UA117038C2 - SELF ADJUSTMENT OF REGULATOR TRANSMISSION COEFFICIENT - Google Patents

Abstract

The regulator's self-adjusting gear ratio is related to the energy, chemical and food industries and can be used to control objects that vary widely in gear ratios. Using the method, it is possible to calculate, at a sliding time interval, the current value of the probabilistic characteristics of the control variables of the control object and its model, for example, variances, which are proportional to their transmission coefficients. This eliminates the effect of phase shifts between adjustable signal variables that distort the values of the probabilistic characteristics. Stabilizing the values of these estimates allows you to approximate the transmission coefficient of the model object in the loop of its self-adjustment to the current value of the transmission coefficient of the control object. The transmission coefficient of the model of the object is calculated by the transmission coefficient of the regulator so that their output remains unchanged. The proposed method improves the accuracy of determining the transmission coefficient of the model of the control object when changing the transmission coefficient of the control object and, as a consequence, improving the accuracy of maintaining the controlled variable signal of the control object at a given value.

Description

The invention belongs to the energy, chemical and food industry and can be used to control objects whose transmission coefficient varies over a wide range.

A known method of optimal automatic adjustment of the control system (Patent

of the Russian Federation Mo 2243584 IPC S2O058 13/00 (2000.01), published on December 27, 2004, bull.

Mo36b|, selected as an analogue, which includes the transfer of a closed control system to an open mode, the application of a test step signal with adjustable amplitude and polarity to the input of the object, measurement of the derivative of the output of the object and determination of the characteristic points of the derivative, calculations based on them parameters of the accepted model of the control object (including - the current value of the transfer coefficient of the object model), calculation of the optimal parameters of the controller setting (including - the transfer coefficient of the regulator) and transfer of the system with optimal setting parameters to the operating mode based on the parameters of the model.

The analog and the method that is claimed have the following general feature (action) - calculation of a new value of the regulator transmission coefficient when the transmission coefficient of the object model is changed.

The analog has the following disadvantages: a) low accuracy of stabilization of the adjustable signal variable at the specified value.

Reasons for low accuracy: - when switching the closed system to the open mode, the deviation of the regulated variable signal from the set value, caused by the influence of external disturbances, will not be compensated by changes in the control influence of the regulator; - when a test signal is applied to the input of the control object, additional deviations of the regulated signal variable from the set value occur. b) low accuracy of calculations of the transfer coefficient of the object model.

The reason for the low accuracy is that after opening the system, external disturbances continue to act on the control object, which distort the value of the regulated signal variable (output of the object).

When measuring the derivative output of the object, which requires differentiation of the regulated variable of the signal, these distortions will be amplified. Accordingly, the results of determining the characteristic points and the results of calculating the parameters of the model of the control object, which are determined by these points, are distorted.

The method of self-adjustment of an automatic control system with a mathematical model of an object on a control channel, the transmission coefficient of which changes over time, is the closest to the proposed method, implemented in a self-adjustment system Patent for a utility model OA 36671 IPC (2006) C2058 13/02, published 11/10/2008 , bull. 21). The method provides stabilization at a given value of the adjustable variable signal of the control object with a transmission coefficient that changes over time due to a change in the control action; filtering of adjustable signal variables of the control object and its model from low-frequency components caused by changes in disturbing influences on the control object; calculation on a sliding time interval of current estimates of probabilistic characteristics of filtered regulated signal variables (in particular, variances); calculation of the current value of the difference in estimates of probabilistic characteristics; a change in the transmission coefficient of the object model in the direction of reducing the difference in estimates of probabilistic characteristics down to zero. This method is chosen by the prototype.

The prototype and the claimed method have the following common features (actions): - stabilization at a given value of the adjustable variable signal of the control object with a transmission coefficient that changes over time due to a change in the control action; - filtering of adjustable signal variables of the control object and its model from low-frequency components caused by changes in disturbing influences on the control object; - calculation on a sliding time interval of current estimates of probabilistic characteristics of filtered regulated signal variables (in particular, variances); - calculation of the current value of the difference in estimates of probabilistic characteristics; - a change in the transfer coefficient of the object model in the direction of reducing the difference in estimates of probabilistic characteristics down to zero.

The set of actions listed above, except for the first one, constitutes the essence of the functioning of the circuit of the self-adjustment of the transfer coefficient of the object model.

The prototype has the following drawback - low accuracy of determining the current value of the transmission coefficient of the object model when the transmission coefficient of the control object changes.

Reasons for low accuracy: - the occurrence of a phase shift between the filtered adjustable variables of the control object and its model, which distorts the current value of the difference in estimates of probabilistic characteristics; - a change in the transmission coefficient of the self-adjustment circuit of the transmission coefficient of the object model because when the transmission coefficient of the control object changes.

The essence of the reasons for the low accuracy of the prototype in determining the current value of the transmission coefficient of the object model when the transmission coefficient of the control object changes is as follows.

The first reason. Under ideal conditions, that is, when the dynamic properties of the control object and its model are identical and there are no high-frequency components in the composition of disturbances, the controlled variables of the control object and its model will be in phase after filtering (they will have the same phase and, in particular, signs) . It should be noted that the change in the transmission coefficient of the object does not cause a phase shift of the filtered adjustable signal variables of the control object and its model, because the amplitude of the adjustable signal variables will change, but the phases (signs) of the variables remain the same. In real conditions, the dynamic properties of the object are always approximately reproduced in its model. At the same time, the unequal delays and inertias of the object and the model will cause a phase shift of their filtered, regulated signal variables. In addition, the filtering of adjustable signal variables of the control object and its model is carried out only from low-frequency components caused by external disturbances. But in real conditions, components that have, for example, a step-like character, may appear in the composition of disturbances. They lead to the appearance of high-frequency components in the composition of the regulated variable signal, which are not filtered by high-frequency filters. This will also cause a phase shift between the controlled variables of the control object and its model.

It is important that all considered phase shifts of the filtered adjustable signal variables change the current difference of estimates of the probabilistic characteristics of these variables without changes in the transmission coefficient of the control object. The method described in the prototype does not take into account the nature of the appearance of the current difference in estimates of probabilistic characteristics, namely, it arose as a result of a change in the transmission coefficient of the object or as a result of phase shifts of the filtered regulated variables of the signal.

This reduces the accuracy of determining the transmission coefficient of the control object model.

The second reason. When the transmission coefficient of the control object is reduced, the increment of the regulated signal variables, caused by changes in both disturbances and control influences, decreases. And this leads to a decrease in the absolute values of the current difference in estimates of the probabilistic characteristics of the filtered variables of the object and its model. In the case of an increase in the transmission factor of the object, the current difference, on the contrary, increases. Including, in the prototype during changes

From the transmission coefficient of the object, the transmission coefficient of the self-tuning loop changes. With high transfer coefficients of the self-tuning circuit, the transient processes in it will have a high fluctuation, and with low ones - they will be slow. This, in both cases, reduces the accuracy of determining the transfer coefficient of the control object model.

The basis of the invention, which is claimed, is the task of developing an improved method of self-adjustment of the transmission coefficient of the regulator, in which, by performing new operations and changing the order of execution of known operations, to ensure an increase in the accuracy of determining the transmission coefficient of the model of the control object when the transmission coefficient of the control object changes and, as a result, increasing the accuracy of maintaining the adjustable signal variable of the control object at the specified value.

The task is solved in the method of self-adjustment of the controller transmission coefficient, which includes stabilization at a given value of the adjustable variable of the control object signal, the transmission coefficient of which changes over time, due to the change in the control action of the regulator, filtering of the adjustable variables of the control object signal and its model from of low-frequency components caused by changes in disturbing influences on the control object, calculation on a sliding time interval of current estimates of probabilistic characteristics of the filtered adjustable variables of the control object signal and its model, calculation of the current value of the difference in estimates of probabilistic characteristics, a change in the transfer coefficient of the object model in in the direction of reducing the value of the difference in estimates down to zero, calculating the transfer coefficient of the regulator based on the variable transfer coefficient of the object model, which is distinguished by the fact that the signs of the filtered regulated variables of the signal are additionally determined and, in the case of the coincidence of these signs, the calculation is continued on sliding time interval of current estimates of the probabilistic characteristics of the filtered regulated signal variables, calculation of the current value of the difference of the estimates of the probabilistic characteristics and calculation of the transfer coefficient of the controller according to the variable transmission coefficient of the object model, and in the case of a discrepancy in the signs of the filtered regulated variables of the signal, the estimates of the probabilistic characteristics and the transfer coefficients are stored on previous level, while the estimate of the probabilistic characteristic of the filtered adjustable signal variable of the control object is stabilized at a fixed value, for which the current value of the difference between this estimate and the fixed value is calculated, the current value of the difference is transformed, for example, integrated, and the estimate of the probability is multiplied by the transformed value because the characteristics of the filtered adjustable variable of the control object signal, reducing the difference to zero due to this value, at the same time multiplying the estimate of the probability characteristic by this transformed value ics of the filtered adjustable signal variable of the control object model.

The invention is explained with the help of drawings, where Fig. 1 shows the structural diagram of the system that implements the proposed self-debug method; in Fig. 2: graph of changes in the transmission coefficient o) of the control object, graph of changes in the delay time To of the control object, graphs of changes in the transmission coefficient Kt of the model of the control object, graphs of changes in the regulated variable of the signal UC) of the control object, the mean square deviation of the error adjustment of the main circuit Te.

The system that implements the proposed self-adjustment method includes: the regulator of the main control loop 1, which consists of the first multiplication block 2 (the first part of the regulator, which implements the proportional part of the control algorithm and determines the variable transmission coefficient Kg ), and the second part of the regulator 3 , which implements the inertial part of the algorithm, for example, the integral and differential components in the PiD algorithm; control object 4; first, second, third and fourth setters 5, 28, 30, 31; the first, second, third and fourth algebraic adders 6, 22, 27, 32; control object model 7, which consists of the inertial part of the control object model 8 and the second multiplication block 9, which implements the proportional part of the control object model and determines the variable transfer coefficient t of the model; identical first and second high-frequency filters 10, 11; first and second controlled keys with memory 12, 13; identical first and second calculators of probabilistic characteristics 14, 17, which consist of the first and second quadrator 15, 18, as well as the first and second low-pass filters 16, 19; the third and fourth blocks of multiplication 20, 21; the first and second limit detectors 23, 24; logic block 25; regulator 26 of the self-adjustment circuit of the coefficient of transfer of the model to objects; division block 29; controller 33 of the scaling contour of estimates of probabilistic characteristics.

In Fig. 1 the following designations are given:

Y - the signal of the regulated variable signal from the output of control object 4; 7

Y - signal from the output of the encoder 5; y -y e - error signal:

Зои - controlling influence of regulator 1 of the main regulation circuit;

ZhK, Ya. external coordinate and parametric disturbances acting on the control object;

Ut - signal from the output of the control object model 7;

U. centered signal of the regulated variable signal of the control object y from the output of the filter 10;

Ut - the centered signal of the regulated variable signal of the control object model Ut from the output of the filter 11;

Uk - centered signal from the output of key 12; tk - centered signal from the output of key 13;

UC - a signal for estimating the dispersion of the regulated variable of the control object signal at the output of the probabilistic characteristics calculator 14;

UT. the signal for estimating the variance of the regulated variable of the control object model signal at the output of the probability characteristics calculator 17; p.s.

K - signal from the output of the encoder 29 - a constant that determines the initial value of the transmission coefficient of the object model; 7 -

A" - signal from the output of the setter 30 - a constant that determines the desired type of transient process in the main control loop; 7.07 - signal from the output of the setter 31 - a constant that determines the set value of the evaluation of the probability characteristic; a - the controlling effect of the regulator 33 of the mass circuit scaling of estimates of probabilistic characteristics, i.e., controlling influence from the output of logic block 25;

AO - the signal of the difference in variance estimates at the output of the algebraic adder 22;

re . млу ую - the adjusted signal of the variance estimation of the regulated variable of the signal of the control object at the output of the third unit of multiplication 20.

The elements listed in the scheme of the system, which implements the proposed self-adjustment method, are interconnected as follows. The output of the first encoder 5 is connected to the first input of the first algebraic adder 6, the output of which is connected to the first input of the regulator 1, namely to the first input of the first multiplication block 2. The output of the first multiplication block 2 is connected to the input of the second part of the regulator 3 , the output of which is connected to the input of the control object 4 and the first input of the control object model 7, namely to the input of the inertial part of the control object model 8. The output of the inertial part of the control object model 8 is connected to the first by the input of the second multiplication unit 9. The output of the control object 4 is connected to the second input of the first algebraic adder 6 and the input of the first high-frequency filter 10. The output of the second multiplication unit 9 is connected to the input of the second high-frequency filter 11.

The output of the first high-pass filter 10 is connected to the first input of the first controlled key with memory 12 and to the input of the first limit detector 23. The output of the second high-pass filter 11 is connected to the first input of the second controlled key with memory 13 and with the input of the second limit detector 24. The output of the first limit detector 23 is connected to the first input of the logic block 25, and the output of the second limit detector 24 is connected to the second input of the logic block 25. The output of the logic block 25 is connected to the second input of the first controlled key with memory 12 and with the second input of the second controlled key with memory 13. The output of the first controlled key with memory 12 is connected to the input of the first calculator of probabilistic characteristics 14, namely to the input of the first squarer 15, the output of which is connected with the input of the first low-pass filter 16. The output of the second controlled key with memory 13 is connected to the input of the second calculator of probabilistic characteristics 17, namely to the input of the second quadrature 18, the output of which is connected with the input of the second low-pass filter 19. The output of the first low-pass filter 16 is connected to the first input of the third multiplication block 20, and the output of the second low-pass filter 19 is connected to the first input of the fourth multiplication block 21. The output of the third multiplication block 20 with connected to the first input of the second algebraic adder 22 and to the first input of the fourth algebraic adder 32. The output of the fourth multiplication block 21 is connected to the second input of the second algebraic adder 22,

The output of which is connected to the input of the controller 26 of the self-adjustment circuit of the transfer coefficient of the object model. The output of the regulator 26 is connected to the first input of the third algebraic adder 27, the second input of which is connected to the output of the second setter 28. The output of the third algebraic adder 27 is connected to the second input of the division block 29 and to the second input of the second multiplication block 9. The first the input of the division block 29 is connected to the output of the third encoder Z0, and the output of the division block 29 is connected to the second input of the first multiplication block 2.

The output of the fourth setter 31 is connected to the second input of the fourth algebraic adder 32, the output of which is connected to the input of the controller 33 of the scaling circuit of estimates of probabilistic characteristics. The output of the regulator 33 is connected to the second input of the third multiplication block 20 and the second input of the fourth multiplication block 21.

The proposed method is implemented by the control system, the structural diagram of which is shown in Fig. 1, in the following order. The controlling influence of the controller stabilizes the controlled variable at the set value in the object of control, compensating for the influence of external coordinate disturbances. In the structural diagram of the system, the stabilization function is performed by regulator 1, control object 4, encoder 5 and algebraic adder 6, with their connections forming the main regulation circuit.

For solving many practical problems, control object 4 can be described by a transfer function

K.

MUR (c) - your "ehr(- tov) - Co M" (c) (1)

Kk p. from T, not ' . where 79 is the transfer coefficient of the control object, 9 is the time constant of the control object when it is approximated by the inertial link of the first order, To is the delay time in the control object, and (c) is the inertial part of the transfer function, and Z is the Laplace operator.

The parameters of regulator 1 can be calculated by various engineering methods.

For example, according to |(Kopelovich A.P. Engineering calculation method for the vibration of automatic regulators |Hext| / M.: Hobs. nauchno-technicheskoe izdatelstoye literaturyi po chernoi i tsvetnoi metallurgii, 1960-190 p., p. 111), the PID parameters -regulators are determined by the known parameters of the control object (1): the transfer coefficient of the regulator "" is calculated from the expression

ATAK Kk, --- then isodrome time - 72 7 then,

Tr Hundred time bias - ; 7. - - where A", a, B, C are constants that determine the desired type of transition process. 7 b. -

If the value A is set, then the transmission coefficient of regulator 1 can be calculated from the expression

Kk, - d- / Ko (2)

By analogy with (1), the transfer function of the control object model is presented in the form

K. and (c) - ep. g ehri ov) 5 Ku MU c) 5 y (3)

Kt g. where is the transmission coefficient of the model of the control object, is the time constant of the model of the control object when it is approximated by an inertial link of the first order, tt is the delay time in those models of the control object, and (c) is the inertial part transfer function.

The variable transmission ratio of the controller 1 implements the first multiplication block 2, which multiplies the error signal 97U by the current transmission ratio of the KE regulator calculated from dependence (2),

Part C of the regulator implements the inertial components of the regulation algorithm, for example, I - and

D - components.

The proposed method is aimed at self-adjustment of the transmission coefficient of the regulator "in case of changes in the transmission coefficient of the object o by adjusting (approaching) to it the current values of the transmission coefficient Кl of the object model. At the same time, it is assumed that: Кк г - changes in the transmission coefficient object 79, which are essentially parametric perturbations З for the control system, can change significantly (by 2...5 times compared to the momentum in value), however, these changes are of a low-frequency character and in the time interval (0 s,

From where Is the cut-off frequency of the system, the transmission coefficient of the object Ko can be considered quasi-stationary; - the parameters of the object are quasi-stationary and change insignificantly; - external coordinate perturbations of the CO that affect the object are random in nature, and their spectral composition is lower-frequency than the spectrum of oscillations of the regulated variable of the UCH signal,

In addition to those listed earlier, other blocks presented in the structural diagram of the system that implements the proposed method are intended for determining variable values of the transmission coefficient 9 of the control object and adjusting (approaching it) values of the transmission coefficient Ka of the model.

The control object model 7 according to dependence (3) in the structural diagram of the system that implements the proposed method consists of block 8 - the inertial part of the control object model

M (c), which includes both the delay link and the second multiplication block 9, which forms the variable transmission coefficient kt.

In accordance with the method that is claimed, the adjustable variables of the signal of the control object and its model are filtered from low-frequency components caused by changes in disturbing influences on the control object. In the structural diagram of the system that implements the proposed self-tuning method, the filtering function is performed by filters 10, 11. In the simplest case, these are differentiators - high-frequency filters, the transfer function of which

Mvt c

It's not. and . where is the time constant of the inertial part of the filter.

Filters (3: 41 blow from the signals of regulated variables of the signal from the output U of the control object 5 U tuhuk and from the output of its model Ut() - Ut 7 Ut() constant components U ; Ut and centered signals U, U are observed at their output ,

Next, in accordance with the claim, the signs of the filtered adjustable variables of the signal Uk are determined; Utii and, in the case of the coincidence of these signs, continue the calculation on the sliding time interval of the current estimates of the probabilistic characteristics of the filtered regulated variables of the signal, the calculation of the current value of the difference of the estimates of the probabilistic characteristics and the calculation of the transfer coefficient of the regulator according to the variable transfer coefficient of the object model, and in the case of a discrepancy of the signs filtered adjustable signal variables estimate probabilistic characteristics and transfer coefficients are kept at the previous level.

In the structural diagram of the system that implements the proposed self-debug method, this function is performed by the first 12 and second 13 keys with memory, the first 23 and second 24 limit detectors, as well as the logic block 25. those )

Signs of the filtered adjustable signal variables of the Uk control object and its model

Utio is determined by the first 23 and second 24 limit detectors according to dependencies in 1uU020 -1U0) "0 1Ut|)" 0

Zt uch (5) -1Ut) "0

Logic block 25 forms a controlling influence depending on 1 5 and vit - 1 "o О, ви - 1 ; ЗИ - 9 тШЖВ З - и. (6)

The first 12 and the second 13 keys with memory implement the following dependencies; in, tsr-1

KA y" 1 tsi -0 - Ut Chi -1

Utk - th n-th (7)

Ut »sh-0 u pi op-1 where uUt are the previous values of the centered components of the controlled variable signal of the control object Ut and its model Ut until the transition of the control signal M of the logic block 25 from the value of one to zero. At those moments of time when the filtered adjustable variables of the control object and its model change in phase and their signs are the same, it is reasonable to calculate the values of the estimates of their probabilistic characteristics (that is, the estimates of the variances uk here because they contain information about the current values of the control transfer coefficients of the object and its models.

At such moments of time, the logic block 25 includes keys 12, 13 with the T control signal, calculators 14, 17 calculate the current values of estimates of the probabilistic characteristics - dispersions of the UK, HERE,

At other time points, when the filtered adjustable variables of the control object and its model change out of phase and their signs are not the same, the information about the current values of the transfer coefficients of the control object and its model is distorted under the influence of the discrepancy between the parameters of the control object and its model, as well as under the influence of external coordinate disturbances of a stepped nature. At such moments of time, the logic block 25 turns off the keys 12, 13 of the control signal of the FM for estimating the probabilistic characteristics of the filtered adjustable variables of the signal of the control object and its model (estimates of dispersion Uk Utk are memorized at the previous level by low-pass filters 16 and 19. re increases accuracy further determination of the transmission coefficient of the model of the control object "PT" when the transmission coefficient changes

To the control object.

Further, according to the method that is claimed, the current estimates of the probabilistic characteristics of the filtered regulated signal variables, in particular the dispersion estimates, are calculated on a sliding time interval. In the structural diagram of the system that implements the proposed self-debug method, this function is performed by the first 12 and the second 13 calculators for estimating probabilistic characteristics. Centered signals У у, Ут of the regulated variable signal of the control object and its model from the outputs of filters 10 and 11 pass through the front shield and the second 13 controlled keys with memory, at the output of which the signals Uk and Utk Y are formed - these signals are fed on the first 12 and the second 13 calculators of probabilistic characteristics, which include the first 15 and the second. gout, the first 16 and the second 19 low-pass filters with transfer function 5) DTZ) It is known (see (Kulikov E.I. Method of measurement of random processes | Texts / M- Radtso and communication, 1986. - I KO 106) that at the output of the filters, estimates of the dispersions of the UV and UV signals of the UV and UV signals are formed on a sliding time interval

Tos - Te Ta. The value of the evaluation interval Tos is selected from the range of Tos - (5..10)/, where іs is the cut-off frequency of the regulation system.

Calculation of the current value of the difference in estimates of probabilistic characteristics, as provided for by the method that is claimed, in the structural diagram of the system is performed by an algebraic adder 22, the inputs of which are fed with signals of variance estimates, which are scaled by the third 20 and fourth 21 blocks of change and at the output of the adder 22, the AO signal of the difference in estimates is formed dispersions of UV signals | Knit M according to the expression

AV - Ok "tsa -Outk "tsa (c) where ba - scaling factor - controlling influence of the controller 33 of the scaling circuit of estimates of probabilistic characteristics.

From In the prototype, based on known dependencies (see (V.S. Pugachev et al. Basic principles of automatic control / Under the editorship of V.S. Pugacheva / Text / M.: Gos. izdatelsto fizikr- matematicheskoi literatur, 1963. - 646 p., p. 264|, it is shown that the estimates of the variance U .nie - b; p. transmission of the control object model Ka,

On the basis of dependencies (7) in the system that implements the proposed self-tuning method, the centered signal Uk from the output of the first controlled key with memory 12 is proportional to the centered signal in the adjustable variable of the signal of the control object y from the output of the filter 10, and the centered signal Utk from of the output of the key 13 is proportional to the centered signal of the adjustable variable signal of the model of the control object Ut from the output of the filter 11. Therefore, the 2 ц2 difference АУ (8) is proportional to Ко Кл and therefore, if you direct АО to zero, then at the same time

Ko from Kt

According to the proposed method, the transmission coefficient of the object model "t is changed in the direction of reducing the value of the difference in the AYU estimates down to zero. In the structural diagram of the system that implements the proposed self-adjustment method, this function is performed by the regulator 26 of the self-adjustment circuit of the transmission coefficient "7 object models, for example with

PI-or PID-algorithm, adder 27 and setter 28, which determines the initial transmission ratio. n. . models of the object K. The parameters of regulator 26 are calculated according to the recommendations, as for regulator 1.

If a non-zero signal 40 is received at the input of the regulator 26, then it changes its control influence through the adder 27, the second multiplication block 9 also changes the transfer coefficient t of the object model until the difference lo (8) at the output of the second algebraic adder 22 will not become equal to zero. At the same time, in fact,

In the structural diagram of the system that implements the proposed self-adjustment method, the calculation of the transmission coefficient of the regulator 1 of the main circuit according to the variable transmission coefficient of the object model is performed by the division block 29 according to dependence (2), the first input of which is 7 - . and. the constant A" is received from the output of the encoder 30, and the second input is the current value of the transfer coefficient "pt of the control object model from the output of the adder 27.

In accordance with the method that is claimed, the estimate of the probabilistic characteristic of the filtered adjustable signal variable of the control object is stabilized at a fixed value, for which the current value of the difference between this estimate and the fixed value is calculated, the current value of the difference is transformed, for example, integrated, and multiplied by the transformed value estimation of the probability characteristic of the filtered adjustable variable of the control object signal, reducing due to this value the difference up to zero. In the structural diagram of the system that implements the proposed self-tuning method, these functions are performed by the third encoder 31, the fourth algebraic adder 32, the scaling controller 33 for estimating the probabilistic characteristic of the filtered adjustable variable of the control object signal, for example, with the I-algorithm, and the third multiplication unit 20 Together, these blocks form a circuit for stabilizing the estimate of the probabilistic characteristic (estimate of the dispersion Uk) of the filtered adjustable variable of the signal of the control object from the output of the third multiplication block. . 4. 20 on the fixed value of Y", which is set at the output of the third setter 31. The current value of the evaluation of UY is fed to the input of the fourth adder 32, at the output of which the difference UZ between these signals is formed. Arc --0, . The regulator 33 transforms (for example, integrates ) the current value of the difference, forming the controlling influence of cha 11 a -- | Arch

That , (9)

IJ. . . where! - the setting parameter of regulator 33, which is calculated according to the recommendations, as for regulator 1.

Zo By the amount ba in the third multiplication block 20, the signal for evaluating the probabilistic characteristic of the filtered adjustable variable signal of the control object UC, which comes from the output of the first low-pass filter 16, is multiplied, i.e.

YuOsyu - YuOzhkots kocha

UYU U I (10)

In order to ensure the symmetry of the estimates of the probabilistic characteristics of the filtered adjustable variable signal of the control object and its model and not to distort the value of the difference AB, according to the method that is claimed, at the same time, the estimate of the probabilistic characteristic is multiplied by the transformed value 79 from the output of the controller 33 in the fourth multiplication unit 21 of the filtered adjustable signal variable Uk of the model of the control object.

Stabilization of the estimated values of probabilistic characteristics by the regulator 33 improves the operation of the self-adjustment circuit of the transmission coefficient 7 of the object model when the transmission coefficient Ko of the control object changes. Reduction of fluctuations in the absolute values of estimates of variances of the UC, re. with. . . ut allows the controller 26 to more accurately set the transfer coefficient Ka of the model, monitoring changes in the transfer coefficient Ko of the control object. Therefore, the accuracy of determining the transfer coefficient of the model of the control object increases.

As the results of computer modeling showed, the proposed method solves the problem - it allows to increase the accuracy of determining the transmission coefficient of the control object model when the control object transmission coefficient changes and, as a result, to increase the accuracy of maintaining the adjustable variable of the control object signal on the given value.

As follows from Fig. 2, when external disturbances of a random nature and linear changes in the transmission coefficient o and delay time To in the prototype control system affect the accuracy of the self-adjustment of the transmission coefficient Ka of the object model, it is affected by the control object. As a result, deviations of the regulated signal variable gradually increase

UF from the set value U, In the control system according to the invention, the influence of phase shifts of the filtered adjustable variables of the signal of the control object and its model, which distort the value of estimates of probabilistic characteristics, is eliminated. The stabilization of these estimates increases the accuracy of determining the current values of the transmission coefficient "pl" of the object model. At the same time, the root mean square deviation of the adjustment error of the main circuit improves from the value of 7792 to the value of 77,

Claims (1)

FORMULA OF THE INVENTION A method of self-adjustment of the transmission coefficient of the regulator, which includes stabilization at a given value of the adjustable variables of the signal of the control object, the transmission coefficient of which changes over time, due to the change of the control action of the regulator, filtering of the adjustable variables of the signal of the control object and its model from low-frequency components , caused by changes in disturbing influences on the control object, calculation on a sliding time interval of current estimates of probabilistic characteristics of the filtered adjustable signal variables of the control object and its model, calculation of the current value of the difference in estimates of probabilistic characteristics, a change in the transfer coefficient of the object model in the direction of decrease the value of the difference in estimates up to zero, the calculation of the transfer coefficient of the controller based on the variable transfer coefficient of the object model, which differs in that, additionally, when filtering the regulated variables of the control object signal and its model from low-frequency components, the signs of fil are determined filtered regulated variables of the signal and, in case of coincidence of these signs, continue the calculation on the sliding time interval of the current estimates of the probabilistic characteristics of the filtered regulated variables of the signal, the current value of the difference between the estimates of the probabilistic characteristics and the transfer coefficient of the regulator according to the variable transfer coefficient of the object model, and in case of discrepancy the signs of the filtered adjustable signal variables, estimates of the probabilistic characteristics and Zo transfer coefficients are kept at the previous level, while the estimate of the probabilistic characteristics of the filtered adjustable signal variables of the control object is stabilized at a given fixed value, for which the current value of the difference between this estimate and the given fixed value is calculated , transform the current value of the difference - integrate, and multiply by the transformed value of the estimate of the probabilistic characteristic of the filtered adjustable variable of the control object signal, reducing due to this value of the difference up to zero , at the same time this converted value is multiplied by the estimate of the probabilistic characteristic of the filtered adjustable signal variable of the model of the control object.