Abstract
This article presents a maximum likelihood approach to identifying the mistuning, aerodynamic forcing functions, and aerodynamic influence coefficients of compressor rotor blades from noisy experimental measurements of blade deflection. The maximum likelihood estimation approach leads naturally to a nonlinear least squares optimization problem that can be solved using numerical optimization techniques such as quasi-Newton methods. An eigenmode decomposition approach is presented that permits the optimization objective function and gradient to be efficiently calculated so that the optimization is computationally feasible, even for experiments involving very large data sets. Simulation results show that the method has the potential to accurately identify the mistuning and aerodynamic nodal diameter forcing functions, but obtaining accurate results for influence coefficients may not be possible if there is significant disk flexibility. We demonstrate the method using both simulated and experimental blade vibration data from rotor 2 of the Purdue University three-stage axial compressor research facility collected using a light probe tip timing system as the rotor is slowly accelerated or decelerated through a Campbell diagram crossing.