Optimal Experimental Design to Improve System-Level Predictions with Mission-Level Data under Uncertainty

While observed mission-level test data plays a critical role in the assessment of system of systems performance, it can be further used to improve prediction confidence in mission-level measures of success through calibration of underlying uncertainty in computational models. These calibrated system-level models can be used for future mission planning and optimization. This work develops a framework to leverage mission-level data to improve system-level predictions through the lens of a hierarchical model structure describing the systems and missions. The system-level simulation models are calibrated to quantify the underlying uncertainty using mission-level data through Bayesian model calibration. The calibrated system-level uncertainty is then used identify the next tests to run that maximize the amount of information gained about the system of systems. The developed framework integrates mission-level data and optimal experimental design techniques to improve system-level model accuracy to support Test and Evaluation and mission planning efforts. As a proof-of-concept, the framework is demonstrated on a multi-drone search and rescue application.