In table 2, γ represents the regulatory factor of Discrete PI controller；The result of table 2 is shown when process model remains unchanged
When, the change of PI controller regulatory factors does not influence model quality index proposed by the present invention, is maintained at 0.99, illustrates to work as
The process model of preceding closed-loop control system is matched with object；However, the change of PI controller regulatory factors is to entire closed loop
The performance of control system is influential；As shown in Table 2, as γ=2, the performance of entire closed-loop control system is optimal.
When the process model of acquisition isRegulatory factor γ=2 of Discrete PI controller, changed
Coloured noise in journey output, discrete transfer function is as shown in table 3, and method proposed by the present invention has been applied to coloured noise and has changed
In closed-loop control system after change, model quality index and system performance index are as shown in table 3.
Model quality index and system performance index after the transmission function change of 3 coloured noise of table
In table 3, Gd(z-1) it is practical interference model,For the contraction of adaptive least absolute value and selection opertor
The interference model of method estimation；As shown in Table 3, when the model of coloured noise changes, adaptive least absolute value is received
Contracting and selection opertor method can correctly estimate practical interference model；The result of table 3 is shown when process model is kept not
During change, the change of the model transfer function of coloured noise does not influence model quality index proposed by the present invention, is maintained at
0.74, show that the process model of current closed-loop control system and object are unmatched.
As it will be easily appreciated by one skilled in the art that the foregoing is merely illustrative of the preferred embodiments of the present invention, not to
The limitation present invention, all any modification, equivalent and improvement made all within the spirits and principles of the present invention etc., should all include
Within protection scope of the present invention.
Claims (10)
1. a kind of model of closed-loop control system and the unmatched detection method of object, which is characterized in that specifically include following step
Suddenly：
(1) close loop maneuver data are acquired, and centralization processing is carried out to the close loop maneuver data；The close loop maneuver data packet
Include the output of process of closed-loop system and process input；
(2) according to closed-loop control system structure, by rectangular projection method, interference update is obtained；
(3) according to reference signal and the output of process, the tracking error of closed-loop control system is obtained, and to the tracking error
Carry out centralization processing；
(4) it according to the interference update and the tracking error after centralization, is shunk by adaptive least absolute value and selects to calculate
Submethod establishes the interference model of closed-loop control system；
(5) according to the interference model of the close loop maneuver data and the closed-loop control system, the mould of closed-loop control system is obtained
Type quality variable；
(6) according to the interference update and the model quality variable, obtain for detect the model of closed-loop control system with it is right
As the model quality index of unmatched index, i.e. closed-loop control system；
(7) according to closed-loop control system structure, using the model quality index, the model and object of closed-loop control system are detected
Whether match.
2. the model of closed-loop control system as described in claim 1 and the unmatched detection method of object, which is characterized in that institute
It states step (1) and includes following sub-step：
(1.1) setting value is added in into closed-loop control system, the setting value is constant；
(1.2) process input and the output of process of closed-loop control system are obtained；
(1.3) centralization processing is carried out to process input and the output of process.
3. the model of closed-loop control system as claimed in claim 2 and the unmatched detection method of object, which is characterized in that root
Process is inputted according to following formula and carries out centralization processing：
Centralization processing is carried out to the output of process according to the following formula：
Wherein, t represents t-th of sampling instant, and y (t) is the output of process of closed-loop control system, and u (t) is closed-loop control system
Process inputs, and N represents sampled data number, the process input after changing centered on u ' (t), and the process after changing centered on y ' (t) is defeated
Go out.
4. the model of closed-loop control system as claimed in claim 3 and the unmatched detection method of object, which is characterized in that institute
It states step (2) and includes following sub-step：
(2.1) according to closed-loop control system structure, the high-order autoregression model of the output of process is established, it is specific as follows：
Wherein, M is the order of high-order autoregression model；
(2.2) according to centralization treated the output of process y ' (t), following data matrix is established：
y′P(t)=[y ' (t) ... y ' (t-P+1)] YP' (t)=[y 'P(t-1)T … y′P(t-M)T]T
Wherein, P represents that newer data window size, y ' are interfered in estimationP(t) centralization treated the output of process y ' (t) is represented
Tie up matrix, Y in the P of composition × 1P' (t) represents M × P dimension matrixes that centralization treated the output of process y ' (t) is formed；
(2.3) according to the high-order autoregression model, by rectangular projection method, the interference renewal vector of estimation is obtained：
Wherein,Represent the interference renewal vector for the estimation that P × 1 is tieed up, wherein, Represent the interference update of estimation, I is the unit matrix of P × P dimensions.
5. the model of closed-loop control system as claimed in claim 4 and the unmatched detection method of object, which is characterized in that step
Suddenly (2.3) obtain newer method of interfering, specific as follows：
(2.3.1) is to matrixMake QR decomposition, byObtain orthogonal matrixDiagonal matrix R11, diagonal matrix R22With row vector R21；
(2.3.2) is according to orthogonal matrixThe characteristics of, by matrix y 'P(t)YP′(t)TWith matrix YP′(t)YP′(t)T
It represents as follows：
(2.3.3) obtains interference update according to the following formula：
6. the model of closed-loop control system as claimed in claim 3 and the unmatched detection method of object, which is characterized in that institute
It states step (3) and includes following sub-step：
(3.1) tracking error of closed-loop control system is obtained：E (t)=y (t)-r (t)；
(3.2) centralization processing is carried out to tracking error：
Wherein, e (t) represents tracking error, and e ' (t) represents centralization treated tracking error, and r (t) is setting value.
7. the model of closed-loop control system as claimed in claim 4 and the unmatched detection method of object, which is characterized in that institute
It states step (4) and includes following sub-step：
(4.1) according to closed-loop control system structure, the ARMA model of the tracking error after centralization is established, is represented
It is as follows：
Wherein, z-1Represent delay factor, φ1..., φIRepresent the coefficient of autoregression part, φI+1..., φI+JRepresent mobile flat
The coefficient of part, I represent the maximum order of autoregression part in ARMA model, and J represents that autoregression movement is flat
The maximum order of rolling average part in equal model；Wherein, interference update is the driving noise of the ARMA model；
(4.2) according to the interference update and centralization treated tracking error, following data matrix is built：
EW(t)=[e ' (t) ... e ' (t-W+1)]T
Wherein, W represents the window size of sampled data, EW(t) the dimension data matrix of W × 1, H are representedW(t) W × (I+J) dimension is represented
According to matrix；
(4.3) according to adaptive least absolute value shrink and selection opertor method, obtain ARMA model coefficient to
Amount：
Wherein,Represent that adaptive least absolute value is shunk and the autoregression of the above-mentioned steps (4.1) of selection opertor method estimation moves
The coefficient vector of dynamic averaging model, i.e., Above-mentioned autoregression for estimation is moved
The coefficient of averaging model, φ represent the coefficient vector of practical but unknown auto regressive moving average type, φ=[φ1,…,
φI+J], λ represents the adjustable parameter that adaptive least absolute value shrinks and penalty factor is controlled in selection opertor method,It represents
J-th of coefficient φ in the coefficient of ARMA modeljWeight factor；
(4.4) according to closed-loop system structure and above-mentioned ARMA model, the interference model of closed-loop system is obtained,
Its discrete transfer function is expressed as：
Wherein,WithThe coefficient of ARMA model for step (4.3) estimation,Interference for estimation
The discrete transfer function of model.
8. the model of closed-loop control system as claimed in claim 3 and the unmatched detection method of object, which is characterized in that institute
It is specific as follows to state step (5)：
According to closed-loop control system structure, the model quality variable of closed-loop control system is obtained, is expressed as：
Wherein, ν (t) is the model quality variable of closed-loop control system,Estimate in above-mentioned closed-loop control system
The discrete transfer function of interference model it is inverse, d for interference, Gm(z-1) for above-mentioned closed-loop control system model discrete transmission letter
Number, m represent model.
9. the model of closed-loop control system as claimed in claim 8 and the unmatched detection method of object, which is characterized in that institute
It is specific as follows to state step (6)：
According to the interference update and model quality variable, model quality index η is obtained：
Wherein,The newer variance of interference for estimation,For
The variance of model quality variable, N are for detection model and the unmatched sampled data length of object.
10. the model of closed-loop control system as claimed in claim 8 and the unmatched detection method of object, which is characterized in that
The step (7) includes following sub-step：
(7.1) according to closed-loop control system structure, the model quality variable ν (t) of closed-loop control system and the interference of estimation are obtained
UpdateRelationship：
Wherein, Gd(z-1) discrete transfer function for practical interference model in above-mentioned closed-loop system, Gp(z-1) it is above-mentioned closed loop
The discrete transfer function of object in control system, p represent process, Q (z-1) for closed-loop control system internal mode controller it is discrete
Transmission function；
(7.2) according to Gd(z-1) withRelationship, i.e.,Model quality variable is updated with interference
Relationship be reduced to：
Wherein,It is the discrete transfer function for the interference model that above-mentioned steps (4) obtain；Model matter after above-mentioned simplification
Quantitative change amount is with interfering newer relationship to show：Model quality variable and the model of process interference are unrelated；When closed-loop control system
When model is with object matching, i.e. Gp(z-1)=Gm(z-1), model quality variable is approximately equal to the interference update of estimation, internal model control
The variation of device on model quality variable in estimation the newer relationship of interference without influence；
(7.3) according to the structure of closed-loop control system, by above-mentioned (1+Q (z-1)[Gp(z-1)-Gm(z-1)]) be transformed to：
Wherein, 1+Q (z-1)[Gp(z-1)-Gm(z-1)] first term be 1, fiRepresentative polynomial 1+Q (z-1)[Gp(z-1)-Gm(z-1)] in
The coefficient of i-th delay item；Model quality variable ν (t) is infinite order moving average process, interference updateIt is that driving is made an uproar
Sound；
(7.4) it is updated according to the model quality variable ν (t) that step (7.3) obtains and interferenceRelationship, obtain model quality
The value range of index η, η ∈ (0,1]；Model quality index η is closer to 1, the model of closed-loop control system and of object
It is higher with spending；For model quality quality index closer to 0, the model of closed-loop control system and the matching degree of object are lower.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610124989.6A CN105807611B (en) | 2016-03-05 | 2016-03-05 | A kind of model of closed-loop control system and the unmatched detection method of object |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610124989.6A CN105807611B (en) | 2016-03-05 | 2016-03-05 | A kind of model of closed-loop control system and the unmatched detection method of object |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105807611A CN105807611A (en) | 2016-07-27 |
| CN105807611B true CN105807611B (en) | 2018-06-12 |
Family
ID=56467660
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610124989.6A Expired - Fee Related CN105807611B (en) | 2016-03-05 | 2016-03-05 | A kind of model of closed-loop control system and the unmatched detection method of object |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105807611B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108536127B (en) * | 2018-04-20 | 2019-08-13 | 华中科技大学 | A kind of model mismatch diagnostic method of the multivariable control system of data-driven |
| CN108646553A (en) * | 2018-04-20 | 2018-10-12 | 华中科技大学 | A method of statistics on-line monitoring closed-loop control system model quality |
| CN109345473B (en) * | 2018-09-12 | 2021-04-13 | 南京医科大学 | Image processing method based on self-adaptive fast iterative shrinkage threshold algorithm |
| CN110045716B (en) * | 2019-04-18 | 2020-08-11 | 中广核工程有限公司 | Method and system for detecting and diagnosing early fault of closed-loop control system |
| CN111176155B (en) * | 2019-12-20 | 2021-07-02 | 华中科技大学 | Process model mismatch detection method of closed-loop model predictive control system |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3966115B2 (en) * | 2002-08-05 | 2007-08-29 | 株式会社デンソー | Design support apparatus, design support method, and design support program |
| US7421374B2 (en) * | 2005-11-17 | 2008-09-02 | Honeywell International Inc. | Apparatus and method for analyzing model quality in a process control environment |
| CN104932488B (en) * | 2015-06-30 | 2017-06-16 | 南京工业大学 | A kind of Model Predictive Control Performance Evaluation and diagnostic method |
-
2016
- 2016-03-05 CN CN201610124989.6A patent/CN105807611B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN105807611A (en) | 2016-07-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105807611B (en) | A kind of model of closed-loop control system and the unmatched detection method of object | |
| WO2016155241A1 (en) | Method, system and computer device for capacity prediction based on kalman filter | |
| CN107561503A (en) | A kind of adaptive target tracking filtering method based on the Multiple fading factor | |
| CN108920863B (en) | Method for establishing energy consumption estimation model of robot servo system | |
| CN105700358B (en) | A kind of modeling quality control method of the model predictive controller of band drift interference | |
| CN108536127B (en) | A kind of model mismatch diagnostic method of the multivariable control system of data-driven | |
| CN108646553A (en) | A method of statistics on-line monitoring closed-loop control system model quality | |
| CN109916388A (en) | Fiber Optic Gyroscope Temperature Drift compensation method based on wavelet de-noising and neural network | |
| Umlauft et al. | Learning stochastically stable Gaussian process state–space models | |
| CN113758890A (en) | Gas concentration calculation method, device, equipment and storage medium | |
| CN107273659A (en) | A kind of trajectory predictions method for space junk photoelectric tracking based on RANSAC algorithm improvements | |
| CN108984851B (en) | Weighted Gaussian model soft measurement modeling method with time delay estimation | |
| Pereira et al. | Adaptive RBFNN versus conventional self-tuning: comparison of two parametric model approaches for non-linear control | |
| CN106569982A (en) | GPR online soft measurement method and system with singular point detection compensation function | |
| CN108827905B (en) | near-infrared model online updating method based on local weighting L asso | |
| CN116184830A (en) | Cage type electric throttle valve opening control method | |
| Shan et al. | Finite-time boundedness of state estimation for semi-Markovian jump systems with distributed leakage delay and linear fractional uncertainties | |
| CN108875962A (en) | Core ridge regression on-line study method based on fixed budget | |
| CN110673470B (en) | Industrial non-stationary process soft measurement modeling method based on local weighting factor model | |
| CN110455902B (en) | Method for rapidly calibrating multiple standard samples in environment detection | |
| CN112925207A (en) | Greenhouse environment temperature self-adaption method based on parameter identification | |
| CN111027611A (en) | Fuzzy PLS modeling method based on dynamic Bayesian network | |
| CN113435106A (en) | Method and system for detecting transition mode operation fault | |
| CN113064401A (en) | Closed loop system micro fault detection and estimation method based on data driving | |
| Fan et al. | Monitoring of nonlinear time‐delay processes based on adaptive method and moving window |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180612 Termination date: 20200305 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |