CN101937219B - Embed data driven intelligent control system and the method thereof of hormone regulating and controlling mechanism - Google Patents

Abstract

The present invention relates to a kind of the data-driven intelligent controller and the method thereof that embed hormone regulating and controlling mechanism, the output data of the controlled device of described controller by feedback transducer group compared with the data-oriented as controlled device reference input, obtain controlled device output error, this error inputs conventional controller simultaneously and error threshold passes judgment on unit.Data-driven computing unit is according to the System History service data of history data store unit record, selected current operating state between optimization routine state and cooperate optimization state, to determine the controller that need optimize, adjustment is optimized to its parameter, the original controlled quentity controlled variable calculated passes judgment on the processing of unit via error threshold, optionally add the controlled quentity controlled variable increment of hormone regulating and controlling unit, synthesize final controller and export, send into controlled device and implement to control.Invention increases sensitivity and the accuracy of data drive control, be conducive to controlled device and remain on continual and steady duty.

Description

Embed data driven intelligent control system and the method thereof of hormone regulating and controlling mechanism

Technical field

The invention belongs to automatic control technology field, particularly relate to a kind of data driven intelligent control system and the method thereof that embed hormone regulating and controlling mechanism.

Background technology

Many industrial process stream (as material, chemical industry, pharmacy etc.) are all the set of series of complex processing step.The randomness contained in these processing steps and their dynamic changes in operational process, make it that " input-output " corresponding relation of standard cannot be adopted to set up complete mathematical model with satisfied control needs; Namely allow to obtain more complete model, changing or fluctuation also may appear along with the operation of production run in its inner structure and parameter.On the other hand, existing various advanced control method, majority all needs to be applied on accurate object model, just can obtain good control effects.The uncertainty of complex industrial production procedure and stochastic and dynamic variation characteristic, add the difficulty of application of advanced control method thereon.

In order to solve the contradiction that high quality control demand and object model lack, people attempt the input and output data utilizing controlled process, therefrom extract the information that can reflect controlled process character, recycle these information design controllers, generate control signal and put on controlled process, to reach control effects, this is the control strategy of data-driven.Several controllers based on data-driven existing and control method thereof are suggested at present, as MFA control (ModelFreeAdaptiveControl, MFAC), pseudo-control (UnfalsifiedControl is removed, UC), Iterative feedback tuning (IterativeFeedbackTuning, IFT) etc.It is worth mentioning that, the PID control method of industrial extensive employing, its principle of work also meets the implication of data-driven, also belongs to the one of data drive control method.

In practice, one of subject matter that controller and method thereof based on data-driven exist adopts the dynamic perfromance of the closed-loop system of this controller not good.Although the control strategy of data-driven largely solves the problem that cannot obtain high mass object model; but also just because of not having accurate object model for referencial use; often there will be controlled variable fluctuation large; the phenomenons such as waste increases, the consideration of some data drive control devices to the dynamic performance index such as response time, overshoot of controlled device is also inadequate.In a lot of situation, the control effects based on the controller of data-driven can reach comparatively satisfied control effects (i.e. " controlled "), but also has very large space to the improvement of the closed-loop system dynamic perfromance that it is formed.

The adjustment of various hormones to himself physiological action secreted by biosome internal system has feature fast and accurately.The secernent official of various hormone can according to the slight change of internal milieu, the secretion speed of dynamic adjustments hormone and the opportunity of secretion, make hormone accelerate increment secretion when needs strengthen physiological action, and deceleration decrement secretion when needing to suppress physiological action, even can by affecting other organ, utilize other organ to act on self generation, thus reach the effect indirectly controlling autologous hormones secreting function.These characteristics can abstract formation bionical regulon, is embedded in other controller, carries out structure improvement and function strengthens to them, thus improves the Control platform of these controllers.That is, hormone regulating and controlling mechanism is embedded in the controller based on data-driven, form the data-driven intelligent controller based on hormone regulating and controlling, the dynamic perfromance of data drive control device can be improved, contribute to the riding quality improving controlled industrial process, improve production stability and product quality.

Summary of the invention

Technical matters to be solved by this invention is to provide a kind of data driven intelligent control system and the method thereof that embed hormone regulating and controlling mechanism, it does not rely on the accurate model of industrial process, and can have the controller and control method thereof that are satisfied with dynamic perfromance to a certain extent.When the accurate model of controlled industrial process is difficult to obtain, this controller can utilize the method for data-driven to apply effective closed-loop control to controlled process, utilize the further improvement of bionical hormone regulating and controlling principle realization to closed-loop system dynamic perfromance, to reach the object of complex industrial process being carried out to effective high quality control simultaneously.

The technical solution adopted for the present invention to solve the technical problems is: provide a kind of data driven intelligent control system embedding hormone regulating and controlling mechanism, comprises controlled device reference input, the intelligent controller of hormone regulating and controlling itself, controlled device and feedback transducer group; Described controlled device reference input compares with the output data of controlled device, generates controlled device output error, is connected with the input end of the hormone regulating and controlling intelligent controller of data-driven; The output terminal of described hormone regulating and controlling intelligent controller is connected with controlled device, and described controlled device is connected with feedback transducer group; Described hormone regulating and controlling intelligent controller comprises conventional controller, hormone regulating and controlling unit, error threshold judge unit, data-driven computing unit, the original output of controller and history data store unit.Described controlled device output error is connected with the input end of conventional controller, the output terminal of conventional controller is connected with hormone regulating and controlling unit input end, the output terminal of hormone regulating and controlling unit exports with controller respectively and is connected with the input end of data-driven computing unit, output terminal and the error threshold of data-driven computing unit are passed judgment on unit and are connected, history data store unit is passed judgment on unit with data-driven computing unit and error threshold respectively and is connected, the output terminal of described feedback transducer group is connected with the comparer of history data store unit with controlled device reference input respectively.

A kind of a kind of method embedding the data-driven intelligent controller of hormone regulating and controlling mechanism used described in claim 1, the data of described controlled device by feedback transducer group compared with the data-oriented as controlled device reference input, obtain controlled device output error, this error inputs conventional controller simultaneously and error threshold passes judgment on unit, data-driven computing unit is according to the system history data of history data store unit record, selected current operating state between optimization routine state and cooperate optimization state, to determine the controller that need optimize, adjustment is optimized to its parameter, the original controlled quentity controlled variable calculated passes judgment on the processing of unit via error threshold, optionally add the controlled quentity controlled variable increment of hormone regulating and controlling unit, synthesize final controller to export, send into controlled device and carry out speed governing.

The job step of data-driven intelligent controller under optimization routine state of described embedding hormone regulating and controlling mechanism comprises:

(1) system electrification, controller is started working;

(2) optimization aim of parameter as data-driven computing unit of conventional controller in described intelligent controller is selected;

(3) within each sampling period, record controlled device corresponding to this sampling period and export y (k), controlled device reference input y in real time _dk (), conventional controller export u (k) and controlled device output error e (k), and be stored in history data store unit;

(4) within a kth sampling period, according to the calculation procedure of described data drive control device, calculate in this sampling period, the estimated value of conventional controller parameter and obtain conventional controller original output u under this estimates of parameters _origin(k); Wherein k>=1;

(5) within a kth sampling period, according to the operation law of described hormone regulating and controlling unit, u is utilized _origin(k), u (k-1), e (k), e (k-1) formation controller output increment Δ u _origink (), concrete grammar is

${\mathrm{\Δu}}_{\mathrm{origin}}\left(k\right)={\mathrm{\Δu}}_c\left(k\right)+\mathrm{\αg}(\frac{{\left(\right|{\mathrm{\Δu}}_f(k-1)\left|\right)}^n}{\mathrm{\λ}+{\left({\mathrm{\Δu}}_f(k-1)\right|)}^n}+\mathrm{\β}){\mathrm{gL}}_1{\mathrm{gL}}_2$

${\mathrm{\Δu}}_c\left(k\right)=G_{\mathrm{ctrl}}({\widehat{\mathrm{\θ}}}_k,e\left(k\right))-G_{\mathrm{ctrl}}({\widehat{\mathrm{\θ}}}_{k-1},e(k-1))$

Wherein Δ u _c(k) in the kth sampling period and (k-1) individual sampling period in, namely described data drive control device controlled quentity controlled variable export difference;

(6) according to current error condition, the timing input of selection control output increment, the controlled quentity controlled variable that formation controller is final, method is

$u_{\mathrm{final}}\left(k\right)=\left\{\begin{array}{ccc}u(k-1)+{\mathrm{\Δu}}_{\mathrm{origin}}\left(k\right)& \mathrm{if}& e\left(k\right)>{\mathrm{\ϵ}}_{\mathrm{threthold}}\\ u\left(k\right)& \mathrm{if}& e\left(k\right)\≤{\mathrm{\ϵ}}_{\mathrm{threthold}}\end{array}\right.$

Wherein ε _thretholdfor the error threshold preset;

(7) with the final controlled quentity controlled variable u generated _finalk the controller in this sampling period of recording in history data store unit in () replacement step (3) exports u (k), for the next sampling period;

(8) by u _finalk () exports controlled device to, complete control action.

The job step of data-driven intelligent controller under cooperate optimization state of described embedding hormone regulating and controlling mechanism comprises:

(1) adjustable parameters of selected conventional controller is as initial data-driven optimization aim, enters optimization routine state; Now the adjustable parameters of hormone regulating and controlling unit puts into operation according to fixing initial value, and choosing of initial value can rely on artificial experience or existing controller parameter automatic setting method to determine;

(2), during system works, error threshold comparing unit, according to the dynamic error level of controlled device, determines whether the optimization aim of the adjustable parameters of hormone regulating and controlling unit as data-driven method be optimized, enter cooperate optimization state in real time; If need to optimize, then hormone regulating and controlling unit is considered as independent control, the data-driven optimization method adopting same conventional controller identical is optimized; Now described intelligent controller changes the structure containing two conventional controllers into, and its controlled quentity controlled variable exports and is expressed as

$u_{\mathrm{final}}\left(k\right)=\left\{\begin{array}{ccc}{p}_{1}u\left(k\right)+p_2{u}_{\mathrm{enhance}}\left(k\right)& \mathrm{if}& e\left(k\right)>{\mathrm{\ϵ}}_{\mathrm{parallel}}\\ u\left(k\right)& \mathrm{if}& e\left(k\right)\≤{\mathrm{\ϵ}}_{\mathrm{parallel}}\end{array}\right.$

p ₁+p ₂＝1，p ₁≥0，p ₂≥0

Wherein, u _enhancek (), under cooperate optimization state, the controlled quentity controlled variable as the hormone regulating and controlling unit of independent control exports; ε _parallelfor the working state of system switching threshold preset; p ₁, p ₂for the scale factor that conventional controller and hormone regulating and controlling unit controls dynamics are distributed;

(3) when controlled device error is less than said system duty switching threshold u _enhancek time (), in time working state of system is switched back optimization routine state, now the Parameter reconstruction of hormone regulating and controlling unit is to its initial value.

The present invention is based on the controller parameter Self-tuning controller that one utilizes simultaneous perturbation stochastic approximation method (SynchronousPerturbationStochasticApproximation, SPSA) to optimize.The method belongs to the one of data drive control device optimization method, mainly comprises following content:

(1) output based on the closed-loop system of described data drive control device is

y(k+1)＝G _inig(k)u(k)

u(k)＝G _ctal(θ _k)e(k)

(1)

e(k)＝y _d(k)-y(k)

θ _k＝{θ _1，k，θ _2，k，...，θ _N，k}

Wherein, y (k) is the output of closed-loop system during kth sampling instant; U (k) is the control signal of described data drive control device; The output error that e (k) is closed-loop system; y _dk () is reference input; G _imgk imaginary transport function that () is controlled process (because exact transfer function cannot obtain, therefore is called " imagination ".In practical application, the inputoutput data of controlled process all can directly obtain, and does not need this transport function, facilitates row to write on herein for expressing); G _ctrl(θ _k) for having the described data drive control device transport function of time-varying parameter; θ _kfor adjustable controller parameter set; N is the number of adjustable controller parameter.

(2) control objectives of described data drive control device is, makes the performance function defined by following formula reach minimum value

J _k(θ _k)＝E{[y(θ _k，k+1)-y _d(k+1)] ²}，k＝1，2，...(2)

Wherein, J _k(θ _k) be about described data drive control device parameter θ _kperformance function; Y (θ _k, k+1) and be the actual output of kth+1 sampling instant closed-loop system under described data drive control device regulation and control; E{} is mathematical expectation.(2) the actual principle of work indicating described data drive control device of formula, namely passes through each sampling time to controller parameter collection θ _kadjustment, setting controller, makes the output of controlled device constantly approach desired output.

For achieving the above object, make controller parameter collection θ _kchange according to following rule

${\widehat{\mathrm{\θ}}}_k={\widehat{\mathrm{\θ}}}_{k-1}-a_k{\hat{g}}_k\left({\widehat{\mathrm{\θ}}}_{k-1}\right)---\left(3\right)$

Wherein, for the estimated value of the controller parameter of a kth sampling instant, a _kit is a scalar factor; be called simultaneous perturbation estimate vector, and have

${\hat{g}}_k(\·)=\{{\hat{g}}_{1,k}(\·),{\hat{g}}_{2,k}(\·),...,{\hat{g}}_{N,k}(\·)\}---\left(4\right)$

This vector visible needs the controller parameter of adjustment to have an independently component for each.For l component (1≤l≤N), its expression formula is

And have

${\hat{J}}_k^{(\±)}={({\hat{y}}_k^{(\±)}-y_d(k+1))}^2$

${\hat{y}}_k^{(\±)}=G_{\mathrm{img}}\left(u{\left(k\right)}^{(\±)}\right)$ (6)

$u{\left(k\right)}^{(\±)}=G_{\mathrm{ctrl}}({\widehat{\mathrm{\θ}}}_{k-1}\±c_k{\mathrm{\Δ}}_k,e\left(k\right))$

Δ _k＝[Δ _1，k，Δ _2，k，...，Δ _N，k] ^T

Wherein, c _kit is an ordered series of numbers gone to zero; Δ _kit is a random number vector meeting the distribution of independent bounded symmetric.Described data drive control device passes through above-mentioned steps constantly adjusting and optimizing controller parameter, to reach object controlled device being formed to effective closed-loop control.

Present invention employs following technical scheme:

1. data drive control device, its principal character is:

(1) described data drive control device does not rely on the mathematical model of controlled device, and uses the input and output information work of controlled device;

(2) described data drive control device uses above-mentioned SPSA method as the optimization method of controller;

(3) described data drive control device is in the present invention as the main body of the data-driven intelligent controller of whole embedding hormone regulating and controlling, complete main controlled quentity controlled variable and generate task, accept the controlled quentity controlled variable regulating and controlling effect of hormone regulating and controlling unit and the controller parameter optimization function of data-driven computing unit simultaneously.

2. hormone regulating and controlling unit, it mainly comprises following content:

(1) scaling is carried out to the original output of controller, can be written as

u _f(k)＝k _f×u(k)(7)

Wherein u (k) is the controlled quentity controlled variable before scaling, and this controlled quentity controlled variable can derive from the output of any type of controller in closed-loop system; u _fk () is the controlled quentity controlled variable after scaling; k _ffor scaling factor.

(2) calculate regulated quantity according to hormone regulating and controlling rule, concrete grammar is:

$f({\mathrm{\Δu}}_f\left(k\right),e\left(k\right))=\mathrm{\αg}(\frac{{\left(\right|{\mathrm{\Δu}}_f(k-1)\left|\right)}^n}{\mathrm{\λ}+{\left(\right|{\mathrm{\Δu}}_f(k-1)\left|\right)}^n}+\mathrm{\β}){\mathrm{gL}}_1{\mathrm{gL}}_2$

Δu _f(k-1)＝u _f(k-1)-u _f(k-2)

$L_1=-\frac{e\left(k\right)}{\left|e\left(k\right)\right|}\·\frac{\mathrm{\Δe}\left(k\right)}{\left|\mathrm{\Δe}\left(k\right)\right|}---\left(8\right)$

$L_2=\frac{{\mathrm{\Δu}}_f(k-1)}{\left|{\mathrm{\Δu}}_f(k-1)\right|}$

Wherein, f () is about Δ u _fthe hormone regulating and controlling rule expression formula of (k) and e (k); α, β, λ are scalar factor, and its value should at least meet as Δ u _f(k-1), when=0, have f ()=0, therefore β is generally zero; L ₁, L ₂for controlling party is to certainty factor, their common guarantee are when error increases, and the controlled quentity controlled variable strengthened by hormone regulation rule changes towards the direction increased, to suppress error fast; When error reduces, controlled quentity controlled variable, towards the direction change reduced, to reduce overshoot, reduces controlled device output pulsation.

3. the data-driven intelligent controller embedding hormone regulating and controlling mechanism is overall, described data drive control device and described hormone regulating and controlling unit are combined, Data-drive mode is utilized to be optimized hormone regulating and controlling unit, utilize the output of hormone regulating and controlling unit to conventional controller to regulate, thus closed-loop control is implemented to controlled device.Described intelligent controller is overall to be passed judgment on unit, data-driven computing unit and history data store unit formed (as shown in Figure 1) by a conventional controller unit, hormone regulating and controlling unit, an error threshold.Its course of work specifically comprises the following steps:

(1) system electrification, controller is started working;

(2) optimization aim of parameter as data-driven computing unit of conventional controller in described intelligent controller is selected;

(3) within each sampling period, this sampling period corresponding y (k), y is recorded _d(k), u (k) and e (k), and be stored in (wherein u (k) is conventional controller output) in history data store unit;

(4) within a kth sampling period (k>=1), according to the calculation procedure of described data drive control device, calculate in this sampling period, the estimated value of conventional controller parameter and obtain conventional controller original output u under this estimates of parameters _origin(k);

(5) within a kth sampling period, according to the operation law of described hormone regulating and controlling unit, u is utilized _origin(k), u (k-1), e (k), e (k-1) formation controller output increment Δ u _origink (), concrete grammar is

(9)

Wherein Δ u _c(k) in the kth sampling period and in (k-1) individual sampling period, the difference that the controlled quentity controlled variable of described data drive control device exports;

(6) according to current error condition, the timing input of selection control output increment, the controlled quentity controlled variable that formation controller is final, method is

$u_{\mathrm{final}}\left(k\right)=\left\{\begin{array}{ccc}u(k-1)+{\mathrm{\Δu}}_{\mathrm{origin}}\left(k\right)& \mathrm{if}& e\left(k\right)>{\mathrm{\ϵ}}_{\mathrm{threthold}}\\ u\left(k\right)& \mathrm{if}& e\left(k\right)\≤{\mathrm{\ϵ}}_{\mathrm{threthold}}\end{array}\right.---\left(10\right)$

Wherein, ε _thretholdfor the error threshold preset;

(7) with the final controlled quentity controlled variable u generated _finalk the controller in this sampling period of recording in history data store unit in () replacement step (3) exports u (k), for the next sampling period;

(8) by u _finalk () exports controlled device to, complete control action.

Conventional controller under data-driven and hormone regulating and controlling unit cooperate optimization, it mainly comprises following content:

(1) optimization routine state: the adjustable parameters of selected conventional controller, as initial data-driven optimization aim, enters optimization routine state.Now the adjustable parameters (being α, β and λ in the present invention) of hormone regulating and controlling unit puts into operation according to fixing initial value, and choosing of initial value can rely on artificial experience or other methods of self-tuning to determine;

(2) cooperate optimization state: during system works, error threshold comparing unit is according to the dynamic error level of controlled device, whether real-time decision is also optimized the adjustable parameters of hormone regulating and controlling unit as the optimization aim of data-driven method, enters cooperate optimization state.If need to optimize, then hormone regulating and controlling unit is considered as independent control, adopts the data-driven optimization method identical with conventional controller to be optimized.Now described intelligent controller changes the structure containing two conventional controllers into, and its controlled quentity controlled variable exports and is expressed as

$u_{\mathrm{final}}\left(k\right)=\left\{\begin{array}{ccc}{p}_{i}u\left(k\right)+p_2{u}_{\mathrm{enhance}}\left(k\right)& \mathrm{if}& e\left(k\right)>{\mathrm{\ϵ}}_{\mathrm{parallel}}\\ u\left(k\right)& \mathrm{if}& e\left(k\right)\≤{\mathrm{\ϵ}}_{\mathrm{parallel}}\end{array}\right.---\left(11\right)$

p ₁+p ₂＝1，p ₁≥0，p ₂≥0

(3) wherein, u _enhancek (), under cooperate optimization state, the controlled quentity controlled variable as the hormone regulating and controlling unit of independent control exports; ε _parallelfor the working state of system switching threshold preset; p ₁, p ₂for the scale factor that conventional controller and hormone regulating and controlling unit controls dynamics are distributed;

(4) when controlled device error is less than said system duty switching threshold u _enhancek time (), in time working state of system should be switched back optimization routine state, described intelligent controller exports controlled quentity controlled variable according to the definition of formula (10).Now the Parameter reconstruction of hormone regulating and controlling unit is to its initial value, and to prevent hormone regulating and controlling unit to be in Optimal State for a long time, parameter constantly changes, and causes final controlled quentity controlled variable concuss, thus causes controlled variable to export deterioration.

The actual generalized application method indicating intelligent controller of the present invention of above-mentioned steps.In fact, in intelligent controller one-piece construction of the present invention, as long as the concrete control module adopted has adjustable parameter (or set of parameter), then the applicable hormone regulating and controlling method based on data-driven of the present invention is optimized.The present invention does not limit the concrete controller type as conventional controller.

Beneficial effect

A kind of data-driven intelligent controller utilizing bio-hormone principle of adjustment and control that the present invention proposes, combine model-free data drive control mechanism and biological regulation mechanism, can under plant model the unknown or inaccurate situation, implement effective closed-loop control, and the controlled quentity controlled variable control method formed by being regulated and controled to inspire by bio-hormone, shorten the response time of closed-loop system, improve the sensitivity and accuracy that control, be conducive to controlled device and remain on continual and steady duty.

Accompanying drawing explanation

Fig. 1 is system architecture diagram of the present invention;

Fig. 2 is algorithm flow chart of the present invention;

Fig. 3 is cooperate optimization process flow diagram of the present invention;

Fig. 4 is hardware principle block diagram of the present invention;

Fig. 5 is the drawing of fiber roller control system interface framework as the specific embodiment of the invention.

In figure:

1: controlled device reference input y _d(k); 2: controlled device output error e (k); 3: the hormone regulating and controlling intelligent controller of data-driven of the present invention; 4: conventional controller (changeable parameters); 5: hormone regulating and controlling unit; 6: controller exports u _final(k) (i.e. u (k)); 7: controlled device; 8: controlled device exports y (k) in real time; 9: feedback transducer group; 10: data-driven computing unit; 11: history data store unit; 12: controller original output u _origin(k); 13: error threshold passes judgment on unit; 14: system electrification starts; 15: the controller parameter that need optimize is selected; 16: record the y (k) in each sampling period, y _d(k), u (k) and e (k); 17: computing controller estimates of parameters 18: calculate u _origin(k); 19: calculate Δ u _origin(k); 20: error threshold is passed judgment on; 21: utilize the original controlled quentity controlled variable of hormone regulating and controlling; 22: generate u _final(k); 23: export controlled quentity controlled variable to controlled device; 24: computing system state switching threshold ε _parallel; 25: working state of system switches to be passed judgment on; 26: optimization routine state; 27: cooperate optimization state; 28: input interface; 29: central processing unit; 30: man-machine interface; 31: output interface; 32: history data repository; 33: draw roll (containing motor); 34: roller speed sensor; 35: roller speed backfeed loop; 36: intelligent controller of the present invention; 37: controlled quentity controlled variable output loop; 38: stretching roller motor controller; 39: drafting assembly casing; 40: stretching roller motor controller exports.

Embodiment

Below in conjunction with specific embodiment, set forth the present invention further.Should be understood that these embodiments are only not used in for illustration of the present invention to limit the scope of the invention.In addition should be understood that those skilled in the art can make various changes or modifications the present invention, and these equivalent form of values fall within the application's appended claims limited range equally after the content of having read the present invention's instruction.

As shown in Figure 1, a kind of data-driven intelligent controller 3 embedding hormone regulating and controlling mechanism of the present invention, passes judgment on unit 13, data-driven computing unit 10 by conventional controller 4, hormone regulating and controlling unit 5, error threshold of a changeable parameters and history data store unit 11 forms.The control system being core with this controller also comprises the signal circuit composition of controlled device 7, feedback transducer group 9 and necessity.When the above-mentioned system cloud gray model as embodiment, draw roll (containing motor) is controlled device 7, its rotating speed by feedback transducer group 9 with as controlled device reference input 1 roller speed given compared with, obtain controlled device output error 2, this error sends into conventional controller 4 on the one hand, be supplied to error threshold on the other hand and pass judgment on unit 13, the standard of final controlled quentity controlled variable is added, also as the standard that controller duty switches as hormone regulating and controlling part.Simultaneously, the system history data that data-driven computing unit 10 records according to history data store unit 11, selected current operating state between optimization routine state and cooperate optimization state, to determine the controller that need optimize, adjustment is optimized to its parameter, the original controlled quentity controlled variable u calculated _origink () passes judgment on the processing of unit 13 via error threshold, optionally add the controlled quentity controlled variable increment Delta u of hormone regulating and controlling unit 5 _origink (), synthesizes final controller and exports 6, send into controlled device 7 (i.e. draw roll) and carry out speed governing.

A kind of data-driven intelligent controller embedding hormone regulating and controlling mechanism of the present invention, its software flow following (as shown in Figure 2):

(1) system electrification, controller is started working;

(2) optimization aim of parameter as data-driven computing unit of conventional controller in described intelligent controller is selected;

(3) within each sampling period, this sampling period corresponding y (k), y is recorded _d(k), u (k) and e (k), and be stored in (wherein u (k) is exported by conventional controller) in history data store unit;

(4) within a kth sampling period (k>=1), according to the calculation procedure of described data drive control device, calculate in this sampling period, the estimated value of conventional controller parameter and obtain conventional controller original output u under this estimates of parameters _origin(k);

(5) within a kth sampling period, according to the operation law of described hormone regulating and controlling unit, u is utilized _origin(k), u (k-1), e (k), e (k-1) formation controller output increment Δ u _origin(k), concrete grammar is;

${\mathrm{\Δu}}_{\mathrm{origin}}\left(k\right)={\mathrm{\Δu}}_c\left(k\right)+\mathrm{\αg}(\frac{{\left(\right|{\mathrm{\Δu}}_f(k-1)\left|\right)}^n}{\mathrm{\λ}+{\left(\right|{\mathrm{\Δu}}_f(k-1)\left|\right)}^n}+\mathrm{\β}){\mathrm{gL}}_1{\mathrm{gL}}_2---\left(12\right)$

${\mathrm{\Δu}}_c\left(k\right)=G_{\mathrm{ctrl}}({\widehat{\mathrm{\θ}}}_k,e\left(k\right))-G_{\mathrm{ctrl}}({\widehat{\mathrm{\θ}}}_{k-1},e(k-1))$

Wherein Δ u _c(k) in the kth sampling period and in (k-1) individual sampling period, the difference that the controlled quentity controlled variable of described data drive control device exports;

(6) according to current error condition, the timing input of selection control output increment, the controlled quentity controlled variable that formation controller is final, method is

$u_{\mathrm{final}}\left(k\right)=\left\{\begin{array}{ccc}u(k-1)+{\mathrm{\Δu}}_{\mathrm{origin}}\left(k\right)& \mathrm{if}& e\left(k\right)>{\mathrm{\ϵ}}_{\mathrm{threthold}}\\ u\left(k\right)& \mathrm{if}& e\left(k\right)\≤{\mathrm{\ϵ}}_{\mathrm{threthold}}\end{array}\right.---\left(13\right)$

Wherein ε _thretholdfor the error threshold preset;

(7) with the final controlled quentity controlled variable u generated _finalk in () replacement step (3), the controller in this sampling period of record exports

U (k), for the next sampling period;

(8) by u _finalk () exports controlled device to, complete control action.

A kind of data-driven intelligent controller embedding hormone regulating and controlling mechanism of the present invention, its cooperate optimization course of work is as follows:

(1) adjustable parameters of selected conventional controller is as initial data-driven optimization aim, enters optimization routine state.Now the adjustable parameters (being α, β and λ in the present invention) of hormone regulating and controlling unit puts into operation according to fixing initial value, and choosing of initial value can rely on artificial experience or other methods of self-tuning to determine;

(2), during system works, error threshold comparing unit, according to the dynamic error level of controlled device, determines whether the adjustable parameters of hormone regulating and controlling unit be also optimized as the optimization aim of data-driven method, enter cooperate optimization state in real time.If need to optimize, then hormone regulating and controlling unit is considered as independent control, the data-driven optimization method adopting same conventional controller identical is optimized.Now described intelligent controller changes the structure containing two conventional controllers into, and its controlled quentity controlled variable exports and is expressed as

$u_{\mathrm{final}}\left(k\right)=\left\{\begin{array}{ccc}{p}_{1}u\left(k\right)+p_2{u}_{\mathrm{enhance}}\left(k\right)& \mathrm{if}& e\left(k\right)>{\mathrm{\ϵ}}_{\mathrm{parallel}}\\ u\left(k\right)& \mathrm{if}& e\left(k\right)\≤{\mathrm{\ϵ}}_{\mathrm{parallel}}\end{array}\right.---\left(14\right)$

p ₁+p ₂＝1，p ₁≥0，p ₂≥0

Wherein, u _enhancek (), under cooperate optimization state, the controlled quentity controlled variable as the hormone regulating and controlling unit of independent control exports; ε _parallelfor the working state of system switching threshold preset; p ₁, p ₂for the scale factor that conventional controller and hormone regulating and controlling unit controls dynamics are distributed;

(3) when controlled device error is less than said system duty switching threshold u _enhancek time (), in time working state of system should be switched back optimization routine state, described intelligent controller exports controlled quentity controlled variable according to the definition of formula (13).Now the Parameter reconstruction of hormone regulating and controlling unit is to its initial value, and to prevent hormone regulating and controlling unit to be in Optimal State for a long time, parameter constantly changes, and causes final controlled quentity controlled variable concuss, thus causes controlled variable to export deterioration.

A kind of data-driven intelligent controller embedding hormone regulating and controlling mechanism of the present invention, its hardware composition comprises a central processing unit 25, input interface 24, output interface 27, history data repository 28 and man-machine interface 26.Wherein

(1) central processing unit 25 adopts kernel to be not less than the flush bonding processor of ARM7 grade or other same treatment abilities, and has at least 2 universal input and output port (GPIO);

(2) input interface 24, output interface 27 have the digital signal interface such as standard USB, RJ45, RS232, RS485, have the industrial standard analog signal interface of the I/O of 4 ~ 20mA electric current, the I/O of 0 ~ 5VDC voltage;

(3) history data repository 28 adopts DRAM as primary memory, adopts non-volatile memory medium (as FLASH memory, 2.5 inches or 3.5 inches of hard disc of computer) storer in support;

(4) man-machine interface 26 has touch liquid crystal display and keyboard.

A kind of data-driven intelligent controller embedding hormone regulating and controlling mechanism of the present invention, its hardware components working power is provided by the external world.

The algorithm flow of described intelligent controller as shown in Figure 2.

The hardware composition frame chart of described intelligent controller as shown in Figure 3.

As shown in Figure 5, the drafting assembly adopted in a kind of fiber production process, by being arranged on draw roll (containing motor) 29, the roller speed sensor 30 of drafting assembly casing 35 inside and forming with the stretching roller motor controller 34 of stretching roller motor type matching, described stretching roller motor controller 34 can be the controller type that the industry such as PLC are commonly used.Be subject to the impact of the factors such as running environment change and mechanical wear, the accurate model of this system is not easy to obtain.When the system is operated, the speed of draw roll sends into intelligent controller 32 (its system architecture, the course of work and hardware composition is illustrated by accompanying drawing 1 ~ 4 respectively) of the present invention by roller speed sensor 30 via roller speed backfeed loop 31, as calculated, the controlled quentity controlled variable generated sends into stretching roller motor controller 34 through controlled quentity controlled variable output loop 33, the latter is according to concrete motor type, generate the motor drive signal corresponding with controlled quentity controlled variable, drive stretching roller motor to carry out speed adjustment.Adopt intelligent controller 32 of the present invention, when drafting assembly accurate model is difficult to obtain, utilize data-driven principle to realize closed speed control, utilize hormone regulating and controlling mechanism to improve speed governing effect simultaneously, to reach the target keeping draw roll speed long-term stability.

Claims (2)

1. one kind embeds the data driven intelligent control system of hormone regulating and controlling mechanism, comprise controlled device reference input (1), controlled device (7), hormone regulating and controlling intelligent controller (3) and feedback transducer group (9), it is characterized in that: described controlled device reference input (1) receives the output data of the controlled device (7) that feedback transducer group (9) is sent here, generate controlled device output error (2) to be connected with the input end of the hormone regulating and controlling intelligent controller (3) of data-driven, the output terminal of described hormone regulating and controlling intelligent controller (3) is connected with controlled device (7), described controlled device (7) is connected with feedback transducer group (9), described hormone regulating and controlling intelligent controller (3) comprises conventional controller (4), hormone regulating and controlling unit (5), error threshold judge unit (13), data-driven computing unit (10), the original output of controller (12) and history data store unit (11), described controlled device output error (2) is connected with the input end of conventional controller (4), the output terminal of conventional controller (4) is connected with the input end of hormone regulating and controlling unit (5), the output terminal of hormone regulating and controlling unit (5) exports (6) respectively and is connected with the input end of data-driven computing unit (10) with controller, output terminal and the error threshold of data-driven computing unit (10) are passed judgment on unit (13) and are connected, history data store unit (11) is passed judgment on unit (13) with data-driven computing unit (10) and error threshold respectively and is connected, the output terminal of described feedback transducer group (9) is connected with the comparer of history data store unit (11) with controlled device reference input (1) respectively.

2. one kind uses a kind of method embedding the data driven intelligent control system of hormone regulating and controlling mechanism described in claim 1, it is characterized in that: the data of described controlled device (7) by feedback transducer group (9) compared with the data-oriented as controlled device reference input (1), obtain controlled device output error (2), this error inputs conventional controller (4) simultaneously and error threshold passes judgment on unit (13), the system history data that data-driven computing unit (10) records according to history data store unit (11), selected current operating state between optimization routine state and cooperate optimization state, to determine the controller that need optimize, adjustment is optimized to its parameter, the original controlled quentity controlled variable calculated passes judgment on the processing of unit (13) via error threshold, optionally add the controlled quentity controlled variable increment of hormone regulating and controlling unit (5), synthesize final controller and export (6), send into controlled device (7) to implement to control.