基于人工神经网络的缝翼凹槽填充降噪设计

摘要

Airframe noise becomes one of the dominant noise sources during landing and tak-ing off phases of a civil aircraft because that jet engine noise is decreased obviously in recent years.High lift device noise is the main source of airframe noise,decreasing high lift device noise makes a significant contribution to the overall noise reduction for a civil aircraft.Slat cove filler (SCF),as a type of slat noise reduction method,may decrease the aerodynamic performances, such as the maximum lift coefficient and the stall angle.To this situation,a database of SCF is built for a multi-element airfoil,and one of the SCFs in the database is selected as the reference configuration,20 optimized configurations are generated through confidence coefficient reason-ing.Aerodynamic performance of each optimized configuration is predicted by a back propagation (BP)artificial neural network,and the one with the best aerodynamic performance is selected as the design configuration for verifying computation,steady Navier-Stokes simulations are executed to compare aerodynamic performances of the design configuration with that of the baseline config-uration.A hybrid method,combining CFD with acoustic analogy,is applied to compare the acoustic performances of the design configuration with that of the baseline configuration.The re-sults indicate that the noises are reduced significantly at the given observation points by adding the designed SCF,while the aerodynamic performances have been maintained.%缝翼凹槽填充技术作为一种缝翼降噪方法，有可能会造成气动性能的损失，如最大升力系数和失速迎角的减小。基于这种情况，针对某多段翼型建立了缝翼凹槽填充构型的数据库，挑选出参考构型，利用置信度推理确定了优化方向，生成了20个优化构型；采用 back propagation（BP）人工神经网络快速预测各优化构型的气动性能，选择其中气动性能最好的构型作为设计构型进行校核计算，求解定常 Navier-Stokes 方程评估其气动性能与基准构型作对比，应用 CFD 和声类比相结合的混合方法评估其气动噪声性能并与基准构型作对比。结果表明：在保持多段翼型气动性能的同时，对于给定观测点，所设计的缝翼凹槽填充构型使得气动噪声明显降低。