A new nonlinear lifting line method for configuration aerodynamics and deep learning based aerodynamic surrogate models

Abstract

Determination of the aerodynamic characteristics of unmanned aerial vehicles (UAVs) is of prime importance from both the design optimization and the flight control system design perspectives. Because many of the small, mini, and micro UAV configurations are operated at flight regimes with low Reynolds numbers, the nonlinear aerodynamics and dominant viscous effects play a key role in aerodynamic performance characterization. The existing approaches to determination of the aerodynamic characteristics of small UAVs use either semi-empirical methods with limited prediction capability to reduce computational complexity or computationally intense and complex computational fluid dynamics (CFD) methods. By contrast, in this work, we present a computationally efficient and high-precision nonlinear aerodynamic analysis method for both design optimization and mathematical modeling of small UAVs. First, a new nonlinear lifting line method is developed for lifting surface configurations using Prandtl's classical lifting line theory. This method is further extended to a complete configuration analysis tool that incorporates the effects of basic fuselage geometries. To be specific, the developed method is able to determine the maximum lift coefficient and the pre- and post-stall aerodynamic behavior of a UAV by using its wing and tail section's nonlinear two-dimensional lift curve obtained experimentally or numerically. The method also gives the induced drag directly, and provides the viscous drag and pitching moment coefficients by using two-dimensional airfoil data on the order of 0.01s using a personal computer. A direct comparison between the results of the current method, experiments, and computationally intensive tools shows good agreement. Moreover, we have also developed a deep learning based surrogate model using data generated by our new aerodynamic tool that can characterize the nonlinear aerodynamic performance of UAVs. The major improved feature of this model is that it can predict the aerodynamic properties of UAV configurations by using only geometric parameters without the need for any special input data or pre-process phase. The obtained black-box function can calculate the performance of a UAV over a wide angle of attack range on the order of milliseconds, whereas CFD solutions take several days/weeks in a similar computational environment. The aerodynamic model predictions show an almost 1-1 coincidence with the numerical data even for configurations with different airfoils that are not used in model training. The developed model provides a highly capable aerodynamic solver for design optimization studies as demonstrated through an illustrative profile design example.