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