A functional approach for the concurrent design of diagnostic systems

This paper presents an approach to develop a diagnostic model of a product or process concurrently with the design stage. This approach is consistent with the principles of design. A functional hierarchy is used to map the system requirements to physical parameters in the design stage to set up a complete and efficient model of diagnosis. The functionality of design is represented with a combination of a functional hierarchy in the form of strata and graph-based methods, and the diagnosis at the level of physical artifacts is performed using failure probabilities and hierarchical search. The failure probabilities are updated by a stochastic approximation method and can be acquired through learning. Generic class diagnostic modules that allow the reuse of diagnostic knowledge are also defined.