Vehicle Integrated Prognostic Reasoner (VIPR) Metric Report
This document outlines a set of metrics for evaluating the diagnostic and prognostic schemes developed for the Vehicle Integrated Prognostic Reasoner (VIPR), a system-level reasoner that encompasses the multiple levels of large, complex systems such as those for aircraft and spacecraft. VIPR health managers are organized hierarchically and operate together to derive diagnostic and prognostic inferences from symptoms and conditions reported by a set of diagnostic and prognostic monitors. For layered reasoners such as VIPR, the overall performance cannot be evaluated by metrics solely directed toward timely detection and accuracy of estimation of the faults in individual components. Among other factors, overall vehicle reasoner performance is governed by the effectiveness of the communication schemes between monitors and reasoners in the architecture, and the ability to propagate and fuse relevant information to make accurate, consistent, and timely predictions at different levels of the reasoner hierarchy. We outline an extended set of diagnostic and prognostics metrics that can be broadly categorized as evaluation measures for diagnostic coverage, prognostic coverage, accuracy of inferences, latency in making inferences, computational cost, and sensitivity to different fault and degradation conditions. We report metrics from Monte Carlo experiments using two variations of an aircraft reference model that supported both flat and hierarchical reasoning.