Numerical Signal Processing in Computational Aeroacoustics

摘要

The paper presents a way of stochastic signal processing using results of the numerical modeling of turbulent noise. For this case a half-stochastic model based on Reynolds averaged Navier-Stokes (RANS) equations closed by k - ε turbulence model is used. The signal under analysis has a form of time changed stochastic fields of acoustic disturbances given as time series. For the purposes of the analysis a special code was developed. The time series give information about the intensity of acoustic radiation in the point given and spectral characteristics of this radiation. The code can evaluate the mean (mathematical expectation, dispersion) and spectral properties (autocorrelation, peak frequencies and width, spectra shape), filter signal from parasite oscillations generated by the imperfection of numerical algorithm. It is proved that mean spectrum width and peak frequencies in particular points of the computational domain are in an inverse proportion to the distance from the sourse of the acoustic radiation to the point.