JP3518436B2 - Control device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a control device equipped with an auto-tuning function, and more particularly to a control device suitable for controlling a controlled object having a non-linear characteristic.

[0002]

2. Description of the Related Art Conventionally, as a control device of this type, FIG.
As shown in FIG. 1, a controller 3 that outputs a manipulated variable MV by performing PID calculation using a PID constant from the deviation between the measured value PV from the controlled object 1 and a predetermined target value SV, and the controlled object 1
Auto-tuning (abbreviated as AT in the present invention)
While observing the auto-tuning signal generator 5 that outputs a signal and the measured value PV obtained based on the auto-tuning signal applied from the auto-tuning signal generator 5 to the controlled object 1, the PID constant and the like are calculated. There has been known a configuration including a response observation constant calculation unit 7 for setting an output to the controller 3 and an operation output correction unit 9 for correcting the operation output from the controller 3 and outputting the corrected operation output to the control target 1.

Reference numeral 11 is an auto-tuning section comprising an auto-tuning signal generating section 5 and a response observation constant calculating section 7, and a reference numeral SW is an auto-tuning signal at the time of auto-tuning and an operation amount MV at the time of actual control.
This is a switch that switches between and.

In the control device having such a configuration, for example, when the limit cycle method is adopted as the auto-tuning method, an on-off signal is output from the auto-tuning signal generator 5 to the controlled object 1 and the response observation constant calculator 7 is used. When the PID constant, which is a control parameter, is calculated from the on / off signal, the target value SV, and the hunting waveform of the measured value PV and set in the controller 3, and the control target 1 is actually controlled, the controller 3 is set. Target value SV and measured value PV
The operation amount MV calculated by PID is output to the controlled object 1 via the operation output correction unit 9.

However, when the process gain (ΔPV / ΔMV) characteristic of the controlled object 1 is non-linear, the PID constant calculated by such an auto-tuning method is generally the target value S at which the auto-tuning is executed.
It was obtained from the process gain in the vicinity of the load factor with respect to V, and was insufficient to obtain good control performance for a wide range of load factors.

In view of this, the operation output correction unit 9 uses a so-called broken line approximation method that uses a correction function obtained by linearizing the load factors at a plurality of locations other than the partial load factors obtained by auto-tuning. The operation output from the was corrected so that good control performance could be obtained for a wide range of load factors. It should be noted that almost the same correction was necessary when using other auto tuning methods.

[0007]

However, in the above-described control device, the control parameters such as the PID constants are different depending on the load factor when the auto tuning is executed for the controlled object 1 having the non-linear process gain characteristic. Therefore, when the operation output correction unit 9 attempts to correct the characteristics of the operation amount MV output from the controller 3 by the broken line approximation method, it is necessary to recreate a correction function for each load factor for which auto tuning is performed. Occurs, and if it is attempted to correct the process gain characteristic approximately accurately, a large number of break points are required.

In this type of control device, since the number of break points that can be set is generally limited due to restrictions such as the memory area, it is difficult to control the controlled object 1 accurately by the broken line approximation method. Met.

Moreover, in order to control the controlled object 1 with high precision, it is necessary to make the difference in the inclination of the folding line before and after the folding point not so large, and in order to properly determine the limited number of folding points, It was necessary to repeat trial and error to set the operation output correction unit 9, which required a long time and skill.

The present invention has been made in order to solve such a problem. Even if the process gain characteristic of the object to be controlled is non-linear, it is possible to easily obtain good control performance for a wide range of load factors. The purpose is to provide a device.

[0011]

In order to solve such a problem, the present invention is such that the deviation between a target value and a measured value from a controlled object is calculated by PID based on a PID constant to reduce the deviation. A controller that outputs a manipulated variable and an auto-tuning signal that outputs a signal for auto-tuning to the control target, and calculates the PID constant related to the control target and the AT load factor during auto-tuning from the measured value and target value obtained from this Part and at least the load factor 0% to 100% as static characteristic data relating to the controlled object .
Setting unit that sets the relationship between the manipulated variable and measured value
The load factor 0% to 100% for the static characteristic data
Divided into a plurality sections at regular intervals to calculate the process gain characteristics in the respective section thereof from relationship between the optimum proportional gain set in controllers and their process gain characteristic ward
The proportional gain characteristic for the load factor between
A controller gain ratio characteristic calculation unit that calculates a proportional gain ratio characteristic that represents the ratio of the proportional gain characteristic to the minimum gain, and a proportional gain obtained by auto tuning based on the proportional gain ratio characteristic and the AT load factor. A constant conversion unit for converting the gain to output and setting the output to the controller.

Further, according to the present invention, the number of break points for correcting the operation amount from the controller by the broken line is set in the setting section, and based on the maximum gain ratio from the proportional gain ratio characteristic, the controller Calculate the slope ratio that reduces the slope difference between each of the polygonal lines when the operation amount is divided into polygonal lines by the number of broken points,
It is preferable to provide a correction function calculation unit that creates a correction function table that can obtain this slope ratio, and an operation output processing unit that corrects the operation output from the controller to match the correction function table.

Further, according to the present invention, it is possible to form the correction function calculating section so as to calculate the same inclination ratio between the polygonal lines and create the correction function table.

[0014]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. The same parts as those in the conventional example are designated by the same reference numerals.

FIG. 1 is a block diagram showing an embodiment of a control device according to the present invention. In FIG. 1, the controller 3 is
A predetermined PID constant is used, PID calculation is performed from the deviation between the target value SV and the measured value PV from the controlled object 1 so that the deviation becomes smaller, and the manipulated variable MV is output. It is connected to the gain adjusting unit 15.

PID used for PID calculation by this controller 3
The constant is set via the constant conversion unit 17 described later, but the details will be described later.

The operation output processing section 13 is composed of a gain adjusting section 15 and a correcting section 19 and is connected to the controlled object 1 via a switch SW, which will be described in detail later.

The auto tuning unit 11 has a switch S.
While observing an auto-tuning signal generator 5 that outputs an auto-tuning signal, for example, an on-off signal to the controlled object 1 via W, and a measured value PV obtained based on this auto-tuning signal, the auto-tuning signal and the target value SV to PID constant (P: proportional band,
(I: integration time, D: derivative time), which is a conventionally known one formed from a response observation constant calculation unit 7 that calculates the AT load factor during auto-tuning. It is connected.

The setting unit 21 is formed to have a static characteristic data setting unit 23 and a break point number setting unit 25. The static characteristic data setting unit 23 stores and stores the static characteristic data as shown in FIG. 2 for the controlled object 1 having the non-linear process gain characteristic, and is connected to the controller gain ratio characteristic calculating unit 27. There is.

Generally, when the input value to a certain system is constant, the state in which the output value becomes a steady state and becomes constant is called the equilibrium state. In the static characteristic data of the present invention, the manipulated variable to the controlled object 1 is It is a relationship between the manipulated variable and the measured value when the response of the controlled object 1 is in a balanced state while being kept constant, and there are several points with respect to the manipulated variable (load factor 0% to 100%) (Fig. 2). Is data obtained by measuring about 7 points).

The folding point number setting unit 25 includes an operation output processing unit 1.
The number of break points used in No. 3 is set and stored, and is connected to the correction function calculation unit 29. In this embodiment, five points will be described.

Input setting of static characteristic data to the static characteristic data setting section 23 and input setting of the number of folding points to the number of folding points setting section 25 are manually performed by a keyboard or the like arranged on an operation panel (not shown).

The controller gain ratio characteristic calculation unit 27
As shown in FIG. 3, the seven points of data set in the static characteristic data setting unit 23 are sequentially connected by a broken line to linearly approximate the static characteristic data to interpolate, and the load factor is 0% to 100%.
% Is divided into a plurality of sections at equal intervals, and the rate of change of the measured value (control amount) of the unit load rate (θ) in each of the sections with respect to the polygonal line, that is, the process gain characteristic per unit load rate (ΔPV / ΔMV) Has a function of calculating as shown in FIG. 4, and is connected to the constant conversion unit 17 and the correction function calculation unit 29. Controller gain ratio characteristic calculation unit 27
Then, it is also possible to use curve approximation such as parabolic approximation.

Since the process gain PG of the controlled object 1 and the proportional gain Kp of the controller 3 defined by the reciprocal of the proportional band P have the following relationship, the controller gain ratio characteristic calculation unit 27 It has a function of obtaining the proportional gain characteristic with respect to the load factor as shown in FIG. 5 by using the following equation and also obtaining the ratio Kp / Kpmin of the proportional gain characteristic to the minimum gain, that is, the proportional gain ratio characteristic, as shown in FIG. is doing.

Kp = a / PG Here, the code | symbol a is a coefficient.

The constant conversion unit 17 uses the proportional gain ratio characteristic from the controller gain ratio characteristic calculation unit 27 and the AT load factor output from the response observation constant calculation unit 7 of the automatic tuning unit 11 in the automatic tuning unit 11. The obtained proportional gain is corrected so as to have a minimum gain, and is connected to the controller 3.

Generally, when the static characteristic of the controlled object 1 is non-linear, the optimum proportional gain differs for each load factor, but the maximum proportional gain that does not cause an unstable state such as a vibrational response in the entire load factor range is The constant conversion unit 17 is determined to be one and becomes the minimum value of the proportional gain with respect to the load factor having the highest gain of the process gain characteristic of FIG. 4, that is, the proportional gain Kp of the proportional gain characteristic of FIG. 1
The controller gain ratio characteristic calculation unit 27 so that the proportional gain obtained in 1 becomes the minimum proportional gain of the proportional gain characteristic.
It has a function of performing correction processing with the proportional gain ratio characteristic from.

That is, as shown in FIG. 6, when the proportional gain ratio with respect to the AT load factor is R, the constant converting section 17 can obtain the value of the gain set in the controller 3 by the following equation. Therefore, this is used to correct the proportional gain. Proportional gain of controller = AT calculation proportional gain / R

The correction function calculation unit 29 corrects the operation output from the controller 3 based on the proportional gain ratio characteristic from the controller gain ratio characteristic calculation unit 27 and the number of break points from the number of break point setting unit 25. It has a function of obtaining a broken line table and is connected to the operation output processing unit 13.

That is, the correction function calculating section 29 inputs the proportional gain ratio characteristic and the number of break points (5 points in this example), and n (5) is set between 0 and 100% of the operation output MV of the controller 3. Divide by n + 1 (5 + 1) main fold lines created by individual fold points.

In general, when the characteristic is corrected by a broken line, if there is a large difference in inclination between the broken lines sandwiching the broken point, the correction ratio may largely change in the vicinity of the broken point and the controllability may be greatly impaired.

Therefore, the correction function calculation unit 29 has a function of minimizing the above-mentioned problem by obtaining the slope ratio Sr from the following equation so that the slope ratios between all the break points are equal. is doing.

Sr = (Kp'max) 1 / n + 1 where the symbol Kp'max is the maximum gain of the proportional gain ratio characteristic.

The inclination ratio between the folding lines sandwiching all the folding points is S
As shown in FIG. 7, if the slope of the bent line with the gentlest slope is Q, then the slope of the next loosest line is Q.
× Sr, followed by Q × Sr2, ..., Q × Sr5.

When the operation amount of the controller is corrected by the correction polygonal line as shown in FIG. 7, the proportional gain set in the controller is set to the minimum gain of the proportional gain characteristic shown in FIG. A product obtained by multiplying the size of the inclination of each broken line by the proportional gain of the controller corresponds to the proportional gain of the controller in each broken line section + the total gain of the correction unit 19.

For example, when the load factor of the controller 3 is in the region of the polygonal line with the gentlest slope in FIG. 7, if the proportional gain set in the controller is Kp, the controller 3 + correction unit 1
The total gain of 9 corresponds to Q × Kp.

Now, the value of the proportional gain set in the controller 3 is set as the minimum value of the proportional gain characteristic, and the operation amount from the controller 3 is corrected by the correction polygonal line set in the correction section 19 of the operation output processing section 13. The controller 3 for each load factor of the controller
Since the total gain of the + correction unit 19 is made to be like the proportional gain characteristic of FIG. 5, the proportional gain ratio characteristic shown in FIG. 6 can be considered as a function of the slope of the correction polygonal line with respect to each load factor.

From this, from the proportional gain ratio characteristic as shown in FIG. 8, the value of the load factor corresponding to the gain corresponding to the inclination of each broken line is set in the correction unit 19 on the actual output side of the broken line on the controller. It can be the coordinates (the output side of the correction polygonal line).

That is, the product of the gain of Sr5 in FIG. 8 multiplied by the proportional gain set in the controller is proportional gain + correction in the controller actual output range from 0% to MVO (1)% of the correction polygonal line. The total gain of the part 19 is obtained, and similarly, the product of the gain of Sr4 and the proportional gain set in the controller is MVO (1).
% To MVO (2)%, which is the proportional gain in the controller actual output range + the total gain of the correction unit 19.
The same applies to Sr3, Sr2, and Sr.

Then, the correction function calculation unit 29 determines that the input / output range of the operation output processing unit 13 is both 0% to 100%, that the slope ratio between the folding lines that sandwich the folding point is Sr, and that the controller actual Using the formula shown in FIG. 10 derived from the fact that the coordinates on the output value side are given, the correction unit 1
It has a function of calculating the control calculation value side (input side) coordinates of the correction polygonal line set in 9, and also has a function of outputting and setting the correction function polygonal line thus obtained to the correction unit 19. Therefore, the correction function calculation unit 29 sets the correction polygonal line shown in FIG. 9 in the correction unit 19.

By the way, regarding the correction polygonal line set in the correction unit 19, when the inclination of the polygonal line with the gentlest inclination is "1", the inclination of the next gentler polygonal line is Sr, and the next is Sr.
2, ..., Sr5, so that proportional gain + correction unit 1
Although the total gain of 9 is equal to the proportional gain characteristic shown in FIG. 5, the slope of the polygonal line with the gentlest slope is actually smaller than “1”, and as a result, the slope of all the polygonal lines is decreased. The total gain will be reduced by the amount of the slope of the softest polygonal line.

Therefore, the correction function calculation unit 29 also has a function of adjusting the gain reduction due to the above-mentioned cause. In other words, the correction function calculation unit 29 outputs the adjustment value obtained by the following equation to the gain adjustment unit 15 to adjust the total gain.

Gain adjustment coefficient = 1 / Q where Q is the size of the slope of the polygonal line with the gentlest slope.

The gain adjusting section 15 of the operation output processing section 13
Is for adjusting the gain of the operation amount from the controller 3 based on the correction coefficient from the correction function calculation unit 29, and is connected to the correction unit 19.

The correction unit 19 refers to the output correction polygonal line table from the correction function calculation unit 29, and corrects the gain-adjusted operation amount so that an output according to the output correction polygonal line table is obtained. And is connected to the controlled object 1 side via the switch SW.

Next, the operation of the control device according to the present invention will be briefly described. When the auto-tuning signal generation unit 5 of the auto-tuning unit 11 outputs an on / off signal to the control target 1 via the switch SW, the response observation constant calculation unit 7 generates an on / off signal, a target value SV, and an auto tuning signal. A PID constant and an AT load factor at the time of auto-tuning are calculated from the obtained measured value PV and output to the constant conversion unit 17.

The controller gain ratio characteristic calculation unit 27
The seven points of data set in the static characteristic data section 23 are sequentially connected by a broken line to linearly approximate the static characteristic data and interpolate, and the load factor is divided into a plurality of sections at equal intervals from 0% to 100%. The measured value (control amount) change rate of the unit load rate (θ) in each of the plurality of sections, that is, the process gain characteristic (ΔPV / ΔMV) as shown in FIG. 4 is calculated.

The controller gain ratio characteristic calculation unit 27
The proportional gain characteristic with respect to the load factor as shown in FIG. 5 is obtained from the relational expression between the process gain and the proportional gain, and the ratio of the proportional gain characteristic to the minimum gain Kp /
Kpmin, that is, the proportional gain ratio characteristic is obtained as shown in FIG. 6, and is output to the constant conversion unit 17.

The constant conversion unit 17 corrects the proportional gain defined by the reciprocal of the proportional band obtained by the auto-tuning unit 11 based on the proportional gain ratio characteristic and the AT load factor output from the response observation constant calculation unit 7. The proportional gain is corrected and set in the controller 3 so that the proportional gain Kp set in the controller 3 becomes the minimum gain Kpmin of the proportional gain characteristic regardless of which load factor the auto tuning is performed.

Further, the correction function calculating section 29 determines the operation output MV of the controller 3 from 0 to 0 based on the proportional gain ratio characteristic and the number of break points.
The 100% interval is divided by 6 broken lines formed by 5 broken points, and the tilt ratio Sr is calculated so that the tilt ratios between adjacent bent lines are equal.

The correction function calculator 29 determines the slope ratio S between the polygonal lines.
From r and the proportional gain ratio characteristic, the coordinates of the output correction broken line on the controller actual output side are obtained as shown in FIG. 8, and the output correction broken line table is obtained based on the controller actual output side coordinates and the slope ratio Sr between each broken line. The correction unit 19 is set. Further, the correction function calculation unit 29 calculates a correction coefficient for adjusting the total gain and sets it in the gain adjustment unit 15.

Then, the controller 3 at the time of control calculates the manipulated variable MV by the PID calculation of the deviation between the target value SV and the measured value PV on the basis of the PID constant set by the constant conversion unit 17, and outputs the manipulated variable MV. Output to.

In the operation output processing unit 13, the gain adjusting unit 1
5 adjusts the operation amount by gain, and the correction unit 19 refers to the output correction broken line table to correct the operation amount. Control target 1
Is controlled by this corrected manipulated variable.

In such a control device of the present invention, the target value SV and the deviation of the measured value PV from the controlled object 1 are PID-calculated based on the PID constant, and the manipulated variable MV is output from the controller 3 for automatic tuning. The unit 11 outputs the signal for auto-tuning to the controlled object 1 and the measured value PV and the target value SV from the controlled object 1 obtained by this.
The PID constant related to the controlled object 1 and the AT load rate at the time of auto tuning are calculated from this, and as the static characteristic data related to the controlled object 1, the operation amount at the load rate of 0% to 100%
And the measured value are set in the setting unit 21 at a plurality of points, and the number of break points for correcting the operation amount from the controller 3 is also set in the setting unit 21, and the controller gain ratio is set. Characteristic calculation unit 2
7, the load factor of the static characteristic data is 0% to 10%.
It is divided into a plurality sections 0% at regular intervals to calculate the flop <br/> Rosesugein characteristics in each such interval, then the relational expression best proportional gain set in the controllers and their process gain characteristics
A proportional gain characteristic with respect to the load factor of each section is obtained, a proportional gain ratio characteristic representing the ratio of the proportional gain characteristic to the minimum gain is calculated, and the constant conversion unit 17 automatically calculates the proportional gain ratio characteristic and the AT load factor. The proportional gain obtained by the tuning is converted to the minimum gain, output is set to the controller 3, and the correction function calculator 29 determines the operation amount from the controller based on the maximum gain ratio of the proportional gain ratio characteristic. When a polygonal line is formed by the number of folding points, a slope ratio is calculated by making the slopes equal to each other, and a corrected polygonal line table is created so that the slope ratio can be obtained. The operation output of is adjusted to match the correction polygonal line table.

Therefore, no matter which load factor is used for the controlled object 1 whose process gain characteristic is non-linear, the auto-tuning result which is the minimum value of the proportional gain of the proportional gain characteristic shown in FIG. 5 can be obtained. It is possible to prevent an unstable control response such as a vibration response over the entire load factor range.

Further, since the operation amount is corrected by the correction polygonal line table created by calculating the same inclination ratio between the polygonal lines, the controllability due to the difference in inclination before and after the folding point can be obtained even with a limited number of folding points. It is possible to minimize deterioration and obtain good control characteristics in the entire load factor range.

[0057]

As described above, the control device of the present invention has a controller for PID-calculating a deviation between a target value and a measured value to output a manipulated variable that reduces the deviation, and an automatic control target. The auto tuning unit that tunes to calculate the PID constant and the AT load factor during auto tuning, and the load factor of 0 as static characteristic data regarding the controlled object.
% To 100%, the setting unit for setting the relationship between the manipulated variable and the measured value at a plurality of points, and the load factor for the static characteristic data.
Divide 0% to 100% into multiple sections at equal intervals and
The process gain characteristics are calculated, and the proportional gain characteristics for the load factor of each section are calculated from the relational expression between those process gain characteristics and the optimum proportional gain set in the controller, and the ratio of this proportional gain characteristic to the minimum gain is calculated. The controller gain ratio characteristic calculation unit that calculates the proportional gain ratio characteristic, and the proportional gain obtained by auto tuning based on the proportional gain ratio characteristic and the AT load factor are converted into the minimum gain of the proportional gain characteristic and output to the controller. Since there is a constant conversion unit to set, no vibration response occurs in the entire load factor range, no matter which load factor is used for auto-tuning a control target whose process gain characteristic is non-linear. Therefore, good control characteristics can be obtained over a wide range of load factors. Then, set the number of break points when correcting the operation amount from the controller in the above setting section, and based on the maximum gain ratio from the proportional gain ratio characteristic, the operation amount from the controller is broken line at those break points. When calculating the same slope ratio between the polygonal lines, a correction function calculation unit that creates a correction function table that obtains this slope ratio, and the operation output from the controller are corrected. With the configuration that includes the operation output processing unit that performs correction to match the function table, the operation amount is corrected by the limited number of break points against the constant determined by auto-tuning, and it is suitable for the process gain characteristic of the controlled object and A control characteristic more suitable for the load factor of is obtained.

[Brief description of drawings]

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

FIG. 2 is a diagram showing static characteristic data used in the control device of FIG.

FIG. 3 is a diagram illustrating a state in which the static characteristic data of FIG. 2 is approximated to a broken line.

FIG. 4 is a diagram showing a process gain characteristic per unit load factor obtained by a controller gain ratio characteristic calculation unit in FIG.

5 is a diagram showing a proportional gain characteristic calculated by a controller gain ratio characteristic calculation unit in FIG. 1. FIG.

FIG. 6 is a diagram showing a proportional gain ratio characteristic calculated by a controller gain ratio characteristic calculation unit in FIG. 1.

FIG. 7 is a diagram for explaining the operation of the correction function calculation unit in FIG.

FIG. 8 is a diagram illustrating an operation of a correction function calculation unit in FIG.

9 is a diagram illustrating an operation of the correction function calculation unit in FIG.

10 is a diagram showing an arithmetic expression executed by a correction function calculation unit in FIG.

FIG. 11 is a block diagram showing a conventional control device.

[Explanation of symbols]

1 controlled object 3 controller 5 Auto tuning signal generator (AT signal generator) 7 Response observation constant calculator 9 Operation output correction unit 11 Auto tuning section (AT section) 13 Operation output processing unit 15 Gain control section 17 Constant conversion unit 19 Corrector 21 setting section 23 Static characteristic data setting section 25 Folding point number setting section 27 Controller Gain Ratio Characteristic Calculation Unit 29 Correction function calculation unit SW switch

─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Noriaki Miyashita 5-16-6 Kugahara, Ota-ku, Tokyo Rika Kogyo Co., Ltd. (56) Reference JP-A-9-311703 (JP, A) JP Japanese Unexamined Patent Publication No. 8-161004 (JP, A) Japanese Unexamined Patent Publication No. 8-286707 (JP, A) Japanese Unexamined Patent Publication No. 58-169203 (JP, A) Japanese Unexamined Patent Publication No. 6-161509 (JP, A) (58) Fields investigated (Int .Cl. ⁷ , DB name) G05B 13/02 G05B 11/36

Claims (2)

(57) [Claims] 1. A controller for calculating a deviation between a target value and a measured value from a controlled object by a PID based on a PID constant and outputting a manipulated variable that reduces the deviation, and a signal for auto-tuning to the controlled object. And an auto-tuning unit that calculates the PID constant for the control target and the AT load factor during auto-tuning from the measured value and the target value obtained from the control target and a static value related to at least the control target. Negative as characteristic data
The relationship between the manipulated variable and the measured value when the load factor is 0% to 100% is
A setting unit for setting several points, and the load factor 0% to 100% relating to the static characteristic data.
Divide into multiple sections at equal intervals to calculate the process gain characteristic in each section, and calculate the process gain characteristic in each section from the relational expression between the process gain characteristic and the optimum proportional gain set in the controller .
A controller gain ratio characteristic calculation unit that calculates a proportional gain characteristic with respect to a load factor and calculates a proportional gain ratio characteristic that represents a ratio of the proportional gain characteristic with respect to a minimum gain; and, based on the proportional gain ratio characteristic and the AT load factor, A control unit comprising: a constant conversion unit that converts a proportional gain obtained by tuning into a minimum gain and sets an output to the controller. 2. The setting unit is set with the number of break points for correcting the operation amount from the controller by a broken line. Based on the maximum gain ratio from the proportional gain ratio characteristic, the operation amount from the controller is set. A correction function calculation unit that calculates an equal slope ratio between each of the polygonal lines when the line is polygonalized at the number of folding points, and creates a correction function table such that the slope ratio is obtained, and an operation output from the controller. The control device according to claim 1, further comprising: an operation output processing unit that corrects the value according to the correction function table.