U.S. patent application number 11/355908 was filed with the patent office on 2006-12-21 for system and method for integrated sensing and control of industrial processes.
Invention is credited to John Coates, Fernando Rathgeb, Rand Stowell.
Application Number | 20060284058 11/355908 |
Document ID | / |
Family ID | 34375232 |
Filed Date | 2006-12-21 |
United States Patent
Application |
20060284058 |
Kind Code |
A1 |
Coates; John ; et
al. |
December 21, 2006 |
System and method for integrated sensing and control of industrial
processes
Abstract
An integrated spectral sensing engine is based on a combination
of energy sources (illumination) and detectors housed within an
integrated package that includes the sample interfacing optics and
acquisition and processing electronics. The focus is on a
miniaturized sensor system that can be optimized for specific
measurements and can be integrated into a manifold-based sample
handling system. Design and fabrication components are selected to
support high volume manufacturing of the sensors. Spectral
selectivity is provided by either continuous variable optical
filters or fabricated filter matrix components. The spectral
response of the primary sensors covers the range from the visible
(400 nm) to the near-infrared (1100 nm), as defined by the
availability of suitable low-cost solid-state detector devices.
Provision is made to extend the range into longer wavelengths, and
to shorter wavelengths for filter-matrix devices. A broad selection
of measurement modes is defined and these include
transmittance/absorbance, turbidity (light scattering) and
fluorescence. On board data processing not only provides the
primary data acquisition, as well as data massaging and the display
and output of computed results. The targeted application of the
spectral sensing devices are for water, pulp and paper, chemical
and petroleum based industries. Alternative packaging regimes and
the production of lower cost sensing devices can lead to the use of
the spectral sensing devices in the medical, clinical, forensic and
consumer based areas of application.
Inventors: |
Coates; John; (Newtown,
CT) ; Rathgeb; Fernando; (Sharon, MA) ;
Stowell; Rand; (Weld, ME) |
Correspondence
Address: |
David Aker;Attorney at Law
23 Southern Road
Hartsdale
NY
10530
US
|
Family ID: |
34375232 |
Appl. No.: |
11/355908 |
Filed: |
February 16, 2006 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10913819 |
Aug 6, 2004 |
7057156 |
|
|
11355908 |
Feb 16, 2006 |
|
|
|
60494977 |
Aug 14, 2003 |
|
|
|
Current U.S.
Class: |
250/226 ;
356/419 |
Current CPC
Class: |
G01N 21/85 20130101 |
Class at
Publication: |
250/226 ;
356/419 |
International
Class: |
G01J 3/50 20060101
G01J003/50; G01N 21/25 20060101 G01N021/25 |
Claims
1. An optical spectral sensing device, for determining properties
of a sample, said device comprising: an integrated energy source
and an integrated spectral sensing detector package, an optimized
sample chamber or cell dimensionally designed to match the active
area of the spectral detection device without the aid of additional
interfacing optics; integrated electronics for providing energy for
said source and for receiving a signal generated by said spectral
detector in response to energy coupled to said detector by said
sample chamber or cell, said integrated electronics providing
direct output of sample properties of said sample; on-board
computer processing means including memory storage for data,
calibration coefficients, methods and results; on-board data
communications including the output to a visual display and
communications of results to a process monitoring computer; and a
manifold-based sampling system to provide the primary interface to
said spectral sensing system, and to provide necessary support for
the analytical system in terms of sample conditioning and reagent
interfacing.
2. The device of claim 1, wherein said optical structure is formed
in configurations that allow for at least one of transflectance,
transmittance/absorption, fluorescence, or light scattering
measurements.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] The present application is a continuation of and claims the
priority benefit of U.S. nonprovisional patent application Ser. No.
10/913,819 filed Aug. 6, 2004, entitled "SYSTEM AND METHOD FOR
INTEGRATED SENSING AND CONTROL OF INDUSTRIAL PROCESSES" of the same
named inventors and assigned to a common assignee, which in turn
claims the priority benefit of U.S. provisional patent application
Ser. No. 60/494,977, filed Aug. 14, 2003, entitled "SYSTEM AND
METHOD FOR INTEGRATED SENSING AND CONTROL OF INDUSTRIAL PROCESSES"
of the same named inventors and assigned to a common assignee; both
of said applications being incorporated herein by reference.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The present invention relates to integrated systems and
methods for sensing parameters associated with industrial processes
and controlling such processes based on the sensed information.
More particularly, the present invention relates to a miniaturized
spectral sensing system, with integrated sensed signal
conditioning, signal exchange, and integration of the sensed
information with appropriate process control for improved
industrial processes.
[0004] 2. Description of the Prior Art
[0005] The processes employed to make products vary widely from
industry to industry. Those processes may involve the use of
complex machinery, interconnected machinery and equipment, and
electrical and chemical exchanges. It is important to ensure to the
greatest extent possible that the features and components of an
industrial process interoperate as effectively as possible.
Achieving that goal results in good products and good productivity.
Additionally, it can minimize the impact of the process on the
surrounding environment. Two related aspects of industrial process
improvement include the need to understand the process itself, and
control of the process based upon that understanding. In
particular, it is important to be able to determine the
characteristics or parameters at each stage of the process and for
the final output of a process, such as the chemical composition,
temperature, and/or pressure of a gas mixture used to make a
chemical compound, for example. Based upon that information, it is
important to be able to adjust the process, such as increasing
process temperature, or changing the ratio of reactants, for
example, if the output information deviates from an established set
of parameters. The inability to monitor the process usually means
that there is a risk of low quality product, which results in the
need to make corrections after the product has been made. This is
both inefficient, and leads to a high level of wastage, which can
often lead to an environmental issue.
[0006] There are two established approaches to the monitoring of a
process for chemical composition or physical properties. They are
the extraction of grab samples followed by remote analysis at a
suitable control laboratory, and the use of on-line
instrumentation. The first option is inefficient and is not
effective for control purposes. The second option is usually
expensive, and as a result, it is normal to implement a single
analyzer at the end of the process. This has limited value for good
process control because it is too late in the process to make
meaningful adjustments. For a complex process, the ideal situation
is to have a multiplicity of measurement points and to monitor the
process from the raw material through to the final product. This
has to be cost effective to make the implementation of multiple
sensing points worthwhile. One solution with optical
instrumentation is to use a single analyzer but to multiplex the
stream or the optical output. While this is an option, it has risks
because it lacks redundancy--one instrument controlling an entire
process. It is also limited in terms of its response, dependent on
the number of points measured (measurements are made sequentially,
not in parallel). The present invention uses multiple low cost
sensing devices, a major advancement, and overcomes issues related
to a lack of redundancy. In fact, one may use a redundancy of
sensing devices to ensure maximum efficiency in the event of the
failure of a single sensing device.
[0007] Many industry standard methods for process control still
rely on Proportional-Integral-Derivative (PID) controllers, a
technology dating back 60 years. Once tuned, the system is only
able to control the process with which it started. Should process
behavior change after start-up, the controller cannot counteract
disturbances and the closed-loop system may become unstable. The
traditional fix for time-varying process dynamics is to start over
and manually retune the loop whenever its performance degrades.
While that may not be particularly difficult, repeated tuning can
be tedious and time consuming, especially if the process takes
hours to respond to a tuning test. Manual tuning may not even be
possible should process behavior change too frequently, too
rapidly, or too much. In the case of model based controllers,
process characteristics such as gain, time constants, and dead time
need to be estimated, often by trial and error, to attempt defining
the shape of the process response curve. It is a long and complex
process that requires significant investment of time and money. The
present invention includes the option of a firmware-based control,
which is also miniaturized, and is reduced in size to a small
electronic board, which overcomes these established problems
associated with both standard control systems and traditional model
based systems.
[0008] When measuring devices are integrated into a process it is
normal to employ a sampling system. The sampling system is
typically a collection of valves and sample conditioning devices
(filters, mixing chambers, temperature control loops, etc.) that
extract the sample from the main stream, and present the sample in
an ideal format to the measurement system. Traditionally, this
collection of valves and components takes up a rather large space,
and can sometimes be as expensive as the measurement device to
implement. In the miniaturization of the sensing devices, it makes
little sense to use such a system, in terms of efficiency and cost.
Significant benefits are gained if the sensing device and the
sampling system can be integrated where the sample volumes are
matched to the sensing device itself. Recent developments in
industrial process improvement initiatives have been centered on
the mechanism for integrating sensing devices into sampling
systems. A good example is the New Sensors/Sampling Initiative
(NeSSI) sponsored by the Center for Process Analytical Chemistry
(CPAC) at the University of Washington, an effort by an industrial
consortium to standardize sensors and the sensing platform used for
process monitoring. Initially, traditional parameters such as
temperature, flow and pressure, which can be important indicators
of process characteristics, have been addressed. The goal of the
NeSSI is to make measurement techniques uniform across industries
with an interest in participating in the initiative. The platform
is a miniaturized version of traditional sample gathering and
handling methodologies and, pursuant to the Instrumentation,
Systems, and Automation Society (ISAS), standard SP76, establishes
the interface of the sample gathering components with sensing
devices used to assess the characteristics of the extracted
sample.
[0009] The benefits of the NeSSI system are its size, the ability
to add components as standard modules, and the ability to integrate
the sensing system to form a single stand-alone unit for sample
extraction, conditioning and measurement. The objective is to
develop sensing devices to be compatible with the goals of NeSSI.
What is also needed is such a system that enables sampling and
sensing at intermediate sites along the way of the process, thereby
permitting process corrections earlier and minimizing defective
product output. Note that NeSSI has been used here as a discussion
point, and is not necessarily the only platform for consideration.
There is a movement in a wide variety of fields that involve the
handling of liquids, gases and vapors, including medical and
clinical applications, where miniaturized valves and sample
handling/conditioning are involved. The approach to integration of
sensing within these platforms is also given consideration.
SUMMARY OF THE INVENTION
[0010] It is an object of the present invention to provide an
integrated spectral sensor and an optional control system for
controlling industrial processes. The system includes a sampling
component, a spectral engine including a sensing component and a
signal conditioner, a signal exchange system, and a controller. The
sampling component can include, where required a multiport manifold
on a miniaturized sample handling platform (such as NeSSI) for
gathering one or more fluid samples (including reagents) related to
the industrial process to be controlled. The sampling component
interfaces with the spectral engine that includes an optical
sensing system for nonintrusive detection of features of the
sampled fluid. The spectral engine further includes a light or
energy source, spectral sensing component for measuring the
characteristic chemical features of the fluid, a sample cell or
chamber that is dimensionally optimized for the light source and
sensing component, and a microprocessor for conditioning the
signals output from the spectral sensor. The signal exchange system
may be a wired or a wireless signal transfer device coupled locally
or remotely to the sensor. The optional controller includes the
signal processor and one or more embedded computer programs to
evaluate the received signal information and make decisions on any
process modifications to be made. The controller is a predictive
controller that maximizes the ability to adjust process conditions
as quickly as possible in response to the sensed process
information. The controller interfaces with one or more process
modification devices, including control valves, chillers and
heaters, for modifying, pressure, flow rates and volumes, and
temperatures of the fluid or fluids forming part of the industrial
process.
[0011] Four example application areas have been identified that can
benefit from this integrated sensor approach, and these include the
water, pulp-and-paper, chemical, and petroleum industries. These
applications require on-line, real-time sensors, sensors that are
capable of operating in harsh environments, and can provide
analytical and physical property measurements. The present
invention addresses these needs. In the case of the chemical
industry, a wide range of applications is envisaged in a wide range
of industry sectors, from specialty chemicals, such as
pharmaceutical products to commodity chemicals, such as
petrochemicals and polymers. The applications go beyond just the
manufacture of the raw material and basic chemicals, and can be
extended to the blending and formulation of final products in key
high energy consuming industries, including those linked to
consumer products. It is to be understood that the present
invention has broader applicability than the application areas
cited.
[0012] In conventional industrial process plants, process sensors
and actuators are hardwired using copper wire or fiber optics
networking. Because of the high costs associated with installation,
maintenance, and constant reconfiguration of a process, there is a
large opportunity for a major cost advantage for using a wireless
communications path to interconnect the sensors and actuators. This
is particularly important when a network of sensors is being
employed. Also, when using an optimizing predictive adaptive
control system, better communications between sensors improves the
overall efficiency. For this reason, the present invention
contemplates the option of employing wireless connectivity to
establish data signal transfer.
[0013] The integrated sample, sensing and control system of the
present invention provides a more granular and immediate picture of
relevant information associated with an industrial process. That
picture is coupled with effective predictive process control to
yield improved productivity with corresponding reduced impact on
the process's surrounding environment. These and other advantages
will become more apparent upon review of the following detailed
description and the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
[0014] FIG. 1 is a simplified representation of the elements of an
example spectral sensing engine: source, sample interface, light
analyzer and detector.
[0015] FIG. 2, comprising FIGS. 2A-2C, show example combinations of
optical filters and detector arrays.
[0016] FIG. 3 shows the electronic components for the integrated
spectral sensor.
[0017] FIG. 4 is a simplified representation of standard package
configurations for on-line sensors: absorbance/transmission and
fluorescence cells.
[0018] FIG. 5, comprising FIGS. 5A-5D, illustrates alternative
packaging configurations of on-line sensors: turbidity, NIR and
immobilized reagent versions.
[0019] FIG. 6 is an example integration of spectral sensor and
reagent dispensing into manifold system.
[0020] FIG. 7 is an example spectral sensor response in the visible
region: colored dyes.
[0021] FIG. 8 is an example spectral sensor response in the near
infrared region: common chemicals.
[0022] FIG. 9 is a simplified flow diagram of the steps associated
with the function of the optional predictive-adaptive
controller.
[0023] FIG. 10 is a schematic representation of an optional
predictive-adaptive control system.
[0024] FIG. 11 is an example schematic representation of the
predictive-adaptive control system in a multi-loop
implementation.
[0025] FIG. 12 shows an example mesh-network for wireless
communications with multiple sensors.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS OF THE
INVENTION
[0026] The present invention is an integrated system for sensing
process characteristics and controlling the process based on the
sensed information. The sensing aspect of the invention preferably
includes one or more miniaturized optical spectral sensors located
at multiple points within a process or an individual process unit.
This provides a means to monitor a process from start to finish,
with key intermediate points also covered, as opposed to the
traditional approach of monitoring the product as it is produced at
the end of the production line. An optional component of the system
is a wireless communication interface, based on a proprietary
adaptation of a standard wireless platform, associated with the
multiple spectral sensors to allow them to interface with the
control component of the system. The system further includes an
optional control component as an adaptive-predictive control system
that makes use of the feedback from the sensors and provides
optimized process control.
[0027] An important component of the spectral sensor technology can
be broadly described as an optical spectrometer on a chip. While
optical sensors have been available, the present invention
integrates an optical filter assembly with a light or energy
sensitive array. The optical filter technology used is either in
the form of a continuous linear variable filter (LVF), or a filter
array (patterned filter or mosaic). In the LVF form the resultant
device, or spectral sensing component, is the most versatile and
can be utilized for many applications. An example format of an
LVF-based spectral sensor is shown in FIGS. 1, 2A and 2C. The
spectral sensing component is preferably implemented as part of a
photodiode or a Complementary Metal-Oxide Semiconductor (CMOS)
array detector package. The sensing component of the present
invention is based on existing optical sensing technology modified
for the present purpose. An example component has been marketed as
a commercial device by OCLI (a JDS Uniphase company), known as a
MicroPac. The device was not produced in a form that was compatible
with the proposed application, and was intended for lab-based
experiments that demonstrated feasibility. As developed, the
MicroPac was based on a silicon photodiode array offering spectral
ranges of 400 nm to 700 nm (visible) and 600 nm to 1100 nm (short
wave near Infrared (NIR)). The MicroPac had a complex construction
featuring a gradient index (GRIN) lens as an optical interface
between the filter and the photodiode assembly. This was required
to preserve the spectral resolution of the filter because the
detector used was a standard commercial package. In the current
embodiment, the LVF is directly bonded to the detector array, and
in this form does not require any form of resolution retaining
optics. Sensors derived from these components based on the LVF can
be used for absorption measurements in the visible and near
infrared (NIR), as well as fluorescence measurements in the
visible. Examples of data have been acquired in all of these modes,
and example spectral response curves for the visible and NIR ranges
are provided in FIGS. 7 and 8. The short wave NIR provides good
differentiation based on chemistry and composition based on
vibrational overtones of the component molecules. However, in
cases, such as the digestions in pulp and paper applications, where
visible absorbing and fluorescence centers are also expected to be
important, the visible version for the spectral sensor is used. For
applications involving chemistry, where the species to be measured
is not normally visible, the analysis may be performed with the
addition of a reactive chemical reagent. The reagent may be added
as an additional chemical to the process stream, or the process
stream can interact directly with versions of the reagent that are
immobilized on or in a solid substrate, which forms part of the
optical path as shown in the examples in FIG. 5.
[0028] Those skilled in the art of optical sensing technology will
recognize that the short-wave Near Infrared (700 to 1100 nm) works
well for a wide range of liquid-based measurements. In this region,
sample pathlengths from 1 to 10 cm are considered to be optimum,
dependent on the material to be measured. Such pathlengths may be
used for direct measurements made on petroleum and liquid chemical
products, while shorter path lengths may be required for darker
materials, such as digestion products and materials with high
aromatic content. Although spectral changes in this region are
subtle, they can be readily correlated with both composition and
key chemical and/or physical properties. Tools such as multivariate
modeling, sometimes known as chemometrics are common for such
applications. These would be used as appropriate, and the
calibration coefficients generated from the modeling are stored on
flash RAM located on-board the sensor.
[0029] It is also possible to consider indirect methods of
measurement while measuring in this silicon spectral region. These
are methods are as noted above, where a reagent interacts with the
stream either as a liquid reagent or on a substrate that interacts
with the process stream, and spectral changes are observed for the
substrate. Examples are pH or chemical reactivity of the stream,
where the stream reacts with a reagent that is immobilized in a
porous solid matrix, such as a sol gel or a membrane (organic or
inorganic). Systems featuring immobilized reagents can also be used
for gas phase measurements where the gas or vapor is chemically
reactive and is able to form a colored species once it interacts
with the immobilized reagent. An example of such a reaction is with
carbon monoxide and a derivative of hemoglobin. If the reaction is
reversible, the sensor may be used for continuous monitoring. in
cases where the resultant reaction is a color change or a change in
a level of fluorescence. The optical sensor system of the invention
may be used to conduct direct measurements, as well as with
immobilized reagents necessary for applications in the water
industry, and in the pulp industry, for example, especially for the
measurement of materials such as sulfides, both in the liquid and
gas phases. For applications that feature liquid based reagents,
the reagent or reagents are metered into a mixing chamber which
exists as one of the components of the manifold system.
Applications based on chemical reagents do not necessarily require
long optical pathlengths, and pathlengths ranging from 0.25 cm to
2.00 cm are expected to be the norm.
[0030] As defined, the spectral sensor can be constructed from
either a continuously variable filter (defined as the LVF) or from
a filter matrix or mosaic. This latter approach is usually
optically more efficient and less expensive than the LVF approach.
It is often more specific in application, but less versatile than
the LVF system. An illustrated example of a matrix-based spectral
sensor is provided in FIGS. 2 and 3. The version shown is a
4-channel RGBW sensor, and is capable of handling a wide range of
color-based applications.
[0031] The sensor hardware for the present invention is not limited
to silicon-based photo-sensing devices, and alternative detector
arrays can be used, including InGaAs, PbS, PbSe and also Micro
Electronic Mechanical System (MEMS)-based devices. Such devices
would be considered for extensions into the longer wavelength NIR
and for the mid-IR. The format of the proposed sensor platform may
be extended into these other spectral regions. In the case of the
mid-IR spectral region, these are expected to be mainly for gas and
vapor phase applications. Note that for some applications, the
silicon-based detectors may be used in conjunction with an
immobilized reagent for certain gas-phase measurements. For
example, vapor analysis for pulp digestions may be handled by this
approach, especially for the detection of sulfides.
[0032] The approach offered is described as being based on a
spectral engine. The spectral engine includes the spectral sensing
device (described above), and the energy source, which can be
either a broadband or narrowband source, dependent on the mode of
measurement (broadband sources are used for NIR and visible
absorption, narrowband sources are used for turbidity and
fluorescence). White LEDs and tungsten bulbs are used as broadband
sources, and individual LEDs are used as narrowband sources.
Another component of the spectral engine is the sample interface,
which is typically a cell or chamber. One of the key benefits
offered by the system are that the cell can be optimized in size
based on the source and spectral sensing system. The sizes of the
detection devices are 1.times.8 mm and approximately 3.times.3 mm
(matrix sensor). Scaling the sample cell to these dimensions
results in a net cell volume of a few microliters up to around 80
microliters. The advantage gained here is that a minimum sample
size is required, which effectively eliminates any sample
temperature effects, and significantly reduces the amount of
reagents that have to be dispensed for reagent-based applications.
The volume reduction on reagents can be up to 1000 times, which
reduces reagent vessel capacities, dramatically reduces
consumption, and cuts operating costs. The final critical set of
components of the spectral engine are the electronic components. An
example of the functional electronics is provided in FIG. 3. Up to
two microprocessors, and possibly more, may handle the initial data
handling (processor #1) and then the data massaging (processor #2).
A final processor may be employed and feature onboard memory to
store methods, calibrations and results, and to handle
communications to displays (if required), external devices via
serial connections and also wireless communications if the option
is used.
[0033] It is known that measurements from industrial processes
require a degree of sample conditioning. Normally this necessitates
the use of a sampling system in conjunction with a process
analyzer. Such an approach negates the cost and size benefits of
the proposed sensor technology. For this reason, the present
invention combines the spectral sensing device with a miniaturized
sample handling platform. The NeSSI generation II platform is used
as a practical example of a miniaturized modular approach to sample
handling, based on a manifold, with the provision for modular
sensor integration. The spectral sensors of the present invention
can be designed to be compliant with the ISAS SP76 standard which
defines the interface and format of the NeSSI platform. The sensors
can also be designed to interface with any other form of
miniaturized sample handling.
[0034] The spectral sensor implementation based on NeSSI may be
implemented in several ways. FIG. 6 provides a view of the sensing
device integrated into a modular base compliant with NeSSI. FIGS. 4
and 5 show example configurations for the spectral sensor that meet
these requirements. This includes a basic transflectance
(absorption/transmission) arrangement wherein the source (B)
radiation is directed into the sample chamber, and the beam is
reflected from the base of the module up into the sensing area of
the spectral sensor(C), which is powered and controlled by onboard
electronics (A). These electronics can be linked to an external
network, which may include wired transmission or wireless
telemetry. An alternative arrangement provides a geometry that can
be used for fluorescence measurements, for light scattering
measurements, or for dark samples. FIG. 5 shows alternative
arrangements that can be used for turbidity, near-infrared and also
for applications involving immobilized agents.
[0035] The optional control system component of the present
invention is an advanced predictive and adaptive model-based
controller suitable for critical and/or inherently unstable
processes. Unlike prior art controllers, it achieves peak
performance in a matter of hours, resulting in major savings of
time, money and energy. The predictive controller does not require
a complex predetermined or theoretical mathematical model of the
process. Instead the system "learns" the process dynamics while in
use and this is achieved regardless of the complexity of the
underlying process behavior. High order and non-minimum phase
transient response characteristics are captured by the system and
are included in the model. A comprehensive model is developed based
on actual process dynamics, not a theoretical model. The controller
then uses this model to make accurate forecasts of process
response. A PC-based controller platform suitable for integration
into the sensor platform is the predictive software made available
by Universal Dynamics Technologies, Inc. of Richmond, British
Columbia, Canada and offered under the name Brainwave.TM..
[0036] The control system includes a form of the Brainwave.TM.
platform adapted for the control component of this invention in the
form of firmware, incorporated into hardware electronics. The
control function is in the form of software, firmware, hardware, or
microcode as part of the overall miniaturized form of the
invention. While the control function may be employed as a control
means for the entirety of an industrial process, it is contemplated
that the controller provides local optimization of control for
small segments of processes, small stand-alone processes and
individual process units. The control system has two main parts:
the adaptive model building component (Process Model
Identification), and the predictive controller. Process Model
Identification is achieved quickly by using a function series
approximation technique called Dynamic Modeling Technology (DMT)
that is based on Laguerre polynomials. The controller monitors
process response to changes in the controller output and other
feedforward input variables. The transient response of the process
is modeled using a Laguerre function series. The coefficients of
this series are automatically calculated so that an accurate model
of the process transfer function is obtained.
[0037] In process control, process transfer functions are transient
and the Laguerre functions are well suited to modeling these types
of transient signals because they have a similar behavior to the
process being modeled. The Laguerre modeling method produces a set
of weights for the Laguerre functions in the series so that when
summed, a high fidelity model of the original transient signal is
obtained. The set of "weights/coefficients" is called the Laguerre
spectrum for the signal. In short, changes in the transient
behavior of the process will trigger the controller to adjust the
model to the new characteristics by automatically generating a new
Laguerre spectrum. The effects of measured process disturbances are
also modeled to incorporate adaptive feed-forward compensation into
the control strategy resulting in further performance improvements.
As illustrated in FIGS. 9 and 10, for each model, there is an
internal representation of the `state` of the system with respect
to the recent changes in each input to the process (i.e., Control
output, FF1, FF2, and FF3). This representation is used together
with the model to make predictions about future changes that may
occur in the process. Each model contributes its component of the
expected change in the process. All of these predicted changes are
summed together to reveal the `net` effect on the process in the
future. The process models, the system state, and the net predicted
change in the process are then used to determine the necessary
control action. This problem is solved by calculating a control
change that (if all conditions remain unchanged) returns the
process to the set point as rapidly as possible. This technique is
known as horizon predictive control whereby the process model is
used to compare the predicted value of the process to the desired
target value. Any difference becomes the amount that the process
has to change and the process models are used to solve the control
action required.
[0038] In an embodiment of the invention requiring control of
multiple functions, a multi-loop approach is used for the control
of large coupled processes as represented in FIG. 11. Advanced
controls can be applied to most complex applications such as
distillation columns, pulp and paper processing, multi-unit batch
reactors, and other complex processes where manipulation of one
process variable disturbs many others or where parallel process are
attempting to achieve multiple objectives simultaneously. In the
proposed control system, these problems become opportunities for
improvement that translate into lower cost, higher quality,
increased production, and a more efficient use of energy.
[0039] The optional control system of the invention is designed to
improve the accuracy and efficiency of an existing industrial
process control arrangement. This is enhanced by the use of the
spectral sensing devices, which are ideally distributed at key
locations throughout the process. There is minimum hardware
required and it readily becomes the technology of choice where PID
controllers cannot deliver increased levels of process control
performance. It may be interfaced to existing Distributed Control
Systems (DCS) using an OLE for Process Control (OPC) server. The
system may be configured to synchronize the predictive-adaptive
controller with the DCS so that safety logic automatically switches
back to traditional PID control in the event of a malfunction.
[0040] In order to make full use of a multiple sensor-based control
system in a manufacturing plant it is necessary to develop an
economical method for interfacing the sensors to the control
system. Conventionally, one would hardwire the sensors in place,
but in a system where a multiplicity of sensors is used, the
efficiency and cost benefits can be rapidly diminished based on the
cost of installation. The present invention provides the optional
feature of wireless signal exchange. It is known that traditional
approaches to the implementation of wireless in a plant environment
are fraught with difficulties because of environmental
interference. However, the present invention minimizes the negative
impact associated with such a signal exchange method by using
redundant wireless connections. As shown in FIG. 12, a mesh based
wireless communications system (Comsys) is designed to provide a
reliable fully connected path between the sensors and the control
system.
[0041] The optional Comsys system has two major elements: 1)
Wireless Network Access Points (NAPs) and 2) Wireless Sensors (S).
In a preferred embodiment, the NAPs are fixed modules in a ring
that loops around the plant or the process unit, and are powered by
AC mains line power with battery backup. The distance between each
NAP is such that they have a reliable wireless path between each
NAP. Even if there are no sensors within range of an NAP (as shown
in right side of FIG. 12), there is an NAP provisioned to ensure
the ability to relay communications traffic from point-to-point.
Reliability is further enhanced in such a mesh network by allowing
for the NAPs to be arranged in a ring that loops into the network
operations center, so that the failure of any one NAP will not
cause a loss of connectivity and control. The sensors shown in FIG.
12 may be line or battery powered with the necessary power control
intelligence to conserve battery power and report battery health
and battery condition as required. Each sensor point communicates
data to the nearest NAP, which in turn will manage contention on
the network and relay the sensor traffic back to the network
operations center and the control system. To further enhance the
system operation and reduce the costs of installation and
reconfiguration, an application layer protocol resides on top of
IEEE 802.11b standard designed to manage wireless traffic, and to
provide for self discovery and configuration of the system. The
application layer optionally provides functions for self-discovery,
geolocation, and test of the network.
[0042] The fundamental aspects of the present invention lead not
only to increased productivity, but also to an energy saving and
process efficiency capability in at least the four target
industries: water, paper, chemical, and petroleum industries. In
the case of water, it also leads to increased public safety. It is
expected to provide similar advantages in consumer-oriented
markets, including food and dairy processing, beverages, and
household products (cleaners, etc.). The technology of the current
invention will be particularly useful in the four noted fields of
application: [0043] Water application: continuous monitoring of
public water supplies for residual chlorine content, and similar
parameters that are important to public safety. Residual chlorine
(disinfectant) is an important marker in drinking water supplies
for the unexpected introduction of toxic and biohazard materials.
The latter can range from the accidental introduction of bacteria
via a breakage in a feeder pipe, to the deliberate introduction of
dangerous materials. The system can also be configure to measure
heavy metals (lead, chromium, mercury, etc.) and also environmental
contaminants, such as phosphate, nitrate and arsenic. While the
systems as described can be configured for public water supplies,
stripped-down versions can also be configured for public buildings,
office buildings, hospitals, and even residential water supplies.
[0044] Petroleum applications: monitoring of process streams for
raw materials, intermediates and final products. In an average
refinery there are many process units, and many of these units have
several critical points where measurements can be made. Examples
include the reformer, cat crackers and the blenders. The
convenience and cost potential of the invention may enable many
more points to be monitored, thereby permitting a higher level of
overall predictive control. Many of these processes are energy
intensive, and so significant savings in terms of improvements in
efficiency and reductions in environmental emissions are
anticipated with the implementation of this technology. [0045]
Chemical applications: there are numerous potential applications
for the invention in the chemical and petrochemical industries.
These can range from the production of raw materials to various
processes used in the pharmaceutical and biotechnology industries.
It is anticipated that it is well suited for the petrochemical
related industry, where a significant amount of energy is involved,
and where good monitoring and control can provide better overall
efficiency and product quality. A practical example is the
production of carbon black, where the composition of the input
streams is used to provide additional information to the control
system. [0046] Pulp and paper products: there are several potential
areas of application for the present invention in the pulp and
paper industries. The most important, from the point of view of
control, are probably in the digestion, pulping, and bleaching
areas. The use of near infrared for the measurement of parameters
such as Kappa has already been demonstrated, and the use of visible
methods are feasible for determining lignin-related information.
Another important area is bleaching. Both direct methods involving
NIR, and indirect methods, involving immobilized agents are
expected to work for this application. Control of both the
digestion and the bleaching are important for the overall process,
and good control parameters for digestion are expected to provide
important energy savings.
[0047] While the present invention has been described with
particular reference to certain specifically described components
and methods, it is to be understood that it includes all reasonable
equivalents thereof.
* * * * *