Enabling in-time prognostics with surrogate modeling through physics-enhanced Dynamic Mode Decomposition method

Computational models provide essential quantitative tools for assessing and predicting the health performance of physical systems. However, high-fidelity are rarely used in real-time operations or large optimization loops, due to their time-intensive nature. A common approach improving computational efficiency prognosis is employ surrogate models. Such can significantly decrease computation time some accuracy loss. In this context, use Dynamic Mode Decomposition (DMD) proposed generate lithium-ion (Li-ion) battery discharge. DMD has been suggested successfully area fluid dynamics over a decade, but it not applied Prognostics Health Management domain, where farahead prediction nonlinear behavior crucial propagate faults predict Remaining Useful Life (RUL). For Li-ion management, standard application using only observable quantities interest was unable capture discharge batteries exhibited lab testing. potential solution found by implementing Koopman theory, which considers theory provides mechanism trade-off low dimensional with high-dimensional linear ones framework, augmenting state variables into system representation. we augmented hidden states higher-fidelity physics model build surrogate. comparison model, improved minimal loss accuracy, enabled long-term prognostics horizons. generalized method implemented ProgPy python packages.