Real-time integrated process supervision

Abstract

This paper presents the use of a micro-controller-based integrated process supervision (IPS) system as a real-time platform for investigative work in structuring expert control. Two different control approaches, based on classical and artificial intelligence techniques, were integrated within IPS and serve as practical examples of the structured approach to expert control. The IPS is a refinement of the expert control architecture. It allows the integration of several control techniques in a single generic framework. Specifically, the paper presents the extensive experimental results derived from a micro-controller-based implementation of IPS on the real-time control of a typical industrial heat-exchanger process. The classical approach, based on auto-tuning techniques, was implemented under the IPS framework. Three auto-tuning techniques, namely Ziegler–Nichols tuning, amplitude tuning and phase tuning were incorporated. In addition, neural-network-based control techniques using the modified cerebellar model articulation controller (MCMAC) were also seamlessly incorporated within the IPS scheme. The real-time experimental results using the IPS architecture significantly demonstrated the effectiveness of IPS in handling varying operating conditions. Furthermore, the inclusion of both AI and classical control techniques within a common supervisory framework adequately shows the generality of the architecture.