Fleet-based System Health Assessment: Reasoning about Change and Differences

Abstract

Monitoring and managing the health of technical systems with advanced diagnosis and prognosis benefits from fleet analytics: insights on the degradation of other but similar systems help, e.g., to forecast actual issues for predictive maintenance as does detecting and correcting anomalies in usage profiles helps to prevent undue wear and tear. Successes in this field usually depend on the similarity of the fleet’s systems: although not necessarily equal, they need to have alike key characteristics, e.g., in the way they age, such that observations on one or many systems constitute an expectation for others.

We introduce fleet-based system health assessments that complement such approaches by reasoning on differences, e.g., those introduced by interventions like upgrades. Given that such change is the only constant for many of today's complex systems, we believe that our addition to health assessment is necessary to cope with the variation in systems that fleet operators face over the lifecycles of all systems.

Technically, our approach is based on probabilistic reasoning. The related modeling and inference techniques allow us to incorporate insights from past observations in individual systems, but also from comparable systems and, at the same time, they permit comparisons between parts of the fleet and estimate possible effects of existing differences. Our work shows that especially such comparisons are valuable for fleet health management over long lifecycles that is typically linked to continuous improvement processes.