Numerical Study of Blended Winglet Geometry Variations on Unmanned Aerial Vehicle Aerodynamic Performance

Abstract

An unmanned aerial vehicle (UAV) is an unmanned aircraft that can be controlled remotely or flown automatically. Nowadays, the use of UAVs is extensive, not only limited to the military field but also in civilian tasks such as humanitarian search and rescue (SAR) tasks, railroad inspections, and environmental damage inspections. Therefore, study on UAV becomes essential to answer the challenges of its increasingly widespread use. This study explores the addition of a blended winglet on the swept-back wing of the UAV. It is to predict the effect of the aerodynamic performance. The backpropagation neural network (BPNN) method helps to predict the aerodynamic performance of the UAV in the form of a lift-drag coefficient ratio (C L /C D ) and drag coefficient at 0 O angle of attack (CD 0 ). It is based on blended winglet parameters such as height, tip chord, and cant angle. The obtained BPNN modeling has a network architecture of 3 inputs, 2 hidden layers, and 1 output with a mean square error (MSE) of 4.9462e-08 and 4.4756e-06 for the relationships between blended winglet parameters with C L /C D and CD 0 , respectively.