JPH0519725B2 - - Google Patents

Description

[Detailed Description of the Invention] [Field to which the invention pertains] This invention relates to a sampled value PID control device that controls a controlled process using sampled values, and in particular identifies dynamic characteristics of the process during closed loop control and uses the identified results to identify the dynamic characteristics of the process. In a system that has the function of automatically optimizing control constants, it is possible to improve the control system, automatically adjust the control constants accordingly, optimize the control constants in response to changes in the dynamic characteristics of the process during operation, and improve the control system. Sample values from which you can select a response
Regarding PID control device.

[Prior art and its problems]

One of the inventors of the present invention invented a sample value PID control device that has the function of identifying the dynamic characteristics of a process during closed-loop control and automatically adjusting the PID control constants optimally based on the identification results. There is a technique described in JP-A-57-23108).

FIG. 1 is a block diagram showing the configuration of this sample value PID control device. In this sample value PID control device, a persistent exciting identification signal is generated during closed loop control, and process 1
The pulse transfer function of is identified, and from the identification result,
Calculate the transfer function of the S region of process 1, calculate the sample value PID control constant so that the characteristics from the target value to the process output match the optimal model characteristics, and control using the calculation results. I was getting used to it.

In other words, this device is a sample value PID control device that aims to respond quickly to changes in the target value, so it has good followability to the target value, but the main pole of the process and the main zero point of the sample value PID control device are almost the same. Since these disturbances may cancel each other out, it has become necessary to have the ability to suppress the process output against disturbances.

In other words, both the quick response of the process output to the target value and the ability to suppress the process output against disturbances,
There was a need to obtain a sample value PID control system with excellent controllability.

Furthermore, in process control, various types of control are carried out, such as controlling the process output so that it does not overshoot when the target value is changed, or allowing it to overshoot, depending on the object to be controlled.

However, with this sample value PID controller, the response of the process output to a change in the target value is approximately 10% no matter what control constant is adjusted.
It was fixed so that it would overshoot. Therefore, there was a need for something that fully satisfies the on-site requirements for process control.

Furthermore, the dynamic characteristics of a process in operation may change due to changes in operating conditions, changes in the composition of raw materials, or aging of the plant. Naturally, if this happens, it is necessary to reset the optimal control constants. Currently, expensive plant monitoring equipment such as recorders and television monitors are used, and operators constantly monitor changes to monitor and respond to changes in the plant.

Therefore, there is a need for a sample value PID control device that can monitor the plant and re-optimize control constants.

[Purpose of the invention]

The purpose of the present invention is to solve the above-mentioned conventional sample value
Improving the controllability of PID control devices, in other words, it is possible to select the response shape of the process output to changes in the control system's target value, in other words, it is possible to decide the control constants so as not to overshoot the control system, or to arbitrarily select the amount of overshoot. It is an object of the present invention to provide a sample value PID control device whose control constants can be determined.

[Summary of the invention]

The sample value PID control device of the present invention calculates a control deviation from a process signal obtained by sampling a target value signal of a process to be controlled and a control amount of the process, and performs a PID control operation on this control deviation signal to control the process. a sample value PID control calculation unit that outputs an operation signal; a pulse transfer function identification unit that identifies parameters of the process from the operation signal and the process signal; and a pulse transfer function of the process obtained by the pulse transfer function identification unit. A transfer function calculation unit that calculates a transfer function in the S (Laplace operator) domain from
It has a sample value control constant calculation part that calculates PID control constants, and in particular, the reference model used in the calculation of the sample value control constant calculation part is unified into a set of shape coefficients of a binomial model and a model with less damping. It is characterized by having means for determining PID control constants generated by varying coefficients so that the characteristics from the control target value to the control amount match this reference model.

〔Effect of the invention〕

The characteristics of the process in operation can be identified by time-series calculations, and the control constants of the sample value PID control system can be automatically adjusted from the identification results.

Further, the control system is adjusted to have almost the same characteristics as the designed reference model. Therefore, by simply specifying (inputting) one parameter, the shape factor α, it is possible to easily realize the response characteristic specifications intended by the designer.

In particular, in the present invention, the user can select the response shape of the controlled variable to a step change in the target value of the process so that control can be performed in accordance with the operating specifications of the process. Moreover, the sample value PID in that case
Control constants related to control can be automatically adjusted. That is, control specifications regarding stability and quick response intended by the user can be easily reflected in the controller, and the adjustment work of control constants related to sample value PID control itself can be made more efficient. Therefore, not only can the manpower and time required to adjust the sample value PID control system, as well as the plant operation loss, be greatly reduced, but it is also practical because it provides a means of adjusting the control system to match the process operation specifications. The effect is great.

[Example of the present invention]

EMBODIMENT OF THE INVENTION Below, one Example of this invention will be described in detail using drawings. FIG. 2 is a block diagram showing a circuit configuration of a sample value PID control system according to the present invention. The controlled object is a process 1 configured to control the temperature, pressure, flow rate, liquid level, etc. of the plant, and the process input u(t) sample-held by the sample-holding section 2 is the input, and y(t ) is the output of process 1.

This controlled object inputs a sampler 41 that operates the process output y(t) at a predetermined sampling period τ, and inputs the sampled process output y ^* (N) outputted from this sampler 41 as a feedback signal.
Through a sampler 42 that synchronizes the target value r(t) with the same sampling period τ, the sampled target value signal r ^* (N) is input to generate the output r ^* ₀ (l) of the target value filter 12. , this r ^* ₀ (N) and said process output y ^*
Deviation calculation unit 3 that calculates the control deviation e ^* (N) from (N)
Based on the control deviation e ^* (N), the sample value PID control unit 5 that controls the process 1 calculates and outputs the operation signal u ^* (l) of the process. It has come to constitute a PID control system. Here, N represents discrete time and t represents real time.

On the other hand, the dynamic characteristics of the process 1 are identified, and the PID of the sample value control calculation unit 5 is determined based on the result.
The identification/tuning function for adjusting the control constant and the filter constant of the target value filter 12 is performed by the identification signal generation section 7, pulse transfer function identification section 9, transfer function calculation section 10, and sample value control constant calculation section 1 shown in the figure.
1 and a filter constant calculation unit 13,
It is connected to the basic sample value PID control system as shown in the figure. To identify the dynamic characteristics of the process 1, a persistent signal is sent from the identification signal generator 7.
Generating an exciting identification signal V ^* (N),
The sample is injected into the process 1 via the sample hold section 2.

As a result, the condition for identifying the closed-loop system is satisfied, so input the data {u ^* (N), y ^* (N), N=1, 2,...} and use the pulse transfer function identification section 9 to The pulse transfer function G _p (Z ⁻¹ ) of process 1 is identified by time series calculation. Next, the transfer function calculation unit 10
From the obtained pulse transfer function G _p (Z ⁻¹ ), low frequency parameters g ₀ , g ₁ , g ₂ , g ₃ of the transfer function G(S) in the S (Laplace operator) domain are calculated. Here g ₀ , g ₁ ,
g ₂ and g ₃ are parameters expressed by the following equation, which represents the transfer function G(S) as a denominator column.

G(S)=1/g ₀ +g ₁ S+g ₂ S ² +g ₃ S ³ ...Equation 1 The proportional gain Kc, integral time constant Ti, and differential time constant Td of the sample value control constants used in the sample value control calculation section 5 are , the low frequency parameters g ₀ , g ₁ , g ₂ ,
Using g ₃ and the sampling period τ, the sample value control constant calculation unit 11 calculates the suppression force against disturbance to the optimum value. In addition, the filter constant calculation unit 1 sets the filter constant of the target value filter 12 from the sample value control constants Ti and Td and the sampling period τ so that the quick response to the target value change is optimal.
It is designed to be calculated using 3.

In this way, process 1 during closed-loop control
Identify the dynamic characteristics of, and based on the identification results,
This provides a dynamic adjustment function for the control constants of the sampled value PID control system that has a strong ability to suppress disturbances and also has excellent quick response to target values.

Further, the sample value PID control constant calculation unit 11
The reference model for control system design used ⁱⁿ
By changing the setting of this shape factor α ^* , it is possible to adjust the amount of overshoot of the response plate shape in response to a step change in the target value of the control system. Moreover, the sample value control constant and the filter constant can be automatically changed in accordance with the adjustment of the shape factor α ^* .

Next, an adaptive function will be described which monitors changes in the dynamic properties of the process 1 during operation and triggers the identification and tuning function in order to readjust the control constants if the changes in the dynamic properties become unacceptable.

The adaptive function consists of an identification end determination section 14, a characteristic change detection section 15, and a control section 16 shown in the figure.
As shown in the figure, it is connected to the identification tuning function section and sample value PID control system.

First, as the identification of the identification/tuning function progresses, the values of the low frequency parameters g ₀ , g ₁ , g ₂ , and g ₃ calculated by the transfer function calculation unit 10 converge. Therefore, the identification end determination section 14 performs a parameter convergence determination as shown in the following equation and outputs an identification end signal.

|g ^* _ii (N)−g ^* _ii (N-1)|≦CLASS(i=1-3)...Second formula Here, CLASS is set to a value such as 0.001, 0.005, 0.01, etc.

When the identification end condition of this second equation is satisfied,
The identification end determination section 14 inputs an identification end signal to the control section 16, and the control section 16 stops the operations of the identification signal generation section 7, the pulse transfer function identification section 9, and the transfer function calculation section 10 of the identification/tuning function. It's becoming like that.

Further, the control unit 16 inputs the values of Kc, Ti, and Td calculated by the sample value control constant calculation unit 11 when the identification/tuning function is stopped to the sample value control calculation unit 5, The filter constant calculated in step 1 is input to the target value filter to complete the adjustment of the sample value PID control device.

Further, upon receiving the identification end signal, the control section 16 causes the characteristic change detection section 15 to input the process parameters calculated by the pulse transfer function identification section 9 at the time of the end of the identification, and activates the characteristic change detection section 15. The characteristic change detection unit 15 uses the input process parameters, the process operation signal u ^* (N), and the process signal y ^* (N) to calculate the characteristic change after the end of identification of the process 1 at every sampling period τ. , the results are output to the control section 16.

Here, as a property of the characteristic change signal calculated by the characteristic change detection section 15, a DC component appears when the dynamic characteristics of the process 1 change or when a disturbance is added to the process 1. Moreover, if the input signal of process 1 is changed while this DC component appears, the DC component will not change in the case of disturbance, but
If the dynamic characteristics of process 1 are changing, the DC component will fluctuate.

Therefore, if a DC component occurs in the calculation result of the characteristic change detection section 15 and exceeds a preset tolerance value, the control section 16 sends a dynamic characteristic change confirmation pulse so that the input signal of the process 1 changes in a pulse-like manner. As shown in the figure, the sampled value is added to the target value signal r ^* (N) of the sample value PID control device at one point in the addition section 17, the operation signal u ^* (N) in the addition section 6, or the deviation signal calculation section 3. Immediately after that, the control unit 16 determines whether there is any variation in the DC component output by the characteristic change detection unit 16.

If there is any variation, this is an unacceptable change in the dynamic characteristics of the process, so the identification/tuning function is activated and the control constants are newly adjusted. If there is no variation, it is a disturbance and the sample value
Since regulation can be performed by the operation of the PID control calculation section 5, a correction value for the disturbance is inputted to the characteristic change detection section 16 so that the output becomes zero.

As described above, changes in the characteristics of the process 1 during operation are monitored, and when the dynamic characteristics change, the identification/tuning function is triggered to automatically adjust the optimal control constants.

In the sample value PID control device of the present invention, the control unit 16 inputs the initial value of the sample value control constant, which is input from a data input setting device (not shown), such as a console typewriter, console display monitor, or a dedicated panel.
The sampling period τ, response shape coefficient α ^* , etc. are input and set to each of the related parts mentioned above. Furthermore, settings for PI control or PID control can be changed, identification and tuning functions can be turned on and off, and adaptive functions can be turned on and off.

Next, each component will be explained in more detail.

The sample value control calculation unit 5 calculates a proportional gain Kc, an integral time constant Ti, a differential time constant Td, (in the initial value Kc ₀ ,
Ti ₀ , Td ₀ ), control deviation e ^* using sampling period τ
From (N), the operation signal u ^* (N) is calculated and output for each sample period as follows.

u ^* (N)=u ^* (N-1) + △u ^* (N) ...3rd formula Here, △u ^* (N) = Kc {(e ^* (N)-e ^* (N-1) )＋τ/Tie ^* (N)
+Td/τ(e ^* (N)-2e ^* (N-1)+e ^* (N-2)}...4th formula In the identification signal generator 7, the persistent exciting signal V ^* (N) is This identification signal V ^* (N) is added to u ^* ₀ (N) in the adder 6 and becomes an input to the sample and hold unit 2. That is, it is expressed by the following equation.

u ^* (N)=u ^* (N)+V ^* (N)...Equation 5 In this embodiment, an M-sequence signal that can be generated by a simple algorithm is used as the identification signal V ^* (l).

The pulse transfer function identification unit 9 is a process output signal
The pulse transfer function representing the dynamic characteristics of process 1 is identified from y ^* (N) and u ^* (N) of the fifth equation by time-series processing.

The pulse transfer function identification section 9 of this embodiment is composed of a well-known extended least squares filter. Finally, the pulse transfer function of Process 1 is determined by the following equation.

The transfer function calculation unit 10 calculates the pulse transfer function parameters a^ _i calculated by the pulse transfer function identification unit 19.
= 1 to m), a^ _i (i = 1 to n), the denominator series parameters (g ₀ ,
g ₁ g ₂ , g ₃ ,…). This calculation method was performed using the method shown in Lecture No. 1110 "Study of method for deriving low frequency characteristics in the S domain from pulse transfer function" at the 20th SICE Academic Conference (July 1981). There is.

The sample value control constant calculation unit 11 performs sample value control based on the denominator series parameters (g ₀ , g ₁ , g ₂ , g ₃ ) of the transfer function G(S) calculated by the transfer function calculation unit 10 and the sampling period τ. A sample value that has good ability to suppress process fluctuations against disturbances used in the system calculation unit 5.
Calculating PID control constants. In this example, Kitamori is the author of Proceedings of the Society of Instrument and Control Engineers, Vol. 15, No. 5, 1979.
The sample value PID constants are calculated based on the method described in ``Design method of sample value control system based on partial knowledge of controlled object'', No. P. 113-136.

First, the transfer function W(S) of the reference model is given by the following equation in which Kitamori's reference model with less damping and the binomial model implemented in the present invention are unified by the shape coefficient α ^* .

W(S)=1/1+σS+α ₂ (σS) ²
+α ₃ (σS) ³ +α ₄ (σS) ⁴ +... Here The coefficients 1/2, 3/20, and 3/100 are from Kitamori's normative model. If the shape coefficient α ^* is 0, the binomial model is the reference model, and if the shape coefficient α ^* is 1, it is the Kitamori reference model. Furthermore, by setting the shape coefficient α ^* to a value of 0≦α ^* ≦2.0, various reference models can be used.

Now, by matching the control staff's response to the step disturbance with the reference model W(S),
A set of equations shown below can be derived.

Here, C ₀ , C ₁ , and C ₂ can be expressed in the following relationship with K _c , T _i , and T _d in Equation 9.

Kc=C ₁ Ti=C ₁ /C ₀ Td=C ₂ /C ₁ Equation 9 Therefore, from Equation 8, the following calculation is performed depending on the operation mode of the sample value PID control device.

In PI operation, the following quadratic equation (g ₁ + τ/2g ₀ ) α ₃ σ ² + (−g ₂ + τ/6 ² g ₀ ) α ₂ σ
−(1/2g ₂ +τ/6g ₁ )τ=0...If the 10th equation is solved and the minimum positive root is σ ^* , then C ₀ and C ₁ are as follows.

_{In PID operation ,} ^{solve the} _{following three} ^equations _{: _} ^_ _{_} 7/12τg
₂ + τ/18 ³ g ₀ ) τα ₂ σ + (−g ₃ /3−τ/4g ₂ −τ/18 ² g ₁ ) τ ² = 0...Equation 12 Similarly, let the minimum positive root be ^* σ ^*. Then, C ₀ , C ₁ , C ₂
becomes as follows.

The above-described series of calculations for the PI operation and the PID operation are performed by the sample value control constant calculation section 11 in accordance with the operation command signal from the control section 16.

The operation flow is shown in FIG.

As described above, the response of the control system to the step disturbance can be made to have an appropriate suppressing force.
On the other hand, if the target value r ^* ₀ (N) of process 1 is changed in steps as it is, the overshoot will be too large and the settling time will become long, so that control with good responsiveness cannot necessarily be performed.

Therefore, a new target value filter 1 is added so that the transfer function y(S)/r(S) from the target value signal to the process signal also matches the reference model W(S).
2 is used for correction. This target value filter H
(Z ^-1 ) is calculated every sampling period as shown in the following equation.

H (Z ^-1 ) = h ₃ /1 + h ₁ Z ^-1 + h ₂ Z ^-2 ...Equation 14 Filter constant h ₁ of this target value filter 12,
h ₂ and h ₃ can be obtained by calculating from the integral time constant Ti, differential time constant Td, and sampling period τ calculated by the sample value control constant calculating section as shown in the following equation.

The calculation of this 15th formula is performed by the filter constant calculation section 13.

From the above identification results, sample value control calculation unit 5
A method for adjusting the proportional gain Kc, an integral time constant Ti, and a differential time constant Td used in the above, and a method for adjusting the filter constant of the target value filter 12 from the results are shown.

By doing this, it is possible to automate the adjustment of the sample value PID control system, which has a strong ability to suppress disturbances and has good quick response to the target value.

Furthermore, since the reference model shown in Equation 7 is used, by setting the shape factor α ^* , it is possible to automatically adjust the control constants from a control system that does not overshoot to a control system that causes appropriate overshoot. It is.

FIG. 4 shows an example of the response waveform of the control system resulting from the setting of this shape factor α ^* . As shown in the figure, it can be seen that the response of the control system can be freely selected depending on the set value of α ^* .

Next, referring back to FIG. 2, a detailed explanation of the adaptive function of the present invention will be explained. At the end of identification and tuning, the pulse transfer function identification unit 9 determines the characteristic parameters a^ _i (i = 1 to m) and b^ _i (i = 1 to n) of process 1.
and the equilibrium point parameter d^ of process 1 becomes clear. On the other hand, if the dynamic characteristics of process 1 during operation change, the characteristic parameters a _i i (i = 1 to m),
b _i i (i=1 to n) should have changed, and if a disturbance is applied to process 1, d will change.

Therefore, the characteristic change detection unit 15 calculates the model error as shown in the following equation.

η(N)=y^(N)−y(N)...Equation 16 Here, y(l) is the output of the process calculated from the characteristic parameters at the end of identification and tuning as follows.

y^(N)=- _n 〓 ⁱ⁼¹ a^ _i iy (N-i) + _o 〓 ⁱ⁼¹ b^ _i iu (N-i) + d^ ...Equation 18 On the other hand, the actual process output is as follows It becomes like the expression.

y(N)+ _n 〓 ⁱ⁼¹ a _i iy(N-i)= _o 〓 ⁱ⁼¹ b _i iu(N-i)+ε(N)+ _o 〓 ⁱ⁼¹ ciε(N-i)+d... Equation 19 where ε is noise.

By substituting Equations 18 and 19 into Equation 16, the model error η(N) becomes as follows.

η(N)=- _n 〓 ⁱ⁼¹ δa _i iy(N-i)+ _o 〓 ⁱ⁼¹ δb _i iu(N-i) δd In other words, the model error η(l) is an equation that includes changes in characteristic parameters.

Now, considering the expected value E [η(N)] of equation 20, When there is no change in characteristics, E[η(N)]=0, If there is a change in characteristics, E[η(N)]≠0.

However, it is practically impossible to calculate the expected value E[η(N)]. Therefore, in this embodiment, the short-time average value (N) of the model error η(N) is calculated as shown in the following equation.

(N)=1/J _J 〓 ⁱ⁼¹ η(N+1-i)(J=4~8) ...Equation 21 This short-term average value of model error (N) is the allowable value
Changes in the characteristics of process 1 are monitored by comparing and determining L ₁ as follows.

|(N)|>L ₁ 1 ...Equation 22 Here, the allowable value L ₁ 1 is calculated from the norm of the characteristic parameter vector shown below.

Here, ‖δθ‖/‖θ^‖ in Equation 23 is the process 1
shows the rate of change in characteristics.

Therefore, ‖δθ‖/‖θ‖=0.1 to 0.3 (10% to 30%)... By setting the allowable rate of change in characteristics of process 1 in % as in Equation 24, the allowable value L ₁ 1 can be calculated.

Now, in the short-time average value (N) of the model error calculated in this way, a DC component occurs both when the dynamic characteristics of process 1 change and when a disturbance is applied.

However, in the case of occurrence due to disturbance, regulation can be performed by PID control itself, so there is no problem even if the tolerance value L ₁ 1 is exceeded.

On the other hand, if the problem is caused by a change in the dynamic characteristics of the process, it is naturally necessary to readjust the control constants.

Therefore, it is necessary to determine the causes of these two occurrences.

In this example, to rewrite and explain the 20th equation, η(N)=- _n 〓 ⁱ⁼¹ δa _i iy (N-i) + _o 〓 ⁱ⁼¹ δb _i iu (N-i) + δd ...Equation 24 From this equation, when the dynamic characteristics of the process change, that is, δa _i i and δb _i i change, and δd remains unchanged and is zero, the process input signal u(N-i) is It can be seen that the model error further changes as the value changes.

On the other hand, if the process disturbance changes, that is, δd changes, and δa _i i and δd _i i do not change and are zero, even if the process input signal u(N-i) is changed, the model It can be seen that the error remains unchanged.

In other words, when the model error exceeds the allowable value _L11 , the dynamic characteristic change confirmation pulse is sent to the deviation signal calculation unit 3, so that the input signal u(N-i) of the process changes.
Alternatively, it can be realized by applying the identification signal from one point of the identification signal addition section 6 or the addition section 17.

In this embodiment, it is configured to be added to the deviation signal calculation section 3. Next, to detect the model error after applying this dynamic characteristic change confirmation pulse, the result of calculating the differential value of the model error (N) as shown in the following equation is used.

｜△(N)｜=｜(N)−(N−N _T )｜>L ₂ N _T =g ₁ 1/τ; Judgment time...Equation 25 Here, L ₂ is the following equation similar to L ₁ Calculate with.

L ₂ =‖θ^‖・‖△φ(N)‖・‖δθ‖／‖θ‖ ‖△φ(N)‖=(y(N-1)-y(N-k-1),
...,y(N-m)-y(N-k-m),u
(N-1)-u(N-k-1),...,u
(N-n)-u(N-k-n) ^T ‖δθ‖/‖θ‖=0.05 (5%) 0.10 (10%) 0.15 (15%) In other words, the amplitude of the applied dynamic characteristic change confirmation pulse‖ It is determined by considering δθ‖/‖θ‖.

As mentioned above, the model error |(N)| is the allowable value
When L ₁ 1 is exceeded, a dynamic characteristic change confirmation pulse is applied, and immediately after that, if the differential value of the model error |△(N)| exceeds the detection level L ₂ , it is determined that there is a change in the dynamic characteristics of the process. and activates the identification tuning function.

On the other hand, even if the dynamic characteristic change confirmation pulse is applied, the differential value of the model error |△(N)| becomes L ₂ within the judgment time N _T
If it does not exceed , it is judged as a disturbance and the model error |η
Correct (N)｜.

In other words, δd in the 20th equation is |(N)
Input | as the disturbance correction value. The operation of this adaptive function is shown in FIG.

As described above, the present invention detects changes in the dynamic characteristics of the process during operation and automatically performs readjustment of control constants.

[Brief explanation of drawings]

FIG. 1 is a block diagram for explaining a conventional example, FIG. 2 is a circuit configuration diagram of a sample value PID control device of the present invention, FIG. 3 is a flow diagram for explaining the inside of the block diagram, and FIG. 4 is a diagram for explaining the present invention. FIG. 5 is an explanatory diagram of the effect of the shape factor setting function, and FIG. 5 is an explanatory diagram of the operation of the adaptive function of the present invention. 1...Process, 2...Sample hold, 41 and 42...Sampler, 5...Sample value control calculation section, 7
...Identification signal generation section, 9...Pulse transfer function identification section,
DESCRIPTION OF SYMBOLS 10... Transfer function calculation part, 11... Sample value control constant calculation part, 12... Target value filter, 13... Filter constant calculation part, 14... Identification end judgment part, 15... Characteristic change detection part, 16... Control part, 17... Addition part.

Claims (1)

[Claims] 1. A control deviation is calculated from a target value signal of a process to be controlled and a process signal obtained by sampling the control amount of the process, and this control deviation signal is calculated.
a sample value PID control calculation unit that performs PID control calculations and outputs an operation signal for the process; a pulse transfer function identification unit that identifies parameters of the process from the operation signal and the process signal; A transfer function calculation unit that calculates a transfer function in the S (Laplace operator) domain from the pulse transfer function of the obtained process, and a PID control constant used in the sampled value PID control calculation unit from the calculation result of this transfer function calculation unit. In a sampled value PID control device having a sampled value control constant calculating section that performs Sample value PID control, characterized in that the sample value PID control is generated by varying a shape factor and includes means for determining the PID control constant so that the characteristics from the control target value to the control amount match the reference model. Device.