JPH0250201A - Process control device - Google Patents

Abstract

PURPOSE:To compensate a vain time, to improve control responsibility and to prevent over shoot or under shoot by specifying command timing to output over shoot prevention operation variable and using the determined manipulated variable. CONSTITUTION:In regard to a process that the relation of 'vain time + primary delay' is approximately obtained between physical variable CPV of a control subject and manipulated variable CMV, a Smith method process model part device 2 calculates an estimating value Kpm of primary delay gain for the process from the sampling data of the process in advance. A means 6 is provided to temporarily change the manipulated variable CMV to CMV=(RSV-DELTAT)/Kpm just after the sum of the physical variable CPV and an output CSM of the Smith method process model part device 2 arrives at a value, for which prescribed quantity DELTAT to be determined in advance is subtracted from a setting value RSV, in case that the setting value RSV is changed. Thus, while a control characteristic is bastened, setting can be speedily executed by simple arithmetic contents without generating the over shoot or under shoot.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention relates to a process having an approximate relationship of "dead time + first-order delay" or The present invention relates to a process control device provided with means for preventing overshoot or undershoot in a process having a relationship of "+integral".

(Prior art) One of the main purposes of automatic control of plant processes is to control changes in set values: to follow l (physical quantity) as quickly as possible, and to prevent overshoots and undershoots when approaching the set value. The goal is to quickly settle to the set value without causing any problems. Particularly, for example, when material heating is controlled in a pre-process for machining, if temperature overshoot occurs, the quality of the final product will deteriorate or the product will be defective.

Conventionally, even in a simple first-order lag system, the two demands of increasing control responsiveness and eliminating overcontrol have contradictory elements.

This is just an analogy, but in a car race, if the finish line is drawn just before a precipice, you must reach the finish line as quickly as possible, but if you cross the line even a little. It's similar to a situation where you have to apply the brakes far in front of you to slow down the car because you might fall off a cliff.

Taking these controllability issues into consideration, in processes that have an approximately first-order lag relationship, it is possible to perform simple calculations without slowing down the control characteristics or causing overshoot or undershoot. , a process control device that quickly stabilizes has been proposed (Japanese Patent Publication No. 62
(Refer to Publication No.-13683).

Processes having approximately a first-order lag relationship are generally known, without giving any examples. Furthermore, if we consider a process in which several -order delays are combined in series and parallel, the range becomes extremely wide. Even in the latter case, the above-mentioned known technique can be applied to each loop as long as the system is capable of forming separate control loops for each first-order delay system.

However, in addition to first-order delay, an important element of process characteristics is the "dead time element" that occurs due to the finite propagation speed of matter and energy. However, there was a problem in that the above-mentioned known technology could not be applied.

Normally, even if the process is approximately a first-order lag system, it includes some dead time elements, but if the length of the dead time is very small compared to the time constant of the first-order lag. can be ignored. However, in processes where the length of the dead time is the same as or longer than the first-order lag time constant, not only does it take time for the controlled variable (physical quantity) to settle, but normal PID control alone Even if the manipulated variable is changed, the controlled variable does not change at all until the dead time has elapsed, which often causes a phase shift between the manipulated variable and the controlled variable, causing vibration. However, adjusting the PID parameters of the controller is extremely difficult.

Therefore, various dead time compensation methods have been devised and applied in the past. Examples include the Smith method and the sample PI method.

In general, normal PID control cannot obtain sufficient control characteristics, and an example of a process in which the relationship is approximately "dead time + first-order lag" is plate temperature control in continuous galvanizing equipment in a steel plant. (See Figure 7)
.

The equipment shown in Figure 7 is an equipment that continuously applies hot-dip galvanizing to materials such as cold-rolled galvanized coils and hot-rolled thin sheet coils to produce galvanized iron sheets and galvanized steel sheets used for automobile bodies, for example. It is. Steel plate to be processed 1
5, while being transferred at a constant speed, the steps of preheating in a preheating furnace 17, heating in a non-oxidizing heating furnace 18, and reduction in a reduction furnace 19 are performed continuously, and finally, the galvanizing liquid is passed through and plated. will be held. Note that a flue 16 is attached to the preheating furnace 17. The steel sheet temperature is measured by radiation thermometers Ta and Tb at the exits of the non-oxidizing heating furnace 18 and the reducing furnace 19, respectively, and a predetermined exit sheet temperature is obtained by combustion control by the burner 20 provided in the non-oxidizing heating furnace 18. However, temperature measurement using the thermometer Ta at the exit of the non-oxidizing heating furnace 18 has extremely low reliability due to the poor gas atmosphere. Combustion feedback control is performed.

For this reason, the dead time due to the steel plate transfer time between the non-oxidation heating furnace 18 and the installation point of the thermometer Ta is large, and the normal PI
It is impossible to expect sufficient control performance with D control.

Furthermore, it is difficult to cause the plate temperature to follow step changes or ramp changes in the set value as quickly as possible and to prevent excessive control. However, excessive plate temperature control must be strictly controlled to ensure the quality of the final product. Note that the relationship from the non-oxidizing heating furnace combustion burner 20 to the plate temperature in its vicinity may be considered to be a first-order lag.

Next, the Smith method for dead time compensation, which is commonly applied to processes with long dead times as described above, will be explained.

FIG. 8 is a block diagram of a general configuration of a control system using the Smith method. Main controller (Gc) 1 is a normal PI
D control device, which includes Smith method process model unit device 2
Together with this, they constitute a control block. Target real process 3
cp in the block representing may be a first-order lag,
A high-order delay system such as a second-order delay may be used. The ratio between the set value RSV and the physical quantity CPV in the control system shown in FIG. Smith's condition, G-G
... (2) pm p L -L ...
(3) If pi pv holds, equation (1) can be transformed as follows.

Therefore, the control system of FIG. 8 can be equivalently expressed as shown in FIG. 9, and the dead time element L − S e pv can be taken out of the feedback loop.

As described above, if equations (2) and (3) approximately hold, the PID control device of the main controller 1 can be similarly stabilized using the same parameters as process control when there is no dead time. This means that control characteristics can be obtained.

Of course, the true physical quantity (control ff1) CPV will be delayed by the dead time Lpv from the virtual physical jlPvM in FIG. 9, but this is unavoidable.

Note that it is often difficult to identify the parameters of the actual process, obtain them with high precision, and set them in the process model unit device 2, but if equations (2) and (3) approximately hold true, then , it is stated in literature, Proceedings of the 20th Society of Instrument and Control Engineers Academic Conference (July 1982) (Yonezawa, Heating), and other literature that the desired results can be obtained. This is where I am.

(Problems to be Solved by the Invention) As already mentioned, there are already known devices that can prevent overshoots and undershoots and quickly settle without reducing response speed using simple calculations. However, the scope of application of this known control system is limited to processes that have an approximately first-order lag relationship, and cannot be applied to processes that include dead time, which is another important element of process characteristics. I can't.

Therefore, in the present invention, approximately "dead time + first-order delay"
For processes that have the relationship ``
To provide a process control device capable of preventing overshoot and undershoot and quickly settling a process having a relationship of "dead time + integral" with simple calculation contents while speeding up control response characteristics. The purpose is to

[Structure of the invention]

(Means for Solving the Problem) In order to achieve the above object, the present invention provides a physical f
iCPV and manipulated variable CMV are approximately “dead time +1
The physical fiC
In a process control device that performs control based on the manipulated variable calculated by a main controller and a Smith method process model unit so that PV matches a set value RSV, the Smith method process model unit performs process control based on process sampling data in advance. In addition to calculating the estimated value Kp11 of the first-order lag gain, when changing the set value RSV, the physical quantity C
The manipulated variable CMV immediately after the sum of PV and the Smith method process model unit 20 output CSM reaches the value obtained by subtracting a predetermined value ΔT from the set value RSV is expressed as CMV'≒(RSV-ΔT) / Kpra-(5)-
It is characterized by providing a means for changing it from time to time.

Furthermore, the present invention provides a physical quantity CPV of a controlled object and an operation i.
For a process in which 1cMV has an approximate relationship of "dead time + integral", control is performed based on the manipulated variable calculated by the main controller and the Smith method process model unit so that the physical ff1cPV matches the set value RSV. In a process control device that performs the
When changing V, the manipulated variable CMV immediately after the sum of the physical quantity CPV and the output CSM of the Smith method process model unit device reaches a value obtained by subtracting a predetermined value ΔT from the set value RSV is set to the first Apm of CMV. It is characterized by providing a means for temporarily changing it.

(Function) - What is important in the above control device is that in the conventional technology, the value of equation (5), that is, the command timing to output the overshoot prevention manipulated variable, is at the point when the physical ff1cPV crosses the set value -ΔT. In contrast, in the case of the present invention, the point is that the value of CPV+C8M crosses the set value -ΔT. Therefore, the value of CPV+CSM will be referred to as PVM and will be referred to as a virtual physical quantity of the Smith method.

By using the manipulated variable determined in this way, dead time is compensated for, control responsiveness is good, and overshoot is achieved. It is possible to construct a process control device that is free from overshoots and undershoots.

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

The present invention is applicable to both overshoot prevention when the set value increases and undershoot prevention when the set value decreases. This section describes in detail how to prevent overshoot when moving.

The process control device shown in FIG. 1 is a device that automatically controls the physical quantity CPV of the target real process 3 that is the control target so that it matches the set value RSV. The system portion consisting of the main controller 1, the Smith method process model unit 2, and the target actual process 3 is substantially the same as that described in FIG. In the control device of the present invention, the output of the main controller 1 is normally guided as the manipulated variable CMV to the target actual process 3, but when the set value RSV is changed, the output of the overshoot prevention manipulated variable calculation device 5 is temporarily guided. A changeover switch 6 is provided on the output side of the main controller 1 to guide the flow. physics ff
1cPV and Smith method process model unit device 2 output C5
The sum PVM-CPV+CSM with M is the value obtained by subtracting a predetermined value ΔT from the set value R5V (RSV-ΔT
), the overshoot prevention manipulated variable output command device 4 detects this and moves the changeover switch 6 to the side opposite to that shown, that is, the overshoot prevention manipulated variable calculation device 5.
Switch to the side. The overshoot prevention manipulated variable calculation device 5 calculates (5 )operate according to the formula ficMV
Calculate and output.

It is assumed that the target real process 3 in FIG. 1 has a process characteristic of "dead time + first-order delay" corresponding to . Therefore, the Smith method model unit 2 has the form. Also, Kp' + "pm *
L pm is respectively Kpn first Kpv...(8)T
pm illusion Tpv...(9) Lp
Assume that aL:Lpv-(10). These values may be given in advance as estimated values for the target real process 3, or they may be statistically processed from the data of the target real process 3 and constantly corrected by learning.

Now, if the relationships of equations (8) to (10) hold,
As stated in the explanation of the Smith method, the relationship among the main controller 1, Smith method model unit 2, and target real process 3 in FIG. 1 can be equivalently expressed as in the block diagram in FIG. 2. Can be done. Here, the virtual physical quantity PVM-CSM+CPV (11) in the Smith method in Fig. 2,
The relationship with Operation 11cMV is: PVM − (Kpa + / (1 + TpIl-8)) CMV
... (12) becomes.

Expressing equation (12) in a mathematical expression using an original function,...(
13) It becomes.

Next, the set value RSV increases and becomes constant, and the virtual physical quantity PVM in the Smith method follows it ((7)
(refer to the formula) exceeds RSV-ΔT, which is close to the set value RSV, in order to prevent overshoot and immediately stabilize, (dpvM(t))/at resistance +0-C
14). Note that "+0" here means a positive value close to 0. The value of the manipulated variable CMV required to satisfy equation (14) may be obtained from equation (13) as follows.

(dPVM(t))/dt-0-(15)PVM (
t)-RSV-ΔT-(16)(15), Substituting equation (16) into equation (13), CMV-(RSV-ΔT)/Kpm... (17) Here, equation (15) It is preferable that the value on the left side be a positive value close to 0, rather than a complete 0. This is because the Smith method virtual physical quantity PVM at the current time t is RSV-ΔT, which is slightly lower than the set value RSV by ΔT, so the virtual physical quantity PVM needs to rise a little more. Therefore, from equation (17), the changeover switch 6 is changed only when the virtual physical quantity PVM(t) crosses RSV-ΔT, and the value added by 2 to 3% is calculated from the overshoot prevention manipulated variable calculation device 5. This is the forced manipulated variable output value. Thereafter, the selector switch 6 is changed back to the state shown in the figure to return to normal Smith method control.

Note that when the manipulated variable is switched from the main controller 1 to the overshoot prevention manipulated variable calculation device 5, the value jumps in steps. Then, the changeover switch 6 immediately returns to the main controller 1 side, but at this time, if the operation amount is set to start from the value of the overshoot prevention operation amount calculation device 5, switching can be performed without shock.

FIG. 4 shows an example of control results obtained by applying the control system shown in FIG. 1. In addition, in collecting the data, the target actual process 3 is expressed as follows.The Smith method process model unit 2 is expressed as follows. Further, ΔT is 2%, the vertical axis has a % scale of 1730, and the horizontal axis represents time in [seconds].

The main controller 1 was assumed to be represented by the following PID control device.

In addition, each parameter is Proportional gain CKP-3 Integral time Ti-2 (seconds) Differential time Td-0 (seconds) It was assumed that

As can be seen from Figure 4, the physical quantity CPV increases without slowing down until it approaches the fixed value after changing the set value RSV, and just before reaching the set value, the slope becomes almost +0 and settles without overshooting. are doing. In addition, the manipulated variable CMV is determined approximately 3 seconds earlier (17) by the actions of the overshoot prevention manipulated variable output command device 4, the overshoot prevention manipulated variable calculation device 5, and the changeover switch 6 shown in FIG. The value is forcibly lowered to a value slightly larger than the value of the expression. At this time, it can be seen that the virtual physical quantity PVM has crossed the set value -ΔT.

FIG. 5 shows that even if the parameters of the Smith method process model device 2 deviate somewhat from the parameters of the target actual process 3 due to identification errors, the control characteristics of the device of the present invention do not change much, and it exhibits a considerable effect. This shows that it is possible.

In the case of FIG. 4, only the Smith method process model section device 2 was changed as follows.

That is, here, the time constant of the target real process 3 is 5 seconds, while that of the Smith method process model unit device 2 is 4.8 seconds, and the dead time of the target real process 3 is 3 seconds. On the other hand, it is assumed that the time for the Smith method process model section device 2 is 3°2 seconds.

FIG. 6 shows the results of control using only the Smith method of FIG. 8, in order to compare the advantages of the control device of the present invention with those of the conventional device. The parameters in this case are the same as in the case of FIG.

In the case of Fig. 6, since the physical quantity CPV reaches the maximum value of the set value RSV at the same speed as in the case of Fig. 4, it overshoots the set value RSV by a large amount.
Furthermore, it takes a considerable amount of time to return to the set value RSV and stabilize.

In conventional devices, in order to prevent this overshoot, the physical quantity CPV approaches the set value RSV (for example, by reducing the control gain of the PID control from a considerable distance before the physical quantity CPV approaches the set value RSV).
They implemented methods to slow down the speed at which V increases. In other words, if you really wanted to prevent overshoot, you had to sacrifice response speed, which is important in control performance.

(Other Embodiments) In the embodiments described above, overshoot prevention control of physical quantities when the set value increases has been described, but the present invention also provides under-shoot prevention control of the physical quantity when the set value decreases. It is clear that it can also be applied in the case of shoot prevention.

Furthermore, even when applied to program pattern control in which the set value increases or decreases, it is possible to perform good control that prevents overshoot and undershoot without reducing control response speed.

Furthermore, the present invention can be applied not only to a process in which the target real process is approximately "dead time + first-order delay" but also to a process in which approximately "dead time + integral" is approximated. This will be explained below.

It is assumed that the target real process satisfies the following formula.

Here, Apv is a constant, but in the present invention, it will be referred to as a "process integral equilibrium constant." At this time, the Smith method process model unit device 2 is as follows. Here, Kpffl-Kpv, Apm=Ap
v, Lpm-Lpv holds, the main controller 1, Smith method process model unit 2, and target real process 3 in FIG. 1 can be equivalently expressed as shown in FIG. 3. . The relationship between the manipulated variable CMV and the virtual physical quantity PVM is expressed in a mathematical expression using an original function as follows. Thus, as in the case of "dead time + first-order delay", when the virtual physical quantity PVM in the Smith method is close to the set value RSV, R
In order to prevent overshoot and immediately settle when SV-ΔT is crossed, to set dPVM (t)/d t”7+0, temporarily set the operation fficMV to be slightly larger than Apll from equation (22). Just make it a value.

It is clear that the effect of the device adopting this method is similar to that when applied to the "dead time + first-order delay" process described above.

As described above, according to the present invention, for any process having a relationship of "first-order lag", "dead time + first-order lag", or "dead time + integral", the control characteristics can be speeded up. With simple calculations, it is possible to quickly settle without causing overshoot or undershoot.

According to the present invention, it is possible to dramatically expand the scope of application of a control device implementing the Smith method, and furthermore, it is possible to dramatically expand the scope of application of a control device that implements the Smith method. The present invention can be applied if a control loop can be inserted separately for each process.

Furthermore, even if it is an inseparable high-order delay system, it is often effective to approximately transform the high-order delay system into the form of "dead time + first-order delay". ,
It is a well-known place.

Therefore, the present invention can be applied to such cases as well.

〔Effect of the invention〕

As detailed above, according to the present invention, for any process having a relationship of "first-order lag", "dead time + first-order lag", or "dead time + integral", the control characteristics can be speeded up. With simple calculations, it is possible to quickly settle without causing overshoot or undershoot.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, and FIG. 2 is a block diagram showing an embodiment of the present invention.
Fig. 3 is an equivalent block diagram of the main part when the target real process is "dead time + integral" and satisfies Smith's condition in the device shown in Fig. 1. FIG. 4 is a characteristic diagram showing a control mode of the control device of the present invention when applied to a target real process having a relationship of “dead time - next delay”;
Fig. 5 is a characteristic diagram showing the control mode when the parameters of the Smith method process model unit device are slightly shifted from the target actual parameters in the case of Fig. 4;
The figure is a characteristic diagram showing the control mode when the same target real process as in Figure 4 is controlled only by the conventional Smith method, and Figure 7 is a relationship between the process characteristics of "during dead time - next delay" FIG. 8 is a block diagram of a known Smith method dead time compensation control device, and FIG. 9 is an equivalent diagram of FIG. 8 when the Smith condition is satisfied. It is a block diagram. DESCRIPTION OF SYMBOLS 1... Main controller, 2... Smith method process model unit, 3... Target actual process, 4... Overshoot prevention manipulated variable output command device, 5... Overshoot prevention manipulated variable calculation Device, 6...Switch, 7...Process control device, RSV...Set value, CMV...Manipulated amount, CPV...Physical quantity, Kpv...First-order lag gain, Kpm...1 Next lag gain estimate, ApIll・
...Process integral equilibrium constant estimate.

Claims (1)

[Claims] 1. For a process in which the physical quantity CPV to be controlled and the manipulated variable CMV have an approximate relationship of "dead time + first-order lag", the physical quantity CPV is equal to the set value RSV.
In a process control device that performs control based on the manipulated variables calculated by a main controller and a Smith method process model unit so as to match the process data, the Smith method process model unit determines the first-order lag of the process from sampling data of the process in advance. While calculating the estimated value Kpm of the gain, when changing the set value RSV, the sum of the physical quantity CPV and the output CSM of the Smith method process model unit device is calculated by subtracting a predetermined value ΔT from the set value RSV. 1. A process control device comprising: a means for temporarily changing a manipulated variable CMV immediately after reaching a value as follows: CMV≈(RSV-ΔT)/Kpm. 2. Instead of the Smith method process model unit calculating the estimated value Kpm of the first order lag gain of the process, the estimated value Kpm of the first order lag gain of the process is given to the Smith method process model unit in advance. Claim 1
Process control equipment as described. 3. For a process in which the physical quantity CPV to be controlled and the manipulated variable CMV have an approximate relationship of "dead time + integral", the main controller and Smith method process model are set so that the physical quantity CPV matches the set value RSV. In the process control device that performs control based on the manipulated variable calculated by the unit, the Smith method process model unit calculates the estimated value Apm of the process integral equilibrium constant from the sampling data of the process in advance, and also calculates the estimated value Apm of the set value RSV. At the time of change, the manipulated variable CMV immediately after the sum of the physical quantity CPV and the output CSM of the Smith method process model unit device reaches a value obtained by subtracting a predetermined value ΔT from the set value RSV is set as CMV≒ Apm
1. A process control device characterized by being provided with a means for temporarily changing. 4. Instead of the Smith method process model unit calculating the estimated value Apm of the integral equilibrium constant of the process, the Smith method process model unit is provided with the estimated value Apm of the integral equilibrium constant of the process in advance. The process control device according to claim 3.