Deriving fault location and control from a functional model

It has been established that, by directly representing a system's components and their interactions, causal models lead to algorithmic methods for the monitoring, diagnosis, and control of fault recovery. Analysis of sensor data with component-oriented models has demonstrated that: (1) sensor failure is easily identified, (2) the absence or degradation of sensor data is easily accommodated once recognized, and (3) emergency signals and warnings can be intelligently selected or generated from the same system description. Control steps can also be dynamically derived, permitting the automation of state maintenance and failure recovery. An implementation (the KATE system) of the ideas presented is discussed.<
>