Innovative Development of Kernel-Based Reduced-Order Models for Predicting LCO Onset

Reducing uncertainty in the prediction of limit cycle oscillations (LCO) and other nonlinear aeroservoelastic phenomena is critical to flight safety. To do so requires nonlinear methods. First-principles based methods (CFD/CSD) have made considerable progress but still cannot predict LCO from the outset. NEAR proposes the development of two innovative nonlinear data-based methods to characterize aeroservoelastic systems using flight-test data. The proposed methods provide a natural extension of existing linear methods, provide uncertainty estimates of the prediction, and are applicable to flight-test data. Both approaches are formally related. However, their practical implementations place various limitations on the physics they represent. The Phase I effort will document, on a benchmark test case, the advantages and disadvantages of each method. Phase II will further develop the most promising approach and demonstrate its use on flight-test data, such as data from the F18-AAW. Special emphasis will be placed on the problem of data generalization across flight conditions, which is key to ensuring safe and efficient envelope-expansion flight testing.