Robust online monitoring of aircraft modal parameters using data fusion-based mode tracking

Abstract

Accurate real-time identification of modal parameters of aeroelastic systems, such as aircraft during flight vibration testing (FVT), remains a major challenge, e.g., due to low signal-to-noise ratios, variations in aerodynamic excitation, and the time-varying nature of flight conditions. This paper presents a robust methodology for online monitoring of aeroelastic systems using output-only modal analysis and data fusion techniques. A clustering-based Automated Modal Analysis framework enables the combination of time-domain (Stochastic Subspace Identification) and frequency-domain (Least-Squares Complex Frequency) methods, providing real-time uncertainty estimates for each identified mode. These results are fused across methods and over time using a Kalman filter, allowing for improved mode tracking and significant reduction of uncertainty. The proposed approach is validated using simulated data, wind tunnel experiments, and flight test data from a modified business jet. Results from wind tunnel testing demonstrate a reduction of the identification errors by up to 63% for eigenfrequency and 33% for the damping ratio. In addition, the results from the flight test demonstrate that the fusion of multiple identification methods with real-time uncertainty assessment and Kalman filtering enhances the robustness and resolution of aeroelastic parameter tracking - enabling more reliable real-time monitoring even under challenging operational conditions.