Abstract
We present a deep learning framework for correcting existing dynamical systemmodels utilizing only a scarce high-fidelity data set. In many practicalsituations, one has a low-fidelity model that can capture the dynamicsreasonably well but lacks high resolution, due to the inherent limitation ofthe model and the complexity of the underlying physics. When high resolutiondata become available, it is natural to seek model correction to improve theresolution of the model predictions. We focus on the case when the amount ofhigh-fidelity data is so small that most of the existing data driven modelingmethods cannot be applied. In this paper, we address these challenges with amodel-correction method which only requires a scarce high-fidelity data set.Our method first seeks a deep neural network (DNN) model to approximate theexisting low-fidelity model. By using the scarce high-fidelity data, the methodthen corrects the DNN model via transfer learning (TL). After TL, an improvedDNN model with high prediction accuracy to the underlying dynamics is obtained.One distinct feature of the propose method is that it does not assume aspecific form of the model correction terms. Instead, it offers an inherentcorrection to the low-fidelity model via TL. A set of numerical examples arepresented to demonstrate the effectiveness of the proposed method.