JPS62163107A - Process controller - Google Patents

Abstract

PURPOSE:To improve the control performance of a process controller by correcting the coefficient of a function in terms of learning in a normal operation mode in order to increase the degree of approximation of the function in case said function that approximates the process characteristics is contained in a control loop. CONSTITUTION:When a coefficient store device 35 is selected, the value of K1' is reduced less than the previous value and supplied to a function generator 40 and a differentiator 37. While K2' larger than the previous value is supplied to the generator 40 and the differentiator 38 respectively in case a coefficient store device 36 is selected. As a result, the output compensating amounts corresponding to the changes of the K1' and K2'' are given to a transport amount controller 32 as the VN value. At the same time, the generator 40 delivers the Fs value based on new K1' and K2' to a booster gas controller 19. The controller 19 controls a control valve 21 for control of the booster gas amount flowing through a pipeline l2. As a result, the flow rate of powdery substance flowing through a pipeline l3 is varied to cause the change of the value of Wa.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a process control device using a function that approximates process characteristics.

(Prior art) For example, in the freeness control of a refiner in a paper manufacturing process or the transport amount control in a powder transport system, the final control variables such as freeness and powder transport amount are not directly controlled, but are kept constant with these control amounts. A method is being used to indirectly control the freeness or transport amount to a target value by controlling the refiner power and booster gas flow rate by replacing them with the refiner power and booster gas flow rate, which have a correlation. The reasons why such an indirect control method is adopted are that it is difficult to measure the controlled amount, and that the measurement system includes a time delay, so direct control deteriorates control accuracy. .

FIG. 3(a) and FIG. 4(a) are block diagrams showing conventional configuration examples of a freeness control system and a powder transportation amount control system, respectively. First, the freeness control system shown in FIG. 3(a) will be explained. In the figure, 1 is a refiner that refines the valve, 2 is a power controller that controls the amount of power supplied to the refiner 1, and 3 is a freeness setting value C3F.
This freeness controller is composed of a setting section 3A that provides -8, and an adjustment section 3B that adjusts the freeness. It is assumed that the freeness controller 3 performs sample value PI control. Furthermore, when the adjustment operation is performed manually, the output from the setting section 3A is directly input to a function generator 4, which will be described later.

4 is the freeness value C given by the freeness controller 3
Pulp flow rate Q detected by 8F and valve flow rate detector 5
It is a function generator that responds to and outputs the final power value. The output of the function generator 4 is the power target value S V
(Setting Value) is given to the power controller 2. 6 is a freeness detector, the output of which is sent to the freeness controller 3 by PV (Process
Value) is given as a value. 7 is a wattmeter that detects the power of refiner 1, and its output is sent to power controller 2.
is given as a p V (Process Value) value. 8 is an indicator that indicates the pulp flow rate.

The function generator 4 has a built-in function generating mechanism with characteristics as shown in FIG. The power control target value of the refiner 1 is output. (B)
The characteristic curve shown in FIG. 2 shows the pulp flow 11Q (horizontal axis) and the finer power KW (vertical axis). There is a curved relationship between pulp flow rate Q and refiner power KW as shown in the figure.
It changes as shown in the figure using the freeness as a parameter. Therefore, one of the characteristic curves can be identified from the pulp flow fiQ and the freeness value C8F, and if the curve can be identified, the point where the straight line extended horizontally from the intersection of the curve and the perpendicular line passing through Q intersects with the vertical axis corresponds. It becomes Sururi Fine Power. Incidentally, among the curves shown in (b), the one shown by a broken line is a curve giving the freeness target value C3F-8, and the finer power corresponding to Q becomes the power target value KWs. That is, by performing power control so that the refiner power becomes KWs, control can be performed to maintain the freeness at an optimal value.

Next, the powder transportation amount control system not shown in FIG. 4(a) will be explained. In the figure, 11 is a tank into which powder such as coal powder is injected and stored, 12 is a pressure regulator that adjusts the pressure inside the tank 11, and a control valve 13 provided on the pressure line /1 is used. By adjusting the pressure in the tank 11, the pressure inside the tank 11 is adjusted. Reference numeral 14 denotes a transport amount regulator comprised of a setting section 14Δ that provides a transport amount set value Ws and an adjusting section 14B that controls the transport amount. It is assumed that the transport amount controller 14 performs sample value PI control as in the case of the freeness controller 3. When the adjustment operation is performed manually, the output from the setting section 14A is directly input to the function generator 15.

15 receives the transportation amount W given from the transportation amount controller 14, the tank internal pressure P1- detected by the pressure detector 16, and the pressure PB in the transportation blowing light 18 detected by the pressure detector 17. This is a function generator that outputs the corresponding booster gas amount F. The output of the function generator 15 is applied to a booster gas amount controller 19 as a booster gas target value Sv. 20 is a flow rate detector for detecting the booster gas flow path, and its output is given to the booster gas amount controller 19 as a PV value. The booster gas amount controller 19 controls the flow rate of the booster gas by adjusting the control valves 21 provided on the pipelines 1 and 2. In addition, 22 in the figure is Pa
It is an indicator that indicates pressure.

23 is a scale that measures the amount of powder in the tank, 24 is an indicator that indicates the integrated value of the output of the scale 23, and 25 is the transport amount W from the maximum change in powder weight ΔW within the unit time ΔT of the powder. This is a calculation unit that calculates W-ΔW/ΔT, and its output is given to the transportation amount controller 14 as the PV value Wa.

The function generator 15 has a built-in function generating mechanism with characteristics as shown in Fig. 4 (b), and calculates a corresponding curve from the powder transport amount W and the tank internal pressure, and uses that curve as a basis. and outputs booster gas ff1lF. The characteristic curve shown in (b) shows the powder transport MW (vertical axis) and booster gas IF (horizontal axis) calculated by the calculator 25. There is a curved relationship between the amount of powder transported w and the booster gas IF as shown in the figure (however, the internal pressure P8 of the transport blowing light 18 is assumed to be constant here), and the pressure inside the tank is taken as a parameter. Changes as shown in the figure. Therefore, one of the curves can be specified from the powder transport MW and the tank pressure value, and if the curve can be specified, a straight line drawn perpendicularly from the intersection of the curve and the horizontal line passing through W intersects the horizontal axis. The dot is the corresponding booster gas IF. Of the curves shown in (b), the dashed line indicates the required booster gas l.
This is a curve that gives F s.

In the powder transport system shown in FIG. 4(a), the tank 11
Rather than directly controlling the powder flow rate transported from the transport inlet light 18, a booster
The flow rate of powder is controlled by blowing gas. For example, increasing the booster gas flow rate F will decrease the powder flow rate, and decreasing the booster gas flow rate F will increase the powder flow rate.

Therefore, by controlling the booster gas flow IF, the amount of powder transported in the pipeline 13 can be controlled.

The function generator 15 is given the transportation amount W and the tank internal pressure PT. The function generator 15 gives the value of the booster gas flow rate F corresponding to W and PT at that time as 5VlaFs to the booster gas m controller 19, and the controller 19 determines that the PV value from the flow rate detector 20 is the target value Sv. The flow rate is controlled so that it becomes equal to . As a result, the powder flow rate can be controlled to be constant.

<Problems to be Solved by the Invention> By the way, in the control system described above, if the degree of approximation of the approximation formula giving the correlation is sufficiently high, the power setting control and booster gas amount setting control described above are sufficient. . However, in reality, the approximation formula includes coefficients representing equipment or process characteristics, and it is difficult to accurately determine these coefficients. Furthermore, these coefficients usually change over time, and even if they can be set to the optimal coefficients at a certain point in time, they change over time. Therefore, the degree of approximation of the coefficients could not be made very high, and no improvement in control characteristics could be expected.

In the embodiments described above, a freeness controller and a transport amount controller are used to correct this error, but the response to set value changes and disturbance changes is slow to converge and has poor controllability.

FIG. 5 is an explanatory diagram of the operation of the powder transport amount controller 14 shown in FIG. 4. Here, the controller is performing an integral operation.

First, when the setting unit 14A sets the SV value Ws of the transport H, the adjusting unit 14B converts this Ws value into the MV value (lvlani
It is output as it is to the function generator 15 as a (pulated value value - operation output value).

The function generator 15 receives this Ws and determines the corresponding booster gas amount Fs+ as shown in (a).

Using this Fs+ as the first control point, the booster gas m controller 19 performs booster gas amount control.

On the other hand, the computing unit 25 calculates the actual transport IW after one sampling period.
a+ is calculated and fed back to the transportation amount controller 14. The adjustment unit 14B receives the feedback of Wa+ and calculates the deviation δW1 such that δW+ = Ws Wa 1 . This δWl is shown in (b). When the deviation δW1 is determined, the adjustment unit 14B adjusts W in the calculation formula.
Reset. Resetting is W−W+δW+=Ws+δW
Calculated according to 1. The function generator 15 receives this W and determines the corresponding booster gas amount Fsz as shown in (c). By setting this Fe2 to the second control point 1, the booster gas amount controller 19 performs booster gas amount control.

If such a sequence is repeated, the second deviation δW
2 becomes smaller as shown in (d), and the third point Fs3 is obtained. In this way, finally l Ws
-wa Converges within ε.

As described above in detail, the response to a change in setting value or a change in disturbance takes a long time to converge and is not desirable.

The present description has been made in view of these points, and its purpose is to realize a process control device with improved responsiveness.

(Means for Solving the Problems) This explanation for solving the problems described above is based on the following explanation: When a function that approximates process characteristics is included in a control loop, the coefficients included in this approximation function are changed during normal operation. It is characterized by being configured to increase the degree of function approximation by making learning corrections.

(Operation) When the control loop includes a function that approximates process characteristics, the present invention learns and modifies the coefficients of the function during operation.

First, the operating principle of the device of the present invention will be explained.

The case of powder flow rate control mentioned above will be explained as an example. Control calculated by function generator 15 in the figure! The relational expression between am and the manipulated variable, % (%) (1), can be given as follows from past operating results and physical knowledge.

Fs = s'W) 2+A-to'W
-(2) However, A-(PT Pa ) (PT +PB +2
) /2' Here, Kl ' HK2 ' is a coefficient and corresponds to a pressure loss coefficient due to powder flow and a pressure loss coefficient due to gas flow, respectively.

The problem with equation (2) is that it is actually very difficult to accurately predict and calculate the coefficients 1' and 2'.

Even if Kl' and 2' were determined experimentally, there is a problem in that these coefficients change depending on the powder properties, clogging, wear, etc. of the conduit components. Therefore, it can be expected that sufficiently good control performance can be obtained by constantly correcting the values of these coefficients using normal operation data and always increasing the degree of approximation of equation (2) within a certain value.

The concept of coefficient correction is as follows. Now, assuming that the coefficients include an error of 21 in L,
As can be easily understood from the explanation of the operation of the controller shown in FIG. 5, the effects of these errors appear intensively in the output value. The output of the transport volume controller 14, that is, the human power W of the function generator 150, is W = W s + δW1 + δW2 + -
δW n=Ws+Δ...(
3). Here, Δ−δW1+δW2+...δWn...(4
), and the error included in the coefficient 1'lK2' is concentrated in Δ. Therefore, corrections are made online during normal operation so that K1'+ and 2' become Δ→0, and that Δgorge O is maintained regardless of any changes in transportation volume settings or changes in pressure. It is intended to do so.

Next, the rationale for coefficient correction will be explained.

Now, if we rewrite equation (2) into a form that includes an error, we get the following equation.

Fs = EWT - (K++ Δ + > (Ws + Δ)...
・(5) However, B= (Kl +ΔKt)2 (Ws+△)2C−(P
T Pa )(PT +Pa +2)/(K2+Δ
2) Here, K1. 2 is the true value of the coefficient, + is the coefficient error, and 2 is the coefficient error. Here, if we transform equation (5) and rewrite it to obtain Δ, we get Δ=(D/E)−Ws...(6
) However, D = (PT - Pa ) (Pl - 10 Pa + 2) F
s2 (2 for K2+Δ) E=2Fs (K1+ΔKt) (2 for K2+Δ) Here, the problem is how to calculate Δ based on the only information Δ.
1=Δ and 2=0.

If we set Δ−〇 in equation (6), then 2△ + + (Kx + (Fs /2Ws > )
Δ to 2 + ΔKt ・Δ to 2 − (P7 Ps) (PT Pa + 2>/2
Fs Ws-(FS K2 /2Ws)
KI K2... (7) The right side of equation (7) becomes O under the condition that it does not include an error of Δ=0. From this, equation (7) becomes [Δ1+(K
l + (Fs / 2Ws) ) ] (Δ 2 + 2) = (Kl + (Fs / 2Ws)) K2 ”
' It can be transformed as shown in (8). When this is represented in a diagram, a hyperbolic coefficient error curve giving Δ-0 as shown in FIG. 6 is obtained. In the figure, Δ is set at 1 on the horizontal axis.

The hyperbola in the figure is a curve showing Δ=O, and is always at the origin (0,
0). This curve is divided into a Δ>0 region and a Δ<0 region. 1 in Δ. The possible range of 2 for Δ is Ki′
Under the condition that 2' is never negative, it falls within the shaded range in the figure. If the set value or pressure changes, the shape of this curve will naturally change, but it will always pass through the origin. Now, even if we correct 2 to ΔKt + Δ in some way and it becomes 8-0, the origin (that is, the coefficient error Δ +
Unless +Δ and 2 both reach 0), Δ changes to either positive or negative due to a change in setting value or a change in pressure. This Δ change information can be used as a means for reaching the origin. Finally, for example, by bringing the operating state at point X in the figure onto a hyperbola passing through the origin shown in the figure (particularly at the origin), optimal control can be performed.

Next, one method for making the operating point reach the origin (coefficient error Δ=1=Δ=2=0) will be described below.

Since the control device does not know the true value of the coefficient + or 2, the shape of the curve Δ=O (hyperbola here),
Not only do we not know the position, but we also cannot know where in FIG. 6 the operating point is located. Therefore, in order to converge to the origin, trial and error is inevitable. Here, the process of converging the operating point to the origin will be referred to as parameter tuning. It is assumed that the initial value of the parameter deconing period always starts from the first quadrant of FIG.

FIG. 7 is a diagram showing the process of parameter tuning. In the figure, f, , fl, f, are all curves of Δ−〇. Hereinafter, explanation will be given along with this figure.

(Step 1) It is assumed that the 8-0 curve at this time is fl in the figure.

The X point in the first quadrant is set as the initial value of the operation.

In this case, as shown in the figure, 2' is decreased until Δ=0, and the operating point approaches the point on the [1 curve].

(Step 2) Assume that the characteristic of the curve 8-0 changes from [1] to [fl] due to a change in setting value, a change in pressure, etc. In this case Δ〈
O, so in order to increase △ in the positive direction, change 1' to △-
Decrease until it reaches 0.

(Step 3) Suppose that the Δ-0 curve changes to fl again. Since Δ>O, 2' is increased by =15- in order to decrease Δ.

(Step 4) Assume that the curve Δ=O changes to f3 again.

In this case, since Δ>0, 2' is further increased.

(Step 5) Suppose that the 8-0 curve changes to fl again. Since Δ<<O, such an operation of decreasing + is repeated until step n.

Step n) Even if the curve of Δ-〇 changes, the change in Δ is minute. That is, it is determined that parameter tuning is complete when 1Δ1≦ε (ε is a small positive constant).

Based on the above parameter tuning sequence, the operating rules for error convergence are as follows except for step 1.

■In the case of Δ<−ε, decrease l until 1Δ1≦ε■Δ>O (in practice, Δ>ε
), by the time 1△1≦ε? ■ If Δ=0 (actually 1Δ1≦ε), do not operate.

...(9) Using normal operation results according to the rules above,
The coefficients can always be converged near the true value. -After convergence, to prevent the coefficient from being destroyed due to an accident,
Turn off the parameter tuning process.

(Example) Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the present invention. The system shown in the figure is an improved version of the powder transportation amount control system shown in FIG. 4, which allows parameter tuning. Components that are the same as those in FIG. 4 are designated by the same reference numerals. In the figure, 31 is the transportation amount target value Ws
The constant setting section 32 which outputs the constant setting section 31 as the Sv value is a transportation amount controller that receives the output of the constant setting section 31 as the Sv value and the output Wa of the calculator 25 as the PV value. It is assumed that the transport amount controller 32 is always connected in cascade and performs sample value control.

33 indicates the output of the constant setting section 31 as the sv value, and the transportation amount controller 3
The coefficient K(K+'
, and 2'). The coefficient controller 33 is a geared sample value I controller with a gap of ±ε as shown in FIG. 2, and stops the adjustment operation when 1Δ1≦ε. Its output MV is switch 3
It's in 4. The switch 34 is configured to: 1) a sequence for on/off scanning the coefficient controller 33 in synchronization with the start/stop of parameter tuning, 2) a switching sequence for the switch, and 2) sample values for the transport amount controller 31 and the coefficient controller 33. It is connected to the sequence table in which the sequence to perform the conversion is written. 35 and 36 are respectively coefficient storage units, and by switching the switch 34, +'
, is increased or decreased by 2'. For example, when △〈0, ＋
'When the storage unit 35 is selected, the value of l is decreased, and △〉
When O, the 2' store 36 is selected and the value of 2' is incremented, each time read by the relationship generator 40.

37 and 38 are differentiators that calculate the difference before and after changing the values of Kl' and 2', respectively, and the coefficient changes obtained by these differentiators are 1' in δ, respectively.

δ becomes 2' and enters the subsequent arithmetic circuit 39. The arithmetic circuit 39 calculates δ'-f (δK + ' + 2' to δ)...(
10) is performed. This δ′ is + ',
δ' is an output compensation value based on a change of 2', and this δ' is inputted to the transportation amount controller 32 as the output compensation value VN.

40 has a coefficient of 11 in addition to the above-mentioned W, PT, and Pe,
, l, performs the calculation shown in equation (2) and outputs the Sv value Fs of the booster gas amount. The transport amount controller 32, the function generator 40, and the coefficient controller 33 are designed to synchronize sample value control with each other.

The operation of the system configured as described above will be explained as follows.

First, the processing order of the system shown in the figure can be roughly divided into the following order: transportation amount controller 32 → function generator 40 → coefficient controller 33 → booster gas amount controller 19 →. Parameter tuning is started and executed according to the sequence table connected to the switch 34.

The transportation amount controller 32 receives the set value Ws and pv*wa, performs an integral adjustment operation, and outputs the result as an MV value W. This W enters the function generator 40, and the function generator 4
0 performs the calculation shown in equation (2) and outputs the booster gas amount target value Fs. Booster gas amount controller 19
controls the flow rate of the booster gas using this Fs as the target value.

On the other hand, the coefficient controller 33 outputs the output Ws@S of the constant setter 31.
The output MV value W of the transport amount controller 32 is received as the PV value. As is clear from equation (3), this W value is W
-Ws+Δ, which is the target value Ws plus Δ. Therefore, the coefficient controller 33 calculates W - W s from the Sv value and the PV value, calculates Δ, and outputs it as the Mv value.

When the switch 34 receives this MV value Δ, it determines whether it is positive or negative based on the sequence table and switches. If Δ<0, a signal is connected to the coefficient storage 35 to decrease l according to the sequence shown in equation (9), and if Δ>O, 2' is connected to the coefficient storage 35. A signal is connected to the coefficient store 36 for incrementing. When Δ=0, no signal is connected.

When the coefficient store 35 is selected, Kr' is made smaller than the previous value by the coefficient controller, this value enters the function generator 40 and the differencer 37, and the coefficient store 36 is selected. In this case, 2', whose value is increased from the previous value, enters the function generator 4o and the differencer 38.

As a result, an output compensation amount corresponding to the change in K1'lK2' is given to the transportation volume controller 32 as a VN value, and
The function generator 40 generates a new value <Fs based on 1' or 2'.
The value is output and sent to the booster gas amount adjustment 19. Booster gas amount adjustment 19 adjusts 21 to connect pipeline a2
Adjust the amount of booster gas flowing through. As a result, the powder flow rate flowing through the pipeline IIs changes, which appears as a change in the Wa value. Based on these amounts of change, the output W of the transportation amount controller 32 changes at the next sampling period. This change is transmitted to the coefficient controller 33, and Δ changes. Depending on whether this Δ is positive, negative, or O, the coefficient change selection sequence described above is entered.

As this learning operation is repeated, +' and 2' gradually change, approaching the true values of + and 2. Then, in the final stage, 1Δ1≦ε, and the coefficient controller 33 does not perform any adjustment operation, and ideal control is performed.

Although the above explanation has been given using powder flow rate control as an example, the same can be applied to freeness control as well. Furthermore, the present invention is equally applicable to any type of process control in which a control loop includes a function that approximates process characteristics.

(Effects of the Invention) As explained in detail above, according to the present invention, when a control loop includes a function that approximates process characteristics, the coefficients of this function are learnedly corrected during normal operation, and the function is The degree of approximation can be increased.

Therefore, according to the present invention, controllability is improved and the practical effects are extremely large.

[Brief explanation of drawings]

FIG. 1 is a configuration block diagram showing one embodiment of the present invention, and FIG.
The figure shows the gap characteristics of the controller, Figure 3 (A) shows an example of the conventional configuration of a freeness control system, (Japanese) shows the relationship between pulp flow rate and finer power, and Figure 4 (I) shows the relationship between pulp flow rate and finer power. ) is a diagram showing an example of the conventional configuration of a powder transportation amount control system, (b) is a diagram showing the relationship between the booster gas gap and the powder transportation amount, and FIG. 5 is a diagram showing the operation of the transportation amount controller. FIG. 6 is a diagram showing a coefficient error curve, and FIG. 7 is a diagram showing a parameter tuning process. 1...Refiner 2...Power controller 3...
Freeness controller 4.15, 4.0... Function generator 5.20... Flow rate detector 6... Freeness detector 7... Power meter 8.22... Indicator 11.
...Tank 12...Pressure regulator 13.21.
...Control valve 14...Transport amount controller 16.17...
Pressure detector 18...Transportation blow-in light 19...Booster gas amount controller 3A, 14A...Constant setting section 3B, 14-B...Adjustment section 31...Constant setting device 32... Transportation amount controller 33
...Coefficient controller 34...Switcher 35.36...
・Coefficient storage 37.38...Differentiator 39...Arithmetic unit 4□~43
... Pipeline patent applicant Yokogawa Hokushin Electric Co., Ltd. (A) (B) Pulp flow rate Q 5゛Flow rate detector 6; Freeness detector 7; --1----- 14・W5 CoSV Transport amount JL-1111 -m-25F
5 = g (W, PI- 24)
. Pr/. Figure 4 (b) i3 9 ff 13.21 ; Kunsetsuben

Claims (1)

[Claims] A process characterized in that when a control loop includes a function that approximates process characteristics, the coefficients included in this approximation function are learnedly modified during normal operation to increase the degree of function approximation. Control device.