基于贝叶斯稀疏网格的SUBOFF模型流激噪声仿真研究OA

This paper proposed a Bayesian-based sparse grid correction method to address the degradation in flow-induced noise prediction accuracy arising from mismatched spatial scales between finite element model grid dimensions and turbulent boundary layer pressure in large-scale sparse grids.By using the DARPA SUBOFF 5470 submarine model as a case study,numerical simulations were conducted to investigate its applicability in predicting far-field acoustic radiation from submarine appendages.The paper employed Corcos's self-spectrum and normalized cross-power spectral density function computational model,utilizing virtual grid refinement to compensate for the spatial correlation characteristics of low-frequency turbulent pulsation pressures.By integrating engineering characteristics and operational environments,the correlation between pulsation pressure influencing factors was extracted,ultimately constructing a Bayesian network model for flow-induced noise excitation forces.By employing wall-resolved LES(WRLES)coupled with the Ffowcs Williams-Hawkings(FW-H)acoustic analogy method,the flow-induced noise obtained from the modified simulation calculations and the results of the refined grid fluid-structure interaction simulation.Acoustic characteristics were contrasted between the SUBOFF 5470 configuration with appendages and the bare hull.Simulation experiments demonstrate that the Bayesian sparse grid correction method effectively compensates for prediction errors arising from mismatches between grid dimensions and turbulence-relevant scales,validating the method's validity and applicability.