Novel Identification Tools for Multiresolution Analysis of Aeroservoelastic Flight Test Data

Aeroelastic studies, such as the prediction of flutter boundaries and limit cycle oscillations, play a critical role in aircraft safety and design. To perform these analyses with the required accuracy, nonlinear aerodynamic, and possibly nonlinear structural, effects must be included. Furthermore, these effects must be considered early in the design cycle in order to reduce life cycle costs. The Volterra theory of nonlinear systems addresses the need for efficient and accurate reduced-order modeling of unsteady aeroelastic effects. A critical problem, however, is the difficulty of identifying the Volterra kernels. We propose a new identification method that is suitable for use with wind-tunnel or flight-test data. The proposed innovation is a time-domain multiresolution Volterra kernel identification method which uses physically realizable inputs, is robust with respect to noise, and minimizes or eliminates the need for analytical assumptions. The Phase I work will integrate an existing kernel extraction technique with new multiresolution signal decomposition approaches and demonstrate the combined method on experimental data. The resulting technology is expected to facilitate linear and nonlinear system identification for aeroservoelastic analysis and design and will be applied, in Phase II, to data from the F18-AAW program.