DIAGNOSTIC MECHANISM MODELING

Diagnostic models of complex mechanisms are designed specifically to support diagnostic reasoning. Diagnostic reasoning is that aspect of troubleshooting which determines a set of component faults that can account for observed abnormalities in system performance. A system’s performance is evaluated by comparing observations of the system’s behavior with its functional specifications. As such, diagnostic models must incorporate elements of both function as well as behavior. We define two necessary properties of diagnostic models reflecting this outlook: architectural fidelity and functional adequacy. We then discuss two general principles of diagnostic modeling, based upon observations from expert protocols: model variables assume values relative to normal levels (the normality principle) and these values are propagated through the model to account for the production of single outputs (the single output principle). These two principles allow significant simplification in the definition of diagnostic models of complex mechanisms. We illustrate our approach to diagnostic modeling and reasoning with examples from xerography.