JPH01111088A - Thickness profile control method - Google Patents

Abstract

PURPOSE: To provide a method for controlling the thickness profile of a to-be- produced sheet based on a special arithmetic result so as to render the convergence tendency of the control favorable at its start-up. CONSTITUTION: This method comprises the following procedure: when the thickness profile in the cross direction of a sheet afforded after being issued from the slice following regulating slice gap with many slice bolts is to be controlled, the thickness deviation difference between sampling operations at the present time and the last time and that between sampling operations at the last time and the preceding last time are determined until the standard deviation of thickness profile deviation comes to a specified value or smaller at least from the beginning of the control operation, and an arithmetic operation is carried out by adding a magnitude proportional to the above difference to a magnitude of the control.

Description

[Detailed Description of the Invention] <Industrial Application Field> The present invention relates to a control method for uniformly controlling the thickness of a film, etc., and in particular to a method for controlling the thickness of a film, etc., which is suitable for controlling the absolute dry basis weight profile in the papermaking process. The present invention relates to a profile control method.

<Prior art> Figure 5 shows a schematic diagram of a paper making device. In this figure, pulp as a raw material is put into a seed box 1, its flow rate is adjusted by a seed port valve 2, and the pulp is supplied to an inlet box 4 by a fan pump 3. The pulp supplied to the inlet box 4 is discharged in the form of a sheet from the gap in the slicing lip 5 to the wire bart 6. The gap profile of the slicing lip 5 is adjusted by a plurality of slicing bolts 7.

The pulp discharged in the form of a sheet onto the wire bar 6 has its moisture removed by a press 8, is heated and dried with steam by a dryer 9, and is then passed through a calender 10 and wound onto a reel 11 to become a product. 12 is a white water silo, which collects the pulp separated from the sheet by the wire bart 6 and returns it to the inlet box 4. The thickness of the paper is controlled by the basis weight, which is the weight per square meter, and the basis weight and moisture profile are detected by the B/M detector 13. Overall control is executed by the control section 14. The control unit 14 inputs the basis weight measurement value of the B/M detector 13, calculates control la1, and controls the slice bolt 7. It also controls the seed opening valve 2 to adjust the flow of pulp, and also controls the speed of the wire part 6 and the steam pressure of the dryer 9.

FIG. 6 shows a part of the configuration of the control section 14. In FIG. 6, the gap opening degree signal of the slice lip 5 and the B/M detector 1
The tsubo m signal outputted from No. 3 is converted into a digital signal by the A/D converter 15, and its profile is displayed on the display 17 via the I10 controller 16. The basis weight signal is also input to the flow filter 18 and filtered,
The signal is input to the PI control calculation section 19. Pr11m calculation unit 1
9 performs proportional and integral calculations on the deviation from the separately input target value, and outputs the results to the distribution control calculation section 20. On the other hand, the output of the flow filter 18 is input to the distribution coefficient operator 21. The distribution coefficient 8Ill calculation unit 21 calculates a distribution coefficient and outputs it to the distribution control calculation unit 20. Distribution control calculation unit 20
calculates the distribution of the output of the PI-based m-taku section 19 based on the distribution coefficient. This calculation result is converted into an analog signal by the D/A converter 22 and output as a control signal for the slice bolt 7.

Next, a method of calculating the control signal will be explained based on FIG. 7.

When the slice bolt 7 is operated, not only the position of the slice bolt 7 but also the gap opening of the slice lip 5 around it changes, so the control signal must be calculated taking into account the influence of the surroundings. In FIG. 7, the horizontal axis 23
indicates the width direction of the paper, Δ mark 24 indicates the measurement point of tsubom, ○ mark 25
represents the position of the slice bolt, and curve 26 represents the deviation between the measured value and the target value of basis weight. At the basis weight measurement point 24, 1-
The slice bolt positions 25 are numbered sequentially from 5 to l+5, and from i-2 to 1+2. Point i-2, knee 1, il i+1, i+2 are points 1-4, l-, respectively
2,! ,! +2,! It matches +4. One slice bolt corresponds to its position and three tsubo m measurement points on the left and right.

The PI control calculation unit 19 controls sm for the slice bolt i.
ut is calculated and output based on equation (1).

LJt = -KpΔBt- to IΔ[3tdt-...(
1) Kp: Proportional gain KI: Integral gain ΔBl: Deviation ΔBL between the measured value and target value of tsubom is (2)
Find it using the formula.

ΔBt = (bl-+ +bJ1+bj++)/
3-btc・・・・・・(2) bj! -+~b1 Ya I: Measured value of basis weight at points 1-1~l+1 btc Target value of basis weight at two points 1 As mentioned above, when the slice bolt 7 is operated, the effect is on the position of the surrounding slice bolts. Since it affects the basis weight, P
IilJIj performance W part 19 output U (with slice bolt i
cannot be controlled. Therefore, U(
The value obtained by multiplying by the distribution coefficient is set as the control amount of the slice bolt 7, and the distribution coefficient calculator 21 calculates the distribution coefficients αLL2 to αL+2 based on equation (3).

αt −2= bm(−2/ P αt −+ −bml −1/P αt −bm/p αt ++ =bmL+I /P 01i 42 =bm(+2 /P ・・・・・・・・・・・・(3) However, bm(-2= (bj!-5+bj!-4+bJ!
-3)/3 bmL-+ = (bj -3+bl -2+bll
−+ )/3 bmL−−(bl −+ +bl +bl +1 )/
3・・・・・・・・・(4) bm(+1- <bl ++ +btz +2 +b
J! +3 )/3 bm(+2- (b4+ +3 +bl +4+l)j
+5 )/3 P-1bmi +2 1+lbm(++ ++lb
ml l+lbmt -1l+lbm (expressed as -21. Distribution coefficient αi-2 to α (the sign of +2 is determined based on the sign of the measured value of basis weight and the ratio of areas s1 and S2 in FIG. 7. Distribution vlj tXl I n group 20 is the distribution coefficient αt-2 to αL +2 and P■ control calculation unit 1
9 or calculates the slice bolt i control ffi (Jt) from -2 to ul+2 based on equation (5) and outputs it to the D/A converter 22.

Ut −α i + 2 'Jt +2
+ α t ++ ut ++
+αt ut +at −+ uL−+ +(X(−
2ut -2 (5) In this way, the interference of the slice bolt can be automatically corrected. Note that the sampling period of the 87M detector 13 is about 40 seconds, and the control period is set to be longer than this sampling period. In addition, in actual control, for example, the number of slice bolts is 4 for 360 measurement points.
0 point, and nine measurement points correspond to one slice bolt.

<Problems to be solved by the invention> However, such a thickness profile control method is
Since only proportional and integral calculations were performed in IυIII Enfeng Gun 19, there was a drawback that it affected connectivity. Especially when starting up the control, the variation of 1g difference in thickness is large, and the proportional
It has been impossible to follow this variation using only integral calculations, and it has been impossible to absorb the variation. Differential control is generally performed to improve responsiveness, but in papermaking processes and the like, the measurement period for measuring the thickness profile is long, making it impossible to perform differential control.

Therefore, at startup, the operator had to look at the variations in the thickness profile displayed on the display section 17 and manually control the thickness profile to absorb the variations, which placed a heavy burden on the operator. This method has the drawback of poor convergence due to the inability to perform accurate control.

<Objective of the Invention> An object of the present invention is to provide a thickness control method that allows control with good convergence at the start-up of control.

Means for Solving the Problems〉 In order to solve the problems mentioned above, the present invention has a method of discharging raw materials such as valves from a slicing lip whose gap profile can be changed by adjusting a plurality of slicing bolts. Thickness profile control is performed by measuring the thickness profile of this sheet, calculating the control amount of the slice bolt from the Q difference between the measured value and the target value, and operating the slice bolt using this control amount. In the method, at least from the start of control until the standard deviation of the deviation of the thickness profile becomes equal to or less than a predetermined value, The difference in thickness deviation is determined, and an amount proportional to these differences is added to the control Wm for calculation.

<Embodiment> FIG. 1 shows the configuration of an apparatus for realizing the thickness profile control method according to the present invention. Note that the same elements as in FIG. 6 are designated by the same reference numerals, and explanations thereof will be omitted.

In FIG. 1, numeral 30 is a control calculation section, into which the output of the flow direction filter 18 is input. Further, the output is output to the distribution control calculation section 20. A target setting value for thickness and a switching target value AXL indicating the timing for switching the control operation are input to the control calculation unit 30. Control force section 30
has almost the same function as the PI control section 11 in FIG. 6, but additionally performs special control to be described later at the initial stage of control.

Next, the operation of this embodiment will be explained based on FIG. Figure 2 shows the control status, and the horizontal axis is time;
The vertical axis represents the basis weight of each slice bolt. In addition, T1 to TTL+2 are the sampling times of the 87M detector 13, btc is the target setting value, and actual 1s32 is the average basis weight at the 111th slice bolt position shown by the above equation (4) at the sampling times T+ to TTL+2. The point 1131 shows the change in the value bmL, and the true change in this average one shift. The measured value of basis weight is obtained at sampling time T+-
Since it can be determined only at T1+2, these measured values are shown by connecting them with a straight line. It is assumed that control begins at time O. When the calculation formula (1) for proportional-integral control is converted to a discrete system, the recurrence formula is uL (n+1>-LJt (n) -Kp (TI
ΔBt(n+1)−ΔB<(n))・----
It is expressed as (6). Here, ΔBt (n)
represents the deviation between the average basis weight value and the target setting value shown in equation (2), and n represents the order in time series. Also, T.I.
= (Kp-Kx)/Kp. As described above, when the deviation is large as shown in FIG. 2, the calculation using equation (6) alone results in poor responsiveness and a delay in reaching the target set value. Therefore, a correction term corresponding to the change in deviation is added to equation (6).

Differential control is generally used as a method to improve responsiveness, but since the sampling period of the B/M detector 13 is as long as approximately 40 seconds, the control period cannot be shortened. I can't do it. Therefore, a term proportional to the value obtained by subtracting the difference between the previous deviation and the previous deviation from the difference between the deviation at the current sampling time and the previous sampling time is added to the control speech formula. That is, the sampling times of this time, last time, and tti time are respectively T TL + 1, dπ, and d'n.
−+, then the recurrence formula of the control lI operation is ut (n+1)−ut (n>−Kp −(TIΔ
Bt (n+1)−ΔSt (n)) −Ko
(m(n) −m (n-1) )・・・・・・・・・
Let it be <7). However, m (n) = [(bml (TTL
+I)bLc(n+1))-(bm
t(TTL) E)+c(n))]/Ts m (n -1) = [(bmL(TTL) -L
lc(n))−(bmt(TTL−+)
btc (nl)]/Ts TX'-(Kp'-KX-)/Kp-. Ts
represents the sampling period of the B/M detection Pi 13, Ko is a constant, and Kp' and KI- are proportional and integral gains different from proportional-integral control. Normally, as shown in Fig. 2, the target set value btC is a constant at the initial stage of control, so m (n) = [(bmt (TTL +)
bm((Tu))/Ts m (n1)-[(bm((TTL)-bm
t(TTL-+))/Ts. That is, m(n) is the sampling time TT1
The slope of the line segment connecting +I and the measured value at TTL, m
(n-1') represents the slope of the line segment connecting the sampling time TTL and the measured value at T1-1.

The control extractor 30 performs the control shown in equation (7) from the start of the control. The cycle of this control calculation is, for example, 1 second,
The value is set to be sufficiently shorter than the sampling period of the B/M detector 13. At the same time, the standard deviation σ of the deviation of the profile of the basis weight measured by the B/M detector 13 at each sampling time is compared with the switching target value AIL input in advance, and twice the standard deviation is AIL. When it becomes smaller, as shown in FIG. 2, the control shown in equation (7) is switched to the normal proportional-integral control shown in equation (6). In this case, the control period is made longer than the sampling period. Note that the output U of the control section 30 is inputted to the distribution control calculation section 20 as shown in FIG.

FIG. 3 shows a flowchart showing the procedure of this control method. This series of flows is repeatedly executed in each control cycle. First, the basis weight profile is measured and filtered, and the average basis weight corresponding to each slice port is determined based on equation (4).

Next, it is determined whether or not it is time to start up the control, and if it is at the time of startup, the proportional and integral gains Kp-1KX-, constant KO and switching target 1a stored in advance are determined.
Read out Ax L. Next, read out the control weight U (N g difference Δ87.average tsubo ffi b m t and the average tsubo l b m t two sampling periods ago).Next, calculate the control ff1Ll according to equation (7). , the distribution coefficient is calculated according to equation (3), and the control tlffiUL to be output to the slice bolt is calculated and output from equation (5).Then, the standard deviation of the deviation of the tsubo m profile is calculated, and twice that is the switching target value. Check whether it is smaller than AIL. If it is not smaller, read UtZΔat and bml again and operate the control jffi. This period is, for example, 1.
The period is as short as seconds. Although the measured value of basis weight does not correspond to this short control period, it does not interfere with the control because of the long-period disturbance in the flow direction and the large time constant of the system. When twice the standard deviation becomes smaller than AXL, the proportional gain KP% integral gain KI for proportional integral control is input. Next, read the necessary data and control Wl according to formula (6).
ut is calculated by vJ, the distribution coefficient is determined, and the controlled product of the sliced bolt is calculated and output using equation (5). That is, in normal proportional-integral control, control is performed with a control period longer than the sampling period of B/M detection 2113, and at the start of control, finer control is performed within this control period. At this time, control m corresponding to the change in deviation tll
By adding it to lfi, control similar to differential control is performed to improve responsiveness.

Note that υJli! in equation (7)! It is better to make the proportional gain Kp' and the integral gain KX' larger than the proportional gain Kp' and the integral gain KX' when carrying out the [proportional-integral control]. The reason for this is that proportional-integral control is in equilibrium mode, so the deviation ΔB+ does not become that large, and the system becomes more stable if the feedback gain is increased when the system characteristics or operating conditions change during steady state. It is for the sake of becoming. Further, the constant Ko needs to be determined so that the slice opening is within limits. For example, the variation in thickness deviation is 4Q/m2
, Ko is 0.5, and the numbering period Ts is 40 seconds, then Uj is 0.05 mm, but this is inappropriate because the opening limit of the slice bolt is usually about 0.02 mm. In such a case, Ko is set to about 0.1.

FIG. 4 is a diagram schematically showing the effects of this embodiment. The horizontal axis is time, and control starts from zero. Further, the vertical axis represents twice the standard deviation σ of the deviation of the profile of tsubo m. 33 is υ1tllll according to the present invention, and 3
4 is conventional manual control. In the control method according to the present invention, the standard deviation σ decreases significantly, and at time
The value of σ becomes smaller than the switching target value AXL, and the control is switched to normal proportional-integral control. At time T3, the standard deviation σ settles down to a stable constant. On the other hand, in the conventional control method, the standard deviation σ does not easily decrease, and falls below the switching target feAML in time T2, which is longer than T1, and shifts to normal proportional-integral control. The standard deviation σ becomes stable at T4, which is much later than T3.

In this embodiment, proportional-integral control is performed when twice the standard deviation σ falls below the switching target value ΔI[, but the same control may be continued even after the deviation falls. In this case, proportional gain Kp-1 integral gain K
Controllability is improved by changing the value of the x-1 constant Ko.

In addition, in this embodiment, control switching is determined based only on the standard deviation of the thickness profile deviation, but in addition to this criterion, the maximum deviation of the thickness profile must be less than or equal to a predetermined value. It is also possible to add a condition for switching when the

Furthermore, although the paper-making process was explained in this example,
It can also be applied to other processes such as the manufacturing process of plastic films.

<Effects of the Invention> As specifically explained based on the embodiments above, in this invention, from the start of control until the standard deviation of thickness profinyl (g difference becomes less than a predetermined value), An amount proportional to the difference between the thickness deviation during sampling and the difference between the thickness deviation during the previous sampling and the previous sampling is added to the controlled product.Therefore, the thickness profile The variation in deviation converges quickly, and the start-up time can be determined.

Furthermore, since the conventional manual operation can be automatically controlled, the burden on the operator is reduced.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of the configuration of an apparatus for carrying out the thickness profile control method according to the present invention, FIG. 2 is a characteristic curve diagram showing an example of the control according to the present invention, and FIG. A flowchart showing an example of the control method, FIG. 4 is a characteristic curve diagram showing the effect, FIG. 5 is a configuration diagram showing the configuration of a paper making device, and FIG. 6 is a configuration of a m control device implementing the conventional control method. FIG. 7 is a block diagram showing an example of the thickness profile. 5...Slice lip, 7...Slice bolt, 1
3...8/M detector, 14...control unit, 18...
Flow direction filter, 20...Distribution υ" control calculation unit, 21
. . . Distribution coefficient calculation section, 30 . . . Control calculation section. Figure 4

Claims (1)

[Claims] The raw material is discharged from a slicing lip whose gap profile can be changed by a plurality of slicing bolts to form a sheet, the thickness profile of this sheet is sampled, and the deviation between the measured value and the target value is calculated proportionally. , in a thickness profile control method in which a control amount of the slice bolt is calculated by an integral calculation and the slice bolt is operated using the control amount, the standard deviation of the deviation of the thickness profile is at least a predetermined value from the start of the control. Until the value falls below the value, calculate the difference in thickness deviation between the current and previous samplings, and the difference between the thickness deviations between the previous and previous samplings, and set an amount proportional to these differences as the control variable. A thickness profile control method characterized by calculation by addition.