A digital twin framework for prognostics and health management

Despite rapid advances in modeling and analysis technology, the manufacturing industry has been slow to implement prognostic and health management strategies. A cause of this delay is the individualized focus of most health monitoring solutions, which makes it difficult to deploy and reuse modeling resources across manufacturing equipment fleets. This paper presents a digital twin-based framework that standardizes communication and organization of modeling resources used for health monitoring, a critical aspect of prognostics and health management. The framework is based on a novel, state-based model of mechanical system health that can be reused across manufacturing machines and components. A set of modular digital twin classes enables the creation of extensible digital twin hierarchies for monitoring the health of complex systems. A case study implements this framework to standardize fault detection results for the seal and bearing systems of an industrial pump. The framework’s standardized DT classes and aggregation relationships allow component-level models to be re-used and aggregated to predict faults in the pump’s bearing system. •Individualized predictive maintenance solutions have limited reach in manufacturing.•Proposal for reusable digital twin classes for prognostics and health management.•Digital twin architecture supports a range of data-driven and physics-based models.•Aggregation supports extensible monitoring of complex manufacturing equipment.