Model-based diagnostics deals with diagnosing systems, it means determining health of system components based on a system description and an observation of system variables. This paper focuses on systems that can be described in propositional logic, particularly on simplification of their diagnostic models for some given conditions of observation. It is known that performing diagnostics on the entire model of a system when only few variables are expected to be observed is not efficient. If we

more »

... ew the limited set of variables which may appear in the observation then they would be used to simplify the diagnostic model before the diagnosis inference takes place. A pruning procedure which systematically removes segments of a model that do not contribute to the overall system diagnosis is proposed. The procedure employs an algorithm deciding component diagnosability which is based on directional resolution. Diagnoses retrieved using the simplified model are equivalent to those retrieved using the original model.