Integration of Data and Physics Based Models for Fault Management

Fault management is a critical component of many systems, especially space systems. As the number and complexity of space systems increases, there is a need to automate fault management. This is a challenging task as automated fault management relies on models and as the complexity of spacecraft increases, reliance on just pure physics-based or pure data-based modeling is shown to be deficient for accurate fault management. Nirvana Technology, Inc. (NT) proposes to develop an integrated approach to address the modeling issues. Here, both data-based and physics-based approaches are integrated, and compared against each other, for providing the most reliable fault management solution, through a Digital Twin (DT) algorithm or architecture. NTrsquo;s DT formulation is based on an integrated algorithm that combines, on a real-time basis, dynamic system models utilizing physics-based low-order models of the system, and statistical and machine learning algorithms applied to data from the various sensors employed.