Design and use of multiple models for time-constrained hierarchical diagnosis

This paper describes the design and use of multiple models for time-constrained diagnosis. Although model-based diagnosis provides benefits in terms of maintainability and fault coverage, when compared with rule-based diagnosis, it also requires more computation time. We propose in this paper, a hierarchical model representation of the system to be diagnosed, in order that the most precise and detailed diagnosis may be obtained in the time available. The diagnostic process starts with a simple model, and uses more detailed models of the suspect areas as time allows.