JP3279745B2 - Controller with fuzzy inference function - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a controller having a fuzzy inference function used for various process controls in general industries, and particularly to a general PID (P: proportional, I: integral, D: integral).
The present invention relates to a controller having a fuzzy inference function in which fuzzy control technology is applied to differential control.

[0002]

2. Description of the Related Art In general, this type of process control employs a control method based on PID arithmetic control and feedback control combined with feedforward control. In practice, PID control parameters and feedforward control are used. An operation output is obtained by executing a PID operation using a word coefficient, and the operation output is applied to a control target, thereby controlling the temperature, pressure, flow rate, level or components of the plant.

[0003] By the way, various control objects are controlled by using such a control method. However, in the case of a process control aiming at complicated reaction control, a control which can be sufficiently satisfied only by using such a control method. Often results are not obtained.

Therefore, recently, a feedforward coefficient which is a main parameter of the process control system is inferred and determined by using fuzzy inference separately from the feedforward operation unit, and the feedforward coefficient determined here is used by using the determined feedforward coefficient. Attempts to control the plant have been made.

FIG. 3 is a block diagram showing a conventional control device with a fuzzy inference function, which is used by switching between a normal control method and a control method using fuzzy inference. That is, this control device takes in the process control variable 2 which is the detection output of the detector 1 into the PID adjustment calculating section 3 and uses the PID control parameter to make the deviation between the process control variable and the target value zero. After the PID adjustment calculation is performed to obtain the adjustment output 4, the adjustment output 4 is introduced into the multiplication means 5.

On the other hand, the feed-forward control system takes in a change in the process control variable 2 and other disturbance signals 7 to obtain a feed-forward coefficient (feed-forward compensation gain) 8,
Similarly, there is provided a fuzzy inference unit 11 for taking in a change in the process control variable 2 and other disturbance signals 7 and inferring a feedforward coefficient 10. It is selectively extracted by the means 12 and sent to the multiplication means 5.

In the multiplication means 5, the adjustment output 4 from the PID adjustment calculation section 3 is added to the calculation section 9 or the inference section 1.
After multiplying the feedforward coefficient obtained by 1 to obtain the operation output 13 with disturbance compensation,
This operation output 13 is applied to a control valve 14 which is an operation terminal,
The opening of the control valve 14 is adjusted.

[0008]

Therefore, the feedforward operation unit 9 and the fuzzy inference unit 1 as shown in FIG.
The method of switching and selecting 1 is, for example, a feed-forward operation unit 9 for backup of fuzzy inference.
Is used, the switching is performed from the fuzzy inference unit 11 to the feedforward operation unit 9 side. However, when the level change of both outputs is large, the switching timing cannot be determined, and the fuzzy inference cannot be used with security. There is. The reason is that the inference of fuzzy control is valid,
In other words, in order to evaluate the suitability and reliability of the fuzzy rule, it is necessary to perform verification over a long period of time using an actual plant, and application to an actual plant does not readily progress.

Therefore, after all, the fuzzy inference unit 11 is applied on a trial basis, the control result is constantly monitored, and the control result is applied if there is no problem in controllability. Switch to 9 side for use.

Therefore, in the above-described control device with a fuzzy inference function, although the feedforward operation unit 9 and the fuzzy inference unit 11 are provided side by side, these components 9 and 11 can be selectively selected appropriately. It is difficult to use, and there is a problem that the fuzzy inference unit 11 cannot be used continuously online.

The present invention has been made in view of the above circumstances, and the output of fuzzy inference can be used online, whereby the feedforward operation and fuzzy inference can be freely set at arbitrary time intervals and used. An object of the present invention is to provide a control device with a fuzzy inference function that can be used.

[0012]

Means for Solving the Problems To solve the above problems, the invention corresponding to claims 1 and 2 is based on PI or PID.
In adjusting the arithmetic and control unit that combines a feedback control and Ford-forward control by the feedforward calculation unit and the fuzzy inference unit for determining the feedforward coefficients from the process control variables and disturbance signals, pre
Feed forward based on a predetermined period
Switching command means for switching between control and fuzzy inference control
Provided on the output side of the feedforward operation unit.
From the switching command means and from the fuzzy inference control.
When a switch command to forward control is given,
Output signal of the fuzzy inference unit at the time of receiving the exchange signal.
The filter according to a ramp-like function with a predetermined slope
Outputs the signal that shifts to the output signal of the feedforward operation unit
When the transition is completed, the feedforward
First switching means for outputting an output signal;
Provided on the output side of the logic unit, and receives a feed from the switching command means.
Switching command from forward control to fuzzy inference control
When given, the feed at the time the switch signal is received
A predetermined gradient from the output signal of
Output signal of said fuzzy inference unit according to a ramp-like function
Output a signal to shift to
Second switching means for outputting an output signal of the inference unit;
The first switching means according to a switching signal from the switching command means;
The output signal of the stage is switched from the output signal of the second switching means.
Third switching means for switching, and an output signal of the third switching means.
Feedback control of the feedforward coefficient
And a multiplication unit for multiplying and compensating for the output of the control unit.

[0013] The switching command means is provided for the first cycle at an appropriate cycle.
And the second switching means can be selected to take out the output of the feedforward operation unit or the output of the fuzzy inference unit. It is preferable to change the selection time and select the second switching unit corresponding to the fuzzy inference unit.

[0014]

[0015]

According to the invention, the first switching means and the second switching means corresponding to the output side of the feedforward operation unit and the output side of the fuzzy inference unit are provided by taking the above means. The switching means is selected at an appropriate cycle predetermined by the switching instruction means.
And the second switching means selects and outputs the output of the arithmetic unit or the inference unit corresponding to itself at the time of selection of the own unit. At this time, the output of the other inference unit or the output of the arithmetic unit at the time of switching is selected. Take the output and send it to the other party's inference unit.
From the output of the
Arithmetic or inference units that correspond according to a loop-like function
The output can be switched smoothly without abrupt change at the time of selection, so that the fuzzy inference unit can be applied to an actual plant with confidence.

Moreover, if the periodic selection time is varied according to the degree of completion of the fuzzy rules of the fuzzy inference unit as in the invention according to claim 2, the fuzzy inference is gradually made without affecting the actual plant. It can be completed while evaluating the suitability of the fuzzy rules by the sections.

[0017]

[0018]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a control device having a feedforward coefficient intermittent correction type fuzzy inference function. Note that the same reference numerals in FIG. 3 denote the same parts as in FIG. In this control device, a detector 1, which is a detecting end for detecting a process control variable 2, executes a PI or PID adjustment operation based on a deviation between the process control variable 2 from the detector 1 and a target value, and executes an adjustment signal. 4 is provided with an adjustment operation unit 3 for obtaining the value 4 and a control valve 14 which is an operation terminal for controlling the flow rate and the like of the process line using the adjustment signal 4 obtained by the adjustment operation unit 3.
The configuration is the same as that of the conventional device.

What is particularly different in the present apparatus is that the feedforward control system has been improved. In this feedforward control system, the process control variable 2
Feed-forward operation unit 23 which takes in a disturbance signal 21 and a feed-forward coefficient 22
And a fuzzy inference unit 25 for taking in a change in the process control variable 2 and other disturbance signals 21 and inferring a feedforward coefficient 24.
And output means of the inference unit 25 are provided with switching means 26 and 27, respectively.

The switching means 26 is an output of the arithmetic unit 23.
Not only the output of the fuzzy inference unit 25
On the other hand, the switching means 27 is connected to the fuzzy inference unit 25.
Not only the output of the
I'm taking it in.

Reference numeral 28 denotes a switching means for selecting the switching means 26 or the switching means 27 and taking out the output of the corresponding arithmetic unit 23 or the output of the fuzzy inference unit 25. Fuzzy inference part 2
The output of 5 is sent to a multiplying means 30 as a feedforward coefficient 29 as a disturbance compensation signal. Here, the control output 4 of the PID adjustment calculation unit 3 is multiplied by the feedforward coefficient 29, and an operation output 31 on which disturbance compensation is performed is taken out. , Into the control valve 14.

Reference numeral 32 denotes a switching command device, which switches and selects the switching means 26 and the switching means 27 at a predetermined cycle, and controls the switching of the switching means 28 synchronously. 33 is a fuzzy inference evaluation means for storing a process control variable in a memory when receiving a signal indicating that the switching means 27 has been selected from the switching command device 32, or for evaluating an inference result using a predetermined fuzzy inference evaluation program. It is.

Reference numeral 32 denotes a switching command device, which switches and selects the switching means 26 and the switching means 27 at a predetermined cycle, and also controls the switching of the switching means 28 synchronously. 33 receives a signal indicating that the switching means 27 has been selected from the switching command device 32, stores the process control variables in a memory, or evaluates the inference result using a predetermined fuzzy inference program.
And a fuzzy inference evaluation means for judging the degree of completion of the fuzzy rule . That is, this fuzzy inference evaluation means 3
Reference numeral 3 denotes a state when the switching means 27 is selected, for example, when a change in the target value
Process control variables, so that the target value and its
Based on the difference (deviation e) from the stored process control variable
The evaluation is performed using the following well-known control characteristic evaluation indices.
Value. If this metric is below the expected value,
Judges that the degree of perfection is high,
By setting the selection time t
The control ratio by the inference unit is lengthened.

(Equation 1)

In this state, the switching command device 32 gives a command to continuously select the switching means 27 of the fuzzy inference unit for the time t, and simultaneously selects the switching means 28 on the switching means 27 side. To the effect.
Further, the switching command device 32 includes a fuzzy inference evaluation unit 33.
Similarly, a signal indicating selection is transmitted to the switching means 27 side.

The switching means 27 includes a feedforward
Since the output of the operation unit 23 is also input,
Is the feedforward operation unit 23 at the time of switching.
A signal corresponding to the output is output, and thereafter, a fuzzy inference unit 25 is gradually used, for example, by using a ramp function or other reducing elements.
Process variable 2 which is the output of the other disturbance signal 21
While shifting to the feedforward coefficient 24 for
This is to compensate for a change in disturbance. This
And the output of the feedforward operation unit 23 and the fuzzy
Even if there is a difference in the output of the logical unit 25, it gradually shifts
Switching without any sudden change
Introducing the output of the fuzzy inference unit 25 to the multiplication means 30
Can be.

At this time, the switching means 26 is in a non-selected state.
, But the fuzzy inference unit 25 on the other side
It takes in the force and prepares for the next switching selection.

When the time t elapses, the output of the switching command device 32 is turned off and returns to the output selection path of the feedforward operation unit 23. During the remaining time (Tt), the feedforward operation unit 23 turns off. A compensation operation for a change in disturbance is performed using the obtained feedforward coefficient 22. Also at this time, the output of the feedforward operation unit 23 is gradually returned to the original output.

Therefore, at the time t, the control is performed by the almost 100% fuzzy inference unit 25, and the remaining time (Tt)
In this case, the control is almost completely performed by the feedforward operation unit 23.

Therefore, according to the apparatus of the embodiment as described above, when fuzzy control is applied, in order to check the suitability and reliability of the fuzzy rules, it is effective to verify using a real plant. For this purpose, an intermittent correction type fuzzy control method is adopted as in this device, and the conventional PI
Based on feedback control and feedforward control based on D adjustment calculation, control is executed while determining the feedforward coefficient by smoothly switching to fuzzy inference at arbitrary intervals, so that the fuzzy inference gradually becomes compatible and reliable. Can be applied while increasing This means that the fuzzy inference time t can be variably set according to the degree of completion of the fuzzy rules, and the feedforward control by fuzzy inference can be utilized in an actual plant with a sense of security.

It is needless to say that this embodiment is a process control and can be applied to any application in which fuzzy inference is applied to feedback control + feedforward control by PID. In addition, the fuzzy inference evaluation unit 33 captures and stores only the process control variables, but may freely capture other signals necessary for the evaluation. Also, the fuzzy inference evaluation means 33
Is provided in the PID adjustment calculation unit 3, but the fuzzy inference unit 2
5 may be provided. In addition, the present invention can be implemented with various modifications without departing from the scope of the invention.

[0031]

As described above, according to the present invention, the following various effects can be obtained. According to the first aspect of the present invention, when the output of the feedforward operation unit and the output of the fuzzy inference unit are alternately switched and output, the output can be switched smoothly without abrupt change, and the fuzzy inference can be safely performed in the actual plant. Applicable and fuzzy inference can be applied continuously online.

Next, according to the second aspect of the present invention, the fuzzy inference can be completed while gradually evaluating the fuzzy inference without affecting the actual plant.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing an embodiment of a control device with a fuzzy inference function according to the present invention.

FIG. 2 is a time chart illustrating the operation of the device of the present invention.

FIG. 3 is a configuration diagram of a control device with a fuzzy inference function in a conventional device.

[Explanation of symbols]

2 ... Process control variable, 3 ... Adjustment operation unit, 21 ... Disturbance signal, 22, 24 ... Feed forward coefficient, 23 ... Feed forward operation unit, 25 ... Fuzzy inference unit, 26-
28 switching means, 30 multiplication means, 32 switching command device, 33 fuzzy inference evaluation means.

Claims (2)

(57) [Claims] 1. A fee based on PI or PID adjustment calculation.
Combining feedback control and Ford forward control
Of the feed-forward function from process control variables and disturbance signals
Feed-forward operation and fuzzy inference
The department,Feed forward based on a predetermined cycle set in advance
Switch command means for switching between the flash control and the fuzzy inference control
When,  Provided on the output side of the feedforward operation unit,
From the switching command means to the feed format from fuzzy inference control.
When a switching command to the word control is given, the switching signal
From the output signal of the fuzzy inference unit at the time of receiving
According to a ramp-like function with a slope of
Outputs a signal that shifts to the output signal of the
When the line is completed, the output signal of the feedforward
First switching means for outputting a signal,The switching command provided on the output side of the fuzzy inference unit
Fuzzy inference control from means to feedforward control
When the switch command is given, when the switch signal is received
From the output signal of the feedforward operation unit
Fuzzy inference according to a ramp-like function with
Outputs a signal to shift to the output signal of the logical unit, and completes the shift.
Then, a second switch for outputting an output signal of the fuzzy inference unit is provided.
Exchange means, The first switching is performed according to a switching signal from the switching command means.
The output signal of the switching means and the output signal of the second switching means.
Third switching means for switching, Feedforward which is an output signal of the third switching means
Multiplies the feedback control output by the output coefficient to compensate
A multiplication unit; Control device with fuzzy inference function characterized by having
Place. 2. The switching command means according to claim 1, wherein a periodic selection time is variably set in accordance with a degree of completion of a fuzzy rule of the fuzzy inference unit, and the switching instruction means corresponds to the fuzzy inference unit based on the variably set selection time. The control device with a fuzzy inference function according to claim 1, wherein the second switching means is selected.