U.S. patent application number 10/134768 was filed with the patent office on 2003-01-02 for methods and systems for controlling paper quality by adjusting fiber filter parameters.
This patent application is currently assigned to Invensys Systems, Inc.. Invention is credited to Fralic, Gregory R., Quinn, G. Holton III, Strand, William C..
Application Number | 20030000669 10/134768 |
Document ID | / |
Family ID | 27495121 |
Filed Date | 2003-01-02 |
United States Patent
Application |
20030000669 |
Kind Code |
A1 |
Strand, William C. ; et
al. |
January 2, 2003 |
Methods and systems for controlling paper quality by adjusting
fiber filter parameters
Abstract
Disclosed is a system that adjusts the operational parameters of
a fiber filter apparatus to keep measured properties indicative of
paper quality, such as the paper's porosity, within desired value
ranges. A feedback loop may be set up in which input properties are
read, such as the paper's porosity and the pollution content of the
filtrate leaving the fiber filter. These inputs go into a
predictive model that determines how to adjust the fiber filter's
operational parameters in order to move the measured paper quality
properties into their desired value ranges while accommodating
other constraints on the fiber filter's operation. The model looks
at positive and negative correlations between the operational
parameters and the measured inputs. Having processed the inputs and
predicted an outcome, the predictive model directs controllers to
change the operational parameters of the fiber filter. The input
properties are again read, and the feedback loop is repeated.
Inventors: |
Strand, William C.; (Moscow,
ID) ; Quinn, G. Holton III; (Moscow, ID) ;
Fralic, Gregory R.; (Waterloo, CA) |
Correspondence
Address: |
LEYDIG VOIT & MAYER, LTD
TWO PRUDENTIAL PLAZA, SUITE 4900
180 NORTH STETSON AVENUE
CHICAGO
IL
60601-6780
US
|
Assignee: |
Invensys Systems, Inc.
33 Commercial Street
Foxboro
MA
02035
|
Family ID: |
27495121 |
Appl. No.: |
10/134768 |
Filed: |
April 29, 2002 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
60290199 |
May 11, 2001 |
|
|
|
60291683 |
May 17, 2001 |
|
|
|
60293806 |
May 25, 2001 |
|
|
|
Current U.S.
Class: |
162/198 ; 162/20;
162/252; 162/49 |
Current CPC
Class: |
D21G 9/0018
20130101 |
Class at
Publication: |
162/198 ; 162/20;
162/49; 162/252 |
International
Class: |
D01B 001/14; D21F
007/00; D21F 007/06 |
Claims
We claim:
1. A method for controlling a property indicative of a quality of
paper by adjusting an operational parameter of a fiber filter
apparatus, the method comprising: measuring a value of the property
indicative of a quality of paper; predicting how adjusting the
operational parameter of the fiber filter apparatus will affect the
value of the property indicative of a quality of paper; and
adjusting, based on the predicting, the operational parameter of
the fiber filter apparatus to bring or to keep the value of the
property indicative of a quality of paper within a desired value
range.
2. The method of claim 1 wherein the fiber filter apparatus
recycles fibers received from a paper machine back to the paper
machine.
3. The method of claim 1 wherein the fiber filter apparatus is part
of a thermomechanical pulp system.
4. The method of claim 1 wherein the desired value range of the
property indicative of a quality of paper is a single value.
5. The method of claim 1 wherein the property indicative of a
quality of paper relates to a porosity of the paper.
6. The method of claim 1 wherein the operational parameter of the
fiber filter apparatus relates to a rotation rate of a filter in
the fiber filter apparatus.
7. The method of claim 1 wherein the operational parameter of the
fiber filter apparatus relates to a ratio of sweetener added to an
input to the fiber filter apparatus.
8. The method of claim 1 further comprising: choosing a maximum
adjustment rate for the operational parameter; wherein the
operational parameter is adjusted no faster than the maximum
adjustment rate.
9. The method of claim 1 further comprising: choosing a desired
value range of a constraint property, the constraint property
distinct from the property indicative of a quality of paper;
measuring the value of the constraint property; predicting how
adjusting the operational parameter of the fiber filter apparatus
will affect the value of the constraint property; and adjusting,
based on the predicting, the operational parameter of the fiber
filter apparatus to bring or to keep the value of the constraint
property within the desired value range of the constraint
property.
10. The method of claim 9 wherein the constraint property relates
to a drop leg consistency and the desired value range of the
constraint property is specified by a maximum desired
consistency.
11. The method of claim 9 wherein the constraint property relates
to a rotation rate of a filter in the fiber filter apparatus and
the desired value range of the constraint property is a minimum
desired rotation rate.
12. The method of claim 1 further comprising: predicting how
adjusting a second operational parameter of the fiber filter
apparatus will affect the value of the property indicative of a
quality of paper; and adjusting, based on the predicting, the
second operational parameter of the fiber filter apparatus to bring
or to keep the value of the property indicative of a quality of
paper within the desired value range.
13. The method of claim 1 further comprising: measuring the value
of an input property distinct from the property indicative of a
quality of paper; wherein the predicting is based in part on the
value of the input property.
14. The method of claim 13 wherein the input property is in the
set: motor load, chip moisture, temperature, pH level, flow
rate.
15. A computer-readable medium having instructions for performing a
method for controlling a property indicative of a quality of paper
by adjusting an operational parameter of a fiber filter apparatus,
the method comprising: measuring a value of the property indicative
of a quality of paper; predicting how adjusting the operational
parameter of the fiber filter apparatus will affect the value of
the property indicative of a quality of paper; and adjusting, based
on the predicting, the operational parameter of the fiber filter
apparatus to bring or to keep the value of the property indicative
of a quality of paper within a desired value range.
16. A system for controlling a property indicative of a quality of
paper, the system comprising: a fiber filter apparatus; a
controller that adjusts an operational parameter of the fiber
filter apparatus; a sensor that measures the value of the property
indicative of a quality of paper; a computing device; and a
predictive control application running on the computing device that
reads measurements produced by the sensor, predicts how adjusting
the operational parameter of the fiber filter apparatus will affect
the value of the property indicative of a quality of paper, and
directs the controller to adjust the operational parameter of the
fiber filter apparatus to bring or to keep the value of the
property indicative of a quality of paper within a desired value
range.
17. The system of claim 16 wherein the fiber filter apparatus
recycles fibers received from a paper machine back to the paper
machine.
18. The system of claim 16 wherein the fiber filter apparatus is
part of a thermomechanical pulp system.
19. The system of claim 16 wherein the desired value range of the
property indicative of a quality of paper is a single value.
20. The system of claim 16 wherein the property indicative of a
quality of paper relates to a porosity of the paper.
21. The system of claim 16 wherein the operational parameter of the
fiber filter apparatus relates to a rotation rate of a filter in
the fiber filter apparatus.
22. The system of claim 16 wherein the operational parameter of the
fiber filter apparatus relates to a ratio of sweetener added to an
input to the fiber filter apparatus.
23. The system of claim 16 wherein the predictive control
application directs the controller to adjust the operational
parameter of the fiber filter apparatus no faster than a desired
maximum adjustment rate.
24. The system of claim 16 further comprising: a second sensor that
measures the value of a constraint property, the constraint
property distinct from the property indicative of a quality of
paper; wherein the predictive control application reads
measurements produced by the second sensor, predicts how adjusting
the operational parameter of the fiber filter apparatus will affect
the value of the constraint property, and directs the controller to
adjust the operational parameter of the fiber filter apparatus to
bring or to keep the value of the constraint property within a
desired value range.
25. The system of claim 24 wherein the constraint property relates
to a drop leg consistency and the desired value range of the
constraint property is specified by a maximum desired
consistency.
26. The system of claim 24 wherein the constraint property relates
to a rotation rate of a filter in the fiber filter apparatus and
the desired value range of the constraint property is a minimum
desired rotation rate.
27. The system of claim 16 further comprising: a second controller
that varies a second operational parameter of the fiber filter
apparatus; wherein the predictive control application predicts how
adjusting the second operational parameter of the fiber filter
apparatus will affect the value of the property indicative of a
quality of paper and directs the second controller to adjust the
second operational parameter of the fiber filter apparatus to bring
or to keep the value of the property indicative of a quality of
paper within a desired value range.
28. The system of claim 16 further comprising: a second sensor that
measures the value of an input property distinct from the property
indicative of a quality of paper; wherein the predictive control
application reads measurements produced by the second sensor and
bases its prediction in part on the value of the input
property.
29. The system of claim 28 wherein the input property is in the
set: motor load, chip moisture, temperature, pH level, flow
rate.
30. A computer-readable medium containing instructions for
providing a system for controlling a property indicative of a
quality of paper, the system comprising: a fiber filter apparatus;
a controller that adjusts an operational parameter of the fiber
filter apparatus; a sensor that measures the value of the property
indicative of a quality of paper; a computing device; and a
predictive control application running on the computing device that
reads measurements produced by the sensor, predicts how adjusting
the operational parameter of the fiber filter apparatus will affect
the value of the property indicative of a quality of paper, and
directs the controller to adjust the operational parameter of the
fiber filter apparatus to bring or to keep the value of the
property indicative of a quality of paper within a desired value
range.
Description
CROSS-REFERENCE TO RELATED APPLICATION
[0001] The present application claims the benefit of United States
Provisional Patent Applications Serial No. 60/290,199, filed on May
11, 2001, No. 60/291,683, filed on May 17, 2001, and No.
60/293,806, filed on May 25, 2001.
TECHNICAL FIELD
[0002] The present invention relates generally to paper production,
and, more particularly, to systems and methods of controlling the
quality of paper produced in an automated paper manufacturing
process.
BACKGROUND OF THE INVENTION
[0003] Paper making is an extremely complicated process, but in its
essentials a thermomechanical pulp (TMP) system takes wood, pulps
it into wood fibers, and creates a slurry of the fibers and water.
A paper machine pours the slurry over a mesh screen (called a
"wire"). The fibers build up into a mat on the wire. When excess
water is removed, the mat becomes paper.
[0004] Throughout the paper-making process, filters are used to
control the amount and size of fibers in the slurry. For example,
the excess water that passes through the wire contains a
significant amount of unused fibers. To increase the efficiency of
the process, and to reduce pollution, these fibers are retrieved
and recycled. A waste fiber filter apparatus (called a "saveall")
retrieves the fibers. The saveall has a filter mesh through which
water containing fibers flows. Fibers collect on the mesh, just as
they do on the paper machine's wire. The fibers are harvested off
the mesh and then join the slurry flowing from the TMP system to
the paper machine. In a typical saveall, the mesh is in the form of
one or more filter disks that rotate while the water passes through
them. The speed of the rotation affects both the efficiency of
fiber recycling and the cleanliness of the water filtrate leaving
the saveall.
[0005] The quality of paper produced by the paper machine is
determined in large part by the characteristics of the slurry in
its input stream. Traditionally, the fiber filters in the saveall
and in the TMP system were adjusted only to optimize efficiency at
harvesting fibers and to reduce pollution. Other aspects of the
slurry-making process were adjusted to control paper quality.
SUMMARY OF THE INVENTION
[0006] The above problems and shortcomings, and others, are
addressed by the present invention, which can be understood by
referring to the specification, drawings, and claims. The present
invention adjusts the operation of a fiber filter in order to
regulate the consistency of paper as it is being produced, all the
while satisfying the original goals of efficiently harvesting
fibers and reducing pollution. The invention adjusts the fiber
filter's operational parameters to keep measured properties
indicative of paper quality, such as the paper's porosity, within
desired value ranges. The adjustments are made subject to other
constraints on the fiber filter's operation, such as its efficiency
in harvesting fibers and its control of the level of pollution in
the filtrate leaving the fiber filter.
[0007] In some embodiments, a feedback loop is set up in which
input properties are read, such as the paper's porosity and the
pollution content of the filtrate. These inputs go into a
predictive model that determines how to adjust the fiber filter's
operational parameters in order to move the measured paper quality
properties into their desired value ranges while accommodating the
other constraints on the fiber filter's operation. The model looks
at positive and negative correlations between the fiber filter's
operational parameters and the measured inputs. Associated with the
operational parameters may be maximum rates at which the parameters
should be changed, weighting factors saying which parameters are
preferentially changed, and the like. Having processed the inputs
and predicted an outcome, the predictive model directs controllers
to change the operational parameters of the fiber filter. The input
properties are again read, and the feedback loop is repeated.
BRIEF DESCRIPTION OF THE DRAWINGS
[0008] While the appended claims set forth the features of the
present invention with particularity, the invention, together with
its objects and advantages, may be best understood from the
following detailed description taken in conjunction with the
accompanying drawings of which:
[0009] FIG. 1 is a simplified schematic of the basic elements
involved in producing paper from wood pulp;
[0010] FIG. 2 adds to FIG. 1 sensors and controllers of some
embodiments of the present invention to show how logical components
of the invention interact;
[0011] FIG. 3 is a flowchart of an exemplary embodiment of the
invention as it operates to control paper quality; and
[0012] FIGS. 4a and 4b are a flowchart of an exemplary procedure
for using information about the operating characteristics of a
paper mill to create predictive models usable by the invention.
DETAILED DESCRIPTION OF THE INVENTION
[0013] Turning to the drawings, wherein like reference numerals
refer to like elements, the invention is illustrated as being
implemented in a suitable industrial environment. The following
description is based on embodiments of the invention and should not
be taken as limiting the invention with regard to alternative
embodiments that are not explicitly described herein.
[0014] In the description that follows, the invention is described
with reference to acts and symbolic representations of operations
that are performed by one or more computers, unless indicated
otherwise. As such, it will be understood that such acts and
operations, which are at times referred to as being
computer-executed, include the manipulation by the processing unit
of the computer of electrical signals representing data in a
structured form. This manipulation transforms the data or maintains
them at locations in the memory system of the computer, which
reconfigures or otherwise alters the operation of the computer in a
manner well understood by those skilled in the art. The data
structures where data are maintained are physical locations of the
memory that have particular properties defined by the format of the
data. However, while the invention is being described in the
foregoing context, it is not meant to be limiting as those of skill
in the art will appreciate that various of the acts and operations
described hereinafter may also be implemented in hardware.
I. The Saveall's Operation in a Paper Mill
[0015] In the simplified schematic of FIG. 1, the TMP system 100
produces a slurry of wood fibers in water. The process of taking
wood, pulping it into fibers, refining the fibers, and controlling
the parameters of the resulting slurry is well known in the art and
is incorporated in box 100. The slurry is piped (flow 102) to the
mixed stock tank 104. From the mixed stock tank, the slurry is sent
(106) to the paper machine 108. The slurry falls onto a constantly
moving mesh screen conveyor, or "wire," of the paper machine and
there forms a mat. Excess water with some fibers are removed from
the mat, and the mat becomes paper. The excess water and fibers
fall (flow 110) into the paper machine wire pit 112 and are pumped
(114) to the white water collection tank 116. From there, the water
and fibers are recycled back (118) into the TMP system.
[0016] This simplified process is enhanced through better control
of the fibers that pass from the paper machine wire pit 112 into
the white water collection tank 116. A waste fiber filter apparatus
called a "saveall" 120 is introduced to capture and recycle these
fibers, thus increasing the efficiency of the paper-making process
while decreasing the amount of pollution created. Some of the
output from the white water collection tank (flow 122) is diverted
to the saveall. The saveall has one or more rotating mesh screens
through which the water containing the fibers flows. Fibers collect
on the mesh screens, just as they do on the wire of the paper
machine 108. The fibers are harvested off the mesh screens and are
put into a flow of water (124) going to the mixed stock tank
104.
[0017] Very small fibers (called "fines") could pass through the
saveall 120's mesh screens and not be captured. To prevent this, a
slurry of "sweetener," water containing larger fibers, is diverted
from the TMP system 100's output and added to the saveall's input
stream (flow 126). The large fibers of the sweetener build up on
the saveall's mesh screens and increase its efficiency at
harvesting the fines. A proportioning valve 128 controls the ratio
of sweetener to water from the white water collection tank 116 in
the input stream to the saveall.
[0018] After the saveall 120 removes many of the fibers from its
input stream, the filtrate water and remaining fibers are sent to a
seal tank 132. This flow 130 is called the "drop leg." The seal
tank contains two sections. The first section is for "clear"
filtrate that comes from the part of the saveall where the mat is
more developed. This clear filtrate is recycled back to the TMP
system 100 (flow 134) or, when the seal tank overflows, is sent
(flow 136) to a sewer 138. The second section of the seal tank is
for "cloudy" filtrate that comes from the part of the saveall where
the mat is less developed. This filtrate is clouded by fibers that
escaped the saveall. It is piped (140) into the white water
collection tank 116 and eventually returns to the TMP system.
[0019] The quality of paper produced by the paper machine 108 is
determined in large part by the characteristics of its input stream
106. Traditionally, components of the TMP system 100 were adjusted
to control the quality of the paper, while the saveall 120 was only
adjusted to optimize its efficiency in harvesting fibers and in
reducing pollution. An embodiment of the present invention adjusts
the operation of the saveall to produce paper of consistent
quality, while satisfying the saveall's original goals of
efficiently harvesting fibers and reducing pollution.
II. An Exemplary Embodiment: Adjusting Saveall Parameters to
Control Paper Quality
[0020] FIG. 2 adds to FIG. 1 a few sensors and controllers while
removing some of FIG. 1's detail for clarity's sake. The discussion
accompanying FIG. 2 shows how logical components of an exemplary
embodiment of the invention interact and presents the methods of
the invention at a high level. Later, the text accompanying FIGS. 3
and 4 delves into details of an implementation of the invention. A
computing device 200 receives from sensors measurements indicative
of the quality of the paper being produced by the paper machine
108. For clarity's sake, the computing device 200 is depicted as a
personal computer in FIG. 2, but its functions could be implemented
on any control technology, including servers, multiprocessor
systems, microprocessor-based systems, minicomputers, mainframe
computers, and distributed computing environments that include any
of the above systems or devices. In FIG. 2, one standard
measurement of quality is shown: the porosity 202 of the paper.
Properties other than porosity, singly or in combination,
indicative of the quality of the paper may be measured and used in
other embodiments of the invention. A predictive model running on
the computing device compares the measured property with a desired
value range set for that property. In order to keep the property
within its desired value range, the model predicts the effect on
the property of adjusting one or more operational parameters of the
saveall 120. Then, the model directs controllers to adjust the
saveall's operational parameters in accordance with its
predictions. In FIG. 2, the saveall parameters adjustable by the
predictive model include the rotation rate 204 of the saveall's
drum filters and the ratio 206 of sweetener (flow 126) to white
water (flow 122) in the input stream to the saveall.
[0021] At the same time that the predictive model is keeping paper
porosity 202 within its desired value range, the model may attempt
to satisfy other constraints on the operation of the saveall 120.
It does this by controlling other properties to keep their measured
values within their own desired value ranges. For example, the
consistency 208 of the drop leg shows the saveall's efficiency at
harvesting fibers. The predictive model may attempt to keep the
measured drop leg consistency below a desired value. The model may
be responsive to other constraints and may be able to adjust other
operational parameters than those shown in FIG. 2, but these two
controlled properties (paper porosity and drop leg consistency) and
two adjustable parameters (rotation rate 204 and sweetener ratio
206) serve for illustrative purposes. The correlations between the
controlled properties and adjustable parameters have been
determined experimentally and are shown in the following table.
1 TABLE 1 Controlled Property: Adjustable Parameter: Porosity Drop
Leg Consistency Sweetener Ratio - - Rotation Rate - +
[0022] Here, a "-" means that as the adjustable parameter
increases, the controlled property decreases. A "+" means that the
adjustable parameter and the controlled property increase and
decrease together. For example, Table 1 shows that the higher the
saveall filter rotation rate, the higher the drop leg consistency
(more fibers in the filtrate), which means the lower the efficiency
of the saveall's fiber harvesting. Conversely, the higher the
sweetener ratio (within the normal operating range), the greater
the saveall's efficiency. These correlations are used by the
predictive model to determine how to adjust the saveall's
operational parameters to keep the controlled properties within
their desired value ranges. Section III discusses how the
predictive model is developed from the correlations.
[0023] The flowchart of FIG. 3 shows how an embodiment of the
invention reads measurements of controlled properties and then
adjusts the saveall 120's operational parameters to keep the
controlled properties within their desired value ranges. In step
300, a desired range is set for each controlled property. The
"range" may take many possible forms. It may be a single value
representing the optimum value for the controlled property. It may
be a more complex structure such as a single preferred value
(called a "soft target") surrounded by a range of less preferred,
but still acceptable, values. Here, a range of acceptable values is
set for the paper porosity 202 while a maximum allowable value is
set for the drop leg consistency 208. The desired value ranges for
controlled properties generally depend upon the operating
characteristics of each particular paper mill and are generally
derived from experiments run at the mill. These ranges can change
over time, such as when the mill makes a different type of paper.
Techniques for setting these ranges are well known in the art, and
the disclosed control concerns keeping the controlled properties
within their desired value ranges rather than determining what
those ranges should be.
[0024] As an additional feature of embodiments of the invention,
step 300 allows for the setting of optimal values for the
operational parameters. In a -manner similar to moving the
controlled properties toward their desired value ranges, the
operational parameters are driven toward their optimal values. The
table below shows two exemplary methods for specifying optimal
values. (Note that these methods are also used to specify desired
value ranges for controlled properties in exemplary embodiments of
the invention.)
2TABLE 2 Adjustable Parameter LP Coefficient QP Coefficient Target
Sweetener Ratio 0.0 100.0 0.07 Rotation Rate 75 0.0 0.0
[0025] In the LP (Linear Product) method, a cost coefficient is
assigned to the parameter. If the cost is positive, then
optimization tends to maximize the value of the parameter. Negative
costs lead to a minimization of the parameter's value. Table 2's
optimization maximizes the saveall 120's rotation rate 204. The QP
(Quadratic Product) method tends to reduce the square of the
difference between the value of the parameter and a set "target"
value. Table 2 optimizes the sweetener ratio 206 in this manner.
The effects of these two methods can be combined by giving a
non-zero coefficient to each.
[0026] In step 302, input values are collected. This collection
includes the values of the controlled properties, but may also
include values of a much more diverse set of properties. Any
property of the paper mill useful in predicting the result of
changing the saveall 120's operational parameters (see discussion
of step 306, below) may become the subject of a measurement. For
example, besides the exemplary controlled properties of paper
porosity 202 and drop leg consistency 208, motor load, chip
moisture, temperature, pH level, and flow rate at various points in
the paper mill can provide information useful in step 306.
Considerations of cost and practicality determine which property
measurements are utilized in a particular paper mill.
[0027] To collect many of these inputs, sensors measure the
properties and provide the values. For some properties, however,
measurements are not always immediately available. For example, in
an embodiment of the invention measurements of paper porosity 202
are only available after a technician runs a sample through a
laboratory process. The technician takes a sample only once an hour
or so, while the feedback loop in FIG. 3 is set to run once every
30 seconds. When an actual measurement of an input property is not
available in step 302, the predictive model takes the values it can
measure and uses them to model a value for the property it cannot
measure. When an actual measurement becomes available in a later
instance of step 302, the modeled value is reset to the measured
value.
[0028] The values of the controlled properties, whether measured or
modeled, are compared in step 304 to their desired value ranges,
set in step 300. Whenever controlled properties are outside of
their desired ranges, the predictive model develops a proposed
course of action in step 306. The model predicts how adjusting the
saveall 120's operational parameters (in this example, rotation
rate 204 and sweetener ratio 206) will change the values of the
controlled properties. Note that the predictive model balances
changes to all of the controlled properties, not simply the ones
beyond their desired ranges. The predictive model attempts to move
the wayward properties back into line with their desired value
ranges without moving other controlled properties beyond their
desired ranges. If a desired value range includes a soft target,
then the predictive model, for example, attempts to move the
controlled property closer to the soft target even if its current
value is within the range of acceptability.
[0029] In embodiments of the invention, the predictive model takes
into account how far a controlled property has deviated from its
desired value range. The greater the deviation, the harder the
predictive model works to eliminate it. Also, the predictive model
need not treat all controlled properties equally. In the following
table, weights are assigned to show how important it is to return
each controlled property to its desired value range.
3 TABLE 3 Controlled Property Minimum Weight Maximum Weight Paper
Porosity 2.0 2.0 Drop Leg Consistency 1.3 1.3
[0030] The minimum and maximum weights indicate the importance of
correcting deviations when a controlled property's value drops
below or rises above its desired value range, respectively. In this
case, it is more important to keep paper porosity 202 within its
desired range than it is to control drop leg consistency 208, and
deviations above and below the ranges are treated equally.
[0031] In some embodiments, the predictive model takes into account
other considerations when developing its proposed course of action.
The following table illustrates two of these.
4 TABLE 4 Controlled Property Noise Filter Rotation Factor Paper
Porosity 1.0 0.0 Drop Leg Consistency 1.0 0.0
[0032] A noise filter may be assigned to each controlled property
to prevent the predictive model from trying to address minor
variations. The predictive model multiplies a change in the value
of a controlled property by its noise filter and reacts as if the
result were the actual change. Thus, a noise factor of 1.0 means
that the predictive model does no filtering while a value of 0.0
means infinite filtering (that is, the predictive model never
responds to changes in this controlled property).
[0033] For an embodiment of the invention, the predictive model
does not wait until a controlled property's value is beyond its
desired range. Instead, the predictive model monitors trends in the
value, uses previous changes in the adjustable parameters to
predict where the controlled property's value is going, and works
proactively to keep the value within its range. The rotation factor
of Table 4 shows the extent to which the predictive model forecasts
a measured change of a controlled property's value into the future.
A value of 1.0 means that the entire change is forecast. A value of
0.0 means that the model does not use changes in the controlled
property's value to forecast changes. In either case, the
predictive model uses changes in the adjustable parameters to
forecast changes in the controlled property's value.
[0034] Also during steps 304 and 306, the predictive model develops
a proposed course of action for optimizing the operational
parameters. Optimization is similar to controlling the controlled
parameters with this difference: optimizing is of secondary concern
and is only performed when it does not move the controlled
properties beyond their desired value ranges.
[0035] Often, the predictive model's proposed course of action is
not implemented unchanged. In step 308, various considerations are
applied to limit or modify the proposal. Maximum and minimum
allowable values are assigned to each adjustable parameter, such as
a maximum rotation rate. Other considerations are illustrated by
the following table.
5TABLE 5 Adjustable Parameter Step High Step Low Move Suppression
Sweetener Ratio 0.01 0.01 3000 Rotation Rate 0.2 0.2 100
[0036] Step High and Step Low are the maximum changes allowed to an
adjustable parameter when increasing or decreasing it,
respectively. Move suppression determines how fast the predictive
model may adjust a parameter to keep the controlled properties
within their desired value ranges.
[0037] In step 310, the predictive model's course of action, as
modified by step 308, is acted upon by sending commands to
controllers that change the saveall 120's operational parameters.
This feedback loop is repeated by returning to step 302 and
collecting new input values. The repetition rate of the loop is
set, usually at about twice a minute, so that deviations are
detected and addressed quickly.
III. Developing the Predictive Model
[0038] Section II shows how the predictive model uses sensor
inputs, weights, noise filters, rotation factors, move suppression
values, etc., to predict how to adjust the saveall 120's
operational parameters to keep controlled properties in their
desired value ranges while, optionally, optimizing the settings of
the operational parameters. This section describes how the
predictive model is created, that is, how those weights, filters,
factors, etc., are determined and how correlations among them are
set. Table 1 and its accompanying discussion hinted at the answer:
these values depend strongly on the operating characteristics of a
particular paper mill and are determined experimentally. As
mentioned above, even the inputs available to the model and the
parameters under its control vary from plant to plant so that Table
1 does not apply as is to every situation.
[0039] The flowchart of FIGS. 4a and 4b is an example of a method
used to experimentally determine the model's factors. In step 400,
the operator of the paper plant ranks the controlled properties
according to the importance of keeping each one in its desired
value range. This ranking is based on experience with paper making
and on the characteristics of a particular plant. In step 402,
weights are assigned that reflect the ranking. (These weights are
discussed with reference to Table 3.) In steps 404 and 406, the
predictive model is run with these weights and its behavior is
compared against what the plant operator wants. The weights are
refined until the operator is satisfied with the operation of the
system under the predictive model. Similarly, in step 408 the plant
operator decides which of the saveall 120's operational parameters
should be changed faster and which slower. Move suppression values
(Table 5) are assigned in step 410 to reflect the operator's
decisions, tested in step 412, and refined in step 414. In this
iterative manner, the plant operator uses the material at his
disposal, that is, the operational parameters put under the
predictive model's control, and refines the model's responses. This
experimental process is also used to develop the predictive model's
ability to model a value when it does not have a current
measurement for an input. (See discussion of step 302 of FIG.
3.)
IV. Embodiments Using Fiber Filters Other Than the Saveall
[0040] The present invention is described in Section II by means of
an exemplary embodiment that adjusts the operation of the saveall
fiber filter 120 in order to regulate paper quality. The invention
is not restricted to operating with the saveall, however. As
mentioned earlier, the paper-making process uses fiber filters
other than the saveall, such as in the TMP system 100. According to
the teachings of the present invention, paper quality can be
regulated by adjusting the operation of these other fiber filters,
either alone or in combination with the saveall. The control
techniques and methods for creating predictive models follow the
examples given above. As emphasized in Section III, details of each
embodiment depend upon the operating characteristics of a
particular paper-making plant, including which sensors and
controllers are available, and parameter settings are
experimentally determined.
[0041] In view of the many possible embodiments to which the
principles of this invention may be applied, it should be
recognized that the embodiments described herein with respect to
the drawing figures are meant to be illustrative only and should
not be taken as limiting the scope of invention. For example, the
predictive model may be split into several distinct applications
which run on separate computing devices. Modeling of input values
may be performed by a process distinct from the one that controls
the operational parameters. Therefore, the invention as described
herein contemplates all such embodiments as may come within the
scope of the following claims and equivalents thereof.
* * * * *