SU1361502A1 - Adaptive control system - Google Patents

Abstract

The invention relates to self-learning automatic control systems and can be used to control an object with transport delay and disturbing effects such as mixtures, for example, for automatic control of coal mining and processing processes. The purpose of the invention is to improve the accuracy of the system. The system contains a control object 1, a disturbance sensor 2, a control unit 3, two keys 4 and 5, two adders 6, 14, a static model 7, two comparison elements 8, 11, a regulator 9, an output coordinate sensor 10, a correction regulator torus 12, setpoint output coordinate 13, generator of a random signal 15, blocks for selecting a minimum of 16 and a maximum of 17 of a control action with sensors 18, 19, 4 or so on. (L ate figs.

Description

FIELD: self-sticking. It can be used for controlling objects with transport delay and disturbing influences such as bulk material mixtures, for example, for automatic control of coal mining and enrichment processes.

The goal is to show the accuracy of the system.

Figure 1 shows the functional diagram of the device; Fig. 2 is a functional block diagram of the control unit; FIG. 3 is a functional diagram of a random signal generator; 4 shows the combined probability distribution density of the disturbing action.

The adaptive control system contains a control object 1 (an actuator is included in the control object), a disturbance sensor 2 connected to the input of control unit 3, the first and second outputs of which through the first 4 and second 5 keys are connected to the inputs of the adder 6, the last output through sequentially included static model

7 control object, first element

8 and the controller 9 is connected to the control input of the control object 1. The output of the latter through the output coordinate sensor 10 connected in series, the second comparison element 11 is connected to the input of the correction regulator 12. The output of the output coordinate setting device 13 of the output coordinate is connected to the second input of the comparison element 1.1. The output of the regulator 12. connected to the first input of the second adder

14, to the second input of which the third output of the control unit 3 is connected. The corresponding outputs of the generator 15 of a random signal are connected to the control inputs of keys 4 and 5; the output of the second adder is connected to the input of the generator 15 of a random signal through the selection blocks for a minimum of 16 and a maximum of 17 of a control action. The output of the setpoint 18 (upper limit) of the minimum of the control action) is connected to the input of the minimum separation unit 16 and the output of the setter

19 (lower limit) is connected to the input 55 P of the successively connected three-block of the selection of the maximum of the controls, the threshold element 46 is connected to the firing action of the control inputs of the counters 39 and

The control unit 3 (FIG. 2) contains 42 pulses, as well as a key with a controllable key 20, to the input of which is connected with the input of the control key 43.

15022

The output of the sensor 2 is a disturbance. The output of the generator 21 is connected to the control input of the switch 20 with pulses, and the output of this switch is connected to the memory unit 22. The output of the latter is connected to the first inputs of the elements 23 and 24 of the comparison. The output of the comparison element 23 is connected to the input of the 10XY multiplier 25 and quadrant 26. The output of the quadrant 26 through the series-connected comparison element 27 and the threshold element 28 is connected to the second input of the multiplier 25, the output 15 of which is through the serially connected amplifier 29, adder 30, discrete the delay element is connected to the second input of the comparator element 33 and the second input of the accumulator 1 of the mater 30. 20 The output of the comparing element 24 is connected to the outputs of the multiplier 32 and the second quad by 33. 34 and the threshold 25 element 35 is connected to the second input of the second multiplier 32, the output of which is connected through the second amplifier 36, the second adder 37 and the second discrete delay element 38 30 to the second input of the comparator 24 and the second input of the adder 37. The quad output 26 It is also connected to the second input of the comparison element 34, and the output of the quadrant 33 to the second input of the comparison element 27.

The output of the generator 21 pulses also under shyuchen to the input of the counter 39 im-. pulses, the output of which is connected 40 to the first input of the comparison element 40 and to the first input of the divider 41, to the second input of which, through the series-connected pulse counter 42, the output of the threshold element 35 is connected. The output of the divider 41 through the serially connected control key 43 and the block 44 memory, an adder 14, blocks for allocating a minimum of 16 and a maximum of 17 of the controlling action gg are connected to a controlled input of a controlled oscillator 15 of a random signal. The unit 45 of the number of pulses is connected to the second input of the comparison element 40, the output of which is 35

The controlled oscillator 15 of a random signal (Fig. 3) consists of a summer 47, to the first input of which the output of the discrete memory block 44 is connected, and to the second input the output of the noise generator 48 is connected. The output of the adder 47 is connected to the control input of the standby multivibrator 49, the output of which is connected to the first input of the element AND 50 and the first input of the element IS-NE 51. To the input of the waiting multivibrator 49 and to the second inputs of the elements 50 and IS-NOT 51 is connected. the output of the generator of rectangular pulses 52. The outputs of the elements 50 and AND-51 are connected to the control inputs of the keys 4 and 5.

Adaptive control system works as follows.

In Fig. 4, denoted: W is the value of the disturbing action, F (W) is the combined density of the probability distribution of the disturbing influence.,

Changes in the magnitude of the disturbance are the sum (mixture) of two random processes with average values and the distribution law of which is close to normal.

When managing such objects, it is important to know the parameters from which the final (observable) mixture is formed - a disturbing effect (for example, a mixture of solid-coal fed to the enrichment process). The expression for the probability distribution functions of the mixture for the case of two classes (two components) of the disturbing action is

F (W,) PF, (W, /,) + (1-P) F2 (W2 / 2,)

(U

where F (W,) is the function of the distribution of the parameters of the perturbing mixture (W, is the ash content, moisture content of the bulk material, etc.);

Fj (W, /,) is the probability distribution function of the first class of the disturbing component with the mean (most probable) value, J

Fj (W, distribution function. Probability of the second class component

five

0

P disturbance with an average (most probable) value of 1;

probability of occurrence of W, from class

F, (w, / 2.);

The control task in this case is set as follows: requires-. c, observe a mixed implementation of the disturbing action F (W,) divide the mixture into classes and determine their parameters 1, 1 P, on the basis of which 5 correct (randomly) the value of the control coordinate of the control object so that the average value of the output coordinate of the object was equal to 0 specified. This allows us to consider the control of an object as a statistical game of two players: nature and the control system, in which player II, the control system, is 5 with a statistician, and player I is nature. The strategies of nature are the average values of the processes, and h J that appear with probabilities P and (l-P), and for the control system, laziness are the values of the input control value of the control object corresponding to these modes.

The control in this case is reduced to defining the strategies of nature 2, 5 and the probability of their occurrence P and (1-P), determining for these strategies the values of the input coordinates of the control object -S, and Sj, which are the strategies of the control system neither, and the mixing (randomization) of the strategies S, and Sj by the control system with the probability P and (1-P), so that the average value of the output coordinate of the control object is equal to the given value of the output coordinate (setting unit 13).

The signal from the disturbance sensor 2 is applied to the input of the control unit 3, in which this signal is divided into two classes (two random processes) and the average values of the disturbing influence for these two classes are determined: and

55

2; and the likelihood of their occurrence of P and 1-P.

The signal from the pulse generator 21 is fed to the control input of the switch 20, which, shortly, sends a signal from the sensor 2 to the disturbing effect 513

Bind to the input of memory block 22. This block memorizes the value of the signal from the disturbing sensor 2 at the time of closing the key 20, If there is no signal from the pulse generator 21, the key 20 opens, while the signal from the disturbing sensor 2 does not pass to the memory block 22 and the output of the block 22 remains corresponding to the signal from sensor 2 at the moment when the key 20 closes. When the key 20 is next closed, the previously received signal is erased in memory block 22 and the new value of the signal from the disturbance sensor 2 is memorized.

Thus, at the output of the memory block 22, a discrete value of the signal from the disturbance sensor 2 is formed, corresponding to the timing of the switch 20, the signals from the output of the memory block .22 are fed to the first: the inputs of the comparison elements 23 and 24, where the signals are subtracted from this signal from discrete elements 31 and 38 of the delay.

Differential signal from the output element. 23 comparison is fed to the input of the quad 26, and the difference signal from the output of the comparison element 24 is fed to the input of the quad 33. The squares of the difference signals from blocks 26 and 33 are compared with each other on the elements 27 and 34 of the comparison.

If the difference of the squares of the error signal at the output of the comparison element 27 is less than zero, then when this signal hits the threshold element 28, a signal equal to one appears at its output and after multiplying this single signal in the multiplier 25 by the difference signal from the comparison element 23 the signal is amplified by amplifier 29 and fed to adder 30, where it is summed with the output signal of discrete delay element 31. However, at the output of the delay element 31, the sum signal from the output of the adder 30 appears the next time the signal arrives at the input of the adder 30 from the multiplier 25, i, e, a discrete signal is delayed by one step.

If the difference of the squares of the error signal at the output of the comparison element 27 is greater than zero, then the output of the threshold element 28 is Sig502

zero, and after multiplying this zero signal by the difference signal from comparing element 23 at the output of multiplication 25, the signal is equal to zero and the signal at the input of the adder 30 is not received, while the signal values at both the input and the discrete element 31 delays do not change. However, if the difference of the squares of the error signal at the output of the comparison element 27 is greater than zero, then at the output of the comparison element 34 it is less than zero, so

5, as signals from quadrants 26 and 33, are supplied to elements 27 and 34 of comparison with different signs.

Therefore, the signal at the output of the comparison element 34 is less than zero, and therefore, a signal equal to one appears at the output of the threshold element 35. After multiplying this single signal with multiplier 32 by the difference signal from element 24

5, this signal is amplified by amplifier 36 and is fed to the input of adder 37, where it is summed with the output of discrete delay element 38. However, at the output of delay element 38, this signal appears with a delay of one step. I

Thus, depending on the ratio of the squares of the difference difference signal between the incoming discrete signal from the disturbing sensor 2 and the output signals from the discrete delay elements 31 and 38, the difference signal from the outputs of the comparison elements 23 and 24 is fed to the corresponding inputs of the discrete delay elements amplification of amplifiers 29 and 36 and adjusts the output values of discrete delay elements 31 and 38 with a delay of one step of operation of the key 20. The magnitude of the correction is set by the value of the amplification factor of the amplifiers 29 and 36 and is selected from the condition of the required averaging interval.

Consequently, at the outputs of the elements 31 and 38, the average values of the disturbing components are formed with 7, and 1 for each class (the strategy of nature), which are the output signals of the control unit 3. The initial values of the output signals of the elements 31 and 38 are selected from the condition J,.

five

0

five

0

P. -.

(2)

where is t

N.

P

7

The output signal from the pulse generator 21 is also fed to the input of the pulse counter 39, which determines the number of pulses generated by the generator 21. The output from the threshold of the element 35 is fed to the input of the pulse counter 42, which determines the number of discrete values of the signal from the disturbance sensor 2, assigned to first grade,,

The signals from the counters 39 and 42 of the pulses are fed to the inputs of the divider 41, the output of which determines the frequency of occurrence of the components of the first class i values of the disturbing effect W, i.e. determined by hyphenation

-. B

P

the frequency of occurrence of the first-class disturbing components; the number of values of the signal from sensor 2, referred to the first class; the total number of discrete values (the number of pulses from the generator 21) received at the input of the control unit 3.

The number of pulses at the output of the pulse counter 39 is compared on the comparison element 40 with a predetermined number of pulses, which is set by the set points. 45 of the number of pulses - and is selected from the required interval over which the frequency is determined (the probability of occurrence of the components from the first class of disturbing influence). If the number of pulses n is greater

At the output of the comparison element 40, a negative signal appears, which leads to the appearance of a single signal at the output of the threshold element 46. The signal from the threshold element 46 closes the switch 43 and the signal from the output of the divider i equal to the frequency of the components of the first class t, is fed to the input of memory block 44, and at the same time the signal from the output of threshold element 46 is fed to the control inputs of counters 39 and 42 with a small time delay and zeroed them.

Thus, the output of memory block 44 produces a signal equal to

and,

- frequency of occurrence of components

perturbing effect

0

1502

15

25

0

five

0

five

0

five

eight

munching to first class. This signal remains unchanged until the next closure of the key 43, at which the program will erase the previously memorized value and memorize the new one.

Consequently, the frequency m is approximately equal to the probability of the occurrence of the components Wj (t) from the first class F ((W, /), i.e.

t, r,

(3)

and the longer the interval is selected (the number

n.), which determines the value of f.

on m ,, the greater the equality (3).

The signals and from the outputs of the control unit 3 through the keys 4 and 5 are fed to the inputs of the adder 6. To the control inputs of the keys 4 and 5, signals are output from the generator 15 of the random signal.

Thus, the keys 4 and 5 are closed at random with probabilities P and (1-P), which are determined by the state of the control input of the generator 15 of the random signal.

Consequently, when key 4 is written, the signal 1 of control unit 3 arrives at the input of adder 6, and at 3 am.} Cank 11 of key 5, the signal and the output of adder 6 successively appear average values of classes (strategies of nature) j and 2g with corresponding the potencies of P and (1-P).

The output signal from the adder 6 is fed to the static model 7 of the control object 1 at the disturbing input, the output of which is a signal corresponding to the output coordinate of the control object when the average value of the disturbance action arrives at the object, or 2 (depending on which of the YuIyuchey 4 or 5 is closed).

The output signal of the static model 7 of the control object is compared with the signal of the setting device 13 on the comparison element 8 and, depending on the error signal, the controller 9 sets two values of the input coordinate of the object S, and S corresponding to two values of the strategies of nature, for which the output from the control Torus 9 is fed to the control input of the control object. The regulator 9 is a direct type regulator and in the simplest case it can be an amplifier.

The input coordinates S, and S are determined from the condition of the invariance of the average value of the output coordinate of the object to the perturbation, where the action is And are calculated by expelling

. )

  K- (J

14) (5)

hg -. static coefficient

object gain perturbation input (static model 7 gain factor);

h ,, is the static gain of the object at the control input;

9-type regulator gain factor; set point output

coordinates of the control object;

, j - mean (most probable) values of the components of the disturbing action. The input coordinates S, and S are strategies of the control system, and since it is not known which of the strategies of nature Ч or 2 will be applied to object 1, the control system must mix (randomize) its strations S, and randomly, to ensure the equality of the average value of the output coordinate (t) j to the specified value.

The calculated value P, m, through the adder 14, the minimum selection block 16 and the maximum selection block 17 is fed to the control input of the random signal generator 15, which determines the probabilities P and (lP, -) j with which the keys 4 and 5 are closed, those. the probabilities of the occurrence of the strategies S S of the control system are established.

A signal from the output of the memory block 14 is fed to the control input of the generator 15 of a random signal, in which this signal is summed with the signal from the noise generator 48 at summer 47. The sum signal from summa gg pa 47 is connected to the control input. of the waiting multivibrator 49 and changing its time constant in a random way. At the entrance of the waiting multivibra25

thirty

40

45

gQ

36150210

The torus 49 is given a signal from. Generator--: Pa 52 square-wave pulses that triggers the standby multivibrator 49, while the duration of the pulses generated by the standby multivibrator is a random variable. The output signals from the generator 52 of rectangular pulses and the waiting multi10 vibrator 49 are fed to the input of the element I 50 and to the input of the element AND-HE51.

If the oscillation periods of the generator 52 square-wave pulses and the standby multivibrator 49 do not coincide,

15 output element and 50 signal not in

But, at the same time, a signal appears at the output of the AND-HE element 51, which via the control input of the key 5 closes it.

If the oscillation periods of the generator 52 square-wave pulses and the standby multivibrator 49 coincide, their signals at the input of the element I 50 arrive simultaneously and at its output a signal appears which, having arrived at the control input of the key 4, closes it.

This ensures the adaptability of the control system to the varying probability characteristics of the disturbing action.

However, in the course of operation, the parameters of the control object can also be changed, which can lead to the appearance of a static error at the output of object 1. To eliminate this effect, the control system uses a probability correction due to deviation of the output coordinate of the object, which also eliminates and other uncontrollable disturbances.

This is carried out by measuring the current value of the output coordinate H t) by the sensor 10 of the output coordinate, the signal of which is compared at the reference element 11 with the signal of the sensor 13.

If the control actions of the control system strategies S ,, S and P are determined exactly, then the static clearance at the output of the control system is zero and the system maintains equality

and. - Mfu (t) 6 0. (6)

The mismatch signal of the comparison element 11 is fed to the controller 12, which in the simplest case

eleven

is an integrating element.

If conditions (6) are fulfilled, then the 12 a signal is equal to the output of the regulator 12, if these conditions are violated, the dP signal appears on the code of the regulator 12, which in the adder 14 is algebraically summed with the signal from the output of the control block 3.

Thus, at the output of the adder 14, the signal is

13

 R. + LR

(7)

This changes the reference to the generator 15 of a random signal and, at its outputs, the probabilities change accordingly.

p, p + ip; (eight)

p, 1 - p ,.

In block 3-control, the separation of the disturbing signal on the classes is carried out in such a way that

Jt) M {w, (t) r, () (9 Therefore, if

. 5 and - M (u (t), (10)

where and is the specified value of the output coordinate of the control object; U (t) - current output value

object coordinates.

i.e. the probability P of the closure of key 4 decreases and, accordingly, the probability P of the closure of key 5 increases. Due to (9), the mathematical expectation of the signal at the output of the adder 6 increases, which leads to an increase in the signal U from the output of the static model 7 of the object and an increase in the signal at the input of the regulator 9 and, accordingly, to an increase in the input coordinate S at the input of the control object:

S S, P, +.

(eleven)

If and, - M (u (t) О, then, i.e. the closure probability of key 4 increases, respectively, the probability P of closure of key 5 decreases. This leads to a decrease in the signal at the output of the adder 6 and the static model. 7 and the object control and a corresponding decrease in the input coordinate at the input of the control object.

This ensures the adaptability of the control system to the changing probabilistic characteristics of the disturbing influence, as well as from 150,212.

changing the parameters of the control object, which allows maintaining the average value of the given coordinate with higher accuracy.

In order to preserve the operability of the system with significant fluctuations in the output coordinate of the control object, there were introduced blocks for selecting a minimum of 16 and a maximum of 17 of a control action at the input of the generator 15 of a random signal. The unit 18 sets the maximum permissible probability P 1.0, and the unit 19 sets the minimum permissible probability m10

15

P o, this allows you to limit the signals from the adder 14 and to ensure stable operation of the generator 15 of a random signal.

20

thirty

35

The technical advantage of this adaptive control system is the more accurate maintenance of the set value of the output coordinate 25 of the control object. This is achieved by the fact that after comparing the output coordinate of the control object with its specified value, the probability of the control strategy for the corresponding class is adjusted. The absence of a closed control loop due to the deviation does not create conditions for the self-oscillations of the output coordinate for large values of the transport delay and its changes in the process of the control object operation.

Claims (1)

Invention Formula 40 An adaptive control system comprising a controller, a disturbance sensor connected to the input of the control unit, the first and second 45 outputs of which through the first and second keys are connected to the inputs of the first adder, the output of which through a static model of the control object is connected to the first input of the first 50 of the reference element, the output of which is connected to the input of the controller, the output of the control object is connected to the first input of the second comparison element through the output coordinate sensor, 55 the output of which is connected via a correction regulator to the first input of the second adder, the control input of the first key is connected to the first output of the controlled random signal generator, the second and third outputs of which are connected to the control inputs of the second key, and the second inputs of the first and second elements comparisons are connected to the output of the output coordinate setting device, characterized in that, in order to improve the accuracy of the system, a block is inserted into the minimum of the control action with the control device connected to it and vscheleni maxim a control signal with a control device connected to it, with the output of the second adder through serially connected blocks of separation of the minimum and maximum of the control action connected to the input of a random signal generator, the second input of the second adder is connected to the third output of the control unit, and the controller output is connected to the input control object. / J . 0.2 32 W,% tpilz. Compiled by V.Peshkov Editor M.Blanar Tehred L.Serdyukova Proofreader O. Kravtsova Order 6281/46 Circulation 863 Subscription VNSHSHI USSR State Committee for inventions and discoveries 113035, Moscow, Zh-35, Raushsk nab., d. /five Production and printing company, Uzhgorod, st. Oroektna, 4