Abstract
M.Sc. (Computer Science)
This study comprises an investigation into and a discussion of Artificial Intelligence
(AI) techniques and methods which are suitable for the implementation of a
computerised intelligent autonomous manufacturing environment. To this end neural
networks, fuzzy logic, genetic algorithms and Petri nets are discussed. The purpose is
to define a broad framework for the implementation and integration of intelligence in
the Computer Integrated Manufacturing (CIM) environment.
A number of examples have been used throughout the study in order to illustrate the
relevant function or concept.
Currently there is a lack of an integrated structured developmental approach towards
the implementation of AI techniques in the CIM industry. In this dissertation we
examine the relevant topics in some detail before progressing to the development of
the Real-time Artificial Intelligence Agent Model (RAIAM) framework.
The RAIAM framework establishes an environment suitable for the insertion and
application of AI techniques and methods into a computerised CIM system structure.
This is accomplished by the establishment of a centralised database and message
handling unit, and by the usage of intelligent modular agents.
An expert database system (EDS) based on a centralised blackboard design is used as
the structural basis for RAIAM. Although it makes use of this centralised database
system, it is a largely modular approach as it makes use of agents to communicate and
process information and actions.