JPS5837561B2 - Simple predictive control device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention uses two manipulated variable calculation circuits that are advantageous for control depending on the magnitude of a change in the state of the system, and automatically determines switching depending on the magnitude of the change in state. In a control device that performs stable control, the switching deviation, which is the standard for switching between the two manipulated variable calculation circuits, is changed according to the change in the system, and a simple prediction function is added that resets the manipulated variable at the time of switching. Regarding a control device.

FIG. 1 is a block diagram showing a conventional simple predictive control device.

1 in FIG. 1 is a controlled object.

The future state quantity of this controlled object 1 (lead time is II
−J t s l is an integer, Jt is the time value in the discrete time system.
Step width) A predicted value 2a is obtained by predicting the value to be generated using the prediction device 2.

This predicted value 2a and the target value 3 output from the target value generator 3
The subtractor 4 calculates the difference from the lead time it-At, that is, the control deviation 4a ahead of the lead time it-At.

As a result, the manipulated variable 6a that makes the control deviation 4a zero is determined using the coefficient unit 5 and the integrator 6.

In addition, in order to improve the accuracy of prediction, the pulse width n-1t,
Using a pulse generator 7 that generates pulses with a pulse interval A-Ji (n<A?) and a multiplier 8, the manipulated variable 6a is changed by n-1t time out of the lead time II-At time. ing.

Such a conventional simple predictive control device has better prediction accuracy than a case where the manipulated variable constantly changes.

That is, at a certain point in time, the lead time l At
When predicting a state quantity in the future, prediction accuracy is better when the manipulated variable is constant and there is no power than when it is completely unknown how the manipulated variable will change up to the next point in time.

In addition, when the target value 3a or the disturbance A is large, the control deviation 4a that is thought to occur by using a highly accurate prediction side
Since the operation amount 6a that makes the (predicted value) zero is applied to the controlled object 1, the controllability is much better than the classical feedback control, which starts the control operation after the TJsp deviation actually occurs. improves.

However, on the other hand, if even a small control deviation occurs when an extremely small disturbance A is added or when a stable state is reached, sensitive control operations may become intermittent (lead time l・At
Since the operation amount is changed for n-At time), the operation amount and the control amount 1a may become slightly oscillatory.

On the other hand, if the sensitivity of the control system is lowered by placing emphasis on minute fluctuations and a stable state, there is a drawback that controllability deteriorates when a large disturbance is applied.

In order to improve these drawbacks, a logic operation circuit is used to switch between the predictive control system shown in Figure 1, which is effective against large disturbances and target value changes, and the control circuit that is stable against small fluctuations in the settled state. It is also conceivable to automatically switch between the two and bring out the advantages of each manipulated variable calculation circuit depending on the magnitude of the fluctuation range.

In this case, the above automatic switching is performed by comparing the predicted value of the control deviation with a preset switching deviation, but when using a predictive control system with a high gain, the level of the controlled object changes and the round-trip gain changes. In some cases, the prediction may not go well due to vibrations in the control system, or disturbances (such as some force not taken into account by the prediction device) may cause the operation signal to suddenly increase. .

However, the predictive control system can exhibit great superiority against large disturbances if the range of movement of the manipulated variable required by the predictive control system is not limited, or if it is limited but not restricted. This is a case where the value is not exceeded.

This invention was made in order to eliminate the above-mentioned drawbacks of the conventional technology, and the control amount of the controlled object, the operation signal, and the disturbance signals are inputted to predict the predicted lead time 13-1t ahead (Il is an integer, At is A prediction device that outputs the predicted value of the controlled variable (separated time system time step width) and predicted lead time! !
- At is used when the predicted value of the control deviation due to the target value ahead and the predicted value of the controlled variable has a large variation, and the value obtained by multiplying the predicted value of the control deviation by a constant is used as the predicted lead time l-A.
a first manipulated variable calculation circuit with a large gain that changes the operation signal by integrating for n-jt time (n<A) out of t, and holds the operation signal unchanged at other times; A second operation with a small gain that is used when the fluctuation of the predicted value of the deviation is small and removes the intermittent nature of the control operation by adding a proportional-integral operation to the predicted value of the control deviation and constantly changing the amount of the operation signal. and determining which of the first manipulated variable calculation circuit and second manipulated variable calculation circuit to use, using the difference between the absolute value of the predicted value of the control deviation and the preset switching deviation as an index. In order to perform automatic judgment and prevent malfunction, it compares the judgment results with those of the past lead time l/jj 1 hour, and if the same result continues twice, either the first or second manipulated variable calculation circuit is activated. a switching logic operation circuit that executes switching of whether to use or not;
I! - When the operation signal reaches both ends of the movable range by sampling every sampling period of -At time, it is determined that an abnormality has occurred, and only when this abnormality continues twice, the value of the switching deviation is increased and the abnormality is detected. A switching deviation calculation circuit that reduces switching deviation when no switching deviation occurs;
The operation signal reset circuit includes an operation signal reset circuit that resets the operation signal to an average value of a value past 11 #t times and a current value when switching between the first operation amount calculation circuit and the second operation amount calculation circuit. By using this method, controllability can be dramatically improved compared to classical feedback control, and stable swinging can be performed when the disturbance change is small or when the stable state is sufficiently approached. In addition to the effects that can be achieved, even if an abnormal condition occurs, it can be overcome by switching between the two manipulated variable calculation circuits, and at the time of this switching, the operation signal at the end of the movable range is reset and stabilized.
The purpose of the present invention is to provide a simple predictive control device that aims to improve control accuracy.

Next, an embodiment of the simple predictive control device of the present invention will be described based on the drawings.

FIG. 2 is a block diagram showing one embodiment thereof.

Reference numeral 11 in FIG. 2 is the object to be controlled.

This controlled object 11 predicts the control amount in the future (lead time is l-At, It is an integer, Jt is the time step width in the discrete time system) when it receives a disturbance 12a from the disturbance source 12. The prediction is made by the device 13.

This prediction device 13 includes information necessary for prediction, that is,
A disturbance 12a from the disturbance source 12 (including a disturbance signal currently applied to the controlled object 11 or a preceding signal scheduled to be applied in the future), a detection signal 11b of the controlled variable 11a that detects the current state of the controlled object 11, and An operation signal 14a of the operation amount generated by the limiter 14 is introduced.

The output signal of this prediction device 13, that is, the predicted value 13a
is led to a code converter 15 where the code is converted and then added to a first input terminal of an adder 16.

The other input terminal of the adder 16 receives a target value 17a ahead of the lead time l·At from the target value advance generator 17.

Therefore, the adder 16 adds the output of the code converter 15 and the target value 17a ahead of the lead time Il-1t.

In other words, the target value 17a ahead of lead time 13-it
A predicted value of the control deviation between the predicted value 13a of the control amount 11a and the predicted value 13a of the controlled amount 11a is calculated.

The output 16a of the adder 16 is sent to each fixed terminal a of two switches 18, 19 and to the input of an absolute value operational amplifier 20.

The switches 18, 19 each have fixed terminals a, b and a movable terminal C, and the fixed terminal b of each switch 18, 19 is connected to the output end of the zero setter 21A.

This zero setter 21A generates the value "O".

Still, the movable terminal C of the switch 18 is controlled by the output 21a of the R-S flip-flop circuit 21 (hereinafter abbreviated as FF).

An output of 2 lb is also sent from this FF 21 to the negative element 22.

The output 22a of this negative element 22 is adapted to control the opening and closing of the movable terminal C of the switch 19.

A movable terminal C of the switch 18 is connected to one input end of the multiplier 23.

The output of the adder 24 is also introduced into the other input terminal of the multiplier 23.

A cutting end of a pulse generator 25 is connected to one input end of the adder 24 .

The output end of the pulse generator 26 is connected to the other input end of the adder 24 .

Outputs 25a and 25b of the pulse generator 25 are sent to switches 27 and 28, respectively, which will be described later.

The pulse generator 25 generates pulses with a height of 1, a pulse width Jt, and a pulse interval A-Jt, and a pulse generator 26 has a height of 1, a pulse width (n-1) Jt, and a pulse interval II. -At pulse is generated.

The rising edge of the pulse from the pulse generator 26 lags behind the rising edge of the pulse from the pulse generator 25 by Jt.

The output terminal of the multiplication circuit 23 is connected through a coefficient multiplier 29 to one input terminal of a limiter addition integrator 30.

On the other hand, the movable terminal C of the switch 19 is connected to the coefficient multipliers 31 and 3.
2 to the other input terminal of the limiter addition integrator 30.

The output terminal of the limiter addition multiplier 30 is connected to one input terminal of an adder 33, and the output terminal of the coefficient multiplier 31 is connected to the other input terminal of the adder 33.

The output 33a of the adder 33 is sent to the input terminal of the limiter 14.

This limiter 14 attaches an upper limit to the output 33a of the adder 33 and sends an operation signal 14a to the controlled object 11 and the prediction device 13 as described above.

force) As shown by enclosing the dotted line with a comb, the multiplier 23,
Coefficient unit 29, limiter addition integrator 30, adder 33,
The limiter 14, the pulse generators 25 and 26, and the adder 24 constitute a first manipulated variable calculation circuit OP1.

Also, as shown surrounded by a dashed line, coefficient units 31 and 3
2. Limiter addition integrator 30, adder 33, and limiter 14 constitute a second manipulated variable calculation circuit OP2.

Note that the coefficient multiplier 29 has a set magnification kl, and the coefficient multiplier 31 has a set multiplier kp, and when the operation signal 14a supplied to the controlled object 11 increases,
If the control amount increases, it has a positive value, and if it decreases, it has a negative value.

As mentioned above, the predicted value of control deviation) and the operation signal 14a
The circuit that makes the circuit includes (a) an adder 16 - a switch 18 - a multiplier 23, a pulse generator 25 . 26 - Adder 24 - Multiplier 23 - Coefficient unit 29 - IJ Mitsuta addition integrator 30 - Adder 33 - I
J Mitsuta 14 system, (b), adder 16 - switch 19 - coefficient unit 31 - coefficient unit 32 - IJ Mitsuta addition integrator 30 - adder 33, coefficient unit 31 - adder 33 - limiter 14. There are two systems;

There are cases where the route goes through the system (a) and there are cases where the route goes through the system (b).
) The set magnifications kl and kp are adjusted so that the round-trip gain is several times larger than that of the system.

1 The setting magnification 〒1 of the coefficient unit 32 gives a positive value, and the upper limit Ul and lower limit L1 of the limiter 14 give a movable range of the converted amount.

Now, the output 20a of the absolute value operational amplifier 20 is introduced into one input terminal of the adder 34, and the output of the code converter 35 is introduced into the other input terminal of the adder 34.

The output 34a of this adder 34 is introduced into the input terminal of a function generator 36.

The function of this function generator 36 is as shown in the figure.
If the output 34a of the adder 34 is negative, it is "1", and the adder 34
If the output 34a of is positive or "O", it is a function r-IJ.

The output 36a of this function generator 36 is introduced into a dead time element 37.

The output y1 of this dead time element 37 and the function generator 36
The output y2 of is introduced into one and the other input terminals of the exclusive OR element 38, respectively.

Furthermore, the output 37a of the dead time element 3T and the output 36a of the function generator 36 are introduced into respective inputs of the adder 39.

The output y3 of the exclusive OR element 38 is sent to the fixed terminal a of the switch 27 via the negation element 40.

Further, the output terminal of the adder 39 is connected to the fixed terminal a of the switch 28.

Both switches 27 and 28 each have a fixed terminal a. b and a movable terminal C.

The output y3 of the zero setter 41, which always outputs zero, is introduced into the fixed terminals b of both switches 27 and 28.

Switch 27. Each of the 28 movable terminals C is connected to the output 25a, 25 of the pulse generator 25 as mentioned above.
It is possible to switch using b as a switching signal.

That is, the outputs 25a and 25b of the pulse generator 25 are "
1'', each movable terminal C of both switches 27 and 28 becomes electrically connected to the fixed terminal a, and the output 25a of the pulse generator 25
, 25b are "0", the movable terminal C and the fixed terminal b are electrically connected.

The movable terminal C of this switch 28 sends out y1 to the first human power end of the FF 21, and the movable terminal C of the switch 27 outputs y1.
2 is sent to the second human power end.

Furthermore, the output y3 of the zero setter 41 is introduced into the third manual power end of the FF 21.

As described above, the output 21a of the FF 21 is sent to the switch 18, and when the output 21a is "1", the movable terminal C of the switch 18 is electrically connected to the fixed terminal a.

When the output 21a of the FF 2'l is "O", the switch 18 is controlled by the output 21b so that the movable terminal C of the i switch 18 is electrically connected to the fixed terminal b.

Furthermore, the output 21b of the FF 21 is inverted by the negative element 22, and when the output 22a of this negative element 22 is "1", the movable terminal 19c of the switch 19 is electrically connected to the fixed terminal a, and the output 22a of the negative element 22 is At "0", the movable terminal C of the switch 19 is electrically connected to the fixed terminal b.

In other words, switches 18 and 19 perform opposite operations.

Thus, as shown by surrounding i with coarsely spaced dashed lines,
Switch 1 8. 1 9, absolute value operational amplifier 20,
Code converter 35, adder 34, function generator 36, dead time element 37, exclusive OR element 38, adder 39, negation element 40, zero setter 41, switch 27.28 FF
21, negation element A and 22, the switching logic operation circuit S
E is constructed.

This switching logic operation circuit SE sets the predicted value of the control deviation 16a appearing at the output terminal of the adder 16 as a target, and
It automatically determines which of the operation amount calculation circuit OP1 or the fifth operation amount calculation circuit OP2 should be used, and switches the switches 18 and 19 under predetermined conditions. 〇F Next, the switching deviation calculation circuit SEE, which is a major feature of this invention, and the operation signal reset at the time of switching are selected. circuit O
The configuration of the L portion will be described.

Of these, the former switching deviation calculation circuit will be explained first.

A part of the operation signal 14a appearing at the output end of the limiter 14 and sent to the controlled object 11 and the prediction device 13 is sent to one input end of the adder 42, and a part of the operation signal 14a is sent to the code converter 43. The signal is sent to one input terminal of the adder 44 through the input terminal.

The output of the constant generator 45 is introduced into the other input terminal of the adder 42.

This constant generator 45 sets a value of -(U1-α).

Further, the output of the constant generator 46 is introduced to the other input terminal of the adder 44.

This constant generator 46 sets a value ('L?+α).

Here, U1, L1. The relationship of α is such that the upper and lower limits of the limiter of the limiter addition integrator 30 (also the upper and lower limits of the limiter 14) are respectively U1 and L1, and a suitable positive value is α.

The outputs of adders 42, 44 are respectively output to function generators 47, 4.
It is connected to the input terminal of 8.

When the outputs of the adders 42 and 44 are introduced into the function generators 47 and 48, the outputs of the function generators 47 and 48 are sent to the adder 49 after performing the function operation described later. There is.

Each of the function generators 47 and 48 is a function whose output is "O" if the input is "negative" or "0", and whose output is "1" if the input is "positive".

The output of adder 49 is sent directly to a first input of AND element 50 .

Further, the output of the adder 49 is sent to a second input terminal via a dead time element 51.

Therefore, the AND element 50 takes the AND of the output of the adder 49 and the output of the dead time element 51, and the contents of this AND element 50 are as follows: When the above two inputs are both "1", The output becomes "1" and one of them becomes "O"
When , and when both are "O", the output is "0".

The output end of this AND element 50 is connected to a fixed terminal a of a switch 52.

This switch 52 has fixed terminals a, bg and a movable terminal C, and the output end of a zero setter 53 is connected to the fixed terminal b.

This zero setter 53 always generates the value ``O.J''.

The switch 52 performs switching using the output 25c of the pulse generator 25 as a switching signal, and if the output 25a of the pulse generator 25 is "1", the fixed terminal a and the movable terminal C are electrically connected. do.

If the output 25a of the pulse generator 25 is "0", the movable terminal C is controlled to be switched so that the fixed terminal b and the movable terminal C are electrically connected.

The movable terminal C of this switch 52 is configured to send a signal y1 to the first input terminal of the RSFF 54.

The pulse generator 25 is connected to the second input terminal of the RSFF 54.
The output y2 (same as outputs 25a to 25c) is introduced.

The output y of the zero setter 53 is connected to the third input terminal of the RSFF 54.
3 is about to be introduced.

The truth table for RSFF 54 is shown in FIG.

The output of the RSFF 54 is inverted by a negation element 55 and then sent to one input terminal of an adder 56.

The output of the constant generator 57 is introduced into the other input terminal of the adder 56.

This constant generator 57 generates a negative value (-r, O<r<1).

The output end of the adder 56 is connected to a fixed terminal a of a switch 58.

The switch 58 has fixed terminals a, b and a movable terminal C.

The fixed terminal b of this switch 58 is connected to the output end of the zero setter 53, and the movable terminal C is connected to the input end of an integrator with limiter 60 via a coefficient unit 59.

The switch 58 is controlled by using the output 25d (same as the outputs 25a to 25c) of the pulse generator 25 as a switching signal.

That is, when this output 25d is "1", the switch 5
8 movable terminal C and fixed terminal a become electrically connected, and the output is 25.
When d is rOJ, switching control is performed so that the movable terminal and fixed terminal b are electrically connected.

Further, the coefficient unit 59 has a set magnification k2>O.

When the output of this coefficient unit 59 is introduced into an integrator with a limiter 60, this integrator with a limiter 60 integrates it.
A switching deviation εset is created.

The upper and lower limit values (U2 and L2, respectively) attached to this integrator with limiter 60 give the movable range of the switching deviation εset.

Then, the switching deviation εset is determined by the switching logic operation circuit S
The signal is sent to the input end of the code converter 35 at E.

Next, the configuration of the operation signal reset circuit OL will be explained.

In this operation signal reset circuit, the operation signal 14a output from the limiter 14 is introduced into the input terminal of the dead time element 61 and one input terminal of the adder 62, respectively.

The output of the dead time element 61 is introduced into the other input terminal of the adder 62.

The dead time element 61 has a dead time of lead time 11-At time.

The output terminal of the adder 62 is connected to a switch 64 via a coefficient unit 63.
The switch 64, which is connected to the fixed terminal a, is composed of a fixed terminal a and a movable terminal b.

This movable terminal b is connected to the reset terminal of the limiter addition integrator 30.

On the other hand, the final output terminal of the switching logic operation circuit SE, ie, the output terminal of the FF 21, is connected to the input terminal of the dead time element 65 and one input terminal of the exclusive OR element 66.

The other input terminal of the exclusive OR element 66 is the dead time element 6
The output end of 5 is connected.

The output of this exclusive OR element 66 turns on and off the switch 64.

The dead time element 65 has a dead time of Jt.

Moreover, the operation content of the exclusive OR element 66 is such that it outputs "O" only when the output of the FF 21 and the output of the dead time element 65 are equal, and outputs "1" in other cases. It is.

This output is used as a switching signal for the switch 64, and the switch 6
4 on/off.

That is, when the output of the exclusive OR element 66 is "1", the switch 64 is turned on and the value of the coefficient multiplier 63 is transmitted to the limiter addition integrator 30.

The limiter-added integrator 30 has a reset function, and the reset function is activated only when the switch 64 is on.

Next, the operation of the simple predictive swinging device of the present invention configured as described above will be explained.

First, when the controlled object 11 receives a disturbance 12a from a disturbance source 12), the predicted value 13a1 of the control amount 11a ahead of the lead time l-Jt, that is, the output of the prediction device 13 is the lead time 12-jt ahead. When the target value 17a1 is smaller than the output of the preceding target value generator 17, the predicted value of the control deviation (that is, the output 16a of the adder 16) becomes positive.

Hereinafter, in order to simplify the explanation, only the case where the predicted value of the control deviation is positive will be described.

Now, when using the first manipulated variable calculation circuit OP1,
When the output 21a of FF21 is "1", switch 1
The movable terminal C of 8 is electrically connected to the fixed terminal a, and the output end of the adder 16 is connected to the multiplier 23, so that the first operation amount calculation circuit OP1 is utilized.

Also, at this time, the output 2lb of the FF 21 is also "1", but this output 2lb is caused by the negation element 22.
Since the output 22a of the switch 19 becomes "0", the movable terminal C of the switch 19 is electrically connected to the fixed terminal b.

Therefore, the value "0" from the zero setter 21A passes through the switch 19 and is applied to the coefficient multipliers 31.32. The outputs of 32 are all zero, that is, the second manipulated variable calculation circuit OP2 is cut off.

In this state, when the output of the adder 24 is "1",
The multiplier 23 receives the output 16a of the adder 16 and multiplies it with the output "1" of the adder 24.

However, since the output of the adder 24 is "1", the output of the multiplier 23 is sent to the coefficient unit 29 by directly outputting the value of the output 16a of the adder 16 as a positive value.

As a result, the coefficient unit 29 outputs a positive or negative output depending on whether the set magnification k1 is positive or negative.

If the set magnification k1 is positive, the output of the limiter addition integrator 30 increases, the output of the adder 33 also increases, and the output of this adder 33 is introduced into the limiter 14.

In this limiter 14, if the output of the adder 33 does not reach the upper limit f)), an output from the output terminal of the limiter 14,
That is, the operation signal 14a increases and is sent to the controlled object 11 and the prediction device 13.

Therefore, the control amount 11a of the controlled object 11 tends to increase, leading to lead time! l-At works to match the target value as much as possible.

Next, describing the case where the output of the adder 24 is "0", the output "0" of the adder 24 is added to the multiplier 23.

The output 16a of the adder 16 is applied to the multiplier 23 through the switch 18, but the output of the adder 24 is "0".
”, the output of the multiplier 23 is also “O”, the output of the coefficient unit 29 is also “O”, and the limiter addition integrator 3
An output of 0 is held.

Therefore, the outputs of the adder 33 and limiter 14 are held at their previous states, so the operation signal 14a does not change.

Next, a case will be described in which the outputs 21a and 2lb of the FF 21 become rOJ and the second manipulated variable calculation circuit OP2 is used.

In this case, since the output 21a of the FF 21 becomes "O", the movable terminal 18c of the switch 18 becomes the fixed terminal 18a.
power) is switched to the fixed terminal 18b.

Therefore, the value "O" from the zero setter 21A is applied to the multiplier 23 through the switch 18, so the output of the multiplier 23 becomes rOJ, and the output of the coefficient multiplier 29 also becomes "O".
becomes.

Therefore, the first multiplication amount calculation circuit OP1 is substantially cut off.

On the other hand, the output 21b of the FF 21 is also "0", but this output 2lb is inverted by the negation element 22, and the output 22a of the negation element 22 becomes "1".

As a result, the movable terminal C of the switch 19 is switched from the fixed terminal b to the fixed terminal a.

As a result, the output 16a of the adder 16 is input to the coefficient multiplier 31 through the switch 19.

Then, in the same manner as described above, the output 1 of the adder 16 is
6a, that is, when the predicted value of the control deviation is positive, the output of the coefficient multiplier 31 becomes a positive or negative value depending on whether the setting magnification kp is positive or negative, respectively.

Now, if this set magnification kp is positive, the coefficient unit 3
1 becomes positive, and the outputs of the limiter addition integrator 30 and the adder 33 increase.

Then, when the output 33a of the adder 33 enters the limiter 14, if it does not reach the upper limit of the limiter 14, the limiter 1
4, that is, the operation signal 14a increases, and the control amount 11a of the controlled object 11 is made to almost match the target value after a lead time l·,{1.

Next, the operation of the switching logic operation circuit SE which performs the switching operation of the switches 18 and 19 will be explained.

Here, first, the switching conditions of the first manipulated variable calculation circuit OP1 and the second manipulated variable calculation circuit OP2 are listed as below, and then, the action of this switching logic calculation circuit SE realizes the conditions. Explain that it is a thing.

(Switching condition 1) First manipulated variable calculation circuit OP1 and second manipulated variable calculation circuit O
The switching of P2 is performed only in the first step of the lead time l-Jt, and is not changed in the remaining (A!-1) steps.

(Switching condition 2) In principle, if the predicted value of the control deviation ahead of lead time II-At is larger than the preset switching deviation, the first manipulated variable calculation circuit OP1 is used; if it is smaller than the switching deviation, the second The operation amount calculation circuit OP2 is used.

(Switching condition 3) However, switching from the first manipulated variable calculation circuit OP1 to the second manipulated variable calculation circuit OP2, or vice versa, takes lead time! ! -At is compared with past signals, and is carried out only when the same switching signal continues twice.

(Switching condition 4) When using the first manipulated variable calculation circuit OP1, the manipulated variable (
(corresponding to the operation signal 14a) is not always changed, but is set only for the first n-Jt time of the lead time II-At time, and thereafter (Il-n)·At time is held constant.

This is to improve the accuracy of predicted values.

When using the second manipulated variable calculation circuit OP2, the manipulated variable is always changed.

Next, it will be explained that the operation of the switching logic operation circuit SE in the embodiment of FIG. 2 realizes the above (switching condition 1).

In this case, it is sufficient to explain one period of the pulse generated by the pulse generator 25, and in the first step in this one period, that is, O≦t<Jt, the output 25a of the pulse generator 25, 25b is "1".

Therefore, the switches 27, 28 each have a movable terminal C
is electrically connected to fixed terminal a.

As a result, the absolute value operational amplifier 20, the sign converter 35,
The contents calculated by the adder 34, the function generator 36, the dead time element 37, the exclusive OR element 38, the adder 39, and the negation element 40 (described in detail later) are input to the input terminal of the FF 21 (the output y1 of the switch 28). is introduced), the second input end (switch 2
3 (a diagram showing the input/output relationship of FF21) and FIG. 4 (exclusive OR element 3
The switching is performed according to the logic shown in the truth table of No. 8).

On the other hand, after the second step during the lead time 13-At, that is, when At≦t<ll-1t, the outputs 25a and 25b of the pulse generator 25 are "O", so the switches 27 and 28 are Each movable terminal C is electrically connected to the fixed terminal b.

Therefore, this switch 27. 28 is a zero setter 41
from the first input terminal to the third input terminal of FF21.
are transmitted to the input terminals of the respective input terminals.

Therefore, the output of FF21 is "O" if the previous step is rOJ, "1" if it is "1", and the output of FF21 is "1" if the previous step is rOJ.
The manipulated variable calculation circuit OP1 and the second manipulated variable calculation circuit OP2
No switching is performed.

Next, I will explain (switching condition 2) and (switching condition 3) together.

However, since it has been confirmed that (switching condition 1) is satisfied as described above, it is sufficient to explain only the first step of each cycle.

Now, the output 25a of the pulse generator 25. 25b is "1" in the first step of each cycle, so the switch 27
．． All of the 28 movable terminals C are electrically connected to the fixed terminal a.

In this state, a case where the output of the function generator 36 was r-IJ after lead time l-jjt will be described.

In this case, before that, the absolute value operational amplifier 20
The absolute value (output 20a) of the predicted value of the control deviation (i.e., the output 16a of the adder 16) obtained in is larger than the input of the code converter 35, the output of the output 34a of the adder 34 is positive, and the function generator This is a case where the output of 3T is "-1", and the output of dead time element 3T is "-1".

Also, the lead time il-At past the function generator 36
A case where the output 20a of the absolute value operational amplifier 20 is larger than the switching error (output 35a of the error setter 35) will be described when the output of the absolute value operational amplifier 20 is "-1".

At this time, the output 34a of the adder 34 is positive, the output of the function generator 36 is "-1", and the input to the exclusive OR element 38 is "-1", so the output is "O".

Therefore, the output of the negation element 40 is "1".

However, the output of the adder 39 is "-2".

As a result, the first input terminal of the FF 21, that is, the output y1 of the adder 39 via the switch 28 is negative, and the FF
21 second input terminal (negation element 40 via switch 2γ
Since the output y2) of FF21 is positive, the output 21a of FF21,
2lb becomes "1", and as already mentioned, the movable terminal C of the switch 18 is electrically connected to the fixed terminal a.

As a result, the first manipulated variable calculation circuit OP1 is utilized.

At the same time, the output 22a of the negative element 22 becomes "O", so the movable terminal C of the switch 19 becomes electrically connected to the fixed terminal b.

As a result, the value "0" from the zero setter 21A is applied to the coefficient unit 31 through the switch 19, so that the second manipulated variable calculation circuit OP2 is substantially cut off.

From this, it is clear that (switching condition 2) and (switching condition 3) are satisfied.

Next, the function generator 36
A case will be described in which the output 20a of the absolute value operational amplifier 20 is smaller than the switching error when the output of the absolute value operational amplifier 20 is "-1".

In this case, the output 34a of the adder 34 is negative, the output of the function generator 36 is "1", the inputs of the exclusive OR element 38 are "-1" and "1", and the output is "1". becomes.

Therefore, the output of the negation element 40 is "0".

This output is applied to the second input terminal of the FF 21 through the switch 27 as y2=0.

On the other hand, the output of the adder 39 is "0", and this output is passed through the switch 28 to the first input terminal of the FF 21 (y1=0).
Join as.

Since y3=0 is added from the zero setter 41 to the third input terminal of the FF 21, the first to third input terminals of the FF 21 eventually become yi=o. yz=o,y
3=o, and the output y4 (corresponding to outputs 21a, 2lb) is y4=o, y if the value y4a of the previous step is O.
If 4a=1, then y4=1.

Therefore, the first manipulated variable calculation circuit OP1 is not switched.

Then, in the past lead time ll-1t, the function generator 36
The case where the output is "1" will be explained.

This means that in the past lead time l-At, the absolute value of the predicted value of the control deviation obtained by the absolute value operational amplifier circuit 20 is the switching deviation (as will be described later, the output of the integrator with limiter 60).
smaller, the output of adder 34 is negative and the output of function generator 36
Since this is the case where the output of is "1", the output of the dead time element 37 is "1".

If the output of the function generator 36 was "1" in the past during this lead time l-At, the absolute value operational amplifier 2
The case where the output of 0 is larger than the switching deviation will be described.

At this time, the output of the adder 34 is positive, and the output of the function generator 36
The output of is "-1" and the inputs of exclusive OR element 38 are "1" and "-1", therefore its output is "
1”.

Therefore, the output of the negation element 40 becomes "O", and this output is passed through the switch 27 to the second input terminal of the FF 21.
Add as 2=0.

Further, the output of the adder 39 is "0", and this output is passed through the switch 28 to the first input terminal of the FF 21, y1=0.
Join as.

Further, Y3=O is applied from the zero setter 41 to the third input terminal of the FF 21.

As a result, the first to third input terminals of the FF 21 become y i=0, y 2=O − '/3=O, and the output of the FF 21 maintains the value of the previous step.

That is, in this case as well, since the determination result (output 36a of the function generator 36) of (switching condition 2) did not take the same value twice, the condition (switching condition 3) of not switching is satisfied.

Next, the lead time l-j: it is past the function generator 3
A case where the output of the absolute value operational amplifier circuit 2c is smaller than the switching deviation when the output of the absolute value operational amplifier circuit 2c is "1" will be described.

In this case, the output of adder 34 is negative and function generator 36
The output of the adder 39 is "1", and the output of the adder 39 is "2".

Therefore, at the first input terminal of FF21, y'i=z>
o), the output of FF21 becomes rOJ, and according to the action described in the section when using the second manipulated variable calculation circuit OP2, the first manipulated variable calculation circuit OP1
is substantially cut, and the second manipulated variable calculation circuit OP2 is utilized.

As is clear from the above description, (switching condition 2) and (switching condition 3) can be realized.

Finally, (switching condition 4) will be described.

In this case, when using the first manipulated variable calculation circuit OP1, the output of the FF 21 is "1", the output of the negative element 22 is "O", and the movable terminal C of the switch 18 is not electrically connected to the fixed terminal a. As described above, the movable terminal C of the switch 19 is electrically connected to the fixed terminal b, so that the first operation amount calculation circuit OP1 is utilized and the second operation amount calculation circuit OP2 is cut off.

As mentioned earlier, the pulses generated by the pulse generator 25 have a height of 1, a pulse width of Jt, a period of 11-1t,
The initial operating time is O, and the pulse generated by the pulse generator 26 has a height of 1, a pulse width (n 1)・Jt,
The period is 11-1t, and the initial operation time is Jt.

Each of these pulses is applied to an adder 24, and the output of the adder 24 is "1" for the first n steps (n-Jt) of each cycle lJt, and is rOJ for the rest.

Therefore, in each cycle l-lt, only the first n steps correspond to the operation when the output of the adder 24 is "1" when the first manipulated variable calculation circuit OP1 is used, that is, the operation amount is This is the case when changes are made.

Other than that, it is possible to realize the manipulated variable hold effect when the output of the adder 24 is "O" when the first manipulated variable calculation circuit OP1 is used.

On the other hand, when the second manipulated variable calculation circuit OP2 should be used, the output of the FF21 is "O" and the output of the negative element 22 is "1".
”.

In this case, the movable terminal C of the switch 18 is the fixed terminal B.
The movable terminal C of the switch 19 is electrically connected to the fixed terminal a.

Therefore, as described above, the first manipulated variable calculation circuit OP1 is cut and the second manipulated variable calculation circuit OP2 is utilized.

In this case, since there are no other conditions, FF2
If the output of OP2 is "O", the second manipulated variable calculation circuit OP2 is always activated.

As is clear from the above description, (switching condition 4) is also satisfied.

Next, the operation of the switching deviation calculation circuit and switching signal reset circuit, which are important features of the present invention, will be described.

Of these, the former switching deviation calculation circuit will be explained first.

CI,l Switching Deviation Calculation Circuit Here, we will first describe the conditions for changing the switching deviation εset, and then explain how the operation of this switching deviation calculation circuit satisfies those conditions.

Conditions for changing the switching deviation (4) The switching deviation εset is changed at each cycle (period l・Jt
) is performed in the first step, and is not performed in the remaining (Il-1)·Jt time.

(B) The manipulated variable y (output of limiter 14) is l・,J
If the sampling time is set to t and enters both end regions (y>U1-α or y>L1+α) twice, the switching deviation εset is increased, and in other cases, the switching deviation εs is increased.
Decrease et.

[Explanation of row crop N]

Regarding condition (4), one period of the pulse generator 25 I
It is sufficient to explain only the I-At time.

In the first step in one cycle, the pulse generator 25
Since the output of the switch 58 is "1", the movable terminal C and the fixed terminal a of the switch 58 are brought into conduction by the output 25d of the pulse generator 25.

As a result, the adder 42, the function generator 47, . Adder 4
4, function generator 48, adder 49, dead time element 51,
AND element 50, switch 52, zero setter 53, RSF
The results calculated by F54, negation element 55, adder 56, and constant generator 57 (the contents will be described in the explanation of condition (B)) are transmitted to coefficient unit 59 and integrator with limiter 60, and the switching deviation εset is Make changes.

At the (x-i)·Jt time after the second step in one cycle, the output of the pulse generator 25 is rOJ, so the switch 58 is activated by the output 25d of the pulse generator 25.
The movable terminal C is electrically connected to the fixed terminal b.

Therefore, the movable terminal C of the switch 58 is connected to the zero setter 5.
Since the value IOj from 3 is transmitted, the output of the coefficient multiplier 59 is "0", and therefore the integrator with limiter 60
The output of is unchanged, and the value of the switching deviation εset is also unchanged.

Condition (4) can be realized by the above action.

[Explanation of condition (B)] First, it will be shown that the output of the RSFF 54 does not change during each cycle except at the first step.

Since the output 25c of the pulse generator 25 is "O" except for the first step of each cycle, the movable terminal C of the switch 52 is electrically connected to the fixed terminal b, and therefore the movable terminal C
The value “0” is added from the zero setter 53.

As a result, the first input terminal of the RSFF 54 is set to "O", the second input terminal is set to "0" by the output 25d of the pulse generator 25, and the third input terminal is set to "0" by the output of the zero setter 53.
”, and the first to third input terminals are all 10j
It is.

Therefore, according to the truth table of FIG. 5, the output will not change to "O" if the value of the previous step is rOJ, and to "1" if it is "1".

As mentioned above, condition (A) is satisfied and RSFF
Since it has been shown that the output of 54 does not change in steps other than the first step, it is sufficient to explain the condition (old) by focusing only on the first step of each cycle.

When the operation signal y (output of the limiter 14, that is, the operation signal 140) is y>U1−α or y−L1+α, y
>01-α (y<L1+α), adders 42 and 4
Since the outputs of 4 are positive (negative) and negative (positive), respectively, the outputs of function generators 47'j6 and 48 are 1 (
0) and O(1).

Note that here, the value in parentheses is the case where y<L 1 +α.

Therefore, the output of the adder 49 becomes "1" both when 3>U1-α and when L1+α.

On the other hand, if the operation signal y is L1+α≦y<u'i−α,
The outputs of adders 42 and 43 are both negative or "0"
It is.

Therefore, the outputs of function generators 47 and 48 are both "0".
Therefore, the output of the adder 49 also becomes "O".

As mentioned above, whether the output of the adder 49 is "1" or
Therefore, it is possible to determine whether or not each operation signal falls within the both end regions (y>U1-α or y<Ll+α).

Also, wasted time l-,! Dead time element 51 with ft
is calculated using the system of adder 42, function generator 47, and adder 49, and adder 44, function generator 48, and adder 49. = A - The value from Jt time ago is set as the current output.

Therefore, the AND element 50 functions to compare the determination result from l-At time ago with the determination result at the present time.

In other words, if the operation signal y enters the both end regions twice in succession,
Both inputs of this AND element 50 are "1", and the output is "1"; in other cases, one or both of the two inputs is "O", so the output is "1". O”.

Next, switch 52, zero setter 53, RSFF 54, negative element 55, adder 56, constant generator 57, switch 58
, the coefficient unit 59, and the integrator with limiter 60. Note that in the first step of each cycle, the outputs 25c and 25d of the pulse generator 25 are "1", and the switches 52 and 58 are all set. , assuming that the movable terminal C and the fixed terminal a are electrically connected, and that the signal y2 introduced to the second input terminal of the RSFF54 is "1", the following two cases are considered. explain.

(1) When the output of the AND element 50 is "1", RSFF
Since the first human power of RSFF 54 is "1", that is, positive, according to the truth table of FIG. 5
The output of the constant generator 57 is "1", and the constant r of the constant generator 57 is 0 <
Since r is 1, the set magnification k2 of the oa1-rs coefficient generator 59 is positive, so the output of the coefficient multiplier 59 is positive, and therefore the output of the integrator with limiter 60 increases.

([1) When the output of the AND element 50 is “0”, RSF
The first and second input terminals of F54 are rOJ [
j (y1=o, y2=x), so RSFF 5
The output of 4 is "1", the output of negation element 55 is "O", the output of adder 56 is negative at -r, and the output of coefficient unit 59 is 1 k2r
becomes negative.

Therefore, the output of the integrator with limiter 60, that is,
The switching deviation εset decreases.

As described above, condition (B) can also be realized.

In contrast to the integrator with limiter 60, the limiter prevents the switching deviation εset from becoming a value that has no practical meaning (for example, the switching deviation εset becomes a negative value or a huge positive value). It is for.

(n) Operation signal reset circuit Next, the operation of this operation signal reset circuit will be explained.

The dead time element 61 outputs the operation signal of l-At time past as the current output, the adder 62 adds the current operation signal to this, and the result is sent to the coefficient unit 63 as J0. 5 j
Multiply by

That is, the average value of the manipulated variable at the current time and the value before It-At time is taken.

This value and switching of the switch 64 reset the output of the limiter addition integrator 30 as a reset signal for the manipulated variable.

This reset is performed when the switch 64 is on, that is, when the movable terminal b and the fixed terminal a are electrically connected, and this state is determined as follows.

Whether the first manipulated variable calculation circuit OP1 or the second manipulated variable calculation circuit OP2 is used depends on whether the output of the FF 21 is "1" or "O".

Therefore, the time point at which switching is to be performed can be determined by finding the time point at which the value of the previous step of the FF 21 and the current value change.

The output of the FF 21 in the previous step is output from a dead time element 65 having a dead time Jt.

The exclusive OR element 66 outputs "1" when the two manpowers are different, and outputs "0" when the two manpowers are the same, so when the output of the exclusive OR element 66 is "1", Switch 64 is turned on.

In other words, the movable terminal b of the switch 64 is electrically connected to the fixed terminal a.

As a result, the output of the coefficient multiplier 63 passes through the switch 64 and resets the limiter addition integrator 30.

As is clear from the above explanation, in this invention,
When the first manipulated variable calculation circuit OP1 with a high gain is used, the level of the controlled object changes and the gain becomes too high, causing large vibrations in the control system and causing problems in the prediction device 13. There may be cases where predictions do not go well due to the addition of disturbances due to forces that have not been taken into account.

Such an abnormal state is determined based on the magnitude of the operation signal, and the second manipulated variable calculation circuit with a low gain is temporarily used to stabilize the abnormal state, thereby producing an effect of escaping the abnormal state.

In other words, when an abnormal state occurs, the operation signal y is within the upper and lower limits (y>Ui-α or y<
L1+α), and in this case, if the first manipulated variable calculation circuit OPI is used, a sensitive control operation with a high gain will be performed, which may aggravate the abnormal state, so the switching deviation εset is increased. It is possible to easily switch to the second operation amount calculation circuit OP2 with a low gain, and also to reset the operation signal that vibrates at the time of switching or is at the end of the movable range to stabilize it and escape from an abnormal state.

On the other hand, when no abnormal condition occurs (L?+α≦y≦U1
-α), by reducing the switching deviation εset, the first manipulated variable calculation circuit OP1 with high controllability can be used for as long as possible, and controllability can be improved.

This means that, as mentioned above, the round-trip gain when using system a, that is, the first manipulated variable calculation circuit, is several times larger than the round-trip gain when using system b, that is, the second manipulated variable calculation circuit. The set magnifications Kl and Kp are adjusted so that the magnifications Kl and Kp become larger than the above.

In other words, even if the gain of the first manipulated variable calculation circuit OP1 is too large and causes what is called hunting in the control system, the second manipulated variable calculation circuit OP2 has a gain that is a fraction of that of OP1. be.

Therefore, if the second manipulated variable calculation circuit OP2 is used, stable control can be performed without causing hunting due to the low round-trip gain.

As described in detail above, γ As described above, according to the simple predictive control device of the present invention, the control amount of the controlled object, the operation signal, and the disturbance signals are inputted and the predicted lead time l Jt ahead (l is an integer, At is a prediction device that outputs a predicted value of the amount of crawling legs (time step width in a discrete time system), and a predicted lead time 11-
The value obtained by multiplying the predicted value of the control deviation by the target value of At destination and the predicted value of the control amount by a constant is calculated as the predicted lead time 11-
a first manipulated variable calculation circuit with a large gain that changes the operation signal by integrating for n-At time (n<A?) of At, and holds the operation signal unchanged at other times; A second manipulated variable calculation circuit with a small gain that eliminates intermittent control operation by constantly changing the amount of the operation signal by adding proportional integral operation to the predicted value of the control deviation, and the absolute value of the control deviation described above The basic judgment condition for determining whether to use the first manipulated variable calculating circuit or the second manipulated variable calculating circuit and to prevent malfunctions is to determine whether the switching deviation is smaller than the switching deviation calculated by the switching deviation calculating circuit. Therefore, the lead time II-At time is compared with the past judgment results, and if the same judgment result continues twice, it is necessary to switch whether to use the first or second manipulated variable calculation circuit. The switching logic operation circuit performs sampling every sampling period of Il-lt time, and when the above-mentioned operation signal comes near both ends of the movable range, it is determined that an abnormality has occurred, and only when this abnormality continues twice, does the above-mentioned? ! 1
a switching deviation calculation circuit that increases the value of the switching deviation, which is a threshold value for determining whether the predicted value of the m deviation is large or small, and reduces the switching deviation when no abnormality has occurred; The gist of the present invention is to include an operation signal reset circuit that resets the operation signal to the average value of the past value of II-It time and the current value when switching between the operation amount calculation circuit and the second operation amount calculation circuit. , it is possible to dramatically improve controllability compared to classical feedback control, and in addition to the effect of being able to perform stable control when the disturbance change is minute or when the stable state is sufficiently approached, it is also possible to prevent abnormalities. Even if a situation occurs, it can be overcome by switching between the two manipulated variable calculation circuits, and when switching, the operation signal at the end of the movable range can be reset and stabilized, greatly improving control accuracy. It is possible.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of a conventional simple predictive control device, FIG. 2 is a block diagram showing the configuration of an embodiment of the simple predictive control device of the present invention, and FIG. 3 is a block diagram showing the configuration of a simple predictive control device of the present invention. Figure 4 shows the operation of the exclusive OR element in the switching logic operation circuit of the simple predictive control device of the present invention. FIG. 5 is a diagram showing a truth table for explaining the operation of the RS flip-flop circuit in the switching deviation calculation circuit of the simple predictive control device of the present invention. 11...Tropod target, 12...Disturbance source,
13... Prediction device, 14... Limiter, 16, 24, 33.34.39.42, 44, 49.
56... Adder, 17... Advance target value generator, 1B. 19.27.2B, 52.58.64・
...Switch, 20...Absolute value operational amplifier, 21A, 41.53...Zero setter, 21.
54...RS flip-flop circuit, 23.
...Multiplier, 25. 26... Pulse generator, 29, 31. 32.59...Coefficient unit, 30...Limiter addition integrator, 36.47.
48...Function generator, 37.51, 61.65
...Dead time element, 38.66...Exclusive OR element, OP1...First manipulated variable calculation circuit, OP2...Second Manipulated amount calculation circuit, SE
...Switching logic calculation circuit, OL...Operation signal reset circuit, SEE...Switching deviation calculation circuit.

Claims (1)

[Claims] 1 Input the control amount of the controlled object, the operation signal, and the disturbance signal to calculate the predicted lead time II-At destination (l is an integer,
Jt is a prediction device that outputs a predicted value of a controlled variable (time step width in a discrete time system), and a predicted lead time jl-A.
The predicted lead time 13-A is calculated by multiplying the predicted value of the control deviation by the target value t ahead and the predicted value of the controlled variable by a constant.
a first manipulated variable calculation circuit with a large gain that changes the operation signal by integrating it for n·Jt time (n<A) out of t, and holds the operation signal unchanged at other times; A second manipulated variable calculation circuit with a small gain that eliminates intermittent control operation by constantly changing the amount of the operation signal by adding a proportional integral operation to the predicted value of the deviation, and the absolute value of the above control deviation is switched as follows. Whether the switching deviation is larger or smaller than the switching deviation calculated by the deviation calculation circuit is used as a basic judgment condition for determining which of the first manipulated variable calculation circuit and the second manipulated variable calculation circuit should be used, and to prevent malfunctions. The above lead time 13-Jt time is compared with the past judgment results, and if the same judgment results continue, execution of switching is performed to use either the first or second manipulated variable calculation circuit. The switching logic circuit performs sampling at every sampling period of l·At time and determines that an abnormality has occurred when the above-mentioned operation signal comes near both ends of the movable range, and performs the above-mentioned control only when this abnormality continues twice. A switching deviation calculation circuit that increases the value of the switching deviation, which is a threshold value for determining whether the predicted value of the deviation is large or small, and reduces the switching deviation when no abnormality has occurred; and the first manipulated variable calculation circuit. and an operation signal reset circuit that resets the operation signal to an average value of a value l-At time past and a current value when switching the second operation amount calculation circuit.