A sequential design of experiments for dual assessment of performance and safety boundaries in conceptual design

The conceptual design phase in aerospace engineering is critical, as early decisions directly influence lifecycle cost and inherent system safety, yet must be made with limited high fidelity test data. A comprehensive assessment requires identifying a concept’s optimal performance while characterizing its critical safety boundaries. This dual objective poses a challenge for standard experimental investigation strategies. This paper introduces an integrated assessment framework featuring a robust composite expected improvement criterion, a sequential sampling strategy that allocates scarce evaluation resources. This criterion balances the investigation of performance optima and failure boundaries to maximize information gain. An integral four dimensional evaluation system facilitates a comprehensive comparison of competing concepts. Numerical validation demonstrates that the proposed active learning investigation improves the accuracy of failure boundary characterization, achieving an approximate 54.5% boundary root mean square error reduction compared to a feasibility weighted benchmark under restricted evaluation budgets. An air-to-air missile guidance case study confirms the framework’s applicability for providing quantitative decision support in concept selection.