Applying artificial neural networks to the tasks of designing airfoil sections of a helicopter´s main rotor

Abstract

An original system for designing aerodynamic profiles using artificial neural networks (ANN) is presented. A panel method was applied for a quick assessment of the quantitative aerodynamic characteristics of the profiles. This approach eliminates the problems of CFD modeling associated with the necessity to use large computational power. The developed methodology and the algorithms based on were applied to the design of aerodynamic profiles. As test NACA23012 base profile has been selected, on the basis of which was generated by a plurality of profiles acceptable to analysis. The analysis carried out with ANN methods revealed a limited area of the most promising candidate profiles for the further optimization of the geometry from the point of view of the chosen criteria. The profile design process is iterative. Already at the first iteration, it can be shown that the resulting profile family in aerodynamic characteristics exceeds the base profile.