Machine learning based digital twin for stochastic nonlinear multi-degree of freedom dynamical system

The potential of digital twin technology is immense, specifically in the infrastructure, aerospace, and automotive sector. However, practical implementation this not at an expected speed, because lack application-specific details. In paper, we propose a novel framework for stochastic nonlinear multi-degree freedom (MDOF) dynamical systems. proposed has four modules — (a) physics-based nominal model, (b) data collection module, (c) algorithm real-time update (d) module predicting future state. prediction are based on so-called gray-box modeling approach, utilizes both physics driven frameworks; enables to generalize predict responses. gray box used within developed by coupling Bayesian filtering machine learning algorithm. Although, can be with any regression algorithm, have Gaussian process study. Performance approach illustrated using two examples. Results obtained indicate applicability excellent performance framework.