Optimization of time and frequency domain methods for real-time modal parameter identification of aircraft

Abstract

Lightweight structures such as aircraft are prone to vibration. In changing flight conditions, the vibration damping may change due to the aerodynamic-structure-interaction. To predict and prevent unstable aeroelastic conditions such as flutter, real-time operational modal analysis can be applied. Due to high noise levels, application to parameter-varying systems and short time data, the uncertainty of such identification is high. In this study, a new method is presented that optimizes the results provided by time and frequency domain methods (Stochasic Subspace Identification, SSI and Least Squares Complex Frequency, LSCF) for real-time modal parameter identification using Bayesian Optimization. Combining the results of both methods decreases the uncertainties and biases of identified modal parameters. The method has been tested on a laboratory aircraft structure which can change its mass dynamically. The whole process of automated SSI and LSCF, their combination and tracking runs in less than two seconds and therefore can be used for continuous real-time modal parameter identification.