Flutter prediction based on dynamic eigen decomposition of flight data with limited actuators and sensors
Typically, during the design, analysis, flight test phases aircraft flutter boundaries are computed using the p-k iterations, k iterations, p iterations, or eigenvalue analysis, all of which require computationally expensive and numerically sensitive procedures. In this work, based on the concept of the Dynamic Eigen Decomposition (DED) and a frequency domain stability theorem, a new flutter prediction methodology is introduced and extended for applications to flight flutter test (FFT). It is shown that the dynamic eigenmodes of the aeroelastic system can be formulated such that they become an intrinsic property independent of dynamic pressure. This makes it possible to predict the aeroelastic instability by simply extrapolating the corresponding dynamic eigenvalues obtained at a low dynamic pressure. The methodology, however, needs a major modification for FFT where the actuators and sensors are limited in numbers and locations. In particular, it is necessary to make up for the insufficient actuators by numerically generating extra responses. The proposed scheme is demonstrated using computational simulations of the Goland Wing with non-collocated and collocated sensors/actuators. It is shown that the new approach can yield very accurate flutter predictions with limited actuators if a sufficient number of sensors are utilized. The flutter information attainable includes flutter mode shape as well as critical flutter dynamic pressure and frequency.
