Physics Enhanced Machine Learning for Monitoring and Twinning

Abstract

To ensure a resource-efficient and resilient operation of engineered systems, it is imperative to understand their performance as-is; a task which can be effectuated through Structural Health Monitoring (SHM). When considering higher levels of the SHM hierarchy, purely data-driven methods are found to be lacking. For higher-level SHM tasks, such as prognosis, or for furnishing a digital twin of a monitored structure, it is necessary to integrate the knowledge stemming from physics-based representations, relying on the underlying dynamics and mechanics principles. This paper discusses implementation of such a physics-enhanced approach to SHM.