Physics-Informed Machine Learning for Uncertainty Reduction in Time Response Reconstruction of a Dynamic System

This study proposes a hybrid method for leveraging physics-based and data-driven modeling. It is used to model dynamic system specifically, the reconstruction of its time response. relies on idea measuring confidence level predictions using proportional uncertainty those predictions. The Gaussian process (GP) as data-efficient machine learning approach modeling ability quantify enables incorporation model. integration controlled by an alignment which determines weight models. In case study, proposed response nonlinear pendulum. Postprocessing results shows notable reduction in addition, improvement model’s performance measures can be observed well.