JPS603707A - Pid controller of sample value - Google Patents

Abstract

PURPOSE:To respond quickly to a change of a target value and to suppress a disturbance by identifying the characteristics of an operating proces with time serial arithmetic and determining a sample value PID control constant in accordance with this identification result. CONSTITUTION:A sampled target value signal gamma(N) is inputted to generate the output of a target value filter 12, and a control deviation is obtained in a deviation arithmetic part 3 in accordance with this generated output and the output of a process 1. On the basis of this control deviation, a sample value PID control arithmetic part 5 outputs the operating signal of the process to control the process 1. For the purpose of identifying the dynamic characteristics of the process 1, an identifying signal is generated from an identifying signal generating part 7 and is charged to the process 1. Then, since an identifying condition is satisfied, the transfer function of the process 1 is identified in a pulse transfer function identifying part 9 by time series arithmetic. Next, a transfer function arithmetic part 10 calculates low-frequency parameters of the transfer function in a region S on the basis of the pulse transfer function. These parameters and a sample period tau are used to calculate a sample value control constant in a sample value conrol constant operating part 11 so that the suppressive force of a disturbance is optimum.

Description

[Detailed Description of the Invention] [Field to which the invention pertains] The present invention relates to a sample value PID control device f, Jt for sample value control of a process to be controlled, and in particular to identifying and identifying dynamic characteristics of a process during closed loop control. In a device that has a function to automatically optimize control constants based on the results, is it possible to improve the control system and automatically adjust the control constants accordingly? and of processes in operation! ! 'II Optimization of control constants for 1% change In addition, the present invention relates to a sample value PID control device Wl that can select the response of the control system.

[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 gift characteristics of a process during closed-loop control and automatically adjusting PID control constants optimally based on the identification results. Publication (Unexamined Japanese Patent Publication No. 57)
-23108).

FIG. 1 is a block diagram showing the configuration of this sample value I'ID control device. In this sampled value PID controller, a versatile and exciting identification signal is generated during closed loop control.
From the input/output data of process 1), the pulse transfer function of process 1) is identified, and from the identification result, the transfer function of the S region of process 1) is calculated, and the characteristics from the target value to the process output are non-optimal model characteristics. In order to match the sample value PIE), the control constant was calculated electronically, and the results of the calculation were used for control.

In other words, since this device is a sample value P I D control device that aims to be responsive to changes in the target value, the addition (; L ) Since the main zero point of the control device may be almost canceled out, it has become necessary to have the ability to suppress the process output against disturbances.

In other words, it was necessary to obtain a sample value PID control system with excellent controllability in both the correspondence of the process output to the target value and the ability to suppress the process output against disturbances.

In addition, in process control, depending on the controlled object, the process output may be overruled when the target value is changed.

- fQ is controlled so as not to overshoot, or to allow overshoot.

However, in this sample value iJ11) system (rl system), the control constants to be adjusted are fixed so that the response of the process output to a change in the target value is approximately 10% in any process. .Therefore,
process? Current 1 of ii' control: The one that satisfies the requirements of j and 1 is essential.

The dynamic characteristics of the process may change due to

Naturally, if this happens, it is necessary to reset the optimal control constants. Currently, expensive plant monitoring equipment such as recorders and monitor televisions are used, and operators monitor changes from time to time to monitor and respond to changes in the plant.

Therefore, it was required that the sample value, P i D flt controller be able to monitor the plant and perform M optimization of the control constants.

[1] The object of the present invention is to solve the above-mentioned conventional samples 1jf4',
It is an object of the present invention to provide a sample value f'lD control device that can improve the controllability of the PID control device mega, that is, to obtain a control system that is both responsive to target value changes and suppressive to disturbances; It is possible to select the response shape of the process output to changes in the target value of the control system, and to determine control constants so as not to overseat the control system.
Or Opage.

- sample value P1 i) control amount +11 device Fi can be used to determine the control constant, and furthermore J: dynamic characteristics of the process in the bundle. It is an object of the present invention to provide a sample value PID control device that can automatically optimize control constants if necessary by monitoring changes in PID.

[Summary of the invention]

The present invention identifies a process excitation condition during closed-loop control, and based on the identification result, determines sample value PID control constants to be matched to a reference model to obtain a suppressing force when a disturbance enters the process. , as a result, the correspondence to the change in the process C value is not taken into consideration, and the inferior part is compensated for by setting the process target value input point and the control deviation calculation point 41 (i
The constant of the target value filter is determined so that the transfer function from the target value to the process output 1 matches the reference model, and the suppression performance against disturbance and the responsiveness to target value changes are improved. The sample value P]) is obtained by selecting the control system and determining the newly introduced target value filter foot table 'I which can be determined from the sample value PID control constant from the process identification result.
”] Confused by the number of controls, M I\- and then decide the self! f'-, and
Thousand volts! In addition to detecting the change in the B('r'fh property of the process in the drain) and the change in the dynamic characteristic exceeding the allowable value, the above-mentioned identification and the 4 prefectures of the written constant are automatically changed to 11. By observing the dynamic characteristics of the process, the control constants can be constantly optimized to adapt to changes in the dynamic characteristics of the process during operation. Select a response shape, and set the sample value PID control constant and target value filter constant according to the response shape.
A sample value 1' l D control device that allows the user to select a reference model that defines the response shape of a control system by adjusting the Ibh circumference, thereby making it possible to adjust the process control override. It is 6. 1 [Effects of the invention] (1) The tri-characteristics of the process during operation can be improved by
:Identified by T, and from the identification result, it is possible to adjust the sample value P l l which has a strong suppressive power against disturbances, excellent compatibility with the target value, and control system constants. .

(2) Processes in the green bundle! Changes in l1Il can be monitored by the sampled value PID control device, and furthermore, control constants can be readjusted automatically in response to changes in dynamic characteristics.

(3) The user can select the shape of the response 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 Pl in that case]) Constraint comparison and target I direct fill 2. :10,000ζ can be used. Therefore, not only can the manpower and time required for adjusting the sample value PID control system and the plant operation loss be greatly reduced, but also the costly equipment and manpower needed to monitor the process during operation can be saved. In addition, the control system can be adjusted to meet the process operating specifications.
, Tll has an integer number of stages, so the practical effect is human.

[Example of the present invention]

An embodiment of the present invention will be described in detail below using the V<+ plane. FIG. 2 shows the sample value P according to the present invention.
]) is a professional diagram showing the circuit configuration of the control system. control? , II is the target of the plant? , ease, pressure tsuba ryuushi,
The process (1) is configured to control the liquid level, etc., and the process human power u (1), which holds the sample at the bottom, is input to the sample hold section (2), and y (t
) is the output of process (1).

This controlled object is a sampler (41) that operates at a sampling period τ with a constant process output Y (t) k Wt;
Input the sampled process output y(N) output from this sampler (41) to the feeder sensor as a number,
Using the sampler (42) that synchronizes the target value r(t) with the same sampling period τ, the output value of the target value filter (12) ro( N), and from this r4(Nl) and the process output y(N), the control deviation e(to)
and the sample value P I D that controls the process (1) based on the deviation calculation unit 6) that calculates the control 1rIl difference e.
Restriction 16'(n Rlj (5) is te of the process
A basic sandal (ljj r I 11 control system) is configured to calculate and output t work 1 work 5 no.
represents refreshing time.

On the other hand, the force 11 characteristics of the process (1) are identified, and based on the results, the sample
PID control constant and the value filter (12
) The identification/chinning function that adjusts the filter constant of the figure is the identification signal generation section (7), the pulse transmission '9=1 number identification section O), and the traditional sample value I) l D control 1 system. It is connected as shown in 1. To identify both characteristics of process (1), a versatile and exciting identification signal 9 is generated from the identification signal generator σ), and then passed through the number hold unit 3) to the process α). Inject into.

As a result, a closed-loop town identification convention is established, so
(U eclipse, yN), N=1.2,...) is input, and the pulse transfer function identification section (9) processes (
1) The pulse transmission function t<G, (Z s ) is identified by time series calculation. Next, transfer IV, l number performance t1: part (
10) is the low frequency parameter g of the S area transmission function o (s) from the obtained pulse transmission 1:I number G(Z-').
. Calculate Sgl, g2, and g3. Here g. ,gl,
g2 and g3 are transfer function G (expressed by a 70th order equation with cloud as a denominator column) parameters.

-y-yb/l/ 1flj tlil, l f1'j
l ijf g Proportional gain Kc of the sampled braking constant used in section (5), integral time constant 1゛11 time constant #J,:
T d c, i'+'+Vn2 low frequency harameter g,
Using gls g2sg3 and sandal period τ, the sample (iXl,
tI control constant calculation section (operated in the river.
From the sampling period τ, the filter constant of the target value filter (12) is calculated as 4'+((ns change; 'A:'
,LE,1? j i: j#; iN K 17: ryoyo74'/"I'~6.~=+<('13)".

In this way, professional wrestling (1
), and based on the identification, it has a strong suppressive power against disturbances, and it has a strong suppressive power against disturbances;
+: Sample value PI with excellent correspondence to '+
Control of L1 control system (IIjl constant θ automatic adjustment JJ-Jj, 1
．． jFj is accomplished.

In addition, the sample value P, 11) control constant Qi9 tracing unit (
The normative model for Proposition 1 of the control system used in 11) is 2.
] J1 model and model with damping in 4 major shapes: i
:416vα to form a set, this form 4, coefficient α
By changing the settings of the control system, the response board type override for step changes in the control system and 1" value can be changed.A-) fig
can be adjusted. Furthermore, the sample value control constant ↓(() and the filter constant are automatically changed in accordance with the adjustment of the shape coefficient α.

Next, let's look at the operation of process (1) during operation! i) Adaptation to monitor the sexual liquefaction and trigger the identification and checking function to readjust the control constants if the 1% sexual change is no longer possible. .

The adaptive function is composed of the identification end determination section (14), characteristic change detection section (15), and control section (16) shown in the figure.
As shown in the figure, the identification checking function section and the sample value P
, I is glued to the JJ control system.

First of all, as we advance along the same path of self-identification, -ning cherry blossoms, j'fll : f +, 1 transmission gate number deception tfl
tj'l 'n12 low frequency parameter g calculated with t(io). , gl, g2, and g3 begin to converge. Therefore, the identification end signal ill''+(14) is used to determine the convergence of the parameters as shown in the following equation (1-j) and output the identification end signal.

Ig, rtJ)-gi (N-1) 14 几Ass (i
=1~3)...1152 formula % formula % Set the value.

When the identification end condition Y1 of the second formula is satisfied, the identification end determination section (14) inputs the identification end No. 111 to the control section (16), and the control section (16) The identification signal generation part σ) of the function, the Norculus transfer function identification part (9), and the transmission 1-9 Δ part (1
0) is stopped.

Furthermore, the control section (16) calculates the values of Kc, Ti, 'I'd calculated by the sample full'1 control constant calculation section (11) when the identification/checking machine is stopped. The above-mentioned summing value control calculation is also input to 1.5), and the filter constant calculated by W'T is input to 1.5) and the calculated filter constant is calculated.
Input sample value to j value filter PIJJ control: PI
It is designed to complete the υ,·1 integral of l′′1−.

In addition, in response to the control unit (16Nd cycle determination (i4), the characteristic change detection unit (15) outputs the pulsed iV l'N, li same foot fit! (9) at the end of the identification. input the process parameters and wake up the characteristic change detection section (15).Characteristic change detection section (15)
calculates the 11% characteristic change after the completion of identification of process (1) at every sampling period τ using the input process parameters, process operation signal LIN) and process signal 7 generations, and calculates the The results are output to a control section (16).

Here, as the characteristic 'J1 of the characteristic change signal calculated by the characteristic change detection section (15), if the dynamic characteristic of process (1) changes and if disturbance is added to process (1), A DC component appears.Moreover, when this iαbIL component appears, the input power of the process (],) i'j 4j
If we change , the DC component does not change in the case of a disturbance, but if the dynamics of process (1) changes, the DC force changes.

So, the property'! iiQ of Shika 7'N Yamabe (15)
Tr' (a DC component is generated in the result, and the process (1
) is activated from the control section (16) so that the number changes in a pulse-like manner. 1゛〒Confirmation pulse for 2 signs 1
(Ransol value as in I)' l D system 1fll 4
1; f7. 's eyes +';! +Direct1i'i ``3' r
If ('l) is a force If, 'C'J-R1 ((17)' is operation No. 48 U. In addition to one point, control the presence or absence of a change iaj in the DC component output at the peak (16) of that Mj, !' characteristic change + t + s
(Determine at 161.

Here, since the J mixture with the Cil peak is the emulsification of the example 71.1 of the process movement 71.1, the above-mentioned identification/tuning h However, if there is no fluctuation, it is a disturbance, - angle value P1]) control operation'>, r: +X (1iI of 51
Since the regulation and ration are generated in the III work, the disturbance is adjusted so that the output to the characteristic change detection section (16) becomes zero.
The second supplement 111 "4" is used manually.

As described above, q':j of process (1) during planting
Monitor t14 changes and more! 17I114', When the i gender changes, 1st tuning 4; (trigger the function,] Ii cow, 1. Ki ffrlli,! 1) Constant adjustment is automatically done. .

In addition, the Zampurui 11″IP i′ of the present invention, 1)
In the A41 device Vl, control unit (16) is used to input sample value control constants manually from a data input setting device (not shown), such as a console typewriter, console display monitor, or dedicated panel. Initial value, sampling period τ, shape factor α
Enter the above relationship! ? It is set to be set in three cities.

Hata et al. reported on the setting history, identification, and tuning of PI subcontrol or P], D;li control (on/off settings and adaptive towers of Yoshino's 1jiJ work, and on/off settings of Noh. There is.

Next, each configuration tτIS will be explained in further detail h1.

Sample value control n part (5) is proportional gain 1<01 W number T at integration! , differential time constant 'It'd, (initial value ke l(
c.

T+o%pd, ), controlled using the sampling period τ (,Il
Calculate and output the l deviation r(operation from N to Q edition [Io (to) as follows: / per pull cycle.

half uoIJ)=uo(N-1)+ΔuoII)...
・, f433 formula % formula % )) ... 2134 formula identification (in the autogenous part (7) of iT1, there is also a versistintri-exciting inn V), and this identification b family name V) and then 1 shishiro x 1ζ (6) and U. (to) and the sword 11p are input, and the input of the number hold part (2) is performed. In other words, it is shown by the formula.

Yoshi 0-♂. N) + 翰・−・−・・−rg 5 formula 2 in this example, 4 with a simple algorithm as identification number 48
1- An M-sequence signal that can be generated was used.

The pulse transmission relation identification part O) is from the process output number half ytJ) and the U (to) of the fifth equation to the lj of the process (1).
This method identifies a pulse transfer function exhibiting l/l\ property by time-series processing.

Honji jmiπI+'s pulse transmission: Kadomisu times) iJ, t
't:i (9) is the well-known expansion shark minimum 29+! : 4j "L is formed with a filter. It is finally 2, Froseth (
The pulse transmission (1ζ11) of 1) is calculated by the following equation.

Old Tsuji; shi>, 1'j%: 4'l 'riii (
1 (1) is a pulse transmission 1 part 11.1 with an identification part (1
9), the pulse transmission a I'/-1 number of parameters ai (i = 1 to m), b ((i = 1 to n
) to 61',' Transmission 1st gear 1st part ()
Parameters of the denominator series of (S) (go+g1g 2 +
g 3+...) to perform -e. this(,:
,::kj,,'For the method, 2120 times S I
Q E Studies M iif; tj Association (Hf (July 1956)
Lecture number 1110r pulse transmission 1), 1 number to S Jun J
,! , : low frequency characteristic center height method tr=i:''.

Sample value control constant
The denominator series parameter of (S) (gg +g1.+g
2 + g2 ) and sample Jt/th jllτ, f
y pull 1+ff, fljll '1111tI:t
:, i% f':! S (5) f ear) Disturbance K
91 and the suppression of process change -IJ
-sible value P I IJ control'), '+14・ξ (Toshie. In this real 7iiQ example, Kitamori is the paper of the Society of Instrument and Control Engineers, $1979, Vol. 15, No. 5 J)・11:
3-136, "7i+11 illusion's ↑71chi partial knowledge N:
``Lean pull 11i based on j, system (+'ll type word 1
Based on the method V mentioned in 1'Yarihoj, sampleite ()) l I) constant value 'l' is calculated.

- First, Jlf, fi'j co, model transmission j7 motion li 10) p mouth iτ-1 north i C river with dumping, range model and the present invention? −i+, :i 2'', the i model is given a set of shape coefficients ζ′tα, and the pre-r order J(, is applied.

■慴）－－－－－－１－－－－'－"－－－－１－－'
−'−−−'−10σ5−+−α, (σS)2+α3(
σS>3+α4(σS)4+, and the coefficient '4' 3
．． ,E,, 3/1ou is that of the j5 model defined by North Street.If the swelling coefficient α is set to 0, then the I 2 ]100 model is used as the reference model, and if the shape coefficient α is 1, then This is a standard model with a 1° angle.Furthermore, the shape factor α is 0≦α≦2.0.
By taking A11, various normative models can be used.

Now, in response to the Stepta Disturbance, we will regulate the application of control 1 security (
・[l model W (By matching with 6'l, here, (' O+ O,,C2keiJ,', !
If i<c, Jlf, , Td, then the following expressions can be used.

Kc=O.

T'1=o110o Old-・: p9 formula % formula % (Therefore, from the 8th formula, sample value P I l) Control 4 to 1
1. The following Y, ECS'1. Do the following.

In the PL operation, the following quadratic equation = 0... If the Equation and a half of Equation 10 and a half is multiplied by 1.jf to be the minimum positive root missing σ, then 0°01 becomes as follows.

In PID operation, the following cubic equation is
+τg, 10τ2go) C4a 3-1- (g 3 +
-τg++2 Xie Tong4. If the minimum positive value of f is d, then Co, Q,
, (j2 is 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 (river) according to the operation command signal of the control section (16).

The operation flow is shown in FIG.

As described above, the response of the control system to the stenopteral disturbance can be made to have an appropriate suppressing force, but on the other hand, the target value r of process (1) remains unchanged. If the concavity is changed in steps, the aperture or torto may become too large or the settling time may become long, so that control with good coordination cannot necessarily be performed.

Therefore, next step is the transmission from the target value signal to the process signal +
9: i, Q Y (81/ r (,'i) also before I!
2 J'A, 1jj11, model #W $), correction is performed using a new 4th office value filter (12). This target value fill i((Z-1)
is calculated every sampling period as shown in the following equation.

3 II (Z-') = −−m=−−−−−−・・
...2fB4 formula % formula % Filter constant of this target value filter (12)... ~2
゜ ~ 3 is the integral time constant Ill i1 calculated in the sample value control fixed ship vIσ part, the integration time constant 1゛d and the sample period τ, or the filter constant H,
tJ'l', f''1lX (13).

As above, based on the identification results, - proportional gain KC1 sum used in sample value control calculation section (5) 1 > -1 - number 'l' i
, i'41 of the filter constant ρ of the target value filter (12) from the adjustment method of the time-load Td and its results. The method was shown.

By doing this, the sample value PID has a strong ability to suppress disturbances and has good correlation with the target value.
This is to convert the key signature 1"'- of the control system to jjjfJ.

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 open up to a control system that does open up neatly. be.

It can be seen that depending on the plan, the response of the control system can be freely selected.

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 transmission relationship identification unit e) determines the characteristic parameters of process (i) to be negative (i = 1 to m), positive (i = 1 to n), and process (
1) The equilibrium point parameter order becomes clear. On the other hand, 1
If the dynamic characteristics of the process (1) during circulation change, Q
”j 'I', 'i parameter + aI (+ = 1-m
), b+' (1 = 1 to n) should have changed, and when a disturbance is applied to pnibrocess (1), d changes K
JEt b arrived.

Therefore, the characteristic detection unit (15) calculates the model tjj4 difference by jifji as shown in the following equation.

η (to) = yN) - y eclipse ・・・・・・・・・ 16th
In the equation, y(f) is the output of the process calculated as follows from the characteristic parameters at the end of identification and tuning.

Q) > stomach box y (N-i) tenu nishiki u (N-i) +? 1 ・
...Equation 18 □+=1 1=1 On the other hand, the process output of the current zP is as shown in the following equation.

yN) Shibata 1y (~1) = Nagisapa (N-i) + gff”+
Δqε(N-i)+d...Equation 19 where ε is noise.

By substituting Equations 18 and 19 into Equation M'16, the model error is 1NJff as follows.

In other words, the model error η(to) is an equation that includes changes in the characteristic parameters.

Here, considering the expected value E[η~] of Equation 20, when there is no change in characteristics, E[η(e)) = 0, and when there is a change in characteristics, E[η~]~0. Become.

However, it is practically impossible to calculate the expected value E[η(to)] to (1). Therefore, in this embodiment, the short-term average value τ(to) of the model error η(to) is is being calculated.

This short-term average value τ of the model error is defined as r[zero value L1
The process (1) is to compare and judge as follows.
Monitor changes in characteristics.

I7tan1〉Ll・... Equation 22 Here, the tolerance value l is calculated from the norm of the characteristic parameter vector shown below.

Here, 11δθ11/11t11 in Equation 23 indicates the rate of change in characteristics of process (1).

Therefore, by setting the allowable rate of change in characteristics of process (1) as shown below, the allowable value can be calculated.

Now, the short-time average value η(to) of the model error calculated in this way is calculated when the dynamic characteristics of process (1) changes and when a disturbance is applied to J(+), in which case the Naoe component is Occur.

However, if the disturbance is caused by a disturbance, the PID control itself can control the leggy rate, so there is no problem even if the tolerance value is exceeded.

On the other hand, if the problem occurs due to a change in the dynamic characteristics of the process, the entire control constant 'F+H'e aVj is naturally required.

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

In the present embodiment, the above equation 20 is written down and explained as follows: Equation 24 From this equation, when the dynamic characteristics of the process change, the equation δ
It can be seen that when a11 and δ〥 change and δd remains unchanged and is zero, the model error further changes by changing the process input signal u (N - i).

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

In other words, when the model error exceeds the allowable value, the dynamic characteristic change confirmation pulse is sent to the deviation signal calculation unit 6) or the identification signal is added so that the input signal u('N-1) of the process changes. This can be realized by applying from any one point of S1 ((6)) or the adder (17).

In this embodiment, the number of digits is added to the deviation Iha performance 3'#, part (3). Next, in order to detect the fluctuation of the model error after applying the super pulse to this motion 4, we can calculate the result by applying the value of the model error η to the model error η as shown in the following equation. using.

Here, ff- is calculated using the following equation in the same way as for L2 beam.

+1Δψ(Nll=(y (N-1)-y (N-to-
1) , -, y (N-m) -y (N-1c -m
), u (N-1) -u (N-to-1),
・ .

``(N'') u (N- to -n)'' As shown above, the model error 1η(to)1 is the allowable value J+ When the dynamic characteristic change value exceeds , a pulse is applied, and immediately after that, the differential value 1Δη(\)I of the model error reaches the detection level Ll.
If the value exceeds 11'', it is determined that Y changes have occurred in the process, and the identification checking function is activated.

On the other hand, even if the dynamic characteristic change confirmation pulse is applied, the judgment time is NT.
If the differential value 1Δη(f)1 of the model error does not exceed L within this period, it is determined that there is a disturbance, and the model error 1η time 1 is corrected.

Finally, input 1η■1 at this time into δd of the 20th equation as the disturbance correction value. More than the behavior of this adaptive function, tree 9
F' + light detects the change in the dynamic characteristics of the process in the J product, and controls the number of control gummies 'M'ii"I m!
This is something that should be done independently.

4. Figure 1 is a block diagram for explaining a conventional example, and Figure 2 is a sample value PID fliil fjll type of the present invention. li, ', +- circuit 4: 'j 1, 1, 3. The effect of the function h[:1 and Figure 5 shows the effect of the adaptive cr function of the present invention! h is '] Explanation port 1.

(1)...process, e)...sample hold,
(41) (42)...Sampler, however)...Sample value control unit, Ip unit, (7)...Identification number 5j generation unit, υ)...Pulse transfer function identification unit (10 )...Denkinmon number calculation unit, (11)...Sample value control fixed value cleaning calculation unit, (12)...Eye frame value filter, (13)...
·fill"'"""? '(If-741(, (”°°°
"J! M7 M'"""is, 0s+-q"
, liquefaction detection section, (16)... control section, <
171...addition section.

Claims (1)

[Scope of Claims] (1) The eyes of the process to be controlled (the target value filter that samples the value signal and performs filter calculation, the output signal of this target value filter, and the control m of the process) are sampled. Calculate the control deviation from No. 146,
A sample value PID control calculation unit performs a PID control calculation on this control deviation signal and outputs an operation signal for the process.
an identification signal generation section that applies an identification signal consisting of an exciting signal, and an operation signal and a process signal obtained by adding the identification signal generated by the identification signal generation section to the output of the sample value PID control calculation section, and The pulse transfer function identification section identifies the parameters of the pulse transfer function fQ IJ, and the transfer function 1' calculates the transfer function in the S (Laplace operation Jl, child) region from the pulse transfer function of the process obtained by the pulse transfer function fQIJ"31 number identification section. 1Jir calculation unit, a sample value control constant calculation unit that calculates a PID control constant used in the sample value PID control calculation unit from the calculation result of this transfer function calculation unit, and P l 1 of this sample value PID control constant calculation unit. ) an identification/tuning function consisting of a filter constant calculation unit that calculates a filter constant of the target value filter from a control constant; an identification end determination unit that calculates and determines the end of the identification from the calculation result of the transfer function calculation unit; a process parameter of the pulse transfer function calculation unit at the time of termination; a characteristic change detection unit that calculates a change in the characteristics of the process from the process operation signal and the process signal after the identification/tuning function has stopped; and this characteristic change detection unit When the calculation result of the characteristic change n' exceeds the preset zero value, a pulse is applied to the target value signal sampled in the process, and after applying this pulse, the transient of the result of the characteristic change detection section tf4+ (, J sample value PID control'l '-lls''. Value PID control device. 6) The sampled value PID control device according to item 1, characterized in that the pulse application point is the addition point of the identification signal. 1'J,'. (4) The standard model used for the calculation of the sample value PID control constant calculation unit is a two-term model and a model with damping, which are combined into a shape coefficient, and a shape coefficient setting unit is provided to vary the shape coefficient. The sample value PID control device (no.