Modified sparse time domain technique for rotor stability testing

Because of nonlinear and complex interactions involved with the dynamic and aeroelastic stability of a rotor system, detailed and accurate knowledge of the stability characteristics are essential to prove design safety and to validate theoretical analysis. The task of determining these characteristics for a rotor from tests becomes complicated because of the presence of substantial amplitudes of the undamped harmonics, the possibility of close modes, and the complexity associated with the excitation of the modes in the rotating environment. Time domain modal parameter estimation methods are very useful for damping estimation when there are close modes, but they are very sensitive to noise. Subspace methods substantially improve the time domain estimates for noisy data, but they require higher computation time. A simple method is developed that retains the low variance estimation property of the subspace methods, but which is comparable in computation cost to methods that do not use the subspace approach. Its performance is evaluated for multi-output and single-output implementations and compared to the standard sparse time domain method. It is found that the modified method is more accurate in terms of the bias and standard deviation of the damping estimates, and it is faster when the number of modes is much less than the order of the data matrix.