Airfoil Self-Noise Prediction Using Artificial Neural Networks for CFD Boundary Layer Parameter

This study presents the development of an artificial neural network-based boundary layer parameter prediction model, trained using two-dimensional Reynolds-Averaged Navier-Stokes (RANS) simulations. The model accurately predicts boundary layer and displacement thicknesses on the trailing edge of diverse airfoil shapes, alongside estimating airfoil self-noise using empirical formulations. Employing this boundary layer model, the study analyzes the self-noise sensitivity of airfoil shapes, exploring variations in maximum thickness and camber across NACA airfoils. The findings revealed discernible trends in maximum thickness and camber of the airfoils with respect to angle of attack, lift coefficient, and lift-to-drag ratio. Furthermore, the model is extended to assess the UH-1B hovering rotor, predicting both tonal noise and airfoil self-noise across parameteric sweeps of tip Mach number, number of blades, rotor solidity, maximum thickness, and camber. The observed trends confirm the influence of these rotor parameters on tonal noise and self-noise levels.