A Methodology for
Integration in Intelligent
Control Systems
I. ALARCON, * J. L. ZACCAGNINI, X. ALAMAN, P. ADARRAGA,
and P.
Instituto de Ingenier(a del Conocimiento, Modulo p. 4, Universidad Autonoma de Madrid,
28049 Cantoblanco, Madrid, Spain
ABSTRACT
At present, most of the distributed control systems present a hierarchical and multilayered
structure where the basic control close to the process is placed in the bottom and the
intelligent control, nowadays performed by a human team, is located on the top. Our
approach proposes a methodology to formalize the activities carried out by the control
and production team of an industrial plant by means of a basic task. This atomic task
has been found to remain constant through all layers of the hierarchical control structure.
We put forward the fact that the three main steps in knowledge-based systems devel-
opment can be carried out by means of the same set of concepts. The paper is divided
into three different parts. The first one presents a theoretical overview from the cognitive
science point of view discussing the main ideas of the theoretical foundations of this
HPM (HINT project methodology) proposal. The second part expounds the conceptual
framework that represents the main part of the proposed approach describing the main
concepts, the pyramidal structure, and the intelligent layers within the proposed control
hierarchy where our approach is focused. Finally, the blackboard architecture is proposed
as the most suitable one to implement the concepts previously outlined. © 1995 John
Wiley & Sons, Inc.
INTRODUCTION
In recent years, the use of artificial intelligence tech-
niques to support computer-aided engineering ap-
proaches to continue process control problems has
shown very promising. Working over and/or in
collaboration with standard computer-aided control
systems (Efstathiou, 1992), intelligent control sys-
tems can add an important plus of advantages (Ala-
man et al., 1992). In fact, nowadays process control
operators cannot make both correct as well as fast
decisions because the necessary information either
is missing, faulty, or not communicated in an un-
derstandable format. More elaborate and sophisti-
cated support systems are needed to provide oper-
ators with higher level information. These new sys-
tems will reduce decision times and will allow for a
stable process control that will ensure quality.
Received January 3, 1994; accepted July 8, 1994
* Author to whom all correspondence should be addressed.
Integrated Computer-Aided Engineering, 2(3) 165-186 (1995)
© 1995 John Wiley & Sons, Inc. CCC 1069-2509/95/030165-22
Nevertheless, up to now, no methodological ap-
proach has been faced in most of these systems;
namely, there is no standard knowledge engineering
framework to deal with this kind of control prob-
lems. In practice, this means that most actual work-
ing intelligent control systems are ad hoc tool-based
solutions. High costs, nonreusability, and tool in-
165