JPH04148203A - Fuzzy combination type two-degree-of-freedom control device - Google Patents

Abstract

PURPOSE:To improve performance by deviating forcedly a PI control target value by using a membership function which varies dependently on both the difference between the input and the output of a target value filtering means and system deviation when the operated target value of the target value filtering means approaches to a control target value. CONSTITUTION:The difference DELTASV between the control target value SV and the operated target value SVo by the target value filtering means 1 is obtained, and when the difference DELTASV becomes below a prescribed value, deviation (e) is fetched and introduced to a substracted output changing means 14. The subtracted output changing means 14 executes the arithmetic operation of delta = DELTASV-e, and when this value deltacomes within a range not to obstruct a two-degree-of-freedom faculty, it outputs the function m(delta) to take the value of '0' to '1' in accordance with the magnitude of delta by a fuzzy coefficient setting means 16, and executes arithmetic operation so that SVa = SVo + m(delta) X (SV-SVo) is established, and changes the proportional integral (PI) control target value in accordance with the response state of the target value filtering means 1 and the following state of control without giving a shock to a control system. Thus, target value setting time can be quickened, and controllability can be improved.

Description

Detailed Description of the Invention [Objective of the Invention (Industrial Application Field) The present invention provides two degrees of freedom PI or PID (P: proportional, I: integral, D: differential) adjustment using a target value filter means. The present invention relates to a fuzzy combination type two-degree-of-freedom adjustment device that simultaneously optimizes suppression characteristics for a single disturbance to a controlled object and follow-up characteristics for changes in a target value, while providing coordination with changes in the target value.

(Prior Art) A conventional target value filter type two-degree-of-freedom PI adjustment device is constructed as shown in FIG. That is, this adjustment device introduces the target value Sv into the target value filter means 1, performs arithmetic processing to make the proportional gain two degrees of freedom, and obtains the calculated target value Sv.
After obtaining vo, this calculation target value S V o and controlled object 2
The control amount Pv from is guided to the deviation calculation means 3, where (
S.V.

-PV) to find the deviation E. Furthermore, the deviation E obtained by the deviation calculation means 3 is calculated by Kp(1-1/(T+
・S)) is guided to the PI adjustment means 4 having a transfer function of
Here, a PI adjustment calculation is performed to obtain a manipulation signal MV. The operation signal MV and the disturbance are added and synthesized by the addition means 5, and then applied to the controlled object 2, thereby controlling the calculated target value 5Vo - the control amount Pv.

In the above equation, is the proportional gain, T is the integration time, and S is the Laplace operator.

On the other hand, the target value filter means 1 has a multiplication means 11 which multiplies the externally introduced target value Sv by a two-degree-of-freedom coefficient α of the proportional gain. means 12, a first-order lag element 11 which performs a first-order lag calculation using an integral time as a time constant on the output of this subtracting means 12, and outputs the result; an output of this first-order lag element 1) and an output of the coefficient means 11; It is constituted by adding means 14 and the like that perform addition and synthesis to obtain the calculation target value SVO.

Therefore, in the control system as described above, the transfer function CPM(S) between Pv and MV, and the transfer function C between S V and M V
SM(S) is respectively CPM(S) - MY/PV = K p (1+ 1 / T t -8)...
(1) Csm(S) -MV/S V =Kp (α+1/TI -8) (2). α is a two-degree-of-freedom coefficient (a constant that can be set between 0 and 1) for the proportional gain. Therefore, if %TI is determined so that the disturbance suppression characteristics are optimized, and then the proportional gain coefficient α for 2 degrees of freedom is determined so that the target value tracking characteristics are optimized, 2 degrees of freedom can be achieved. can.

(Problem to be Solved by the Invention) By the way, the target value filter type two-degree-of-freedom adjustment device as described above has an excellent feature of being able to simultaneously optimize the disturbance suppression characteristic and the target value tracking characteristic. There is a problem that takes a long time to settle down.

Therefore, when considering the cause of this, it is found that the target value filter means 1 has at least one stage or two or more first-order delay elements, and when the target value Sv is changed in a stepwise manner, the target value of the step change is This is because it takes time for the value Sv to reach the final value due to the influence of the first-order lag element.

Furthermore, the influence of the first-order delay will be explained from the response characteristics shown in FIG. In other words, when the target value Sv is changed stepwise in the apparatus shown in FIG. 5, only the output (SV·α) of the coefficient means 1 changes stepwise as shown in FIG. 12 output (SV・(1−α
)) gradually increases under the influence of the first-order lag element 13 and approaches the target value Sv.

Assuming that 5v-x and 5V0-y, the digital arithmetic expression for the target value filter means 1 shown in FIG. 5 is obtained as shown below. First, the transfer function of the target filter means 1 is as follows, and if this equation (3) is expressed as a differential equation, )l +
T.

y t -ax+aT+ i t ten (1-a) ...(4) becomes.

here, this For the expression, Substituting the relational expression, we get can be obtained.

moreover, Transform this formula and, can be obtained.

When the target value Sv changes stepwise at time n-1, if n≧2, xIl ■ It becomes like this.

In this equation, Δt is very small compared to TI, and (X.-ye-+) is also small, so the value in the latter half of the open equation becomes very small.

Moreover, the closer the output ys-r is to the input X, the more △
The change in y becomes smaller. As a result, the output y is the input
, it will take a considerable amount of time to match.

Naturally, since this output y is the target value of the PI adjustment means 4, it takes a very long settling time.

The present invention has been made in view of the above-mentioned circumstances, and it is possible to finely vary the target value to the PI or PID adjusting means while ascertaining the response state of the target value filter and the follow-up state of the control, thereby achieving successive control and controllability. An object of the present invention is to provide a fuzzy combination type two-degree-of-freedom adjustment device that can improve the performance.

[Configuration of the Invention (Means for Solving the Problems) In order to solve the above problems, the present invention is provided with a target value filter means, and a target value obtained from the target value filter means side and a control from a controlled object. In a two-degree-of-freedom adjustment device that executes at least a PI or PID adjustment calculation using a deviation from the amount and applies the obtained operation signal to the controlled object, the control target value introduced into the target value filter means and this target subtracting means for extracting the difference from the calculated target value obtained by the calculation of the value filter means; discriminating means for calculating the deviation when the subtracted output of the subtracting means becomes less than a predetermined value; and from the subtracted output of the subtracting means. Subtraction output variable means for varying the output of the subtraction means according to the polarity and magnitude of the deviation by subtracting the deviation taken in by the discrimination means, and the output of the subtraction output variable means or the absolute value of the output. Fuzzy that outputs a predetermined function in response to
The fuzzy coefficient is applied to the output of the coefficient setting means and the subtraction means.
and a multiplication means for multiplying the function output from the coefficient setting means, and by adding and synthesizing the multiplication output obtained from the multiplication means and the calculation target value obtained by the calculation of the target value filter means, the control target is PI or P depending on the difference between the sliding value and the calculated target value and the state of the deviation.
This is a configuration in which the target value for ID adjustment is varied.

(Function) Therefore, by taking the above-mentioned measures, the present invention obtains the difference ΔSV between the control target value Sv and the calculated target value SVo by the target value filter means, and calculates the difference ΔSV when the difference ΔSv becomes a predetermined value or less. At this time, the deviation e is taken in and introduced into the subtraction output variable means. In this subtraction output variable means, δ−Δ5V−e
When the value δ falls within a range that does not inhibit the function of two degrees of freedom, the fuzzy coefficient setting means calculates δ
Outputs a function m(δ) that takes a value between 0 and 1 corresponding to the size of SV a −Svo+ m (δ) x (SV
SVo ), and by varying the target value for PI adjustment according to the response state of the target value filter means and the follow-up status of the control without giving a shock to the control system, the target value settling time can be reduced. This speeds up the process and improves controllability.

(Example) Hereinafter, an example of the present invention will be described with reference to FIG. In this figure, the same parts as in FIG. 5 are given the same reference numerals, and a detailed explanation thereof will be omitted, and hereinafter, only the parts that are different from the conventional apparatus will be explained. In this device, the control target value SV at the input terminal of the target value filter means 1 is
A subtraction means 11 is provided for subtracting the calculated target value Svo at the output end of the target value filter means 1 from
Discrimination means 13 is provided on the output side of , within a range that does not interfere with the two-degree-of-freedom function, and operates when the output of the subtraction means 11 becomes less than a predetermined value, turning on the switch means 12 to selectively take in the deviation e. Moreover, the output ΔS V of the subtraction means 11
= SV - S V o A subtraction output variable means 14 is provided for subtracting the deviation e taken in through the switch means 12 from o and obtaining a subtraction output δ-Δ5V-e that changes depending on the polarity and magnitude of the deviation e. Was it. Furthermore, after the absolute value 1δ1 of the subtraction output δ of the subtraction output variable means 14 is taken out by the absolute value means 15, a fuzzy coefficient setting means 16 is provided for outputting a membership function m(δ) corresponding to this absolute value 1δ1. Furthermore, in addition to the multiplication means 17 that multiplies the output δ of the subtraction means 11 by the membership function m(δ), the multiplication value obtained by the multiplication means 17 and the calculation target value S V .

This configuration is provided with an adding means 18 for adding the values to obtain the PI adjustment target value SVa.

Note that the addition means 18, fuzzy coefficient setting means 16, etc. may be realized by hardware using resistors, capacitors, or semiconductors, or addition and subtraction, absolute values, outputs of membership function m(δ), etc. may be realized by a computer. It may also be processed using software.

Therefore, according to the configuration of the embodiment as described above, for example, in response to a step-like change in the control target value Sv, the subtraction means
The difference Δ5v between the control target value SV and the calculated target value S V o
-sv-sv, the difference ΔSv is determined by the discriminating means 1
3. Send it to the subtraction output variable means 14 and the multiplication means 17. This determining means 13 determines whether the output of the subtracting means 11 has become small enough to be within a range that does not interfere with the two-degree-of-freedom function, that is, whether it has become less than a predetermined value. By turning on the switch means 12, the deviation e, which is the output of the deviation calculation means 3, is introduced into the subtraction output variable means 14 via the switch means 12.

In this subtraction output variable means 14, the output Δ of the subtraction means 11 is
Since the deviation e is subtracted from Sv, when the deviation is -0, the output ΔSv of the subtraction means 11 is output as is, while when the deviation is other than 0, it changes depending on the polarity. That is, when the deviation e has a negative polarity, the deviation e is added to the output ΔSv of the subtraction means 11, so that the subtraction output variable means 14 becomes large, and conversely, when it has a positive polarity, it becomes small.

Then, the absolute value 1δ1 of the subtraction output δ obtained by the subtraction output variable means 14 as described above is determined by the absolute value means 15, and then introduced into the fuzzy coefficient setting means 16.

In this fuzzy coefficient setting means 16, the subtraction output ΔS
When V-(SV-5Vo') becomes smaller than a value, e.g., δ2, which does not inhibit the original two-degree-of-freedom function, it becomes exponential, for example, from the value δ2 to the difference δ1 shown in the figure. A membership function m(δ) taking a value of 0-1 is output and sent to the multiplication means 17.

At this time, when the negative polarity deviation -e is added to the subtraction output variable means 14, the absolute value 1δ1 moves in the 62 direction, and conversely, when the positive polarity deviation e is added, the absolute value 1δ1 moves in the 6 direction. . As a result, the fuzzy coefficient setting means 16 outputs a membership function m(δ) that changes according to the response state of the target value filter means 1 and the followability of control, and is introduced into the multiplication means 17. This multiplication means 17 adds a membership function m to the subtraction output △Sv of the subtraction means 11.
After multiplying by (δ), this multiplied value δ・ΔSv is led to the addition means 18 and added to the calculation target value Sv0 to obtain the PI adjustment target value S V a = S V o 10 (δ・ΔSV) It is.

Next, another embodiment of the present invention will be described with reference to FIG. In this case as well, the same parts as in FIG. 5 are designated by the same reference numerals, and detailed explanation thereof will be omitted, and only the different parts will be described below.

This embodiment has a configuration in which the absolute value means 15 is removed from the apparatus shown in FIG. The membership function m(δ) is set separately. Of course, it goes without saying that the membership function m(δ) can be set to different values depending on whether it is positive or negative.

Next, the operation of the apparatus configured as above will be explained. In this device, the target value SVa for PI adjustment and the subtraction output δ of the subtraction output variable means 14 are S V a = S
V o + m (δ) l (Sv-5vo)...(
8) δ■Δ5V-e...(9)
It is expressed as

Therefore, when the control target value Sv changes in a stepwise manner in the above-mentioned state, the change in the PI adjustment target value SVa will be considered by dividing it into cases where there is no deviation and cases where there is a deviation.

(1) In the case of no deviation, that is, emo, the subtraction output δ of the subtraction output variable means 14 immediately after the control target value Sv changes in a stepwise manner.

In other words, the absolute value 1δ1 is a large value, and δ2≦1δI
...(lO), so the membership function is m(δ)-〇, and the above (
From equation 8), it becomes 5Va-5Vo. That is, SVa, which is determined only by the transfer function of the target value filter means 1, is introduced into the deviation calculation means 3 as a target value for PI adjustment. Therefore, in this case, the same response characteristics as the conventional device can be obtained as shown in FIG. 3(a).

Thereafter, the subtraction output δ obtained from the subtraction output variable means 14
becomes smaller than 62, and the relationship of δ□≦δ×δ2 (12) is established, the membership function from the fuzzy coefficient setting means 16 becomes 0×m(δ)×1, and the above (8) is satisfied.
From the formula, S V a = S V o + m (δ)
x (SV-5Vo)...(13) This SVa is sent to the deviation calculation means 3 as a target value for PI adjustment. Therefore, in this case, the difference δ is δ2
When the control target value S becomes smaller, the response characteristic becomes faster than the conventional response characteristic as shown in (b) in Fig. 3, and the control target value S becomes smaller.
approaching v.

Furthermore, the subtraction output δ obtained from the subtraction output variable means 14
becomes smaller and the relationship becomes 0≦δ×δ1 (14), the membership function becomes m(δ)−1
From the above equation (8), 5Va-SV
-(15), and the control target value SV is directly sent to the deviation calculation means 3 as the PI adjustment target value. As a result, at the time point tb when the PI adjustment target value reaches the difference δ1, the control target value S■ is forcibly set, so that it can be stabilized in a much shorter time than in the past.

(2) Regarding the case where there is a deviation As described above, it is assumed that in the case where there is no deviation, a response characteristic as shown in (c) of FIG. 4, for example, is obtained.

On the other hand, when e<0, that is, when the deviation e has negative polarity, the controlled variable Pv responds to the target value Sv in an overshoot manner, so in this state, △Sv becomes small and the 5Vo Bringing it to -8V will further promote overshoot and the hunting phenomenon. Therefore, based on the above formula (9), the subtraction output δ
By forcibly increasing the absolute value 1δ1 and shifting the absolute value 1δ1 in the 62 direction, the membership function causes m(δ) to decrease or become zero, and the response characteristics decrease as shown in (d) in Figure 4. The target value is set to the target value.

Next, when e>0, that is, when the deviation e is positive, the controlled variable Pv responds with a delay with respect to the target value Sv, so if △Sv becomes small in this state, sv
, -sv, there is no problem, so the structure is configured as shown in equation (9) above, and the timing for forcibly setting 5Vo=SV is accelerated in accordance with the size of the deviation e (Fig. 4). (e)).

In other words, by using the deviation e as a reference and forcibly shifting the timing of setting it to 5vo-sv in accordance with the positive or negative magnitude of the deviation e, both the response state of the target value filter means 1 and the follow-up state of the control can be checked. At the same time, we have carefully designed continuous windows and improved controllability.

Therefore, as shown in Fig. 3, this device exhibits the same response characteristics as the conventional device until the difference between the control target value and the calculated target value reaches a certain value in response to a step change in the target value, but the subtraction output is variable. When the subtraction output δ of the means 14 becomes less than δ2,
A membership function m(δ) between "0" and "1" is output from the fuzzy coefficient setting means 16, and the target value SV for PI adjustment is
Since a is varied, the response characteristics become faster as shown in FIG. 3 (b). Further, when the subtraction output ΔSV of the subtraction means 11 falls within a range that does not inhibit the function of creating two degrees of freedom, the fourth
As shown in the figure, since the response characteristics are shifted according to the polarity and magnitude of the deviation e, it is possible to quickly settle to the target value without causing overshoot or the like.

Moreover, the response characteristics are determined by fuzzy coefficient setting means 16.16'
It can be changed freely by setting. In other words, δ−Δ
If the membership function m(δ) of the fuzzy coefficient setting means 16.16' is set to an appropriate value in the region where the value of 5V-e is small and there is no influence on the two degrees of freedom, δ will be set to the predetermined value. If the value becomes below, the function of the target value filter means 1 will be forced to be faster than the original target value filter means 1 (5Va-8V, and P
It is possible to shorten the target value response settling time of the I adjustment device and improve controllability.

In the above embodiment, the multiplication value of the output of the subtraction means 11 and the membership function is supplied to the addition means 18, but for example, the multiplication means 17 may be omitted and a value corresponding to the multiplication value may be supplied to the fuzzy coefficient setting means 16. .16' and added to the adding means 18. Further, although the PI31 clause calculation has been described, it goes without saying that the same can be applied to the PID adjustment calculation. In addition, the present invention can be implemented with various modifications without departing from the gist thereof.

[Effects of the Invention] As explained above, according to the present invention, when the calculated target value of the target value filter means approaches the control target value, the calculation result depends on both the input/output difference and the control deviation of the target value filter means. Since the P1 adjustment target value is forcibly shifted using a changing membership function, the target value tracking speed can be increased while determining the control response situation without interfering with the two-degree-of-freedom function. As a result, the performance of the two-degree-of-freedom adjustment device can be greatly improved, and by distributing these functions throughout the plant, the plant operating characteristics can be revolutionized, and the fuzzy combination type two-degree-of-freedom adjustment device can greatly contribute to the industry. Can be provided.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the fuzzy combination type two-degree-of-freedom adjustment device according to the present invention, FIG. 2 is a block diagram showing another embodiment of the device of the present invention, and FIG. 3 is a conventional device. FIG. 4 is a diagram showing the response characteristics of the target value for PI adjustment that changes depending on the presence or absence of control deviation and the polarity. FIG.
The figure is a block diagram of a conventional device. DESCRIPTION OF SYMBOLS 1...Target value filter means, 2...Controlled object, 3.
... Deviation calculation means, 4 ... PI or PID adjustment means, 11 ... Subtraction means, 12 ... Switch means, 13
. . . Discrimination means, 14 . . . Subtraction output variable means, 15.
... Absolute value means, 16.16'... Fuzzy coefficient setting means, 17... Multiplication means, 18... Addition means.

Claims (1)

[Claims] Target value filter means is provided, and the deviation between the target value obtained from the target value filter means and the controlled amount from the controlled object is used to calculate PI or PID (P: proportional, I: integral, D
In a two-degree-of-freedom adjustment device that executes an adjustment calculation (differential) and applies the obtained operation signal to the controlled object, the control target value introduced into the target value filter means and the control target value obtained by the calculation of this target value filter means a subtracting means for extracting the difference from the calculated target value; a determining means for acquiring the deviation when the subtracted output from the subtracting means is equal to or less than a predetermined value; subtraction output variable means for varying the output of the subtraction means according to the polarity and magnitude of the deviation by subtracting the deviation; and a predetermined function corresponding to the output of the subtraction output variable means or the absolute value of the output. fuzzy coefficient setting means for outputting , and multiplication means for multiplying the output of the subtraction means by the function output from the fuzzy coefficient setting means, the multiplication output obtained from the multiplication means and the calculation of the target value filter means Fuzzy combination type two degrees of freedom characterized in that the target value for PI or PID adjustment is shifted according to the difference between the control target value and the calculation target value and the state of the deviation by adding and synthesizing the target value and the target value. Regulator.