Research on aerodynamic characteristics prediction of folding-wing aircraft based on improved PointNet++

Folding-wing aircraft enhance flight performance across diverse mission profiles, representing a key development trend for next-generation variable-configuration aircraft. However, current research predominantly focuses on structural design and aeroelastic vibrations, leaving the prediction of their aerodynamic characteristics in its infancy. Engineering typically relies on computational fluid dynamics (CFD) for analysis, but high-fidelity CFD simulations demand substantial computational resources and time, creating a bottleneck for rapid design iterations. To address this, a method for fast aerodynamic performance prediction is crucial. This study constructs an aerodynamic database using Fluent's dynamic mesh technology. It integrates the PointNet ++ model to build an intelligent mapping from the aircraft's dynamic shape to its surface pressure field distribution, achieving near real-time (second-level) prediction. The model's accuracy is enhanced by processing CFD-calculated surface pressure data through techniques like random sampling and grouping. Compared to CFD results, the predicted lift coefficient error remains within 6%, demonstrating good accuracy and promising application value for improving aircraft design efficiency.