In this project, the approach to building accurate diagnostic and prognostic models for components is based on predictive analytics and modeling. Diagnostic models focus on identification of individual components’ current condition. Prognostic models are used for predicting the health of a component in the future. In order to carry out this research, an integrated approach that uses both measurement based and physics based modeling techniques is used. The measurement based portion is data driven with a focus on collecting and analyzing historical data. The physics based portion is uses theoretical analysis and modeling to better understand the physics of the failures. The integrated system model will greatly enhance the development of predictive tools while also helping to support and inform the historical and test data collected. Simulation data supplements historical aircraft data not feasibly obtainable and allows exploration of various failure modes and root causes. Conversely, the reliability and robustness of the modeling and simulation capability of the proposed algorithm can be demonstrated and correlated with sensor data from historical or test data. The integration of the two parts allows for predictive tools to be created for remaining useful life, health condition, and component design. This is a direct outcome of the approach. The intersection of design, models and simulations, and real world systems is where new tools are being developed to predict health, life, and performance of components, subsystems, and complete systems.
Alex Cao, Joshua Tarbutton, Rhea McCaslin, Erin Ballentine, Lester Eisner, Abdel E.Bayoumi. "Condition-Based Maintenance at the University of South Carolina: A Smart Predictive System." Proceedings of AHS 69th Annual Forum, May 21–23, 2013.