JPH1185214A - Process control device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a process control device that does not require a model of wasted time to be incorporated in the control device and makes use of a merit of a smith method. SOLUTION: A model prediction signal generation means 9 has a model of transfer function that ignores wasted time of a process of a control object 2, inputs a feed forward control signal 22 and outputs a model prediction signal 24 that is obtained by performing a control operation on the basis of the transfer function. A feedback control means 10 outputs a feedback control signal 23 that can be obtained by performing the control operation of a control deviation signal 15, to be inputted, by a differential element. A feedforward control means 13 outputs the feedforward control signal 22 that can be obtained by performing the control operation so that the model prediction signal 24 becomes a set value signal 3.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a process control device for controlling a control target so that a control amount of a process coincides with a set amount, and more particularly to a process control device for a process including a dead time of the target process.

[0002]

2. Description of the Related Art Process control including dead time is performed by PI
The control is generally performed, but the control gain must be reduced due to a dead time, and the control performance is poor because the quick response is impaired. As a countermeasure against the dead time system, the Smith method for predicting a state at a time ahead of the dead time ((1) A Controller to Overc)
homeDead Time J. M. Smith
ISA. J. 6-1 pp. 28-33, 1959,
(2) Control of dead time system Keiji Watanabe (Corporation) Society of Instrument and Control Engineers, pp. 8-12, 1993) is an effective method.

A description will be given below with reference to a schematic block diagram of a Smith method control device shown in FIG.

Here, the controlled object 2 is represented by dead time + first-order delay. The process control device 1 outputs an operation amount signal 5 to control the values of the control amount signal 4 and the set value signal 3 which are outputs of the control target 2, and uses the operation amount signal 5 as an input of the control target 2. The control target 2 changes its state in response to the operation amount signal 5 to a new state, and the control amount signal 4 also changes.
In this way, the control amount signal 4 is controlled to be equal to the set value signal 3. The adding means 6 calculates a deviation between the set value signal 3 and the control amount signal 4, and outputs a control deviation signal 15. The dead time model 12 predicts the effect of the dead time with a model that extracts only the dead time portion of the control target 2, inputs the manipulated variable signal 5, and outputs the dead time model signal 20.
Is output. The adding means 10 outputs a dead time correction signal 21 which is the difference between the manipulated variable signal 5 and the model signal 20 on the dead side.

[0005] The model 11 without dead time is a model in which only the dead time portion of the controlled object 2 is removed (first-order delay model).
Then, the dead time correction signal 21 is input and the control deviation correction signal 18 is output. The adding means 9 calculates the difference between the control deviation signal 15 and the control deviation correction signal 18 and calculates the PI controller input signal 19
Is output.

The proportional-integral control means 25 receives the PI controller input signal 19 as input, performs a proportional / integral operation on the PI controller input signal 19, and outputs the manipulated variable signal 5. At this time, the control object 2 and the prediction model (the model 11 without dead time)
When the dead time model 12) completely matches, the control system becomes equivalent to the schematic block diagram shown in FIG. 18, the dead time can be canceled, and the controlled object 22 can be quickly and stably controlled by the proportional-integral control means 25. Can control.

[0007]

The Smith method shown in FIGS. 17 and 18 is an effective means as a countermeasure for the dead time, but it is necessary to incorporate the dead time model 12 into the process control device 1. In such a case, mismatch between the model 12 and the control target 2 in the dead time may greatly impair controllability. Also, depending on the target system,
The dead time may change dynamically, which makes modeling difficult. Furthermore, the size of the dead time L
Becomes very large as compared with the control cycle of the control device, the storage space of the model 12 with the dead time increases. As described above, it may be difficult to realize the model 12 with the dead time, which is a practical bottleneck.

Accordingly, an object of the present invention is to provide a high-performance process control device which does not require incorporating a dead time model into a control device and makes use of the advantages of the Smith method.

[0009]

According to the first aspect of the present invention, a process including a dead time element is controlled, and an operation amount signal is output to the control object so that the control amount signal of the process becomes a set value signal. A process control device that has a transfer function model that ignores the dead time of a process to be controlled, and inputs a feedforward control signal for predictive control and performs a control calculation based on the transfer function to obtain a model prediction signal. And a feedforward control means for outputting a feedforward control signal obtained by performing a control operation so that the model prediction signal output by the model prediction signal generation means becomes a set value signal. While outputting the addition signal of the feedforward control signal and the feedback control signal as a manipulated variable signal to the controlled object Switching means for switching so as to output only the feedforward control signal to the control target from when the control deviation signal between the model prediction signal and the set value is equal to or more than a predetermined value to a dead time corresponding to the control target is provided. It was done. According to this means, when the control deviation signal, which is a deviation signal between the set value signal and the control amount signal, is small,
A controlled object including a dead time is stably controlled by an operation amount based on a feedback control signal that is control-calculated to be a set value signal and a control amount signal. On the other hand, when the set value signal or the control amount signal changes and the control deviation signal exceeds a predetermined value, a large control deviation signal is output, and the output of the feedback control signal is prohibited by the switching means. As a result, when the control deviation signal is equal to or more than the predetermined value and during a time corresponding to the dead time, a predicted signal is generated by a control target using a transfer function ignoring the dead time, and feedforward control is performed based on the predicted signal. Signals are generated and controlled at high speed. Eventually, when the control deviation signal becomes small, becomes less than a predetermined value, and a time equivalent to a dead time elapses,
Prohibition of the output of the large control deviation signal is released, and the feedback control signal is output. Therefore, it is not necessary to incorporate a dead time model based on the Smith method in the apparatus. As a result, it is possible to eliminate the possibility that controllability may be impaired due to mismatch between the dead time model and the control target. Further, it is possible to cope with a case where the dead time of the control target dynamically changes. Further, it is possible to solve the problem that if the size of the dead time of the model of the dead time is much larger than the control cycle of the process control device, the memory space of the model of the dead time increases, which is not practical.

According to a second aspect of the present invention, in the process control device according to the first aspect, the feedback control means outputs a feedback control signal obtained by controlling and calculating an input control deviation signal by an integral element. It is. According to this means, it is not necessary to provide a model of the dead time by the Smith method in the apparatus, and it is possible to avoid the adverse effects caused by providing the model of the dead time in the apparatus. In addition, since the feedback control means is constituted by the integral element, the control deviation signal is normally calculated and controlled by the integral element to perform the feedback control so that the stable altitude control is performed.

According to a third aspect of the present invention, in the process control device according to the first aspect, the feedback control means outputs a feedback control signal obtained by controlling and calculating an input control deviation signal by a proportional and integral element. It was done. According to this means, it is not necessary to provide a model of the dead time by the Smith method in the apparatus, and it is possible to avoid the adverse effects caused by providing the model of the dead time in the apparatus. In addition, since the feedback control means is constituted by the proportional integral element, the control deviation signal is controlled and calculated by the proportional integral element at the normal time, and the feedback control is performed, so that the stable altitude control is performed.

According to a fourth aspect of the present invention, in the process control device according to the first aspect, the feedforward control means includes:
A feedforward control signal obtained by performing a control operation using a control constant determined by fuzzy inference according to the model prediction signal and the set value signal is output. According to this means, it is not necessary to provide a model of the dead time by the Smith method in the apparatus, and it is possible to avoid the adverse effects caused by providing the model of the dead time in the apparatus. Further, since the feedforward control means is constituted by the fuzzy control means, when the predicted control deviation signal is equal to or more than a predetermined value, the control constant is determined according to the value of the predicted control deviation signal, and the response can be made at high speed.

According to a fifth aspect of the present invention, in the process control device according to the first aspect, the feedforward control means includes:
A feedforward control signal obtained by performing a control operation using a control constant determined by self-organizing fuzzy inference according to the model prediction signal and the set value signal is output. According to this means, it is not necessary to provide a model of the dead time by the Smith method in the apparatus, and it is possible to avoid the adverse effects caused by providing the model of the dead time in the apparatus. Further, since the self-organizing fuzzy control means is configured as the feedforward control means, the control constant is determined so that the feedforward signal corresponding to the dead time responds at high speed when the predicted control deviation signal is equal to or more than a predetermined value.

According to a sixth aspect of the present invention, there is provided a process control apparatus for controlling an output of an operation amount signal to a control target such that the control amount signal of the process becomes a set value signal as a process control target including a dead time. Predictive signal generating means having a transfer function model ignoring the dead time of the process of claim 1, inputting a feedforward control signal for predictive control, and outputting a model predictive signal obtained by performing a control operation based on the transfer function Feedforward control means for outputting a feedforward control signal obtained by performing control operation so that the model prediction signal output by the model prediction signal generation means becomes a set value signal; and a control deviation signal and a prediction control deviation signal. When the control deviation signal is smaller than the predetermined value, the control deviation signal is controlled by the fuzzy inferred control constant. A fuzzy signal is output to select the feedback control signal obtained by the calculation, while selecting the feedback control signal to be zero from the time when the control deviation signal is equal to or more than a predetermined value to the time corresponding to the dead time of the controlled object. There is provided a selection control means, and an addition means for adding the feedback control signal and the feedforward control signal to output an operation amount signal to a control target. According to this means, it is not necessary to provide a model of the dead time by the Smith method in the apparatus, and it is possible to avoid the adverse effects caused by providing the model of the dead time in the apparatus. Further, the switching means according to the first to fifth aspects of the present invention is not required, and a fuzzy operation is executed according to the value of the predicted control deviation signal and the control deviation signal to switch between the feedback control signal and the feedforward control signal. Switching can be performed smoothly, and fluctuation of the control amount signal accompanying the switching can be avoided.

[0015]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram of a process control device according to a first embodiment of the present invention.

In FIG. 1, the same reference numerals as those in FIG. 17 showing the conventional example indicate the same or corresponding parts. The process control device 1A includes a model prediction signal generation unit 9, a feedback control unit 10, and a feedforward control unit 1.
3 and switching means 17 and the like.

Here, the model prediction signal generating means 9 has a transfer function model ignoring the dead time of the process of the controlled object 2, and inputs a feedforward control signal 22 for predictive control to the transfer function. It outputs a model prediction signal obtained by performing a control operation based on the model prediction signal. The feedback control means 10 outputs a feedback control signal 23 obtained by controlling and calculating an input control deviation signal by an integral element. The feedforward control unit 13 outputs a feedforward control signal 22 obtained by performing a control operation so that the model prediction signal 24 output by the model prediction signal generation unit 9 becomes the set value signal 3.

With the above configuration, the control deviation signal 15 obtained by inputting the control amount signal 4 and the set value signal 3 from the control target 2 to the adding means 6 is output via the contact 17 a of the switching means 17. I have. As a result, the feedback control unit 10 performs a predetermined control operation, and outputs the obtained feedback control signal 23 to the adding unit 8. On the other hand, the feedforward control signal 22 output from the feedforward control unit 13 is input to the model prediction signal generation unit 9, and is operated by a model of a transfer function ignoring the dead time of the control target 2, and the model prediction signal 24 is output.
Is output.

Subsequently, a prediction control deviation signal 16 obtained by inputting the model prediction signal 24 and the set value signal 3 to the addition means 7 is output to the feedforward control means 13.
The feedforward control unit 13 to which the prediction control deviation signal 16 has been input outputs a feedforward control signal 22 obtained by executing a predetermined operation to the addition unit 8 and the model prediction signal generation unit 9. The adder 8 receives the feedback control signal 23 and the feedforward control signal 22 and outputs the manipulated variable signal 5 to the control target 2.

When the predicted control deviation signal 16 is smaller than a predetermined value, the contact 17a is closed and the control deviation signal 15 is output to the feedback control means 10 by the switching means 17 for inputting the predicted control deviation signal 16. On the other hand, when the prediction control deviation signal 16 is greater than or equal to a predetermined value by the switching means 17,
The contact 17a is opened for a time equivalent to the dead time of the control target 2, and the input to the feedback control means 10 is set to zero.

As a result, the normal controlled object 2 becomes stable,
Further, when the set value signal 3 is not changed, the contact 17a is closed when the control deviation signal 15 is small and less than a predetermined value, and the control deviation signal 15 is
And a predetermined control operation is performed, and stable feedback control is performed by the feedback control signal 23. Thereafter, when the set value signal 3 changes as required, the predicted control deviation signal 16 becomes a predetermined value or more, the contact 17a is opened by the switching means 17, and the predicted control deviation signal 16 is calculated by the feedforward control means 13. The feedforward control signal 22 is controlled by the feedforward control signal 22 for a time equivalent to the dead time of the control target 2. As a result, the prediction control can be performed with good response at high speed, ignoring the transfer function of the dead time. When the predictive control is performed and the predictive control deviation signal 16 decreases, the feedback control means 10 performs stable and high-precision control in accordance with the control deviation signal 15.

As described above, according to the first embodiment, it is not necessary to incorporate a dead time model based on the Smith method in the process apparatus. As a result, there is a mismatch between the dead time model and the control target 2. Thus, the risk of impairing controllability can be eliminated. Further, it is possible to cope with a case where the dead time of the control target 2 dynamically changes. Further, it is possible to solve the problem that if the size of the dead time of the model of the dead time is much larger than the control cycle of the process control device, the memory space of the model of the dead time increases, which is not practical.

FIG. 2 is a block diagram of a process control device according to a second embodiment of the present invention.

In FIG. 2, the same reference numerals as those in FIG. 17 showing the conventional example indicate the same or corresponding parts. The process control device 1B comprises a dead time model 11, an integral control means 14, a feed forward control means 13, a switching means 17 and the like. The second embodiment of the present invention is the same as the first embodiment of FIG. Is characterized in that the feedback control means 10 is constituted by an integral element to form an integral control means 14.

With the above configuration, the control deviation signal 15 obtained by inputting the control amount signal 4 and the set value signal 3 from the control target 2 to the adding means 6 is output via the contact 17 a of the switching means 17. I have. As a result, the integration control means 14 performs the integration operation of the following equation (1) and outputs the obtained feedback control signal 23 to the addition means 8.

1 / Ti1 · s (1)

On the other hand, a feedforward control signal 22 output from the feedforward control means 13 is input to the dead timeless model 11 and the transfer function of the transfer function represented by the following equation (2) ignoring the dead time of the controlled object 2 An operation is performed by the model, and a model prediction signal 24 is output.

1 / (1 + T · s) (2)

Subsequently, a prediction control deviation signal 16 obtained by inputting the model prediction signal 24 and the set value signal 3 to the adding means 7 is output to the feedforward control means 13.
In the feedforward control means 13 to which the prediction control deviation signal 16 has been input, the feedforward control signal 22 obtained by executing the control operation represented by the following equation (3) is added to the addition means 8 and the model prediction signal generation means 9. Output.

Kp2 {1+ (1 / Ti2 · s)} (3)

The adder 8 receives the feedback control signal 23 and the feedforward control signal 22 and outputs the manipulated variable signal 5 to the control target 2.

When the predicted control deviation signal 16 is smaller than a predetermined value, the contact 17a is closed by the switching means 17 for inputting the predicted control deviation signal 16, and the control deviation signal 15 is output to the integral control means 14. On the other hand, when the prediction control deviation signal 16 is equal to or greater than the predetermined value
The contact 17a is opened for a time equivalent to the dead time, and the input to the integration control means 14 is set to zero.

As a result, the normal controlled object 2 becomes stable,
When the set value signal 3 is not changed, the contact 17a is closed when the control deviation signal 15 is small and less than a predetermined value, the control deviation signal 15 is input to the integration control means 14, and a predetermined control operation is performed. Then, stable feedback control is performed by the feedback control signal 23. Thereafter, when the set value signal 3 changes as required, the predicted control deviation signal 16 becomes a predetermined value or more, the contact 17a is opened by the switching means 17, and only the predicted control deviation signal 16 is calculated by the feedforward control means 13. And
The feedforward control signal 22 is controlled by the feedforward control signal 22 for a time equivalent to the dead time of the control target 2. As a result, the prediction control can be performed with good response at high speed, ignoring the transfer function of the dead time. When the predictive control is performed and the predictive control deviation signal 16 becomes small, the contact 17a is closed and the integral control means 14
As a result, stable and highly accurate control is performed according to the control deviation signal 15.

As described above, according to the second embodiment, it is not necessary to incorporate a model of the dead time by the Smith method into the apparatus, and as a result, the model of the dead time is inconsistent with the control target 2. The risk of impairing controllability can be eliminated, and the adverse effects caused by incorporating the dead time model in the device can be eliminated.

FIG. 3 is a block diagram of a process control device according to a third embodiment of the present invention.

In FIG. 3, the same reference numerals as those in FIG. 17 showing the conventional example indicate the same or corresponding parts. The process control device 1C includes the dead timeless model 11, the proportional-integral control means 25, the feedforward control means 13, the switching means 17, and the like.
The first embodiment is characterized in that the feedback control means 10 of the first embodiment is constituted by a proportional-integral element to form a proportional-integral control means 25.

With the above configuration, the control deviation signal 15 obtained by inputting the control amount signal 4 and the set value signal 3 from the control target 2 to the adding means 6 is output via the contact 17 a of the switching means 17. I have. Thereby, the proportional-integral control means 2
In step 5, the control operation represented by the following equation (4) is executed, and the obtained feedback control signal 23 is added to the adding means 8
Output to

Kp1 {1+ (1 / Ti1 · s)} (4)

On the other hand, the feedforward control signal 22 output from the feedforward control means 13 is input to the dead timeless model 11, and is calculated by the transfer function model ignoring the dead time of the controlled object 2, and the model prediction signal 24 is generated. Is output.

Subsequently, a prediction control deviation signal 16 obtained by inputting the model prediction signal 24 and the set value signal 3 to the addition means 7 is output to the feedforward control means 13.
The feedforward control means 13 to which the prediction control deviation signal 16 has been input, adds the feedforward control signal 22 obtained by executing the same operation as the equation (3) according to the second embodiment to the addition means 8 and the model prediction signal generation means. 9 is output. The adder 8 receives the feedback control signal 23 and the feedforward control signal 22 and outputs the manipulated variable signal 5 to the control target 2.

When the predicted control deviation signal 16 is smaller than a predetermined value, the contact 17a is closed and the control deviation signal 15 is output to the proportional-integral control means 25 by the switching means 17 for inputting the predicted control deviation signal 16. On the other hand, when the predicted control deviation signal 16 is equal to or larger than the predetermined value by the switching means 17, the contact 17a is opened for a time equivalent to the dead time of the control target 2,
The input of the proportional integral control means 25 is set to zero.

As a result, the normal controlled object 2 becomes stable,
When the set value signal 3 is not changed, the contact 17a is closed when the control deviation signal 15 is small and less than a predetermined value, and the control deviation signal 15 is inputted to the proportional-integral control means 25, and the equation (4) is obtained. Is performed, and stable feedback control is performed by the feedback control signal 23. Thereafter, when the set value signal 3 changes as necessary,
When the prediction control deviation signal 16 becomes equal to or more than a predetermined value, the switching means 1
7, the contact 17a is opened and the predicted control deviation signal 1
6 is calculated by the feedforward control means 13,
The feedforward control signal 22 is controlled by the feedforward control signal 22 for a time equivalent to the dead time of the control target 2. As a result, the prediction control can be performed with good response at high speed, ignoring the transfer function of the dead time. And
When the prediction control is performed and the prediction control deviation signal 16 becomes small, the control deviation signal 15
, Stable and high-precision control is performed.

As described above, according to the third embodiment, it is not necessary to incorporate a dead time model by the Smith method into the apparatus, and as a result, the mismatch between the dead time model and the control target 2 is caused. It is possible to eliminate the risk of impairing the control system and to cope with a case where the dead time of the control target 2 dynamically changes.

FIG. 4 is a block diagram of a process control apparatus according to a fourth embodiment of the present invention.

In FIG. 4, the same reference numerals as those in FIG. 17 showing the conventional example indicate the same or corresponding parts. The process control device 1D includes the dead time model 11, the integral control means 14, the fuzzy control means 28, the switching means 17, and the like. The fourth embodiment is different from the feed control apparatus of the second embodiment shown in FIG. A feature is that a fuzzy control means 28 is used instead of the forward control means 13.

With the above arrangement, the control deviation signal 15 obtained by inputting the control amount signal 4 and the set value signal 3 from the control target 2 to the adding means 6 is output via the contact 17 a of the switching means 17. I have. As a result, the integration control means 14 executes the control calculation represented by the equation (1), and outputs the obtained feedback control signal 23 to the addition means 8. On the other hand, the feedforward control signal 22 output from the fuzzy control unit 28 is input to the model prediction signal generation unit 9, and is operated by the transfer function model ignoring the dead time of the control target 2, and the model prediction signal 24 is output. .

Subsequently, the prediction control deviation signal 16 obtained by inputting the model prediction signal 24 and the set value signal 3 to the addition means 7 is output to the fuzzy control means 28.

As shown in FIG. 5, the fuzzy control means 28 uses a velocity type comprising a differential element 28a, a fuzzy processing means 28b and an integral element 28c, and the predicted control deviation signal 16 is controlled by the differential element 28a. Differentiation 1
6 and the predictive control deviation signal 16 are input as they are, and the output value obtained by fuzzy inference by the fuzzy processing means 28b is integrated by the integration element 28c to obtain the feedforward control signal 22.

FIG. 6 shows the membership function of the fuzzy processing means 28b provided in the fuzzy control means 28. In FIG. 6, the prediction control deviation signal 16 (deviation 16),
The differential value (differential value 16) of the prediction control deviation signal 16 and the increment (increase 22) of the feedforward control signal 22 are shown, and a rule corresponding to this is shown in FIG. 7, for example.

E16 in FIGS. 6 and 7 is the deviation 16, D16
Corresponds to the derivative 16, and D22 corresponds to the increment 22. A summary of this is the rule table shown in FIG.

In the fuzzy processing means 28b of the fuzzy control means 28, the (deviation 16) (E16) of the predicted control deviation signal 16 and the differential value (differential 16) (D16) of the predicted control deviation signal 16 are obtained according to the rule table shown in FIG. (Increase 22) of the feedforward control signal 22 according to the input value of
(D22) is inferred to determine a proportional value that is an increase.
Then, the increase is integrated by the integration value in the integration element 28 c and output to the addition means 8 as the feedforward control signal 22.

Here, when the predicted control deviation signal 16 is smaller than a predetermined value, the contact 17a is closed and the control deviation signal 15 is output to the integral control means 14 by the switching means 17 for inputting the predicted control deviation signal 16. On the other hand, when the prediction control deviation signal 16 is equal to or greater than the predetermined value
The contact 17a is opened for a time equivalent to the dead time, and the input to the integration control means 14 is set to zero.

As a result, the normal controlled object 2 becomes stable,
Further, when the set value signal 3 is not changed, the control deviation signal 15 is small, and the contact 17a is closed when the control deviation signal 15 is less than a predetermined value. The control deviation signal 15 is input to the integration control means 14, and the equation (1) is used. The control operation is performed, and the integral control means 1
4 provides stable feedback control. Thereafter, when the set value signal 3 changes as required, the predicted control deviation signal 16 becomes a predetermined value or more, the contact 17a is opened by the switching means 17, the predicted control deviation signal 16 is calculated by the fuzzy control means 28, The forward control signal 22 is controlled by the feedforward control signal 22 corresponding to a dead time of the control target 2. As a result, the prediction control can be performed with good response at high speed, ignoring the transfer function of the dead time. Then, when the predictive control is performed and the predictive control deviation signal 16 becomes small, the integral control means 14 stabilizes according to the control deviation signal 15, and
High-precision control is performed.

As described above, according to the fourth embodiment, it is not necessary to incorporate a dead time model based on the Smith method into the apparatus, and as a result, there is a mismatch between the dead time model and the control target 2. It is possible to eliminate the risk of impairing the control system, to cope with the case where the dead time of the control target 2 changes, and to make the dead time of the dead time model larger than the control cycle of the process control device. If it is very large, the problem that the memory space of the model with a dead time increases and it is not practical can also be solved. In addition, since the feedforward signal is generated by the fuzzy control means, an accurate control constant is determined uniformly according to the predicted control deviation signal 16, so that a high-speed response can be achieved.

FIG. 9 is a block diagram of a process control device according to a fifth embodiment of the present invention.

In FIG. 9, the same reference numerals as in FIG. 17 showing the conventional example indicate the same or corresponding parts. The process control device 1E includes the dead time model 11, the self-organizing fuzzy control means 29, the switching means 17, and the like. The fifth embodiment is different from the feed-forward control of the second embodiment shown in FIG. The feature is that the self-organizing fuzzy control means 29 is used instead of the means 13.

With the above configuration, the control deviation signal 15 obtained by inputting the control amount signal 4 and the set value signal 3 from the control target 2 to the adding means 6 is output via the contact 17 a of the switching means 17. I have. As a result, the integration control means 14 executes the control calculation of the equation (1) and outputs the obtained feedback control signal 23 to the addition means 8. on the other hand,
The feed-forward control signal 22 output from the self-organizing fuzzy control unit 29 is input to the dead timeless model 11, is operated by a transfer function model ignoring the dead time of the control target 2, and outputs a model prediction signal 24. .

Subsequently, the prediction control deviation signal 16 obtained by inputting the model prediction signal 24 and the set value signal 3 to the addition means 7 is output to the self-organizing fuzzy control means 29.

The self-organizing fuzzy control means 29 is a
0, the reference model 29a and the fuzzy rule 29
b, a fuzzy controller 29c, a predictor 29d, a rule construction 29e, and a target plant 29f.

The self-organizing fuzzy control means 29 automatically generates and modifies rules and membership functions online without any expert experience, and automatically constructs a fuzzy control system. That is,
The control of the target plant is performed by fuzzy control. This fuzzy control rule is dynamically modified (on-line) to finally control the target plant optimally. As an example, the input range of the target plant is calculated by the fuzzy rule, and the actual plant input (control output) is obtained from the desired output yr of the target plant and the expected output ya. On the other hand, the response of the target plant at this time is fetched online for each control step, and a combination of input and output of the target plant is used as a candidate for a new rule. The rule is updated such that the difference between the desired output yr at that time and the actual output result y is minimized. The rule in this case is to divide the control space of the target plant and fill one rule in each space. At first, nothing is in all rule spaces, but (in the image with all Z) When starting from this state, in most spaces, new rules will give better control results, and rules will be added one after another. Is created.

Here, when the predicted control deviation signal 16 is smaller than a predetermined value, the contact 17 a is closed and the control deviation signal 15 is output to the integral control means 14 by the switching means 17 for inputting the predicted control deviation signal 16. On the other hand, when the prediction control deviation signal 16 is equal to or greater than the predetermined value
The contact 17a is opened for a time equivalent to the dead time, and the input to the integration control means 14 is set to zero.

As a result, the prediction control can be performed with a good response at a high speed, ignoring the transfer function of the dead time. Then, when the prediction control is performed and the prediction control deviation signal 16 becomes small,
This time, stable and highly accurate control is performed according to the control deviation signal 15.

As described above, according to the fifth embodiment, it is not necessary to incorporate a dead time model based on the Smith method in the apparatus, and as a result, there is a mismatch between the dead time model and the control target 2. The possibility that controllability is impaired can be eliminated, and the adverse effects caused by incorporating the dead time model in the device can be eliminated.

FIG. 11 is a block diagram of a process control device according to a sixth embodiment of the present invention.

In FIG. 11, the same reference numerals as in FIG. 17 showing the conventional example indicate the same or corresponding parts. The process control device 1 </ b> F includes a dead time model 11, a feedforward control unit 13, and a fuzzy selection control unit 27.
The sixth embodiment is characterized in that a fuzzy selection control means 27 is provided instead of the switching means 17 and the integration control means 14 of the second embodiment shown in FIG. ing.

With the above configuration, the control deviation signal 15 obtained by inputting the control amount signal 4 and the set value signal 3 from the control target 2 to the adding means 6 is output to the fuzzy selection control means 27.

Subsequently, the prediction control deviation signal 16 obtained by inputting the model prediction signal 24 and the set value signal 3 to the addition means 7 is output to the feedforward control means 13 and the fuzzy selection control means 27.

As shown in FIG. 12, the fuzzy selection control means 27 is composed of a calculating means 28a and an integral element 28b, and receives the predicted control deviation signal 16 and the control deviation signal 15 and executes a fuzzy operation. The increase is integrated by the integration element 28b and output as the feedback control signal 23.

FIG. 13 shows an example of the membership function. In FIG. 13, the predicted control deviation signal 16 is represented by deviation 16 or E16, and the control deviation signal 15 is represented by deviation 15 or E16.
15 and the increment with respect to the feedback control signal 23 represents the increment 23 or D23.

The labels of the membership functions shown in FIG. 13 are defined as shown in FIG. 14, and the rules in this case are defined in FIG. 15 to create a rule table, as shown in FIG. Based on this rule table, the calculation is performed by the calculation means 27a of the fuzzy selection control means 27 in accordance with the input values of the predicted control deviation signal 16 and the control deviation signal 15,
The obtained signal is output as a feedback control signal 23 according to the integration request 28b.

As a result, the normal controlled object 2 becomes stable,
In addition, when the set value signal 3 is not changed, and when the control deviation signal 15 is small, the feedback control signal 23 calculated by the fuzzy selection control means 27 performs stable feedback control. Thereafter, the set value signal 3 changes as necessary, and when the predicted control deviation signal 16 becomes a certain value or more, the feedback control signal 23 is reduced from the fuzzy selection control means 27, and the feedforward control signal 22. As a result, the prediction control can be performed with good response at high speed, ignoring the transfer function of the dead time. Then, when the prediction control is performed and the prediction control deviation signal 16 becomes small, this time,
Stable and highly accurate control is performed by the fuzzy selection control means 27 in accordance with the control deviation signal 15.

As described above, according to the sixth embodiment, it is not necessary to incorporate a dead time model by the Smith method into the apparatus, and as a result, there is a mismatch between the dead time model and the control target 2. The disadvantage of incorporating a model of a dead time, such as eliminating the risk of damaging the control system, can be eliminated. Further, since the switching means 17 according to the first to fifth embodiments is not required, the feedforward control signal 22 and the feedback control signal 23 can be extremely smoothly controlled according to the values of the control deviation signal 15 and the predicted control deviation signal 16. Can be switched.

[0074]

As described above, according to the first aspect of the present invention, since it is not necessary to incorporate a dead time model by the Smith method in the apparatus, there is a mismatch between the dead time model and the control target. Can reduce the risk of impairing controllability and respond to the case where the dead time of the controlled object changes dynamically.The size of the dead time in the model of the dead time can be compared with the control cycle of the process control device. If it is very large, the problem that the memory space of the model with a dead time increases and it is not practical can also be solved.

Further, according to the second aspect of the present invention, there is no need to provide a dead time model by the Smith method in the apparatus, and it is possible to avoid the adverse effects caused by providing the dead time model in the apparatus. Therefore, the control deviation signal is normally calculated and controlled by the integration element, and the feedback control is performed, so that stable high-precision control can be performed.

According to the third aspect of the present invention, there is no need to provide a dead time model by the Smith method in the apparatus, and it is possible to avoid the adverse effects caused by providing the dead time model in the apparatus. Therefore, the control deviation signal is normally calculated and controlled by the proportional integration element, and the feedback control is performed, so that stable high-precision control can be performed.

According to the fourth aspect of the present invention, it is not necessary to provide a dead time model by the Smith method in the apparatus, and it is possible to avoid an adverse effect caused by providing a dead time model in the apparatus. Since the fuzzy control means is used as the means, when the predicted control deviation signal is equal to or more than a predetermined value, a control constant is determined according to the value of the predicted control deviation signal, and a high-speed response can be achieved.

According to the fifth aspect of the present invention, it is not necessary to provide a dead time model by the Smith method in the apparatus, and it is possible to avoid an adverse effect caused by providing a dead time model in the apparatus. Since the self-organizing fuzzy control means is used as the means, when the predicted control deviation signal is equal to or more than a predetermined value, the control constant can be accurately determined so that the feedforward signal corresponding to the dead time responds at high speed.

Further, according to the invention of claim 6, it is not necessary to provide a model of the dead time by the Smith method in the apparatus, and it is possible to avoid the adverse effect caused by providing the model of the dead time in the apparatus. The fuzzy operation is executed in accordance with the values of the predicted control deviation signal and the control deviation signal, and the switching between the feedback control signal and the feedforward control signal can be smoothly performed without the need for the switching means according to the fifth aspect of the present invention. ,
It is possible to avoid a change in the control amount signal due to the switching.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a process control device according to a first embodiment of the present invention.

FIG. 2 is a configuration diagram of a process control device according to a second embodiment of the present invention.

FIG. 3 is a configuration diagram of a process control device according to a third embodiment of the present invention.

FIG. 4 is a configuration diagram of a process control device according to a fourth embodiment of the present invention.

5 is a configuration diagram of fuzzy control means provided in the process control device of FIG.

FIG. 6 is a diagram showing an example of a membership function of the fuzzy control means of FIG.

FIG. 7 is a diagram showing rules of the fuzzy control means of FIG. 4;

FIG. 8 is a diagram showing a rule table of the fuzzy control means of FIG. 4;

FIG. 9 is a configuration diagram of a process control device according to a fifth embodiment of the present invention.

FIG. 10 is an explanatory diagram showing a self-organizing fuzzy control means provided in the process control device.

FIG. 11 is a configuration diagram of a process control device according to a sixth embodiment of the present invention.

FIG. 12 is a configuration diagram of fuzzy selection control means provided in the process control device of FIG. 11;

FIG. 13 is a diagram showing an example of a membership function of the fuzzy selection control means of FIG.

FIG. 14 is a diagram showing a label corresponding to FIG. 13;

FIG. 15 is a diagram showing a rule corresponding to FIG.

FIG. 16 is a diagram showing a rule table corresponding to FIG. 13;

FIG. 17 is a first configuration diagram of a process control device showing a conventional example.

FIG. 18 is a second configuration diagram of a process control device showing a conventional example.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Process control apparatus 2 Control object 3 Set value signal 4 Control amount signal 5 Manipulation amount signal 6, 7, 8 Addition means 9 Model prediction signal generation means 10 Feedback control means 11 Model without dead time 12 Model with dead time 13 Feedforward control Means 14 Integral control means 15 Control deviation signal 16 Predictive control deviation signal 17 Switching means 18 Control deviation correction signal 19 PI controller input signal 22 Feed forward control signal 23 Feedback control signal 24 Model prediction signal 25 Proportional integral control means 26 Integrator setting Means 27 Fuzzy selection control means 28 Fuzzy control means 29 Self-organizing fuzzy control means

Claims (6)

[Claims] 1. A process control device for controlling a process including a dead time element as a control target by outputting an operation amount signal to the control target so that a control amount signal of the process becomes a set value signal. Generating a model prediction signal having a transfer function model ignoring the dead time of the process, inputting a feedforward control signal for predictive control, and outputting a model prediction signal obtained by performing a control operation based on the transfer function Means, feedforward control means for outputting a feedforward control signal obtained by performing the control operation so that the model prediction signal output by the model prediction signal generation means becomes the set value signal, and the feedforward control signal And outputting the addition signal of the feedback control signal and the feedback control signal to the control target as the operation amount signal. On the other hand, from the time when the control deviation signal between the model prediction signal and the set value is equal to or more than a predetermined value to the time corresponding to the dead time of the control target, only the feedforward control signal is output to the control target. A process control device comprising: switching means for switching. 2. The process control device according to claim 1, wherein said feedback control means outputs a feedback control signal obtained by controlling and calculating said input control deviation signal by an integral element. 3. The process control apparatus according to claim 1, wherein said feedback control means outputs a feedback control signal obtained by controlling and calculating said input control deviation signal by a proportional and integral element. 4. The feedforward control means outputs a feedforward control signal obtained by performing a control operation using a control constant determined by fuzzy inference according to the model prediction signal and the set value signal. The process control device according to claim 1, wherein 5. The feedforward control means outputs a feedforward control signal obtained by performing a control operation using a control constant determined by self-organizing fuzzy inference according to the model prediction signal and the set value signal. The process control device according to claim 1, wherein: 6. A process control apparatus which outputs an operation amount signal to a control target so that a control amount signal of the process becomes a set value signal as a process control target including a dead time, and controls the process. A model predictive signal generating means having a transfer function model ignoring dead time, inputting a feedforward control signal for predictive control, and outputting a model predictive signal obtained by performing a control operation based on the transfer function; Feedforward control means for outputting a feedforward control signal obtained by performing the control operation so that the model prediction signal output by the model prediction signal generation means becomes the set value signal; and the control deviation signal and the prediction When the control deviation signal is smaller than a predetermined value in response to the control deviation signal, the control constant is fuzzy inferred. The control deviation signal is output so as to select a feedback control signal obtained by controlling and calculating the control deviation signal. Process control comprising: a fuzzy selection control means for selecting the value to be zero; and an addition means for adding the feedback control signal and the feedforward control signal to output an operation amount signal to a control target. apparatus.