Note: Descriptions are shown in the official language in which they were submitted.
" CA 02258766 1998-12-18
GR 96 P 3468 FILE, ~?~~. :, TIf4S AMC1~ID~I.~
Tai °t a'Atd~LATi~3iN
Description
Method and system for commissioning industrial plants, in
particular in the basic materials industry
The invention relates to a method and system for
commissioning industrial plants, in particular in the
basic materials industry, having a plant control system
which carries out non-control functions and control
functions and whose control systems operates with process
models, in particular control engineering models, for
example in the form of mathematical models., neural
network models, expert systems etc., in a control system
computing unit.
In the control of industrial plants, in particu
lar plants in which very rapid processes, very slow
processes or which run in leaps and bounds, or processes
for which there is no suitable state sensors, proceed,
operations are mostly carried out using controlengineer-
ing models. As a rule, such plants have a basic automa-
tion system and a process management system (non-control
and control). Experience shows that the commissioning of
relatively large plants is very time-consuming and costly
and requires specialists who are experienced in plant
engineering. This also applies to the control engineering
plant design and to the control project engineering of
the individual components.
It is the object of the invention, for plants of
the type outlined above, in particular for plants in the
basic materials industry, but also for plants in the
chemical industry and for producing electrical power, to
specify a commissioning method and a system suitable for
this with which, given an optimum commissioning result,
a reduction in time and costs can be achieved. In this
case, the continuous operation of the plant which has
been commissioned is intended to be able to be
subsequently continually improved,
CA 02258766 2004-07-15
20365-3961
- 2 -
and easily evaluable knowledge for the control project
engineering and the design of corresponding plants are to be
obtained. In summary, the aim can be described as reducing
the engineering costs with a simultaneous improvement in the
plant function.
The object is achieved in that the commissioning
is carried out subdivided into commissioning the non-control
functions, with extensive initialization of the control
functions, by means of personnel located on site, and
extensive commissioning of the control functions by means of
remotely-transmitted data via data lines from at least one
site remote from the plant, preferably from an engineering
center. As a result of this subdivision of the
commissioning into a so-called basic commissioning and
engineering commissioning, it is advantageously largely
possible to dispense with having to use control engineering
specialists, in particular specialists for the setting of
parameters and improvement of control engineering models, on
site. The commissioning costs can thus be reduced
considerably. Furthermore, the commissioning can be carried
out more rapidly and more reliably, since for the
engineering commissioning a specialist team can be made
available to whom all the aids of an engineering center and
external consultants are available.
In particular, according to one aspect of the
present invention, there is provided a method for
commissioning in sub-divided fashion industrial plants, in
particular in the basic materials industry, having a plant
control system with at least one non-control part for
controlling and regulating basic functions of the plant and
with a technology-specific control part for influencing the
quality of the manufactured product wherein the control part
operates with control engineering models in a control system
CA 02258766 2004-07-15
20365-3961
- 2a -
computing unit, the method comprising the steps of: basic
commissioning of the non-control part with extensive
initialization of the control part by means of personnel
located on site; technological commissioning of the control
part following the basic commissioning wherein model
calculations for technological optimization are implemented
by means of remotely-transmitted data via data lines from at
least one site remote from the plant; and implementing an
ongoing technological quality optimization after the
technological commissioning.
It is already known to equip PCs by means of
programs which are input into the PC via data transmission.
Furthermore, diagnosis of PCs but also of individual
automation devices, such as machine-tool controllers or
programmed logic controllers, for example, is known. The
known procedure for the equipment, diagnosis and functional
improvement of individual devices cannot, however, be
transferred to the commissioning of entire.plants, in
particular entire plants which are as complex as those
CA 02258766 1998-12-18
GR 96 P 3468 - 3 -
in the basic materials industry. For this purpose,
learning routines are just as necessary as the use of the
computing intelligence of the plant, long access times
and a dialog in the sense of "trial and error". This was
previously held to be impossible to implement to the
extent necessary.
In a refinement of the invention, it is envisaged
that engineering optimization is carried out while
commissioning the control functions. The optimization is
preferably carried out "step by step" under remote
influence in at least one control system computing unit
of the plant, that is to say the individual optimization
steps run on a computing unit of the process control
system, so that those problems cannot result which can
result in the case of taking over an optimization step
carried out on an external computing unit into the
computing unit of the control system. From the point of
view of the complexity of the programs in the plant
control systems, software errors would otherwise always
be expected in taking over optimized program parts. The
avoidance of implementation problems is a considerable
advantage of the system according to the invention.
In addition to remote commissioning, remote
functional improvement and remote optimization of the
control part, provision is also made of a remotely
influenced improvement of the non-control part. Even the
basic automation of an industrial plant is so complex
nowadays that the remote optimization according to the
invention is worthwhile for this. In this case, the
appropriate level of the plant control system is advan-
tageously used.
Following the commissioning of the control func-
tions with the initial optimization, a further improve-
ment of plant operation is carried out continuously by
means of engineering optimization with the aid of the
engineering center. It is thus ensured that the plant is
further operated in an optimum manner in control engin-
eering terms. This is
CA 02258766 1998-12-18
GR 96 P 3468 - 4 -
important in particular in the case of changes in the
product program, for example as the result of accepting
further material grades into the product program.
The optimization relates in particular to setting
parameters for models, in particular in the form of algo
rithms or artificial neural networks (ANN), and to
further development of the algorithms of the models or of
the design of the ANN, but also of expert knowledge
evaluated by computer, for example in the form of limit
ing curves, etc. Thus, the most important modules of a
model-based controller can be continuously improved, in
order to achieve optimum plant behavior.
Provision is advantageously made that, in the
case of using neural networks as process models, the
adaptation takes place in parallel with network training.
Thus, account is taken particularly well according to the
invention of the properties of artificial neural networks
(ANN). They are always in the state which is most advan-
tageous for the optimization. It is furthermore of
particular advantage if the artificial neural networks
(ANN) are used for the improvement of algorithms and/or
models, and if in so doing a closed loop is formed which
is designed as a directly closed loop in a control system
computing unit or as a loop which is closed indirectly
via the engineering center. In particular, the loop which
is closed via the engineering center in this case ensures
that the newest control knowledge and computing knowledge
can always be included in the optimization and improve-
ment of parameters and/or models. In this case, it is
advantageously also provided that the further development
of models is carried out with the aid of an evolution
strategy, for example via genetic algorithms, It is thus
also possible for any necessary further development of
the models to optimize the plant behavior
CA 02258766 2004-07-15
20365-3961
_ 5 -
and, if appropriate, also to optimize the plant itself, to
take place.
In order to carry out the method in an
advantageous manner, a commissioning or plant operational
improvement system is provided which has at least one
engineering center installed remote from the plant - in
particular a commissioning and/or operational improvement
center - which is connected via remote data transmission
means to at least one control system computing unit of a
plant which is to be commissioned in engineering terms or to
be improved further in engineering terms. By this means,
the advantages of the method according to the invention may
be achieved.
According to an aspect of the invention, there is
provided a commissioning or plant operational improvement
system for industrial plants, in particular in the basic
materials industry, the system comprising: at least one
non-control part for controlling and regulating basic
functions of the plant; a technology-specific control part
for influencing the quality of the manufactured product
wherein the control part operates with control engineering
models in a control system computing unit; at least one
engineering center installed remote from the plant, the
engineering center connected via remote data transmission
means to at least one control system computing unit of the
plant; and wherein basic commissioning of the non-control
part occurs with extensive initialization of the control
part by means of personnel located on site, technological
commissioning of the control part occurs following the basic
commissioning wherein model calculations for technological
optimization are implemented by means of remotely-
transmitted data via data lines from the engineering center,
CA 02258766 2004-07-15
20365-3961
- 5a -
and an ongoing technological quality optimization is
implemented after the technological commissioning occurs.
In a refinement of the system, it is provided that
it has, in an engineering center, an internal network which
is preferably equipped as an Ethernet with twisted-pair
connections, having at least a 10 megabit per second
transmission speed, which preferably operates according to
the TCP-IP protocol. This results in a secure internal
network which can be implemented in a cost-effective manner
and which has all the properties which are necessary for a
remote commissioning and optimization center. The system
furthermore has a remote commissioning or operational
improvement network connected to the internal network for
communication with industrial plants which has known,
conventional data transmission components (ISDN, telephone,
modem, Internet connections) and is connected to the
engineering center by means of at least one security data
transfer device (firewall). Thus, using conventional cost-
effective components, the construction of a remote
commissioning network, which has a construction necessary
for reliable operation of the plant and for defense against
operational espionage actions etc., is possible.
_ CA 02258766 1998-12-18
GR 96 P 3468 - 6 -
In a refinement, the system has, in the region of
the engineering center, external sites, for example
project offices, which are physically separate but
connected in terms of data, for example via ISDN lines,
to said center and, together with the latter, form the
engineering expertise. The discussion and/or solution of
optimization tasks by external specialist personnel is
thus possible, whose teams etc. are included in the
engineering expertise. In this case the cooperation of
completely different teams is possible.
In a further refinement of the invention, it is
provided that the commissioning or operational improve-
ment center has an administrator unit, in particular
having evaluation software for collected data, and is
simultaneously designed to be suitable for logbook
maintenance. For the purpose of optimization, in particu-
lar of operating points and specific operational steps,
it is necessary to track the effects of control interven-
tions in the case of earlier optimization measures. This
is advantageously possible by means of the method accord-
ing to the invention.
For ISDN and Internet connections, there are
advantageously routers which set up the desired connec-
tions. With the aid of these routers, which optionally
operate automatically, the connection which is optimum in
each case can be set up for the dialog between the
operators of the individual plant components and the
components of the engineering center, and for the connec-
tion between the individual computing units . Depending on
the day of the week and the time of day, these may be
different.
In order to carry out the invention there are in
the control system of the plant computing engineering
modules, a data collection unit etc., and in the center
personnel with engineering knowledge, databases for the
respective clients, general and client-specific engineer-
ing modules, in particular in a form which can be input
like modules into the plant control system.
CA 02258766 1998-12-18
GR 96 P 3468 - 7 -
Stepwise improvement with simple input of the new data
into the overall system is thus possible.
Provided in the plant control system are comput
ing devices for the adaptation of plant-specific parame
ters, for the storage of models designed specifically for
the plant, for the storage of prior calculation
algorithms, for the storage of trend sequences and for
the storage of adaptation algorithms. The plant control
system is thus able to carry out the engineering
optimization in accordance with the predefinitions of the
engineering center.
Provision is in particular made that for neural
networks such as are often used for example in rolling
mills or in electric-arc ovens, that is to say in the
basic materials industry, optimization and training. are
carried out in parallel. For this, specific, advantage-
ously favorably designed software modules are present.
Likewise, diagnostic memories and further computing
devices required for the engineering optimizat~.on of the
plant. These software-controlled computing devices can be
influenced via the data transmission means which are used
in accordance with the invention.
The hardware and software devices of the commis-
sioning or operational improvement center comprise both
nonspecific hardware devices, software tools, commis-
sioning tools, software development tools, software
evolution tools, ANN training tools, statistical evalu-
ation programs, etc. and also special plant-specific
software tools, recourse being made as far as possible to
plant-neutral modules, and plant-specific, specially
developed modules only being used whenit is necessary.
The communication and optimization system for
plant engineering optimization which is present according
to the invention is
_ CA 02258766 1998-12-18
GR 96 P 3468 - 8 -
in particular designed to be capable of dialog and
advantageously has, in particular, optical acquisition
components both for the personnel handling it and for the
plant parts to be optimized or to be diagnosed. It is
therefore possible for optimization proposals, change
proposals, diagnoses etc. to be carried out in a manner
which largely corresponds to the presence of the special-
ists on site. It is therefore actually no longer person-
nel who travel but information. The commissioning center
and the plant, as well as the plant control station,
therefore advantageously has monitors and also cameras.
The same applies to external sites of the engineering
center, for example project offices or specific~software
development units, so that it is actually possible to
operate as though all the personnel involved in .the
optimization and the further development were located at
one site, in particular at the site of the plant. This is
of considerable advantage in particular when working with
artificial neural networks (ANN, as well as neuro-fuzzy
and fuzzy applications), in which uniform handling is
necessary from the collection of the training data up to
the output of new parameters.
The invention is explained in more detail using
drawings from which, as well as from the subclaims,
further details which are also essential to the invention
can be inferred. In detail:
FIG 1 shows a symbolic illustration of the principle of
the invention with significant details,
FIG 2 shows the significant parts of the engineering
center in a symbolic illustration,
FIG 3 shows the system formed in a symbolic illustra-
tion,
FIG 4 shows an illustration relating to the optimiza-
tion of a rolling process, with the use of a
neural network for optimizing the rolling-force
calculation and
CA 02258766 1998-12-18
GR 96 P 3468 - 9 -
FIG 5 shows a simple interaction of a mathematical
' model with a model in the form of a neural net
work.
In FIG 1, 1 denotes the control system of the
client, which contains the engineering module 3 and a
data collection 4. The control system of the client is
commissioned by commissioning engineers 5. In actual
fact, there is a team of commissioning engineers on site.
Via the transmission plane 6, which is illustrated as a
bubble and [lacuna] ISDN connections, in particular
having ATM components for image transmission, but also
has telephone modems or the Internet, components 1, 3, 4
and 5 which serve for example for commissioning a rolling
mill 2, are connected to the components 7 to 11. The
components 7 to 11 are the components in or connected to
the energy center with its personnel 7. In the engineer-
ing center are general engineering modules 9, client-
specific engineering modules 8, a database for various
clients and project engineering components 11 for differ-
ent clients, as well as further components which can be
inferred in their details from the description. The
separation of the components located on site and the
components in or connected to the engineering center is
clearly visible. In contrast to the known diagnostic or
equipment routines, this is not a connection which is
closely limited in time, but a relatively permanent
connection of the components which are on site and in the
engineering center. The components which are connected to
the engineering center do not in this case need to be
physically concentrated, instead they extend, if necess-
ary, to different continents. Because of the time shift,
there is an optimization, consultancy and diagnostic
procedure at the client which under certain circumstances
takes place 24 hours per day. Plants in the basic
materials industry, in chemistry and in energy production
are certainly also constructed in such a way
CA 02258766 1998-12-18
GR 96 P 3468 - 10 -
that they enable continuous 24-hour operation.
The plant itself advantageously always remains
autonomously serviceable, since it is only the intelli-
gence present at the plant which is further improved.
Interruption to the line or satellite connections between
the plant and the engineering center therefore have no
effect on the production. The individual optimization
steps are only carried out later.
In FIG 2, 12 denotes an input station for pro-,
jest-specific parameters 13, with which data from the
client's plant, said data being optimized in control
engineering terms in single-pass and in loops, are
determined in a prior calculation unit 14, an adaptation
unit 15 and a unit 16 which takes into account forwarding
laws, with the aid of a network training unit 18 and a
diagnostic part 17. In the case of a rolling mill, for
example, operations are carried out with the models
specified individually in 19, such as the rolling-force
model, flatness model, bending model and roll-gap model,
said models being able to be further improved in the unit
20 by means of genetic algorithms and new model parame-
ters. Together with the project-specific parameters from
the unit 13, they make possible the optimization comput-
ing processes running in the units 14, 15 and 16. The
generated data pass via the transmission plane 21, here
ISDN stands symbolically for all data transmission means,
into the database 25, which is divided up in a client-
specific manner and where the data lead, with the aid of
tools for project engineering, diagnosis, from remote
commissioning tools and generally valid engineering
models, to engineering, project-specific models 22. The
latter pass via the plane 21 into at least one computer
unit of the client's control system.
CA 02258766 1998-12-18
_ GR 96 P 3468 - 11 -
FIG 3 shows the core of the engineering center 24
with external sites 25. The core of the engineering
center 24 is connected to clients 26 to 29, for which
purpose various connecting means can be used. The abbre-
viation SCN in this case stands for companies' Intranets,
which can be extended to specific clients. Furthermore,
the core of the engineering center 24 is connected to
plant service stations 30, in order to be able to give
direct instructions to the regional service areas and to
be able to evaluate their experiences. The core of the
engineering center 24 has an office network with the
computing units 31, which are connected to one another
via a bus 35. Furthermore, the computing units~35 which
ensure the connection with the clients and likewise among
themselves , are connected by a bus 3 6 . Between the . two
buses 35 and 36 there is a firewall 33 with a monitoring
station 34. The firewall 33 also prevents an unauthorized
through-access to the internal office network being able
to take place from outside. The internal network is
advantageously an Ethernet with twisted-pair connections
and advantageously has a transmission speed of
10 megabit/second. It operates, for example, with the
TCP/IP protocol and may contain up to 3000 terminals. It
is also possible to supply an entire department or an
entire division of a company with the data of the inter-
nal network. In this case, an FDDI home ring with up to
fifteen servers and a transmission speed of 100 megabit/
second serves as the backbone . In this way, the data from
workstations or true large computers, such as are advan-
tageous, for example, for the rapid training of neural
networks, can advantageously be transmitted very rapidly.
The cooperation of a large number of coworkers in the
final stage of remote commissioning is also possible
without difficulty in order to shorten the time.
A firewall is of importance in that no viruses or
sabotage commands can be introduced. Suitable programs
CA 02258766 1998-12-18
GR 96 P 3468 - 12 -
for firewalls are known; their updating and the monitor-
ing for unauthorized accesses is carried out via station
34.
The programming within the framework of the
plant-specific and general modules is advantageously
partially carried out in an object-oriented manner, for
example by means of the programming language C++, a work
flow system with case tools being able to be used with
advantage.
FIG 4 shows in schematic form, using the example
of a rolling mill, the cooperation and training of a
neural network with an algorithm for the rolling force.
In this case, for example, the strip thickness, the
thickness reduction, the strip width, the temperature,
the roll radius and the strip tension are taken into
account. These values are supplied both to the algorithm
and to the neural network. Furthermore, the details of
the chemical analysis and the roll speed are supplied to
the neural network, and the values from the rolling-force
algorithm and from the neural network are fed together in
the point denoted by X. This results in a set point for
the rolling force, which is compared with the actual
rolling force value. The difference is in turn supplied
to the neural network as the feedback value, so that an
adaptive feedback loop results. The arrow 37 is intended
to signify that an adaptation of the individual network
weightings takes place, being carried out in accordance
with the specified difference. This leads to a continuous
adaptation of the behavior of the neural network in
accordance with the actual behavior of the roll train in
the case of the material currently being rolled_ A "daily
shape" of the roll train can also be taken into account
in this way. The production results of the roll train are
better than in the case of conventional control.
FIG 5 shows, finally, a forward-acting example of
the interaction of a mathematical model (algorithm)
CA 02258766 1998-12-18
' . GR 96 P 3468 - 13 -
with a neural network. Here, a simple additive improve-
ment of the control variable generated takes place.
It goes without saying that, in addition to
neural networks, further software modules, for example
S with expert knowledge, possibly with the aid of limiting
curves, may be employed for managing the plant. This is
carried out, in particular, for casting and smelting
processes. The commissioning and engineering optimization
method and system according to the invention can in this
case be used for different plants irrespective of the
individual control engineering modules. However, it is
particularly advantageous for a control technology which
operates with the aid of neural networks.
