The use of Knowledge

1
The use of Knowledge Based Expert Systems
for Fume Extraction System Control
Corus Consulting Limited, Teesside Technology
Centre, P.O. Box 11, Grangetown,
Middlesbrough, Teesside TS6 6UB U.K. Telephone
44 (0)1642 467144 Fax [phone]
460321 E-mail paul.marsden_at_corusgroup.com
Expert systems are computer programs that contain
expert knowledge stored in a knowledge base.
This information or knowledge can be used to
examine plant data and form conclusions about the
way the plant, or piece of equipment within that
plant is performing. The advantage of this
approach is that it does not require detailed
mathematical understanding of plant operation in
order to provide important information to plant
engineers. Steelmaking processes generate fume
that must be captured and controlled to prevent
emissions to atmosphere. This is done with large
extraction fans and bag filter cleaning systems.
The operation of these systems is critical to
ensure good environmental performance of the
steelmaking process and optimisation is necessary
to ensure that the fume extraction systems are
run economically. Fume capture also helps to
recycle the metallic oxides back into the
steelmaking process. Owing to the complexity of
large extraction systems it is important
to pre-empt plant malfunctionso that site
engineers are aware of potential
problems that may cause system failure or
fugitive particulate emissions to atmosphere. A
prototype expert system has recently been
installed at a local electric arc based
steelworks. The model accesses real-time data
from the melting shop, extraction system and bag
filter plant and is able to perform the two roles
of information and advice. Real-time data are
displayed as dynamic plant mimics so that site
engineers are able to see the condition of the
extraction system without need to be on plant.
An expert system containing a knowledge base of
plant rules informs engineers of any plant
operational anomalies. Typical expert rules
might check for the correct operation of dampers
controlling the filter cleaning sequences, or to
check for damaged filter bags that might
eventually cause roof emissions. The installation
of this prototype expert system, has shown
potential savings of over 60,000p/a based on
energy costs alone.
74