On the specification of multiple models for diagnosis of dynamic systems

Article type: Research Article

Authors: Coghill, George Macleod | Shen, Qiang

Affiliations: Department of Computer Science, University of Wales, Aberystwyth SY23 3DB, Wales, UK E‐mail: gmc@aber.ac.uk | Division of Informatics, University of Edinburgh, 80 South Bridge, Edinburgh EH1 1HN, UK E‐mail: qiangs@dai.ed.ac.uk

Abstract: In this paper the issue of utilising multiple models for diagnosis of dynamic systems is explored. Models are defined by their properties; which are selected from the three categories: variables, relations, and structures. Also, Model‐based Diagnosis is generally perceived to consist of three tasks: fault detection, fault isolation, and fault identification, dealing with the existence, location, and degree of a fault, respectively. In order to utilise multiple models for diagnosis it is necessary to have a correlation between the model properties and the diagnostic tasks. This provides a coherent means of guiding the switching process according to the task to be performed. A proof of concept of the process is demonstrated with reference to fault identification of a laboratory scale process system rig.

Journal: AI Communications, vol. 14, no. 2, pp. 93-104, 2001