A Modified Noise-Resistant Trend Estimation Method Based on EMD and SSA for Aeroelastic Aircraft Systems

Aimed at the nonlinear system identification of aeroelastic aircraft, the signal decomposition methods are required to extract the contributing natural and non-standard flight modes from flight test data, especially in the presence of flight noise. To this end, the SSA-EMD algorithm is proposed in this paper as a noise-tolerant signal decomposition method. The SSA-EMD is an improved Empirical Mode Decomposition (EMD) in which the sifting process is implemented by a direct approach to the signal trend extraction as a substitute for the envelope concept. In the proposed method, Singular Spectrum Analysis (SSA) is used for extraction of the signal trend in order to improve the mathematical foundation of the EMD. The proposed method is verified by decomposing some benchmark signals. Numerical results demonstrate that the proposed method outperforms the original one, especially in handling noisy signals. Afterwards, a novel gray-box non-parametric system identification method is proposed for considering extracted flight mode in the aircraft dynamics. The performance of the SSA-EMD is studied for the aircraft system identification from real flight test data of an aeroelastic aircraft in the transonic regime. It can be observed that the average fitness values of 60.01% and 88.41% are obtained for the lateral flight parameters using the EMD and SSA-EMD, respectively. Moreover, the RMSE values of the flight parameters predicted by the EMD and SSA-EMD are 1.85 and 0.65, respectively. Therefore, the SSA-EMD can achieve better results than the original EMD for the aircraft system identification due to its noise rejection properties.