JPH0429797B2 - - Google Patents

Description

Detailed Description of the Invention (Field of Industrial Application) The present invention relates to a device for controlling a paper machine, which is the final process in the paper manufacturing industry, and particularly relates to a paper machine control device in which the control characteristics in the paper width direction are improved using a Kalman filter. Regarding equipment.

(Prior Art) FIG. 4 is a schematic diagram of a paper machine control device known from Japanese Patent Publication No. 57-24435. In the figure,
1 is a seed box for storing raw material liquid containing pulp etc., 2 is a seed opening valve that controls the amount of raw material liquid supplied from seed box 1, 3 is a silo for storing white water, and 4 is a seed box for mixing raw material liquid and white water. The feeding fan pump 5 is an inlet box equipped with a plurality of discharge ports (slice lips) corresponding to the width of the paper, and stores the liquid sent from the fan pump 4. 6 is a wire part for extracting the pulp part of the liquid released from the slice lip, and 7 is a suction box for dehydrating the raw material on the wire part 6, which makes the white water mixed with the raw material fall into the silo 3 under negative pressure. 8 is a press part that thins the dehydrated raw material, 9 is a dryer that dries the paper to a preset moisture content, 10 is a calender that performs polishing, and 11 is a reel that winds up the finished paper. 12 is a basis weight moisture content detection unit (hereinafter referred to as BM meter); 13 is a control unit that controls the absolute dry amount and moisture content of paper detected by BM meter 12 to be equal to their respective set values; 2, the steam pressure of the dryer 9, and the papermaking speed.

The device configured in this manner operates as follows. The raw material liquid supplied from the seed box 1 passes through the seed port valve 2 and is sent to the inlet box 5 by the mixing fan pump 4 with white water supplied from the silo 3. The liquid discharged onto the wire part 6 from the inlet box 5 is dehydrated and dried while passing through the wire part 6, the suction box 7, the press part 8, the dryer 9, etc., and then wound onto the reel 11. The basis weight of the paper is kept constant by controlling the seed opening valve 2 and the like using a control signal from the control unit 13 based on the output of the BM meter 12.

FIG. 5 is an explanatory diagram of paper width direction profile control, where A is an explanatory diagram showing a developed paper from inlet box 5 to a total of 12 BM scanning lines, and B is an explanatory diagram showing a spread of paper from inlet box 5 to a total of 12 scanning lines of the BM, and B is an explanatory diagram showing a plurality of profiles (for example, 40) provided in the width direction of the paper. C is a distribution map of absolute dry basis weight measured by a BM meter having a plurality of observation points (for example, 360 points) in the width direction of the paper.

The degree of opening of each slice lip is controlled independently to keep the target absolute dry basis weight constant in the width direction.

FIG. 6 is a functional block diagram showing control of the paper width direction profile. In the figure, reference numeral 21 denotes an operating section for independently operating the opening degrees of a plurality of slice lips. Reference numeral 22 denotes a papermaking process, which includes multiple steps such as pressing, drying, and polishing. 23 is
The BM meter detection unit measures bone dry basis weight, moisture content, paper thickness, etc. while scanning at a rate of, for example, once every 40 seconds. 24 is a display section that displays, for example, the basis weight profile shown in FIG. 5C. 25 is BM meter detection section 2
A control signal U is sent to the operation unit 21 based on the data Y of No. 3.
The control unit outputs , allowing the operator to switch between automatic and manual control.

(Problems to be Solved by the Invention) FIG. 7 shows the absolute dry basis weight distribution when only a single slice lip is opened. With respect to the width w of the slice lip, the basis weight is also distributed in the area of adjacent slice lips. If the bone dry basis weight at a certain point deviates from the target, it is necessary to control a plurality of slice lips, and there is a problem in that the automatic control of the control unit 25 cannot achieve the same control as manual operation by a skilled operator.

Furthermore, as shown in the figure, there was a problem in that if a cross flow occurred due to some kind of disturbance, the automatic system could not follow it at all.

In order to solve these problems, the applicant has separately proposed a paper machine control device that identifies the characteristics of the paper machine in a steady state and stores and controls the operating procedures of the paper machine by an expert. There is.

Although this control device solves the above problems, the following problems remain.

Since it is fixed value control, it takes time to settle.

Since the dead time is large, constant value control must be adopted in order to obtain stable control characteristics.

Since we ignore the influence of external disturbances, we are directly affected by this.

Since the output measurement value is fed back as is, it is affected by detection noise.

The present invention has solved the above problems, and aims to provide a paper machine control device that dynamically predicts and estimates paper quality profile using a so-called Kalman filter and has a short settling time.

(Means for Solving the Problems) The present invention achieves the above object by providing a plurality of slice lips in the width direction of the paper and supplying an amount of raw material corresponding to the opening degree, and a plurality of slice lips supplied from the slice lips. It is equipped with a process section that dehydrates and dries raw materials to obtain a predetermined paper quality, and a detection section that decomposes and detects the paper quality of the paper sent from this process section in the paper width direction, and the paper quality profile detected by this detection section. In a paper machine that operates the opening degree of the slice lip based on the paper quality profile, an interference matrix between the paper quality profiles detected by the detection unit, an interference matrix between this paper quality profile and the slice lip operation amount, and an interference matrix between the paper quality profiles detected by the detection unit; an identification unit that estimates an interference matrix between the paper quality profile and the disturbance to determine the dynamic characteristics of the paper quality profile;
a command profile generating unit that generates a transient paper quality profile that is passed through when the paper machine is operated by a skilled person to reach a target paper quality profile; a transient paper quality profile that is generated by the command profile generating unit; A model matching section which inputs the target paper quality profile and the three interference matrices outputted by the identification section and calculates a gain when operating the slice lip opening, and inputs the paper quality profile detected by the detection section. a prediction estimating section that predicts the paper quality profile and disturbance ahead by the dead time from the slice rip to the detecting section; and a prediction estimating section that feeds back the paper quality profile predicted by the prediction estimating section and filters the predicted disturbance. The present invention is characterized in that it includes a control section that controls the slice rip by forwarding the yield and inputs these gains from the model matching section.

(Function) Each of the above-mentioned components has the following function. The identification unit stores the identified dynamic model of the paper machine control. The command profile generating section stores desirable transient paper quality profiles performed by an expert and transmits them in chronological order. The model matching section calculates the gain of the slice rip operation amount necessary to realize the transient paper quality profile in a dynamic state. The prediction estimation section compensates for the time delay due to the movement of paper from the slice lip to the detection section, and estimates the paper quality profile and disturbance as values generated at the slice lip. The control section operates the slice lip according to the deviation between the gain commanded by the model matching section and the target value determined by the prediction estimation section.

(Example) The present invention will be explained below using the drawings.

FIG. 1 is a block diagram showing an embodiment of the present invention. In FIG. 1, parts having the same functions as those in FIG. 6 are given the same reference numerals, and their explanations will be omitted.

In the figure, 31 indicates the opening degree of the slice lip;
The identification section identifies the relationship between the opening degree and the paper quality profile in the dynamic state.Identification here refers to finding a dynamic model when the paper machine operating conditions and paper quality are constant. For example ARMA
(Auto Regulating Moving Average) model is used. Reference numeral 32 denotes a command profile generating section that stores a transient paper quality profile that is passed through when a skilled person operates the paper machine to reach a target paper quality profile, and generates this transient paper quality profile in chronological order. 33 is a model matching unit that inputs the ARMA model of the paper machine, the transient paper quality profile, and the target paper quality profile, and calculates the gain of the slice rip operation amount necessary to realize the transient paper quality profile with the dynamic ARMA model. do. 34 is a prediction estimation unit which inputs the paper quality profile detected by the detection unit 23 and predicts the paper quality profile and disturbance at the slice rip by compensating for the dead time Td from the slice rip to the detection unit 23; Estimation refers to extrapolating the current value based on the measurement data measured at the current time t to obtain measurement data that does not include disturbances, and prediction refers to obtaining measurement data that does not include disturbances based on the measurement data measured at the current time t. It is obtained by extrapolating the data measured at the previous time t+Td. Reference numeral 35 denotes a control unit that operates the slice rip using the gain of the model matching unit 33 and the deviation between the paper quality profile predicted by the prediction estimation unit 34 and the target paper quality profile. For feed forward control.

Uop is the manipulated variable (identification manipulated variable) under conditions (identification conditions) that include the amount of information necessary to estimate the unknown parameters of the dynamic ARMA model, and Yop is the paper quality obtained in a steady state for the identified manipulated variable Uop. The profiles A^ and B^ are interference matrices representing the dynamic ARMA model of the paper machine, A^ is the interference matrix between the paper quality profiles detected by the detection unit 23 (hereinafter referred to as paper quality interference matrix), and B^ is the interference matrix representing the dynamic ARMA model of the paper machine. This is an interference matrix between the paper quality profile and the slice lip operation amount (hereinafter referred to as operation interference matrix). F^ is the disturbance interference matrix, Y^ is the measurement data predicted from the current measurement data Y, and r^ is the disturbance predicted from the current measurement data Y. Y^
is the target paper quality profile, Yc is the transient paper quality profile stored in the command profile generation section 32, and the M steps until the paper quality profile is stabilized by one slice rip operation are the steps required to reach the target paper quality profile. It is saved for N slice rip operations up to the end.

FIG. 2 is a specific example of the transient paper quality profile Yc stored in the command profile generation unit 32, in which (A) shows the amount by which a skilled operator (expert 5 operates the slice lip), and (B) shows the amount by which a skilled operator (expert 5) operates the slice lip. Correspondingly, the detection unit 2
The paper quality profiles detected in No. 3 are shown in chronological order. This amount of operation is taken, for example, when changing from low-quality paper with many variations to high-quality paper with uniform paper quality.

First, for the first slice rip operation U ¹ _i ,
The paper quality profile is established through M steps. Here, the step refers to, for example, the detection section 23.
It refers to one measurement of , and is performed every 40 seconds on a BM meter. Next, the expert performs the l-th slice rip operation ^l _i and repeats this operation N times to converge to the target paper quality profile.

In FIG. 1, the broken line indicates offline processing and the solid line indicates online processing, but all processing may be performed online.

The operation of the device configured in this manner will be described next. FIG. 3 is a flowchart showing the operation of the apparatus shown in FIG. 1, in which (A) represents an offline process for identifying the paper machine as an operating system, and (B) represents an online process for operating the paper machine.

The three transfer matrices A^, B^, F^ are identified online based on the measured data Yop for the open-loop manual slicing operation Uop of the paper machine.
Note that offline processing is performed because the amount of identification calculation is enormous, but online processing may also be used if there is a sufficient degree of slice rip operation and calculation can be performed quickly enough.

First, the identified manipulated variable Uop is inputted to the operating section 21 (S11), and the corresponding measurement data Yop in a steady state is read out from the detection section 23 (S12). here,
The components of Uop and Yop are expressed as follows.

U ^T _pp = (u ₁ , u ₂ ,..., u _r ) (1) Y ^T _pp = (y ₁ , y ₂ ,..., y _o ) (2) Here, r is the total number of slice lips, for example 40
There are three slice lips, and u _i (i=1, 2, . . . , r) is the opening degree of each slice lip. n is the total number of measurements by the detection unit 23, for example, 360 points, and y _i (i=1, 2,...,
n) is a paper quality measurement quantity such as basis weight at each measurement point.

The identification unit 31 uses the identification operation amount Uop and the paper quality profile Yop to obtain three transfer matrices A^, B^, and F^ of the ARMA model expressed by the following equations by the least squares method (S13).

A(q ^-1 )Y(t)=B(q ^-1 )U(t)+〓(t) (3) Here, A(q ^-1 ) and B(q ^-1 ) are defined as follows. be done.

A (q ^-1 ) = 1 0...0 0 1...0 : : : 0 0...1+ _n 〓 ^i=1-i _q・a ⁱ ₁₁ a ⁱ _i2 …a ⁱ _io a ⁱ ₂₁ a ⁱ ₂₂ …a ⁱ _2o : : : a ⁱ _o1 a ⁱ _o2 …a ⁱ _oo (4) B (q ^-1 )=q ^-Td B ^* (q ^-1 ) (5) B ^* (q ^-1 )= _l 〓 ⁱ⁼¹ q ^-i・b ⁱ ₁₁ b ⁱ ₁₂ …b ⁱ _1r b ⁱ ₂₁ b ⁱ ₂₂ …b ⁱ _2r : : : b ⁱ _o1 b ⁱ _o2 …b _or (6) Subscript of each element a ⁱ _jk , b ⁱ _jk To explain, a ⁱ _jk is the interference coefficient between the jth and kth paper quality profiles, and b ⁱ _jk is the interference coefficient between the jth paper quality profile and the kth slice rip operation amount. be.
q is an operator that shifts the time forward by one step, for example as shown in the following equation.

q・y _k (t)=y _k (t-1) (7) m is a symbol that displays data m steps before the current time t, l is a symbol that displays data l steps ago, and m, l can be arbitrarily selected.

〓(t) is an n-dimensional colored noise matrix including S-dimensional independent disturbances, and satisfies the following equation.

F(q ^-1 )〓(t)=W(t) (8)〓 ^T (t)=(r ₁ , r ₂ ,..., r ^S , r ^S+1 ,..., r ⁿ )(S
n) = (r ₁ , r ₂ ,…,. ^S , 0,…, 0) (9) Here, W(t) is the S-dimensional white noise matrix, F(q ^-1 )
is defined by the following equation.

The k-th component y _k (t) (k=1, 2, . . . , n) of equation (3) is expressed in the following vector format.

y _k (t)=φ ^T _K (t-1) Θ _k + r _k (t) (11) Here, φ ^T _K is a matrix consisting of measured data and manipulated variables, and Θ _k is an interference matrix A^, B A matrix consisting of coefficients of ^ satisfies the following equation.

φ ^T _K (t-1)] {-[y ₁ (t-1), y ₂ (t-1)
，
..., y _o (t-1)], ..., -[y, (t-m), y ₂ (t-m), ..., y _o (t-
m)]}〓 {[u ¹ (t-1-Td), u ₂ (t-1-Td),...,
u _r (t-r-Td)],..., [u ₁ (t-l-Td), u ₂ (t-l-Td),..., u _r
(t-l-Td)]} (12) Θ ^T _K = {[a ¹ _k1 , a ¹ _k2 , ..., a ¹ _ko ], ..., [a ^m _k1 , a ^m1 _k
2 ,…,
a ^m _ko 〕｝〓 {[b ¹ _k1 , b ¹ _k2 ,…, b ¹ _kr ],…, [b ^l _k1 , b ^l _k2 ,…
，
b ^l _kr ]} = {〓 ^T 〓〓 ^T } (13) Next, the colored noise r _k (t) in equation (11) is expressed in the following vector form.

r _k (t)=〓 ^T _k (t-1)〓 _k +W _k (t) (14) Here, 〓 ^T _k is a matrix composed of disturbances, 〓 _k is a matrix composed of coefficients of interference matrix F^ satisfies the following equation.

〓 ^T _k (t-1)] {[r ₁ (t-1), −r ₂ (t-2),
...,
−r _o (t−1)],…, [−r ₁ (t−s), −r ₂ (t−s),…, −r _o (t−
s)] (15) 〓 ^T _k = {[f ¹ _k1 , f ¹ _k2 ,..., f ¹ _ko ],..., [f ^S _k1 , f ^S _k2 ,..., f ^S _ko ]} (16) Time t Based on the data, the interference coefficient parameters Θ _k and 〓 _k with the manipulated variable y _k are estimated using the following recurrence formula. Note that when Θ^ _k is estimated, the interference matrices A^ and B^ are simultaneously determined.

Θ^ _o (t)=Θ^ _k (t-1) +Z(t-2)φ _k (t-1)/1+φ ^T / _k (t
−1) Z(t−2)φ _k (t−1)(y _k (t)−y^ _k (t)(17)
y^ _k (t)=φ ^T _k (t-1)Θ^ _k (t-1)+r^ _k (t) (18) r^ _k (t)=〓 ^T _k (t-1)〓^ _k (t-1) (19) Here, Z(t-1)=Z(t-2)-Z(t-2
) φ ^T / _k (t-1) Z (t-2) / 1 + φ ^T / _k (t-1
)Z(t-2)φ _k (t-1)(20) Also, the interference coefficient 〓 _k of the disturbance model is estimated by the following equation.

〓 _k (t)=〓 _k (t-1) +zf (t-2) ψ _k (t-1)/1+ψ ^T / _k (t
−1) Z _f (t−2) ψ _k (t−1)・[r^ _k (t)−ψ ^T _k (
t-1)〓＾ _k (t-1)〕(21) Here, r^ _k (t)=y _k (t)−φ ^T _k (t-1)Θ^ _k (t-1) (22 ) Z _f (t-1)=Z _f (t-2)-Z _f (t-2
)〓 ^T / _k (t-1)Z _f (t-2) /1+〓 ^T / _k (t-1
) Z _f (t-2) 〓 _k (t-1) (23) The initial conditions of equations (13), (20), (21), and (23) are as follows, respectively. .

Θ^ _k (-1) = Θ^ _k (0) = Θ^ ₀ Z (-1) = Z ₀ (>0) (24) (25) 〓^ _k (-1) = 〓 _k (0) = 〓＾ ₀ Z _f (−1)=Z _f0 (>0) (26) (27) Next, the model matching unit 33 calculates the interference matrices A^, B^ and the excess paper quality profile obtained by formula (17). Required slice rip operation amount U using Yc
(S14).

U=(B^ ^T B^) ^-1 B^ ^T A^Yc (28) Here, disturbances are ignored and F^=0.

Next, the model matching unit 33 determines the gain matrix 〓 from equation (28) (S15).

〓=(Φ ^T Φ) ^-1 Φ ^T U (28) Here, the gain matrix 〓 is calculated by the control unit 35 in equation (9).
Displays the gains K _P and K _I when performing PI control,
When expressed in terms of components, it becomes as shown in equation (10).

U=K _P ΔY−K _I ΣΔY (30) u ^j _i = _N 〓 ^K=1 {−kK ^jk _P Δy ⁱ _ik ^- (M) −K ^ik _IM 〓 ^l=1 Δy ⁱ _ik (l)} (31 ) Here, i is the slice lip number, j is the number of cycles, and k is the measurement point number, i = 1, 2,
..., r, j=1,2,...,N, k=1,2,...,
Consists of n.

K ^ik _P : P control gain between the i-th slice lip and the k-th measurement point.

K ^ik _I : I control gain between the i-th slice lip and the k-th measurement point.

y ^j _ik (M): deviation between the i-th slice lip and the k-th measurement point in the j-th round, M is the number of settling steps in one round of operation, and It's the same. Note that the deviation is the measured value y ^j _ik
and the target value _ik . _M 〓 ^l=1 Δy ^j _ik (l): Sum of the deviations between the i-th slice rip and the k-th measurement point in the j-th round, which is the cumulative deviation between the target value and the transient paper quality profile. ing.

When the gain matrix 〓 is expressed by equation (32) and the manipulated variable matrix U is expressed by equation (33), equation (30) can be expressed by equation (34).

〓 ^T = ([(K ¹¹ _P K ¹¹ _I ) (K ¹² _P K ¹² _I )…(K ¹ⁿ _P K ¹ⁿ
_I )] [(K ²¹ _P K ²¹ _I ) (K ²² _P K ²² _I )…(K ²ⁿ _P K ²ⁿ _I )]… [(K ^r1 _P K ^r1 _P ) (K ^r2 _P K ^r2 _I )…( K ^rn _P K ^rn _P )〕
)
(32) U ^T = {[U ¹ ₁ U ² ₂ …U ^N ₁ ] [U ¹ ₂ U ² ₂ … U ^N ₂ ] … [U ¹ _r U ² _r …U ^N _r ]} (33) U= Φ〓 (34) Here, the small matrix Φ _pq that constitutes Φ is expressed as (32), (33)
It is defined corresponding to the parentheses [ ] in the expression.

Φ＝Φ ₁₁ Φ ₁₂ …Φ _1o Φ ₂₁ Φ ₂₂ …Φ _2o : : : Φ _r1 Φ _r2 …Φ _ro (35) Here, P=1, 2,..., r Φ _pq =0 (for p≠q) In this way, the gain matrix 〓 corresponding to the expert's operation is determined off-line.

Next, the identified gain matrix 〓 is used to process the paper machine online. The prediction estimation unit 34 inputs the measurement data Y, first estimates a value that does not include disturbance, and then predicts a value after the dead time Td (S21).

First, in the estimation calculation, equation (3) is transformed as shown in the following equation using equations (11) and (14), and the operation amount Y and disturbance 〓 are estimated by the predictive estimator 34. Note that the number of expressions that are passed through during the transformation of the expression is enormous, so they will be omitted.

〓(t)=ΦΘ^(t-1), 〓^(t-1)〓(t-
1) +Γ・U(t-1−Td)+Gw・W(t) (37) Here, Γ is the coefficient matrix applied to the manipulated variable U, Gw
is a coefficient matrix related to disturbance, and 〓(t) is a state vector consisting of elements of paper quality profile measurement data Y, operation amount measurement data U, and disturbance time series data r, which satisfies the following equation.

〓(t) ^T 〕{[y ₁ (t),…, y _o (t)], [y ¹ (t-1),…, y _o (t-1)], …, [y ₁ (t -m+1),...,y _o (t-m+1)]}
〓 {[u ₁ (t-1-Td),..., u _r (t-1-Td)],
…, [u ₁ (t-l-Td),…, u _r (t-l-Td)]}
〓 {[r ₁ (t),...r _o (t)],..., [r ₁ (t-s+1),...
，
r _o (t−s+1)]} (38) Further, Φ is a matrix composed of the interference parameter vectors Φ^, 〓^ estimated by the identification unit 31.

Estimation of 〓(t) using the Kalman filter is performed using the following equation.

〓^(t)=(Θ^(t-1), 〓^(t-1), (t-
1)〓(t-1) +ΓU(t-1-Td)+〓 _k [Y _n (t-1)-H _n 〓
＾(t−1)〕 (39) Y _n (t)＝H _n 〓(t)＋V(t) (40) Here, 〓 With _k Kalman gain, set the filter gain so that the following formula holds. do.

Y^ _n (t-1)=Y _n (t-1) (41) V(t) is white noise with an average value of zero, and Hm is a coefficient matrix that satisfies the equation.

Next, a predictive calculation is performed. In order to compensate for the dead time Td that exists between the operation amount U and the paper quality profile Y, the control unit 3 calculates the predicted value Td ahead of the current time t.
This is to send it to 5th.

〓(t+Td 1 t)=Φ(Θ^(t+Td), 〓＾
(t+Td), Td)〓＾(t) +Γ・U(t−1) (43) However, since Θ^(t+Td), 〓＾(tTd) are estimated values of constant parameters, their setting values are used. Then, the following equation is obtained.

Θ^((t+Td)=Θ^ss (44) 〓^(t+Td)=〓^ss (45) ∴Φ(Θ^(t+Td), 〓^(t+Td), t+Td)=
Φss (46) Here, the subscript ss means the setting value.

The control unit 35 uses the gain matrix 〓 obtained by the model matching unit 33, the target paper quality profile, and the paper quality profile predicted by the prediction estimation unit 34 using equation (43) to calculate the operation amount U according to the following equation. Calculate (S22).

U(t)−〓 _P (Y^(t+Td)−(t+Td)) −K _I Σ(Y^(t+Td)−(t+Td))−K _F 〓＾
(t+Td) (47) The operating section 21 operates the slice lip according to the command from the control section 35 (S23). The process section 22 makes paper (S24), and the detection section 23 outputs the paper quality profile Y of this paper (S25). As needed,
This paper quality profile Y is displayed (S26).

In the above embodiment, all the deviations Δy _ik (k=1, 2,..., n) are fed back to each slice lip, but the feedback is given to each two slice lips that have a large influence adjacent to the slice lip in question. The corresponding deviation (about 40 points if n=360 and r=40) may be fed back. This has the advantage of shortening calculation time.

If the paper quality or operating conditions change, the interference matrix
The gain matrix 〓 can be determined by re-identifying A^, B^, and F^.

Furthermore, in this embodiment, the prediction estimator 34 is configured using a Kalman filter, but the detector 23
If the noise level is low, the predictive estimation unit 34 may be configured using a deterministic estimation method.

(Effects of the Invention) As explained above, the present invention has the following effects.

Since the control unit 35 performs control based on paper quality profile control by an expert, the reliability of the papermaking process is improved.

Since the dynamic system identification of the paper machine is performed, the settling time is short, without the long settling time that occurs in the case of steady system identification.

Since the prediction estimator 34 compensates for dead time, the controllability of the papermaking process is improved.

When the disturbance is estimated using a Kalman filter in the prediction estimator 35 as in the embodiment, the influence of noise can be minimized and the reliability of the papermaking process is improved.

Since disturbances to the process are reduced by feedforward, the influence of disturbances is reduced and the reliability of the papermaking process is improved.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a specific example of data used in the device shown in FIG. 1, and FIG. 3 is a flowchart showing the operation of the device shown in FIG. FIG. 4 is a schematic configuration diagram of a conventional device, FIG. 5 is an explanatory diagram of paper width direction profile control, FIG. 6 is a functional block diagram of paper width direction profile control, and FIG.
The figure is an explanatory diagram of the problem. 21...Operation section, 22...Process section, 23...
...Detection section, 31...Identification section, 32...Command profile generation section, 33...Model matching section, 3
4... Prediction estimation unit, 35... Control unit.

Claims (1)

[Claims] 1. A plurality of slice lips provided in the paper width direction and supplying an amount of raw material corresponding to the opening degree, and a process section that dehydrates and dries the raw material supplied from the slice lips to obtain a predetermined paper quality. and a detection unit that separates and detects the paper quality of the paper sent from the process unit in the paper width direction, and operates the opening degree of the slice lip based on the paper quality profile detected by the detection unit. In this step, an interference matrix between the paper quality profiles detected by the detection unit, an interference matrix between the paper quality profile and the slice rip operation amount, and an interference matrix between the paper quality profile and the disturbance are estimated, and the paper quality profile is estimated. an identification unit that determines the dynamic characteristics of the isle; a command profile generation unit that generates a transient paper quality profile that is passed through when the paper machine is operated by an expert to reach a target paper quality profile; a model matching unit that inputs the transient paper quality profile generated by the generation unit, the target paper quality profile, and three interference matrices output from the identification unit, and calculates a gain when operating the slice lip opening; a prediction estimating section that inputs the paper quality profile detected by the detecting section and predicts the paper quality profile and disturbance ahead by the dead time from the slice lip to the detecting section; and the paper quality predicted by the predictive estimating section. A paper machine control comprising: a control unit that feeds back a profile and feeds back predicted disturbances to control the slice rip, and inputs these gains from the model matching unit. Device.