A Supervisory Control Framework for Set Point Tuning in Industrial Processes Christoforos E. Economakos Department of Automation Halkis Institute of Technology 34400 Psachna, Evia, Greece ecece@hol.gr Fotis N. Koumboulis Department of Automation Halkis Institute of Technology 34400 Psachna, Evia, Greece koumboulis@teihal.gr Abstract The problem of adjusting the set point of industrial processes with unknown characteristics is studied. An Agent based Supervisory Control System Framework is proposed. Modeling Agents as well as Intelligent Agents are included. The Modeling Agents derive local steady – state models around operating points of the process. These models are considered to be constructed during normal process operation using various identification techniques, each of which corresponds to a different instantiation of our system. The Intelligent Agents are proposed to be implemented using finite automata units. The performance of the proposed scheme is illustrated through simulation results to some well-known benchmark problems of chemical industry. 1. Introduction In industrial process operation, either manual or automatic, there is often a need to make a step transition from one operating point to another. An operator or the process automation logic must apply some control action, which will send the steady-state values of the performance variables to a new desired level. Such actions are applicable to both open loop and closed loop systems. In spite of its apparent simplicity, automatic implementation of a transition between operating points is not always a trivial task, especially if we don't have enough knowledge of the process characteristics. In this paper we study this problem. In particular, we present the challenges that may be encountered in a practical situation and The paper has been funded by the General Secretariat for Research and Technology, International Cooperation, Eureka Project E!3219 - AADSS, EU. propose ways for dealing with them. A distinctive feature of our approach is that we make no assumption about the existence of an a priori known mathematical model of the process behavior. Instead, we propose a generic rule based scheme, which constructs a number of local steady state models, by using measurements of the process variables around a current operating point. The proposed scheme is quite generic, in the sense that it does not specify any particular modeling method for identifying and tuning these local models. Upon choosing a particular modeling method, which is more suitable for a given application, or a set of competitive methods, one can get a concrete rule based supervisory control system that can be analyzed to appropriate modeling subsystems and intelligent subsystems, which operate independently among themselves in the form of automation agents that manipulate the same process data. The intelligent agents are proposed to be implemented using finite automata. Supervisory Control Systems of this kind can be very useful in a number of process control applications, especially if the behavior of the process is characterized by high uncertainty and/or temporal variability. Such processes are, for example, those operating under big random disturbances or at changing environment conditions [1]-[7]. It is obvious that in these cases it can be extremely difficult to construct a priori some reliable mathematical model of the process behavior. Existing results in the field are benefited by different tools e.g. [1] where functional analysis tools are used as well as [8]-[10] where Fuzzy Logic tools are used. 2. Problem Statement Consider a nonlinear process, whose dynamical behavior is either unknown or too complex to be Proceedings of the 2005 International Conference on Computational Intelligence for Modelling, Control and Automation, and International Conference on Intelligent Agents, Web Technologies and Internet Commerce (CIMCA-IAWTIC’05) 0-7695-2504-0/05 $20.00 © 2005 IEEE