An architecture for the integrated supervision of complex dynamic processes is presented. This allows the supervised evolution of the control from normal feedback control during acceptable behaviours to fault diagnosis techniques when the behaviour is considered unacceptable. A behaviour classification is presented and used to identify the appropriate generic control task. This results in an integrated approach to process supervision that allows the co-ordinated use of different modelling techniques, including artificial intelligence (AI) approaches. Extensive experimental results from the supervision of a laboratory scale process rig are presented.
|Number of pages||11|
|Journal||IEE Proceedings D (Control Theory and Applications)|
|Publication status||Published - May 1992|