JP3259112B2 - controller - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an IMC (Internal Model)
More particularly, the present invention relates to a controller that can automatically set a dead time of an internal model and perform good control even when a dead time of a process to be controlled is unknown.

[0002]

2. Description of the Related Art Conventionally, there has been proposed a controller using a control algorithm having an IMC structure for performing control by incorporating an internal model expressing a process to be controlled mathematically.
If this IMC controller is used, a large dead time (time from when hot air is released from the air conditioner to when the indoor temperature rises) is increased in a process to be controlled (for example, when the controller is an indoor air conditioner, this corresponds to an indoor environment). There is an excellent advantage that it can cope even if it exists.

FIG. 20 is a block diagram of a control system using a conventional IMC controller. 33 is a first subtraction processing unit for subtracting a feedback amount to be described later from a target value (room temperature set value), and 32 is a filter unit for preventing a change in the output of the first subtraction processing unit 33 from being transmitted suddenly. ,
Reference numeral 34 denotes an operation unit for calculating an operation amount (the temperature of warm air or cold air emitted from the indoor air conditioner) which is an output of the controller based on the output of the filter unit 32. Reference numeral 36 denotes a control process approximated by a mathematical expression. An internal model that outputs a reference control amount corresponding to a control amount (indoor temperature) as a control result, and a second subtraction process that outputs a feedback amount by subtracting the reference control amount from the internal model from the control amount. A unit 40 is a process to be controlled.

Further, F, Gc, Gm, and Gp are transfer functions of the filter unit 32, the operation unit 34, the internal model 36, and the controlled process 40, r is a target value, u is an operation amount, and d
Is, for example, a disturbance corresponding to an outdoor environment relative to an indoor environment,
y is a control amount, ym is a reference control amount, and e is a feedback amount.

Next, the operation of such an IMC controller will be described. First, the feedback amount e is subtracted from the target value r by the first subtraction processing unit 33, and the result is output to the filter unit 32. Next, the operation amount u is calculated from the output of the filter unit 32 by the operation unit 34 and output to the control target process 40 and the internal model 36 of the controller. Then, the reference control amount ym from the internal model 36 that performs the approximate operation of the control target process 40 is subtracted from the control amount y of the control target process 40 by the second subtraction processing unit 38, and the result is referred to as the feedback amount e. Thus, a feedback control system that feeds back to the first subtraction processing unit 33 is configured.

Ideally, the internal model 36 of such an IMC controller is mathematically expressed so as to be exactly the same as the process 40 to be controlled.
Ideally, the inverse characteristic (1 / Gm) of the transfer function of the internal model 36 is obtained, but since the dead time element of the internal model 36 cannot be reciprocalized, the dead time element is usually ignore. Therefore, the control amount y can be obtained from the target value r and the disturbance d by the following equation using such a configuration. y = F × Gp × Gc × r / {1 + F × Gc × (Gp−Gm)} + (1−F × Gm × Gc) × d / {1 + F × Gc × (Gp−Gm)} (1) )

Here, the transfer function Gm of the internal model 36 is equal to the transfer function Gp of the process 40 to be controlled, and the transfer function Gc of the operation unit 34 is the reciprocal of the transfer function of the internal model 36 (1 / Gm = 1 / Gp). Assuming an ideal state equal to, Equation (1) becomes: y = F × r + (1−F) × d (2)

Further, under ideal conditions where there is no rapid change in the target value r, the filter unit 32 becomes unnecessary, and F = 1 can be set, so that the control amount y becomes equal to the target value r (y =
r), it is possible to realize control without any influence of the disturbance d. Focusing on the disturbance d, the controlled process 4
0 and the internal model 36 have the same characteristic with respect to the manipulated variable u even if there is a large dead time, so that the feedback amount e, which is the output of the second subtraction processing unit 38, is the disturbance d
It can be seen that the disturbance d can be suppressed.

In such an IMC controller, an internal model 36 that performs an approximate operation of the control target process 40 is determined by a known model identification technique. However, when the dead time, which is a parameter of the internal model 36, is unknown, the control is performed. There is no corresponding method while performing. Even if the dead time can be set, a model identification error of the internal model 36 with respect to the control target process 40 cannot be avoided to some extent.
Since the control when the estimation of the model identification error is incorrect does not operate as expected, a countermeasure in that case is taken by an expert having control knowledge.

[0010]

Since the conventional IMC controller is configured as described above, it cannot cope with the unknown dead time of the process to be controlled while performing control, and the time constant of the filter unit is reduced. Since it cannot be set, there is a problem that an operator other than an expert having control knowledge must give up using the IMC controller. Even if the internal model can be determined by estimating the dead time of the controlled process, if the estimation of the model identification error is incorrect or the dead time of the controlled process fluctuates irregularly, the controller There was a problem that the operation did not work as expected. The present invention, in order to solve the above-mentioned problem, provides a controller capable of automatically setting a dead time of an internal model and performing good control even when a dead time of a process to be controlled is unknown. Aim.

[0011]

According to the present invention, there is provided a target value filter section which outputs a target value of an input control with a time-delay characteristic of a transfer function, and a method of subtracting a feedback amount from an output of the target value filter section. No. 1 subtraction processing unit, a target value / disturbance filter unit which outputs the output of the first subtraction processing unit with a transfer function having a time delay characteristic, and an output of the target value / disturbance filter unit based on the parameters of the internal model. An operation amount calculation unit including an operation unit that calculates and outputs an operation amount; an internal model storage unit that stores internal model parameters; and an internal model output that calculates a reference control amount from the operation amount based on the internal model parameters. An operation unit, a second subtraction processing unit that subtracts the reference control amount output from the internal model output operation unit from the control amount of the process to be controlled, and outputs a feedback amount, and a control unit that controls the control amount. While the change as a response does not appear,
A sequential dead time updating unit that repeats increasing and updating the dead time in the parameters of the internal model and changing the dead time in the parameters stored in the internal model storage unit to the updated dead time; a target value and control Quantity and reference control quantity
Based on the step width, which is the difference between the initial value and the set value
A step width calculation unit for calculating each of the control amount and the reference control amount, and a first response start area in which the control amount starts to change.
The start time and end time are detected based on the step size of the control amount.
And the second response opening when the reference control amount starts to change.
Reference the start time and end time of the start area to the step width of the control amount
Response start area detection unit for detecting based on a first response and a first response
Control amount specified by start time and end time of start area
And the start time and end time of the second response start area
Time based on the amount of change in the reference control amount specified by
Is calculated geometrically, and the
The correction time is calculated from the dead time in the parameters of the
Estimate the dead time of the controlled process by subtracting
And a dead time detecting unit that changes the dead time in the parameters of the internal model stored in the internal model storage unit to the estimated value.

[0012]

[0013]

Further, the input control target value is transferred to a transfer function.
Target value filter section that outputs with a time delay characteristic,
Subtract the amount of feedback from the output of the value filter
Transfer function of the first subtraction processing unit and the output of the first subtraction processing unit
Target / disturbance filter section that outputs with time-delay characteristics
And the target value / disturbance noise based on the parameters of the internal model.
An operation unit that calculates and outputs an operation amount from the output of the filter unit;
And the parameters of the internal model.
Internal model storage and internal model parameters
Outputs an internal model that calculates the reference control amount from the operation amount based on the
Internal model based on force calculation unit and controlled variable of controlled process
Subtract the reference control amount output from the output operation unit and
A second subtraction processing section for outputting the feedback amount and a control amount.
As long as there is no change in response, the parameters of the internal model
Increase and update the dead time in the meter and store the internal model
The dead time in the parameters stored in this updated
Repeatedly updating dead time repeatedly to change to dead time
Unit, a noise processing unit that performs a process of reducing noise of the control amount of the control target process, and sequentially outputs the noise-reduced control amount to the dead time updating unit, a target value, a control amount,
Based on the reference control amount, the difference between the initial value and the set value
A step width calculator for calculating the step width for each of the control amount and the reference control amount, and a noise output from the noise processor.
Of the first response start area in which the control amount after the
The start time and end time are detected based on the step size of the control amount.
And the second response opening when the reference control amount starts to change.
Reference the start time and end time of the start area to the step width of the control amount
Response start area detection unit for detecting based on a first response and a first response
Control amount specified by start time and end time of start area
Change rate and start time and end time of the second response start area
Time based on the change rate of the reference control amount specified by
Is calculated geometrically, and the
The correction time is calculated from the dead time in the parameters of the
Estimate the dead time of the controlled process by subtracting
And a dead time detecting unit that changes the dead time in the parameters of the internal model stored in the internal model storage unit to the estimated value.

The rate of change of the control variable and the change of the reference
Based on the conversion rate
An internal module is used instead of the confidence calculation unit that calculates the confidence for evaluating the estimation accuracy of the dead time , and the dead time detection unit.
In the parameters of the internal model stored in the Dell storage
The result obtained by subtracting the correction time from the delay time is used as the confidence
By correcting the error, the dead time of the process to be controlled
And a modified dead time detection unit that calculates the estimated value of the dead time and changes the dead time in the parameters of the internal model stored in the internal model storage unit to the estimated value.

A target value filter unit for outputting the input target value of the control with a time-delay characteristic of a transfer function; a first subtraction processing unit for subtracting the feedback amount from the output of the target value filter unit; A target value / disturbance filter unit that outputs the output of the subtraction processing unit 1 with a time delay characteristic of the transfer function, and an operation amount is calculated and output from the output of the target value / disturbance filter unit based on the parameters of the internal model. An operation amount calculation unit including an operation unit; an internal model storage unit for storing parameters of the internal model; an internal model output calculation unit for calculating a reference control amount from the operation amount based on the parameters of the internal model; A second subtraction processing unit for subtracting the reference control amount output from the internal model output calculation unit from the control amount of the second model to output a feedback amount, A noise processing unit that performs processing for reducing's target value, based on the control amount and the reference controlled variable, first
The step width, which is the difference between the initial value and the set value, is
A step width calculating unit that calculates for each of the control amounts, Neu
The control amount after noise processing output from the noise
Control the start time and end time of the first response start area to be started
Detection based on the step size of the
Start and end of the second response start area where the response starts to change
A response start area detection unit for detecting a point based on the step width of the reference control amount; a start point of the first response start area;
A start region time difference detection unit for detecting a time difference between the start time of the first response start region and the start time of the first response start region;
Change rate of the control amount specified by the
Reference control specified by the start time and end time of the start area
Correction time based on rate of change of amount and time difference, noise processing
Computes geometrically while correcting the displacement due to
Dead time in the parameters of the internal model stored in the storage unit
Error was corrected by subtracting the correction time from
A dead time correction unit that calculates a dead time and changes the dead time in the parameters of the internal model stored in the internal model storage unit to the calculated dead time.

[0017]

According to the present invention, the target value is input to the target value filter unit, the feedback amount is subtracted from the output of the target value filter unit in the first subtraction processing unit, and the first subtraction is performed in the operation amount calculation unit. The manipulated variable is calculated from the output of the processing unit and output to the control target process and the internal model output calculation unit. Next, the reference control amount from the internal model output operation unit is subtracted from the control amount of the control target process in the second subtraction processing unit,
A feedback control system in which the result is output to the first subtraction processing unit as a feedback amount is configured. Then, while no change as a control response appears in the control amount, the dead time in the parameters of the internal model is sequentially updated by the dead time updating unit and output to the internal model storage unit is repeated, thereby repeating the internal model. The dead time is set. In addition, the control amount and reference control are performed by the step width calculation unit.
The step width of the amount is calculated and the response start area
Response of control amount and reference control amount based on step width
The area is detected and the dead time detector controls the response start area.
Of the controlled process from the amount of change in the
Is estimated and output to the internal model storage.
This corrects the dead time of the internal model.

[0018]

In addition, as long as there is no change in the control amount after the noise processing output from the noise processing unit, the dead time in the parameters of the internal model is successively updated by the dead time updating unit and output to the internal model storage unit. By repeating this, the dead time of the internal model is set.

The response start area detecting section detects a response start area of the control amount and the reference control amount based on the step width and the control amount after the noise processing output from the noise processing section. An estimated value of the dead time is calculated from the change rate of the control amount and the reference control amount in the response start area, and the dead time of the internal model is corrected.

A certainty factor, which is a dead time estimation accuracy, is calculated by the certainty factor calculating unit based on the rate of change of the control amount of the response start area and the change rate of the reference control amount. An estimated value of the dead time is calculated based on the rate of change of the quantity and the certainty factor, and the dead time of the internal model is corrected.

The start area time difference detecting section detects the time difference between the control amount and the reference control amount at the start of the response start area from the detection result of the response start area detecting section, and the dead time correcting section detects the control amount and the reference time. The dead time corrected for the error is calculated from the change rate of the control amount and the time difference detected by the start area time difference detection unit, and the dead time of the internal model is corrected.

[0023]

FIG. 1 is a block diagram of a controller having an IMC structure showing a reference example of the present invention, and FIG. 2 is a block diagram of a control system using the controller having the IMC structure. In FIG. 1, reference numeral 1 denotes a target value input unit for inputting a target value r set by an operator (not shown) to the controller;
Reference numeral 2 denotes a target value filter unit which outputs a target value r from the target value input unit 1 with a transfer function having a first-order lag characteristic, and reference numeral 3 denotes a first unit for subtracting the feedback amount e from the output of the target value filter unit 2.
Are subtraction processing units 4 and 4 are operation amount calculation units that calculate an operation amount u from the output of the first subtraction processing unit 3 based on parameters from an internal model storage unit described later.

Reference numeral 5 denotes a signal output unit for outputting the manipulated variable u output from the manipulated variable operation unit 4 to a process to be controlled (not shown in FIG. 1), and 6a denotes an internal model storage for storing parameters of an internal model of the controller. And 6b, an internal model output operation unit for performing an operation as an internal model based on the parameters output from the internal model storage unit 6a and outputting a reference control amount ym, and 7 for controlling the control amount y from the process to be controlled by this controller. Is a control amount input unit for inputting the control amount.

A second subtraction processing unit 8 subtracts the reference control amount ym output from the internal model output calculation unit 6b from the control amount y output from the control amount input unit 7 and outputs a feedback amount e. , 9 repeatedly increase the dead time in the parameters of the internal model and update the dead time in the parameters stored in the internal model storage unit 6a while no change as a control response appears in the control amount y. A sequential dead time updating unit.

In FIG. 2, reference numeral 4a is inside the manipulated variable calculator 4, and the output of the first subtraction processor 3 is 1
Target value / disturbance filter unit that outputs with next delay characteristics, 4b
Is the target value / disturbance filter section 4a
, An operation unit for calculating an operation amount u from the output of the internal model storage unit 6a and an internal model output operation unit 6b, F1 is a transfer function of the target value filter unit 2, and F2 is a target value / disturbance filter unit 4a is a transfer function. Also,
du is a manipulated variable disturbance, and can be treated equivalent to the control variable disturbance d by setting the disturbance d = Gp × du.

FIG. 2 shows the target value filter unit 2 of FIG.
The basic configuration of the controller having the IMC structure including the first subtraction processing unit 3, the operation amount calculation unit 4, the internal model storage unit 6a, the internal model output calculation unit 6b, and the second subtraction processing unit 8 includes a process to be controlled. 40, the control system including the disturbance d and the manipulated variable disturbance du is rewritten.

Next, the operation of the basic configuration of such a controller will be described. The target value r is set by an operator of the controller or the like, and is input to the target value filter unit 2 via the target value input unit 1. The target value filter unit 2 outputs the target value r as a characteristic of a transfer function F1 as shown in the following equation with its time constant being T1. F1 = 1 / (1 + T1 × s) (3)

The time constant T1 is equal to the dead time L of the internal model 6, which will be described later, excluding a preset initial value.
According to the change of m, the following equation is set. T1 = 4 × α × Lm (4) Here, α is a proportionality constant, for example, α = 0.3.

Next, the first subtraction processing section 3 subtracts the feedback amount e output from the second subtraction processing section 8 from the output of the target value filter section 2. The target value / disturbance filter section 4a in the manipulated variable operation section 4 includes a first subtraction processing section 3
Is output with the characteristic of the transfer function F2 as shown in the following equation with its time constant being T2. F2 = 1 / (1 + T2 × s) (5)

The time constant T2 is also set to the target value filter unit 2.
Similarly to the time constant T1, the following equation is used in accordance with the change in the dead time Lm excluding the initial value. T2 = α × Lm (6) That is, the time constant T1 is set to four times the time constant T2 as a standard setting.

The operation unit 4 in the operation amount calculation unit 4
b is the manipulated variable u from the output of the target value / disturbance filter unit 4a.
Is calculated, and its transfer function Gc is stored in the internal model storage unit 6
According to the gain and the time constant of the internal model 6 output from “a”, the following equation is obtained, and the reciprocal of the transfer function Gm of the internal model 6 excluding the element of the dead time Lm is similar to the example of FIG. Gc = (1 + Tm × s) / Km (7) where Km and Tm are gains of the internal model 6, respectively.
It is a time constant.

Therefore, the transfer function of the entire operation amount calculating section 4 is as follows. F2 × Gc = (1 + Tm × s) / {Km × (1 + T2 × s)} (8) In this way, the manipulated variable u is calculated from the output of the first subtraction processing unit 3 and the signal output unit 5 to the control target process 40 and to the internal model output operation unit 6b.

Next, the process 40 to be controlled is assumed to have elements of first-order delay and dead time, and its transfer function Gp
Can be represented by an approximate transfer function such as Gp = Kp × exp (−Lp × s) / (1 + Tp × s) (9) where Kp, Lp, and Tp are a gain, a dead time, and a time constant of the controlled process 40, respectively.

The internal model 6 is a mathematical expression of the control target process 40 described above using these parameters including the gain Km, the time constant Tm, and the dead time Lm stored in the internal model storage unit 6a. In the internal model output operation unit 6b, the reference control amount ym is calculated from the operation amount u output from the operation amount operation unit 4. The transfer function Gm is as follows. Gm = Km × exp (−Lm × s) / (1 + Tm × s) (10)

Next, the second subtraction processing unit 8 subtracts the reference control amount ym from the internal model output operation unit 6b from the control amount y from the control target process 40 input via the control amount input unit 7. And outputs the feedback amount e. Then, the feedback amount e is input to the first subtraction processing unit 3 as described above. Thus, a feedback control system, which is a basic configuration of the controller having the IMC structure, is established.

In such a control system, the sequential dead time updating section 9 changes the dead time Lm of the internal model 6 stored in the internal model storage section 6a as follows. FIG. 3A is a diagram illustrating the target value followability of the control variable y for explaining the operation of the sequential dead time updating unit 9, and FIG. 3B is a diagram illustrating a change start portion of the control variable y. . y0 is the initial value of the control amount y, ST is the step width which is the difference between the target value r and the initial value y0 in the control amount y, Lm1 is the dead time Lm changed by the sequential dead time updating unit 9, and Δt is the actual time. This is a control cycle (sampling cycle) of the control system, and a circle on the control amount y indicates a sampling time.

In FIG. 3A, at time 0 (initial state), a target value r, which is a step input, is input, and an operation amount u is output from the controller to the process 40 to be controlled. As a result, the control amount y becomes the initial value y0. , And finally reaches the target value r and shifts to the set state.
FIG. 3B shows a case where the control amount y in FIG.
Corresponds to an enlarged view of the first part that changes from.

The successive dead time updating unit 9 determines whether or not a change as a control response has appeared in the control amount y according to the following equation. | Y−y0 | <ε × ST = ε × | r−y0 | (11) Here, ε is a proportionality constant, for example, ε = 0.05.

That is, as shown in FIG. 3 (b), when ε × ST is used as a threshold value and the difference between the current control value y and the initial value y0 is smaller than this threshold value, there is no change in the control value y. It is determined that a change has occurred in the above case. The reason why the threshold value of 5% of the step width ST is provided for the determination of the change in the control amount y is to prevent the change in the control amount y due to disturbance or the like from being erroneously detected.

If it is determined that there is no change in the control amount y, the control cycle Δt is added to the dead time Lm stored in the internal model storage unit 6a. At first, since the dead time Lp of the control target process 40 is unknown, the initial value of the dead time Lm is set to 0, so that the dead time Lm is Δt
Becomes The added dead time Lm is sequentially output from the dead time updating unit 9 to the internal model storage unit 6a, so that the current dead time Lm stored in the internal model storage unit 6a is updated.

The change of the dead time Lm as described above is repeated while no change is detected in the control amount y as determined by the equation (11), and stopped when the change is detected. As a result, the finally obtained dead time Lm is shown in FIG.
Lm1 in (b). In actual control, the dead time Lm
Is added to the internal model storage unit 6a each time the control cycle Δt is added, and is output to the internal model output calculation unit 6b after the stored dead time Lm is changed, and the reference control amount ym is changed. The calculation is performed based on the dead time Lm.

Accordingly, the control amount y of the control target process 40
The reference control amount ym is held at the initial value y0 by repeatedly increasing the dead time Lm until a change appears in
A change due to the control response appears in the control amount y, and a change appears in the reference control amount ym from the time when the change of the dead time Lm ends. The dead time is a time from the input of the target value r to the appearance of a change in the control amount y. Therefore, as described above, the dead time Lm of the internal model 6 is sequentially changed while the change in the control amount y does not appear as a control response. By updating, the dead time Lm of the internal model 6 can be set even when the dead time Lp of the control target process 40 is unknown.

In the example shown in FIG. 1, since the change due to the control response of the control amount y is detected by a threshold value for preventing erroneous detection, the detection accuracy is determined by the threshold value, and the dead time Lp of the process 40 to be controlled is determined. In order to detect with higher accuracy, another measure is required. FIG. 4 is a block diagram of a controller having an IMC structure showing one embodiment of the present invention, and the same parts as those in FIG. 1 are denoted by the same reference numerals.

Reference numeral 10 denotes a step width for calculating a step width, which is a difference between the initial value of the control amount y and the reference control amount ym from the set value at the end of the change, based on the target value r, the control amount y, and the reference control amount ym. It is a calculation unit. Reference numeral 11 denotes a response start area detection unit, which outputs the operation amount u from the operation amount calculation unit 4 in response to the input of each response start area, that is, the target value r, based on the step width of the control amount y and the reference control amount ym. As a result, a region where the control amount y and the reference control amount ym start to change from the initial values is detected. Reference numeral 12 denotes the control target process 4 based on the change amount of the control amount y in the response start area and the reference control amount ym.
The dead time detecting unit calculates an estimated value of the dead time Lp of 0, and changes the dead time Lm of the internal model 6 stored in the internal model storage unit 6a to the estimated value.

The configuration of the controller of this embodiment is exactly the same as that of the example shown in FIG. 1 except for the step width calculator 10, the response start area detector 11, and the dead time detector 12. The basic operation is the same as in the example of FIG. 1, and the sequential dead time updating unit 9 changes the dead time Lm stored in the internal model storage unit 6a to Lm1. In such a controller,
The step width calculation unit 10, the response start area detection unit 11, and the dead time detection unit 12 calculate an estimated value of the dead time Lp of the control target process 40 as described below.

FIG. 5 is a diagram illustrating the target value followability of the reference control amount ym for explaining the operation of the step width calculation unit 10. ym0 is the initial value of the reference control amount ym, and rm is the target value r.
Is a set value of the reference control amount ym and STm is a step width which is a difference between the set value rm and the initial value ym0 in the reference control amount ym. FIG. 5 shows a state in which the reference control amount ym changes from the initial value ym0 and is settled to the set value rm by inputting the target value r as in FIG. 3A.

Then, the step width calculating unit 10 calculates the step width ST of the control amount y by the following equation, similarly to the sequential dead time updating unit 9. ST = | r−y0 | (12) Further, the step width STm of the reference control amount ym is calculated as in the following equation. STm = | rm-ym0 | (13)

Here, the control amount y of the control target process 40 and the reference control amount ym output from the internal model 6 can be obtained from the target value r and the disturbance d by the following equation. y = F1 × F2 × Gp × Gc × r / {1 + F2 × Gc × (Gp−Gm)} + (1-F2 × Gm × Gc) × d / {1 + F2 × Gc × (Gp−Gm)} (14) ym = F1 × F2 × Gm × Gc × r / {1 + F2 × Gc × (Gm−Gp)} + (− F2 × Gm × Gc) × d / {1 + F2 × Gc × (Gm−Gp)}・ ・ (15)

Then, assuming that the initial value of the gain Km of the internal model 6 is Km0, the initial value ym0 of the reference control amount ym is given by the following equation from equations (14) and (15). ym0 = (y0−d) × Km0 / Kp (16) From the equation (16), if the disturbance d = 0, the control amount initial value y0
Is equal to the ratio between the gain Kp (here, an estimated value) of the control target process 40 and the gain initial value Km0 of the internal model 6, as shown in the following equation. Kp / Km0 = y0 / ym0 (y0 ≠ 0, ym0 ≠ 0) (17)

Therefore, the gain K of the process 40 to be controlled is
p can be estimated as follows: Kp = Km0 × y0 / ym0 (18) Similarly, the target value r and the set value r of the reference control amount ym
The ratio to m matches the ratio between the gain Kp and the initial gain value Km0. Kp / Km0 = r / rm (19)

Therefore, equation (13) is obtained by replacing equations (17) to
From (19), the following equation can be used. STm = | rm−ym0 | = | r × Km0 / Kp−y0 × Km0 / Kp | = | r−y0 | × Km0 / Kp = | (r × ym0 / y0) −ym0 | (20) , The step width calculation unit 10 calculates the step width ST of the control amount y and the step width STm of the reference control amount ym by the formula (1).
2) and (20) are calculated and output to the response start area detection unit 11.

Next, the response start area detecting section 11 determines the control amount y and the reference control amount y based on the step widths ST and STm.
A response start region of m, that is, a region where the control amount y and the reference control amount ym start to change from the initial values y0 and ym0, respectively, when the operation amount u is output from the operation amount calculating unit 4 in response to the input of the target value r. Is detected.

FIG. 6A is a diagram showing a response start area of the control variable y for explaining the operation of the response start area detecting unit 11.
FIG. 6B is a diagram showing a response start area of the reference control amount ym. i1 and i2 are the start and end times of the response start area of the control amount y, and im1 and im2 are the start and end times of the response start area of the reference control amount ym and y1 and y, respectively.
2 is the start time i1 and the end time i of the response start area, respectively.
2, the control amounts ym1 and ym2 are also the start time i
m1 is the reference control amount at the end time im2.

Also, 1, 2,... J, j + 1.
Each number of n + 1 is a sampling point, and the start point of the response start area is set to 1. FIGS. 6A and 6B correspond to enlarged views of the change start portion in FIGS. 3A and 5, respectively.

First, the response start area detecting section 11 sets the sampling time detected by the following equation as the start time i1 of the response start area of the control amount y. | Y−y0 |> β × ST (21) Here, β is a proportionality constant, for example, β = 0.05.

That is, at the start time i1 of the response start area of the control amount y, as shown in FIG. 6A, the threshold value is β × ST, and the difference between the current control amount y and the initial value y0 is this threshold. This is the first sampling point that exceeds the value. Thus, the control amount y1 at the start time point i1 is obtained.

Then, the start time im1 of the response start area of the reference control amount ym is similarly detected. | Ym−ym0 |> β × STm (22) That is, the start point i of the response start area of the reference control amount ym
m1 is the first sampling time point at which the difference between the current reference control amount ym and its initial value ym0 exceeds this threshold value with β × STm as the threshold value as shown in FIG. 6B. Thus, the reference control amount ym1 at the start time im1 is obtained.

Next, the end time point i2 of the response start area of the control amount y is n from the start time point i1 of the response start area of the control amount y.
It is assumed that the point is detected after sampling (for example, n = 9) or the first sampling point satisfying the following expression, whichever is detected first. | Y−y0 |> δ × ST (23) where δ is a proportionality constant, for example, δ = 0.20. When the sampling time according to the equation (23) is the end time i2, the number of samples from the start time i1 to the end time i2 is j.

That is, the end point i2 of the response start area of the control amount y is the point after 9 samplings in this embodiment (the point n + 1 in FIG. 6A), or the current control amount y and its initial value y0. At the first sampling point (j + 1 in FIG. 6A) when the difference from
6A, the time point j + 1 is set as the end time point i2 of the response start area in FIG. 6A. Thus, the control amount y2 at the end time point i2 is obtained.

The end time im2 of the response start area of the reference control amount ym is the time point n sampling after the start time im1 of the response start area of the reference control amount ym, or the same based on the sampling number j obtained above. The time after j sampling from the start time im1 is the earlier one. Thus, the reference control amount ym at the end time im2
2 is required.

Therefore, the number of samplings from the start time i1 to the end time i2 of the response start area of the control amount y and the number of samplings from the start time im1 to the end time im2 of the response start area of the reference control amount ym match. And the obtained sampling number of n or j is Ny.

As described above, the end time points i2 and im2
The reason why the detection of (i) is the earlier of the two points is that if only the detection with the fixed sampling number n is performed, the control amount y approaches the settling state earlier, and the detection of the end time points i2 and im2 may not be in time. Because. The response start area detection unit 11 controls the control amounts y1 and y2 and the reference control amounts ym1 and y in the response start area detected in this manner.
m2, the start time (time) i of the response start area of the control amount y
1. The sampling number Ny is output to the dead time detection unit 12.

The dead time detecting section 12 calculates an estimated value of the dead time Lp of the controlled process 40 from these values as follows. FIG. 7A is a diagram illustrating a response start region of the control amount y for explaining the operation of the dead time detection unit 12, and FIG. 7B is a diagram illustrating a response start region of the reference control amount ym.

In FIG. 7A, Lm2 is an estimated value of the dead time Lp of the process 40 to be calculated, i0 is the time when the control variable y starts to change, and A is a straight line on the control variable y from this point i0. B is the time from the intersection of the extension line of H2 to H1 to the initial value y0, and B is the time from this intersection to the start point i1 of the response start area. In FIG. 7B, i
m0 is a point in time when the reference control amount ym starts to change, C is an extension of the straight lines H4 to H3 on the reference control amount ym and the initial value y
D is the time from the intersection with m0 to the start time im1 of the response start area, and D is the time from the point im0 to the start time im1.

Here, the dead time Lm2 can be estimated as the elapsed time from the time 0 to the change start time i0 of the control amount y. Then, the proportional constant ε of the change start determination threshold value (Equation (11)) by the sequential dead time updating unit 9 and the proportional constant of the detection threshold value (Equation (21)) at the start time point i1 by the response start area detection unit 11 Since β is ε = β = 0.05, the dead time Lm1 obtained by the sequential dead time updating unit 9 is a time from time 0 to the start time point i1 as shown in FIG. 7A.

Since the reference control amount ym starts to change after the dead time Lm1 is determined as described above, Lm1 is the time from time 0 to the change start time im0 as shown in FIG. 7B.

Therefore, the estimated value Lm2 of the dead time is calculated as shown in FIG.
As shown in (a), the dead time Lm1 and the times A and B, which are sequentially obtained by the dead time updating unit 9 and are currently stored in the internal model storage unit 6a, can be calculated by the following equation. Lm2 = Lm1-BA (24)

The time from the start time i1 to the end time i2 of the response start area of the control amount y is equal to the sampling number N.
Since y becomes Ny × Δt from the control period Δt, the time B is given by the following equation. B = Ny × △ t × (y1-y0) / (y2-y1) (25) The time A can be obtained by the following equation. A = {(ym2-ym1) / (y2-y1)} ^1/2 × (D−C) (26)

The time from the start time im1 to the end time im2 of the response start area of the reference control amount ym is also Ny × △ t, and the time C is given by the following equation. C = Ny × △ t × (ym1-ym0) / (ym2-ym1) (27) The time D is expressed by the following equation. D = im1-Lm1 (28)

Therefore, the equation (24) can be replaced by the equations (25) to (2)
8) from the following equation. Lm2 = Lm1-Ny × △ t × (y1-y0) / (y2-y1)-{(ym2-ym1) / (y2-y1)} ^1/2 × ｛im1-Lm1-Ny × △ t × (ym1 −ym0) / (ym2-ym1)｝ (29)

The dead time detector 12 calculates the estimated value Lm2 of the dead time of the process 40 to be controlled by the equation (29). The dead time Lm2, which should be a value equal to or greater than 0, is calculated to be a negative value. When Lm2 <0, the following equation is used. Lm2 = Lm1-2 × B = Lm1-2 × Ny × △ t × (y1-y0) / (y2-y1) (30) According to equation (30), if Lm2 is still a negative value, Lm
2 = 0.

The dead time detector 12 calculates the dead time Lm2 calculated in this way after the end of the control response, that is, after the control amount y has settled to the target value r, the internal model storage 6a.
Output to In this way, the dead time Lm (currently Lm1) stored in the internal model storage unit 6a is changed to the estimated value Lm2 of the dead time of the control target process 40, but since it is changed after settling, it is actually used for control. This is after the next input of the target value r.

After that, the change to the dead time Lm1 by the sequential dead time updating unit 9 and the change to the dead time Lm2 by the dead time detecting unit 12 are not normally executed. As described above, the dead time Lm2 having higher accuracy than the dead time Lm1
Can be calculated.

FIG. 8 is a view showing a target value follow-up property when the controller of this embodiment is used for controlling the liquid level in the tank, and FIG. 9 is a view showing a target value follow-up property of the conventional IMC controller. It is. FIGS. 8 and 9 show a control value y (solid line), which is a target value r (dotted line) input at 0 seconds as a step input of a liquid level of 4 cm, and the control level is the liquid level.
This is a simulation result of obtaining. The conventional IMC controller used here does not change the dead time Lm of the internal model 6 in the controller of the present embodiment.

Here, the gain Kp of the process 40 to be controlled, which is the liquid in the tank, the time constant Tp is 20 seconds, the dead time Lp is 15 seconds, and the gain of the internal model 6 of the present embodiment and the conventional IMC controller is set. Km is 8, time constant Tm
Is set to 20 seconds, and since the dead time Lp of the control target process 40 is unknown, the dead time Lm is set to 0 second. Also,
The time constant T1 of the target value filter unit 2 of the present embodiment and the conventional IMC controller is 24 seconds, the time constant T2 of the target value / disturbance filter unit 4a is 6 seconds, and the control cycle Δt is 1 second.

FIG. 8 shows the first control response, which is the result of the change to the dead time Lm1 by the successive dead time updating unit 9. Therefore, the result is the same in the controller of FIG. Also,
The resulting dead time Lm1 is 18 seconds. As is clear from the comparison between FIGS. 8 and 9, it can be seen that the controller of the present embodiment can provide a more stable response than the conventional IMC controller. The calculated estimated value Lm2 of the dead time of the control target process 40 is 15 seconds.

In the example of FIG. 1, when noise is added to the control amount y output from the control amount input unit 7, the detection of the start of the change of the control amount y and the calculation of the dead time Lm1 become inaccurate, and the control becomes unstable. Become. Such noise of the control amount y includes, for example, a measurement noise by a sensor that measures the control amount y and outputs it to the control amount input unit 7.

[0079] Figure 10 shows another reference example of the present invention IMC
FIG. 2 is a block diagram of a controller having a structure, and portions similar to those in FIG. 1 are denoted by the same reference numerals. 9a is a sequential dead time updating unit that performs the same processing as the sequential dead time updating unit 9 of FIG. 1 based on the control amount y after the damping process output from the noise processing unit described later. This is a noise processing unit that performs a process of reducing noise.

[0080] Next, the operation of the controller of the present embodiment. The basic operation is the same as in the example of FIG. 1, but the noise processing unit 13 as a low-pass filter reduces the noise by damping the control amount y as in the following equation. yd (i) = {y (i) + Td × yd (i-1)} / (1 + Td) (31)

Here, yd (i) is the control amount y, y after the damping process at the sampling time i of the present control system.
(I) is the control amount y at the sampling time i.
Td is a damping time constant. For example, if the control period Δt is 1 second, Td = 5 seconds. Then, the control amount yd (i) after the damping process is sequentially input to the dead time updating unit 9a.

The operation of the sequential dead time updating unit 9a is the same as that of the sequential dead time updating unit 9 in FIG. 1, but the control amount yd (i) after the damping process is input, so that the change of the control amount y is detected. Is performed by the following equation corresponding to equation (11). | Yd (i) −yd (0) | <ε × ST2 (32) In the equation (32), ST2 is a step width corresponding to noise in the control amount y. And calculate.

ST2 = | r−r0 | (33) Here, r0 is an initial value of the target value r. Since the initial state of the control is also a settling state, if no noise is added to the control amount y, r0 = y0 and ST = | r−y0 | =
| R−r0 |. However, if noise is added to the control amount y, r0 ≠ y0, so the above ST2 is used as the step width corresponding to the noise. In this way, it is possible to correctly detect whether or not the control amount y has changed even if noise is added to the control amount y.

The operation of the sequential dead time updating unit 9a except for the detection of such a change in the control amount y is the same as that of the sequential dead time updating unit 9. Therefore, as in the example of FIG. 1, while the change in the control amount y does not appear as a control response, the dead time Lm of the internal model 6 is sequentially updated and finally changed to the dead time Lm1, thereby controlling the process to be controlled. 40 dead times L
Even if p is unknown and noise is added to the control amount y, deterioration of control characteristics can be avoided.

Also in the example of FIG. 4, if noise is added to the control amount y, the detection of the response start area and the calculation of the dead time Lm2 become inaccurate, so that it is necessary to deal with noise.
FIG. 11 is a block diagram of a controller having an IMC structure showing another embodiment of the present invention, and the same parts as those in FIGS.

Reference numeral 10a denotes a step width calculator for calculating the step width of the control amount and the reference control amount corresponding to the noise, similarly to the step width calculator 10 of FIG. 4, and 11a denotes the step width of the control amount and the reference control amount, and the noise. A response start area detection unit 12a that detects the start time i1, im1 and the end time i2, im2 of the response start area based on the control amount yd (i) after noise processing and the reference control amount ym output from the processing unit 13;
Are the start time i1 and im1 of the response start area and the end time i
2, the control amount y specified by im2 and the reference control amount y
A dead time detection unit that calculates an estimated value Lm2 of the dead time from the change rate of m, and changes the dead time Lm stored in the internal model storage unit 6a to this estimated value.

The step width calculation unit 10a calculates the step widths of the control amount y and the reference control amount ym in the same manner as the step width calculation unit 10.
For the same reason as in a, ST2 and STm2 are used as step widths corresponding to noise.

Therefore, the step width ST2 of the control amount y is calculated by the equation (33), and the step width STm2 of the reference control amount ym is calculated by the following equation based on the equation (20). STm2 = | (r × ym0 / r0) −ym0 | (34) Thus, the step width calculator 10a calculates the step width ST2 of the control amount y and the step width STm2 of the reference control amount ym, and calculates these. Output to the response start area detection unit 11a.

Next, the response start area detection section 11a outputs the step widths ST2 and STm2, the control amount yd (i) after the damping processing output from the noise processing section 13, and the output from the internal model output calculation section 6b. A response start area is detected based on the reference control amount ym.

The operation of the response start area detecting section 11a is almost the same as the operation of the response start area detecting section 11 in FIG. 4, but the sampling point detected by the following equation corresponding to the equation (21) is determined by the response of the control amount y. The start time of the start area is i1. | Yd (i) −yd (0) |> β × ST2 (35)

That is, at the start point i1 of the response start area of the control amount y, the difference between the current noise-processed control amount yd (i) and its initial value yd (0) exceeds the threshold value β × ST2. This is the first sampling time. Then, the start time im1 of the response start area of the reference control amount ym is similarly detected by the following equation. | Ym−ym0 |> β × STm2 (36)

Next, an equation corresponding to the equation (23) in the example of FIG. 4 for detecting the end point i2 of the response start area of the control amount y is as follows. | Yd (i) −yd (0) |> δ × ST2 (37) Thus, the end time point i2 is the start time point i as in the example of FIG.
The time after 1 to n samplings, or equation (37)
If the first sampling point that satisfies is detected, whichever is detected first, and the sampling point according to equation (37) is the end point i2, the start point i1 to the end point i
The sampling number up to 2 is j.

The end time im2 of the response start area of the reference control amount ym is n times after the start time im1 of the response start area of the reference control amount ym, or the same based on the sampling number j obtained above. The earlier of j sampling points from the start point. The response start area detection unit 11a calculates the start time i1, im1 and the end time i of the response start area thus detected.
2, and outputs im2 to the dead time detector 12a.

Next, the dead time detector 12a calculates the estimated dead time Lm of the process 40 to be controlled as follows.
2 is calculated. FIG. 12A is a diagram illustrating a response start region of the control amount y for explaining the operation of the dead time detection unit 12a, and FIG. 12B is a diagram illustrating a response start region of the reference control amount ym.

Since the control amount y and the reference control amount ym that are not subjected to the noise processing are input to the dead time detection unit 12a,
These values are affected by noise. FIG.
2 (a) and 2 (b) are similar to FIGS. 7 (a) and 7 (b), but show the effect of this noise in the response start region as at the time of sampling with a circle. , Only a straight line approximating these is shown.

The estimated value Lm2 of the dead time is shown in FIG.
At times A2 and B2 defined in the same manner as in FIG.
Can be calculated from the following equation. Lm2 = Lm1-B2-A2 (38)

The dead time detector 12a stores the control amount y after the initial value y0, and sets the start time i of the response start area.
1. The control amount y specified by the end time point i2, that is, the control amounts y (i1) to y (i2) in the response start area.
Is analyzed by the least-squares method to obtain a linear function expression such as the following expression. y (i) = Ay × i + By (39)

That is, equation (39) is an equation representing the straight lines H6 to H5 in FIG. 12 (a), Ay is a control amount change rate which is the slope of this straight line, and By is a constant term which is also an intercept. In addition, the dead time detection unit 12a similarly calculates the initial value ym
The reference control amounts ym after 0 are stored, and the reference control amounts ym (im1) to ym (im2) in the response start area are analyzed by the least square method to obtain the following equation. ym (i) = Aym × i + Bym (40)

This is an equation representing the straight lines H8 to H7 in FIG. 12B by the reference control amount change rate Aym and the constant term Bym. Thus, the control amount change rate Ay, the reference control amount change rate Aym, and the constant terms By and Bym can be calculated.

Next, the time B2 is given by the following equation from these obtained values. B2 = i1- (r0-By) / Ay (41) The time A2 can be obtained by the following equation. A2 = (Aym / Ay) ^1/2 × (D2-C2) (42)

The times C2 and D2 are given by the following equations. C2 = im1- (ym0-Bym) / Aym (43) D2 = im1-Lm1 (44)

Therefore, Expression (38) is obtained by Expressions (41) to (4).
4) From the following equation: Lm2 = Lm1- {i1- (r0-By) / Ay}-(Aym / Ay) ^1/2 * {-Lm1 + (ym0-Bym) / Aym} (45)

The dead time detector 12a calculates an estimated value Lm2 of the dead time of the process 40 to be controlled by the equation (45).
When is calculated to be a negative value, the following equation is used. Lm2 = Lm1-2 × {i1- (r0-By) / Ay} (46) If Lm2 is still a negative value according to equation (46), Lm
2 = 0.

The dead time detector 12a calculates the dead time L
m2 is output to the internal model storage unit 6a after the settling of control similarly to the dead time detection unit 12, and the dead time of the internal model 6 is changed to Lm2. As described above, the dead time Lm2 that is more accurate than the dead time Lm1 can be calculated even when noise is added to the control amount y.

FIG. 13 is a diagram showing the target value follow-up property when the controller of this embodiment is used for controlling the level of the liquid level in the tank similarly to the example of FIG. 8, and FIG. 14 is the control amount of this controller. FIG. 1 shows a control amount y output from the input unit 7;
FIG. 5 is a diagram showing the target value followability of the controller of FIG. In FIG. 14, the control amount yt measured by the sensor and the control amount y input from the sensor to the control amount input unit 7 is shown, and a state in which measurement noise by the sensor is added is shown (FIG. 8). Then yt = y).

Here, the control target process 40, the parameters of the present embodiment and the controller of FIG. 4 are the same as those in the examples of FIGS. 8 and 9, the measurement noise has a maximum amplitude of 0.4 (10% of the step width ST2), The random noise has an average value of 0.
As is clear from the comparison between FIGS. 13 and 15, it can be seen that the controller of the present embodiment can provide a more stable response than the controller of FIG. 4 even when the control amount y includes noise. Further, while the dead time Lm1 calculated by the controller of FIG. 4 is 4 seconds, the dead times Lm1 and Lm2 calculated by the controller of the present embodiment are 21 seconds and 1 second.
4 seconds, a value close to the dead time Lp of the control target process 40 is obtained.

In the examples of FIGS. 1 and 10, the dead time Lm1 is calculated. In the examples of FIGS. 4 and 11, this dead time Lm1 and the dead time Lm2 with higher accuracy are calculated, and the dead time Lm of the internal model 6 is calculated.
However, if the identification error of the other parameters of the internal model 6, that is, the gain Km and the time constant Tm is large, the errors of the calculated dead times Lm1 and Lm2 may be large. Since the control after the correction becomes unstable, it is necessary to evaluate the dead time estimation accuracy.

FIG. 16 shows an IMC showing another embodiment of the present invention.
FIG. 17 is a block diagram of a controller having a structure, and FIG. 17 is a diagram showing a certainty factor calculated by a certainty factor calculation unit of the controller. Reference numeral 12b denotes a correction dead time detection unit, which is a start time point i of the response start area output from the response start area detection unit 11a.
1, a control target process 4 based on a control amount y specified by an end time i2, im2, a change rate Ay of the reference control amount ym, Aym, and a certainty factor that is a dead time estimation accuracy.
The estimated value of the dead time of 0 is calculated, and the dead time Lm of the internal model 6 is changed to this estimated value. Reference numeral 14 denotes a certainty factor calculation unit that calculates a certainty factor based on a change rate Ay of the control amount y and the reference control amount ym.

The configuration of the controller of this embodiment is the same as that of FIG. 11 except for the correction dead time detecting unit 12b and the certainty factor calculating unit 14.
This is exactly the same as the example. The basic operation is the same as that of the example of FIG. 11, but the certainty factor calculating unit 14 calculates the certainty factor which is the dead time estimation accuracy as follows.

The certainty factor calculation unit 14 stores the control amount y after the initial value y0 output from the control amount input unit 7 and the reference control amount ym after the initial value ym0 output from the internal model output calculation unit 6b. 11 based on the start times i1 and im1 and the end times i2 and im2 of the response start area output from the response start area detection unit 11a by the least square method.
The control amount change rate Ay and the reference control amount change rate Aym in the response start area are calculated in the same manner as the dead time detection unit 12a.

Then, the confidence factor CF is calculated by the following equation.
It outputs to the correction dead time detection part 12b. CF = exp {− (1-Ay / Aym) ² / AC} (47) where AC is a positive value, for example, AC = 10.
0. The certainty factor CF is Ay / A as shown in FIG.
This is a bell-shaped function where CF = 1 when ym = 1, and indicates that the dead time estimation accuracy is the highest when Ay / Aym = 1.

The reference control amount ym changes as a control response from the time when the dead time Lm of the internal model 6 is sequentially changed to the dead time Lm1 calculated by the dead time updating unit 9a as described above. Therefore, the above ratio Ay / Aym
Calculating the confidence factor CF based on
The result of the control response by m1 will be evaluated.

Next, the operation of the corrected dead time detecting section 12b is substantially the same as that of the dead time detecting section 12a, but by performing calculation based on the confidence factor CF, control is performed by the following equation corresponding to the equation (45). The estimated value Lm2 of the dead time of the target process 40 is calculated. Lm2 = CF × [Lm1-i1 + (r0-By) / Ay- (Aym / Ay) ^1/2 × {-Lm1 + (ym0-Bym) / Aym}] (48)

When the dead time Lm2 is calculated to be a negative value, the following equation is used. Lm2 = CF × [Lm1-2 × {i1- (r0-By) / Ay}] (49) According to equation (49), if Lm2 is still a negative value, Lm
2 = 0. The reason why the calculation is performed by multiplying the certainty factor CF in the IMC controller is to set the dead time Lm of the internal model 6 to be smaller than the dead time Lp of the control target process 40, rather than to be set longer. Because it is safe.

The dead time detection unit 12b outputs the dead time Lm2 to the internal model storage unit 6a after the settling of control similarly to the dead time detection unit 12a to change the dead time of the internal model 6 to Lm2. . In this way, by using the certainty factor CF, it is possible to prevent improper dead time correction from being performed and to obtain a safe response.

In addition, the dead time Lm1 by the sequential dead time updating unit 9a and the dead time detecting unit 12a in the example of FIG.
Although the calculation of Lm2 is performed only once, the detection result may still be insufficient even with the dead time Lm2 based on the above-described certainty factor CF.
b is the reliability evaluation by the following equation, and the dead time Lm
1. Determine whether to calculate Lm2. Ay / Aym> Alm1 (50) Ay / Aym <Alm2 (51) where Alm1 is the upper limit of the rate of change in the response start region,
lm2 is the lower limit of the change rate ratio of the response start region, for example, Alm1 = 2 and Alm2 = 0.1.

If both equations (50) and (51) do not hold, that is, if the change rate ratio Ay / Aym of the response start area is within the upper and lower limits, it is determined that the dead time detection result is sufficiently reliable and the subsequent control is performed. In response, successively dead time updating section 9a
Calculation of the dead time Lm1 is stopped, and the dead time Lm
The calculation of 2 is not executed. If the equation (50) or (51) holds, it is considered unreliable, so that the dead time updating unit 9a sequentially calculates the dead time Lm1 even when the next target value r is input, and then the dead time Lm2
Is calculated.

For example, in the example of FIG. 11, the gain Km of the internal model 6 of the controller is 80, and the time constant Tm is 4
If the dead time Lm is 0 second, the gain Kp of the control target process 40 is 8, the time constant Tp is 20 seconds, and the dead time Lp is 10 seconds, the finally obtained dead time Lm2 is 6
Seconds. This is the dead time Lp of the process 40 to be controlled.
Since the error rate is 66.7%, which is not a value that can be adopted as a fixed value of the dead time Lm of the internal model 6, it should be determined that it is better to continue the dead time detection without fixing this value. is there.

When control is performed by applying the above parameters to this embodiment, Ay / Aym = 0.006 / 0.097 =
Since 0.062 <Alm2, it is possible to avoid setting the inappropriate result as described above as a fixed value of the dead time Lm of the internal model 6. The gain Km and the time constant Tm of the internal model 6 are determined by the operator or the like.
If the control is executed again after the correction, the more reliable dead time Lm2 can be determined as the fixed value of the internal model 6.

In the present embodiment, the calculation of the certainty factor CF and the dead time Lm2 based on the certainty factor CF and the reliability evaluation are applied to the example of FIG. 11, but may be applied to the example of FIG.
At this time, the calculation of the certainty factor CF performed by the certainty factor calculation unit is based on the control amounts y1 and y2 and the reference control amounts ym1 and ym2 in the response start area output from the response start area detection unit 11. CF = exp [-{1- (y2-y1) / (ym2-ym1)} ² / AC] (52)

The calculation of the dead time Lm2 performed by the dead time detecting unit similar to the dead time detecting unit 12 is represented by the following equation. Lm2 = CF × [Lm1-Ny × △ t × (y1-y0) / (y2-y1) − {(ym2-ym1) / (y2-y1)} ^1/2 × ｛im1-Lm1−Ny × △ t × (ym1-ym0) / (ym2-ym1)}] (53)

When the dead time Lm2 is calculated to be a negative value, the following equation is used. Lm2 = CF × {Lm1−2 × Ny × {t × (y1-y0) / (y2-y1)} (54) According to the equation (54), if Lm2 is still a negative value, Lm
2 = 0.

The reliability evaluation is performed by the following equations (50) and (51).
Is performed according to the following equation. (Y2-y1) / (ym2-ym1)> Alm1 (55) (y2-y1) / (ym2-ym1) <Alm2 (56) Thus, the present invention is also applied to the example of FIG. be able to.

In any of the examples of FIGS. 1, 4, 10, 11, and 16, the dead time Lm of the internal model 6 can be set when the dead time Lp of the process to be controlled is unknown, but the estimated dead time Lm1 Alternatively, a time delay may occur in Lm2 with respect to the actual dead time Lp of the process 40 to be controlled.

When the detection of the dead time Lm1 by the dead time updating unit 9 or 9a is continuously performed, for example, by the operator's judgment for the control target process 40 in which the dead time Lp fluctuates irregularly, the fluctuation amount is changed. The control including the dead time error is performed, and the dead time Lm of the internal model 6 is detected.
If it is fixed to m2, there is a possibility that the dead time error will increase due to the accumulation of the fluctuations, so it is necessary to take measures to avoid such a time delay.

FIG. 18 shows an IMC showing another embodiment of the present invention.
It is a block diagram of the controller of a structure, The same code | symbol is attached | subjected to the part similar to FIGS. Reference numeral 15 denotes a start region time difference detection unit for detecting a time difference between start times i1 and im1 of the response start region in the control amount y and the reference control amount ym from the detection result of the response start region detection unit 11a. Of the internal model 6 stored in the internal model storage unit 6a, the dead time Lm of which the error was corrected is calculated from the rate of change Ay, Aym, and the time difference detected by the start area time difference detection unit 15. This is a dead time correction unit for changing to a dead time.

The internal model storage section 6a of this embodiment stores, for example, the dead time Lm detected by the controller shown in FIG.
(Lm1 in the example of FIG. 10) is stored in advance, and the dead time Lm is defined as Lm0. As a result of the control response based on Lm0, the step width calculation unit 10a similar to the example of FIG. 11 executes the step width ST of the control amount y and the reference control amount ym.
2. The response start area detecting unit 11a calculates STm2, and the response start area detection unit 11a starts the response start area at start time i1, im1, and end time i2, i.
m2 is detected.

Then, the start area time difference detecting section 15 determines the start time i1 of the response start area of the control amount y and the reference control amount ym.
Is calculated by the following equation. Δi = im1-i1 (57) Next, the dead time correction unit 16 calculates the dead time after the error correction as follows.

FIG. 19 is a diagram showing a response start area of the control amount y and the reference control amount ym for explaining the operation of the dead time correction unit 16, and is similar to FIGS. 12 (a) and 12 (b). Are given the same reference numerals. FIG. 19 shows FIG.
FIG. 4B is a diagram similar to (b), but is shown on the same time axis, where Lm3 is a dead time to be calculated, and B4 is the time from the time i0 when the control amount y starts to change to the start time i1 of the response start area. Time.

The dead time correcting section 16 is provided with the dead time detecting section 1.
Similarly to 2a, the control amount y after the initial value y0 is stored, and the control amounts y (i1) to y (i) in the response start area are stored.
2) is analyzed by the least squares method, and the control amount change rate Ay and the constant term By are calculated. Similarly, the reference control amount ym after the initial value ym0 is stored, and the reference control amounts ym (im1) to ym (im2) in the response start area are analyzed by the least square method, and the reference control amount change rate Aym, Constant term By
Calculate m.

Here, the following equation is established between the dead time Lm3 and the preset dead time Lm0 from FIG. Lm0 + C3 = Lm3 + B4 + △ i (58) Further, the time B3 is given by the following equation from the time B2 in FIG. B3 = 2 × B2 = 2 × {i1- (r0-By) / Ay} (59)

Therefore, the time B4 is given by the following equation from the equation (59). B4 = 2 × {i1- (r0−By) / Ay} × (1 + 0.1 × Td) (60) Here, the reason for using the damping time constant Td is that the damping process of the noise processing unit 13 is performed. This is to correct the displacement due to

The time C3 is given by the following equation from the time C2 in FIG. C3 = 2 × C2 = 2 × {im1- (ym0−Bym) / Aym} (61)

Therefore, the dead time Lm3 is calculated by the following equation (58).
From (61), the following equation is obtained. Lm3 = Lm0 + C3- △ i-B4 = Lm0 + 2 × ｛im1- (ym0-Bym) / Aym｝-△ i-2 × ｛i1- (r0-By) / Ay｝ × (1 + 0.1 × Td) (62) If Lm3 according to equation (62) is a negative value, Lm3 = 0.

Then, the dead time correction unit 16 outputs the dead time Lm3 to the internal model storage unit 6a after the control is settled, and changes the dead time Lm0 of the internal model 6 to Lm3. Therefore, the dead time Lm0 stored in the internal model storage unit 6a is equal to the dead time L of the process 40 to be controlled.
Even when the delay time p is delayed or the dead time Lp fluctuates irregularly, the dead time Lm0 can be calculated by correcting the error, and the dead time Lm0 can be corrected. it can.

In this embodiment, the case where noise has been added to the control amount y has been described. However, if there is no noise, the step width calculating unit 10a and the response start area detecting unit 11a are replaced with the steps shown in FIG. Using the width calculation unit 10 and the response start area detection unit 11, the equation (6) of the dead time correction unit 16 is used.
This can be dealt with by making 2) the following equation. Lm3 = Lm0 + 2 × {Ny × {t × (ym1-ym0) / (ym2-ym1)}} − {i−2 × {Ny × {t × (y1-y0) / (y2-y1)}} (63)

[0137]

According to the present invention, the dead time of the internal model can be set during the operation of the target value tracking control by providing the sequential dead time updating unit even if the dead time of the process to be controlled is unknown. Therefore, highly accurate and reliable control can be performed, and it is possible to cope with a change in characteristics of a process to be controlled. In addition, since it is possible to prevent the occurrence of trouble due to the identification error of the process to be controlled,
It is possible to reduce the workload of an operator who does not have specialized control knowledge. In addition, the step width calculator, the response start area
By providing the area detection unit and the dead time detection unit,
Dead time of the internal model set by the dead time update unit
Estimation of Dead Time of Controlled Process with Higher Accuracy
Value can be calculated and the dead time of the internal model
It can be modified to an estimate.

[0138]

[0139]

Further, by providing a successive dead time updating section, a noise processing section, a step width calculating section, a response start area detecting section, and a dead time detecting section, the dead time of the process to be controlled is unknown and the control amount is reduced. Even if noise is added, a highly accurate estimated value of the dead time of the process to be controlled can be calculated, and the dead time of the internal model can be corrected to the estimated value.

Further, by providing a successive dead time updating unit, a noise processing unit, a step width calculating unit, a response start area detecting unit, a certainty factor calculating unit, and a corrected dead time detecting unit, inappropriate dead time correction can be performed. And a safe response can be obtained.

A noise processing unit, a step width calculating unit,
By providing a response start area detection unit, a start area time difference detection unit, and a dead time correction unit, a case where the dead time of the preset internal model is delayed with respect to the dead time of the control target process or the control target process Even when the dead time fluctuates irregularly, the dead time of the internal model can be corrected by calculating the dead time in which the error is corrected, so that a safe response can be obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram of a controller having an IMC structure showing a reference example of the present invention.

FIG. 2 is a block diagram of a control system using the controller having the IMC structure of FIG. 1;

FIG. 3 is a diagram illustrating a target value followability of a control amount and a change start portion of the control amount.

4 is a block diagram of a controller of the IMC structure showing an embodiment of the present invention.

FIG. 5 is a diagram showing target value followability of a reference control amount.

FIG. 6 is a diagram illustrating a response start area of a control amount and a reference control amount.

FIG. 7 is a diagram illustrating a response start area of a control amount and a reference control amount.

FIG. 8 is a diagram showing target value followability of a controller having the IMC structure of FIG. 4;

FIG. 9 is a diagram showing target value followability of a conventional IMC controller.

Figure 10 is a block diagram of a controller of the IMC structure shown another exemplary embodiment of the present invention.

FIG. 11 is a block diagram of a controller having an IMC structure according to another embodiment of the present invention.

FIG. 12 is a diagram illustrating a response start region of a control amount and a reference control amount.

FIG. 13 is a diagram showing target value followability of a controller having the IMC structure of FIG. 11;

14 is a diagram illustrating a control amount y output from a control amount input unit in FIG. 11;

FIG. 15 is a diagram showing target value followability of the controller of FIG. 4;

FIG. 16 is a block diagram of a controller having an IMC structure according to another embodiment of the present invention.

17 is a diagram showing a certainty factor calculated by a certainty factor calculation unit of the controller in FIG. 16;

FIG. 18 is a block diagram of a controller having an IMC structure according to another embodiment of the present invention.

FIG. 19 is a diagram illustrating a response start region of a control amount and a reference control amount.

FIG. 20 is a block diagram of a control system using a conventional IMC controller.

[Explanation of symbols]

 2 target value filter section 3 first subtraction processing section 4 manipulated variable calculation section 4a target value / disturbance filter section 4b operation section 6 internal model 6a internal model storage section 6b internal model output calculation section 8 second subtraction processing section 9, 9a Sequential dead time updating unit 10, 10a Step width calculating unit 11, 11a Response start region detecting unit 12, 12a Dead time detecting unit 12b Modified dead time detecting unit 13 Noise processing unit 14 Confidence calculating unit 15 Start region time difference detecting unit 16 Dead time correction section

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-1-106103 (JP, A) JP-A-58-146901 (JP, A) JP-A-3-111904 (JP, A) JP-A-3-106 152601 (JP, A) (58) Field surveyed (Int. Cl. ⁷ , DB name) G05B 13/02-13/04

Claims (4)

(57) [Claims] An operation amount to be output to a control target process is calculated from a control target value, and a reference control amount corresponding to a control amount of the control target process, which is a control result, is calculated by an internal model expressing the control target process in a mathematical expression. Calculate and feed back the difference between the control amount and the reference control amount.
A controller having an MC structure, wherein a target value filter unit that outputs a target value of the input control with a time delay characteristic of a transfer function; a first subtraction processing unit that subtracts a feedback amount from an output of the target value filter unit; A target value / disturbance filter unit that outputs the output of the first subtraction processing unit with a transfer function having a time delay characteristic, and an operation amount is calculated from the output of the target value / disturbance filter unit based on the parameters of the internal model. An operation amount calculation unit comprising: an operation unit for outputting the internal model; an internal model storage unit for storing the parameters of the internal model; and an internal model output calculation for calculating a reference control amount from the operation amount based on the parameters of the internal model. And a unit for subtracting the reference control amount output from the internal model output calculation unit from the control amount of the control target process to output the feedback amount. And a subtraction processing unit, while a change as a control response does not appear in the control amount, increases and updates the dead time in the parameters of the internal model and updates the dead time in the parameters stored in the internal model storage unit. A successive dead time updating unit that repeats changing to the updated dead time , and an initial value and an adjustment based on the target value, the control amount, and the reference control amount.
The step width, which is the difference from the fixed value, is set for each of the control amount and the reference control amount.
A step width calculation unit for calculating each of them, a start time of a first response start area where a control amount starts to change, and
Detecting the end point based on the step width of the control amount
And the second response start area in which the reference control amount starts to change.
The start time and end time of the
A response start area detection unit that detects the first response start area based on the start time and the end time of the first response start area.
The amount of change in the controlled variable and the second response start area
Change of reference control amount specified by start time and end time
The correction time based on the quantity is geometrically calculated and the internal model is calculated.
In the parameters of the internal model stored in the
By subtracting the correction time from the time,
A controller configured to calculate an estimated value of the dead time of the process, and to change the dead time in the parameters of the internal model stored in the internal model storage unit to the estimated value; 2. A control target value is output to a process to be controlled.
Calculate the manipulated variables to be applied, and formulate the controlled process
Control of the process to be controlled,
Calculates the reference control amount corresponding to the control amount, and calculates the control amount and the reference control.
I which controls by feeding back the difference from the quantity
In the controller of MC structure, the transfer target function receives the target value of the control with the time delay characteristic.
A target value filter for outputting, and a feedback amount reduced from the output of the target value filter.
A first subtraction processing unit for calculating the output of the first subtraction processing unit, and a transfer function having a time delay characteristic.
The target value / disturbance filter section to be output and the parameters of the internal model
The output of the target value / disturbance filter unit based on the meter
Control unit that calculates the control input from the
Arithmetic unit and an internal model storage for storing parameters of the internal model
And the operation amount based on the parameters of the internal model.
An internal model output calculation unit for calculating a reference control amount; and the internal model output calculation unit based on a control amount of a process to be controlled.
Subtracting the reference control amount output from the
A second subtraction processing unit that outputs a control amount, and a control unit that outputs no change as a control response to the control amount.
Update by increasing dead time in parameters of internal model
In the parameters stored in the internal model storage unit,
Change the dead time to this updated dead time.
A continuous delay updating unit, a noise processing unit that performs processing to reduce noise of the control amount of the control target process, and outputs the noise-reduced control amount to the sequential dead time updating unit, a target value, and a control amount. And the initial control value based on the
The step width, which is the difference from the fixed value, is set for each of the control amount and the reference control amount.
And a control amount after noise processing output from the noise processing unit.
Start and end times of the first response start area where the response starts to change
When a point is detected based on the step width of the control amount,
Then, the second response start area in which the reference control amount starts to change is opened.
The start time and the end time are determined based on the step width of the reference control amount.
And a response start area detecting section for detecting the first response start area and a start time and an end time of the first response start area.
The rate of change of the controlled variable and the second response start area.
Change of reference control amount specified by start time and end time
The correction time based on the ratio is geometrically calculated, and the internal model is calculated.
In the parameters of the internal model stored in the
By subtracting the correction time from the time,
A controller configured to calculate an estimated value of the dead time of the process, and to change the dead time in the parameters of the internal model stored in the internal model storage unit to the estimated value; 3. The controller according to claim 2 , wherein a change rate of the control amount and a change rate of the reference control amount are determined.
And the dead time changed by the sequential dead time updating unit.
A confidence calculation unit that calculates a confidence for evaluating the estimation accuracy of time; and the internal model storage unit instead of the dead time detection unit.
From the dead time in the parameters of the internal model stored in
The result obtained by subtracting the correction time is erroneously calculated using the certainty factor.
By correcting the difference, the dead time of the controlled process
A controller configured to calculate an estimated value and change a dead time in parameters of the internal model stored in the internal model storage unit to the estimated value; 4. An operation amount to be output to a control target process is calculated from a control target value, and a reference control amount corresponding to a control amount of the control target process, which is a control result, is calculated by an internal model expressing the control target process in a mathematical expression. Calculate and feed back the difference between the control amount and the reference control amount.
A controller having an MC structure, wherein a target value filter unit that outputs a target value of the input control with a time delay characteristic of a transfer function; a first subtraction processing unit that subtracts a feedback amount from an output of the target value filter unit; A target value / disturbance filter unit that outputs the output of the first subtraction processing unit with a transfer function having a time delay characteristic, and an operation amount is calculated from the output of the target value / disturbance filter unit based on the parameters of the internal model. An operation amount calculation unit comprising: an operation unit for outputting the internal model; an internal model storage unit for storing the parameters of the internal model; and an internal model output calculation for calculating a reference control amount from the operation amount based on the parameters of the internal model. And a unit for subtracting the reference control amount output from the internal model output calculation unit from the control amount of the control target process to output the feedback amount. And subtraction processing section, a noise processing unit for performing processing of reducing the control amount of the noise of the controlled process, the target value, based on the control amount and the reference controlled variable, initial value and integer
The step width, which is the difference from the fixed value, is set for each of the control amount and the reference control amount.
And a control amount after noise processing output from the noise processing unit.
Start and end times of the first response start area where the response starts to change
When a point is detected based on the step width of the control amount,
Then, the second response start area in which the reference control amount starts to change is opened.
The start time and the end time are determined based on the step width of the reference control amount.
A response start area detecting section for detecting the start time of the first response start area;
Japanese and initiation region time difference detector for detecting a time difference between the start time of the start area, the end time and start time of the first response start region
The rate of change of the controlled variable to be set, and the opening of the second response start area.
Change rate of reference control amount specified by start time and end time
And the correction time based on the time difference is used for the noise processing.
The geometric model is calculated while correcting the displacement, and the internal model is calculated.
In the parameters of the internal model stored in the
Error correction by subtracting the correction time from the time
Controller; and a is the calculated dead time, the internal model storage section into the dead time correction unit for changing the dead time to the calculated dead time in the parameter of the stored internal model.