Structured-Singular-Value-Based Optimal Aeroelastic Uncertainty Quantification using Surrogate Models and Flight Test Data

摘要

The onset of aircraft flutter, resulting from the interaction between aerodynamic and structural forces, can be destructive and potentially explosive. Analytical prediction of the flutter instability can be associated with a high degree of uncertainty. An aeroelastic model can be formulated with structured uncertainty and cast into a linear-fractional-transformation framework. With known uncertain bounds, the structured singular value (μ) can be used to determine least conservative robust flutter boundaries that take the uncertainty into account. This is a powerful analysis tool but it requires the uncertain bounds to be known a priori. An existing μ-based theorem can be used to determine if a given uncertain model structure with known bounds is invalidated by existing flight test data. In this work, a μ-based optimization problem is formulated that efficiently quantifies uncertain bounds given a nominal aeroelastic model with a known uncertainty structure and real-world flight test data. Since the determination of μ-bounds is an NP-hard problem, a response-surface-based surrogate model is produced and incorporated into the optimization framework. This surrogate model allows for rapid and guaranteed convergence. This uncertainty quantification algorithm is successfully demonstrated with the use of models and real-world flight test data from the NASA Aerostructures Test Wing.