Harmonic Retrieval and Simulation of Aircraft Engine Noise

Engine aerodynamic noise is a complex component of flight simulation, consisting of multiple groups of harmonics and broadband noise signals based on passing frequency. Harmonic simulation can effectively convey human auditory perception but lacks related study in flight simulation. To accurately simulate the acoustic behavior of harmonic noise, first, a high-order cumulant based pre-filtering method is applied to retrieve harmonic signals from the real engine noise recordings by exploiting the blindness of the third-order cumulants to the harmonic signal. Then, targeting the lower precision of the autoregressive (AR) coefficients in the pre-filtering calculations, a maximum likelihood method is proposed to estimate the ARMA parameters of the high-order cumulants. The simulation results show that the accuracy of AR parameter estimation is significantly improved compared to the moment estimation method. Finally, the exact frequency is verified and obtained based on first-order cyclic statistics, and then the harmonic signal in the time domain is extracted.