INTRODUCING DEVIATIONS AND MULTIPLE ABSTRACTION LEVELS IN THE FUNCTIONAL DIAGNOSIS OF FLUID TRANSFER SYSTEMS

We have recently experimented the FDef (Functional Diagnosis with effo rts and flows) approach on a real-world problem(1) (Chittaro, L., Fabb ri, R. and Lopez Cortes, J. Functional diagnosis goes to the sea: appl ying FDef to the heavy fuel oil transfer system of a ship. In Proceedi ngs of FLAIRS-96, Key West, FL, USA. Florida Artificial Intelligence R esearch Society, 1996, pp. 419-423), i.e. the diagnosis of multiple fa ults in the heavy fuel oil transfer system (HFOTS) of a modern contain er ship. This paper builds on that preliminary work, extending it in s everal directions by: (i) analysing its Limitations: (ii) generalizing the proposed techniques from the specific HFOTS case to a wide class of hydraulic systems in the domain of Fluid Transfer Systems; (iii) si gnificantly increasing the diagnostic capabilities of the approach by introducing representation and reasoning about deviations from nominal values; (iv) adopting a hierarchical organization for representing th e functional model to improve efficiency and to reason at multiple lev els of abstraction; and (v) providing a formal validation of the emplo yed diagnostic knowledge. (C) 1998 Elsevier Science Limited. All right s reserved.