New Fault Containment Algorithms

Associated with any nondeterministic decision, such as fault detection in a noisy environment or medical diagnostics, there are two decision errors, a false positive and a false negative. In the case of fault detection, a false negative means that the algorithm failed to identify the presence of a fault and a false positive means that the fault detection algorithm mistakenly declared the existence of a fault. Typically in the design of fault detection algorithms, a threshold is set that finds the best compromise between these two decision errors.

The consequence of a false negative fault diagnosis in safety critical applications may be life threatening. To mitigate this, a fault tolerant signal consolidation algorithm (ftsc) is placed in parallel with the fault monitor. The ftsc relieves the fault monitor from the fault containment task. This architecture eliminates the need for a compromise threshold, allowing the fault monitor to be optimized for fault identification only. Accordingly, safety is easier to guarantee and the problems and costs associated with falsely condemning resources are minimized.